Prague Linguistic Circle Papers: v. 4

TRAVAUX DU CERCLE LINGUISTIQUE DE PRAGUE n.s. PRAGUE LINGUISTIC CIRCLE PAPERS

TRAVAUX DU CERCLE LINGUISTIQUE DE PRAGUE n.s. PRAGUE LINGUISTIC CIRCLE PAPERS VOLUME 4 Edited by

EVA HAJICOVÁ PETR SGALL
Í HANA JIR TOMÁŠ HOSKOVEC

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984.

Eva Hajicová, Petr Sgall, Jir
í Hana and Tomáš Hoskovec (eds) Prague Linguistic Circle Papers, Vol. 4 ISSN 1383-7583 ISBN 90 272 5444 3 (Eur.) / 1 58811 175 X (US) (alk. paper) © 2002 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. • P.O.Box 36224 • 1020 ME Amsterdam • The Netherlands John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA

Table of contents Preface

vii

Section I: The Prague tradition in retrospect Prolegomena to the history of the Prague School of Linguistics †Josef Vachek Anton Marty’s philosophy of language †Oldrˇich Leška Die Typologie des Ungarischen †Vladimír Skalicˇka

3 83 101

Section II: Grammar Theoretical description of language as a basis of corpus annotation: The case of Prague Dependency Treebank Eva Hajicˇová

111

‘‘Conditionals’’ in Hebrew and English: same or different? Yishai Tobin

129

Sur la paradigmatisation du verbe indo-européen (deuxième partie) Tomáš Hoskovec

143

Section III: Topic–focus articulation The Russian genitive of negation in existential sentences: The role of Theme–Rheme structure reconsidered Vladimir Borschev and Barbara H. Partee

185

Synonymy vs. differentiation of variant syntactic realizations of FSP functions Libuše Dušková

251

Topic–Focus articulation as generalized quantification Jaroslav Peregrin

263

Information structure and the partition of sentence meaning Klaus von Heusinger

275

vi

Contents

Section IV: General views Freedom of language: Its nature, its sources, and its consequences Petr Sgall

309

The natural order of cognitive events Philip A. Luelsdorff

331

Section V: Poetics The principle of free verse Miroslav Cˇervenka

365

Preface

The present volume consists of two parts, the first of which (Section 1) brings three contributions connected with the classical period of the Prague School. J. Vachek looks back at the times of the beginnings of the Circle, which he witnessed as a student and a young researcher. O. Leška characterizes the work of the philosopher A. Marty, who as a professor of the German University in Prague, was one of the teachers of V. Mathesius, so that his ideas had an important impact on the activities of the Circle. The last paper of this section, written by V. Skalicˇka, the typologist of the School, brings some very important insights into the study of linguistic typology. The main body of the volume contains papers delivered at the conference “Function, Form, and Meaning: Bridges and Interfaces’’, organized in Prague in 1998 on the occasion of the celebrations of the 650th anniversary of Charles University. The contributions in Section 2 are devoted to issues of grammar of different languages including a syntactic annotation of a large Czech text corpus, a comparison of Hebrew conditionals with English, a characterization of the typology of the Indo-European verb. Section 3 is focused on topic–focus articulation (information sentence structure, functional sentence perspective), with a concept of ‘perspective’ introduced as close to but distinct from ‘topic’ and with three different viewpoints on the semantics of information structure. Two broader essays on the nature of language are presented in Section 4, while the last Section analyzes the structure of free verse. We offer this volume as a contribution to the continuing fruitful interaction between the work of the Prague School and the more and less closely related approaches of our colleagues elsewhere. This was also the aim of the mentioned conference, which took place in connection with the twelfth cycle of Vilém Mathesius lecture series. We are grateful to all those who made these two events possible, especially to Charles University, to the Joint Institutes of Advanced Studies and to the Higher Education Support Program of the Open Society Institute, as well as to Alena Böhmová and other members of the Institute of Formal and Applied Linguistics of the Faculty of Mathematics and Physics, Charles University in Prague, who actively and enthusiastically participated in organizing the conference. Eva Hajicˇová Jirˇí Hana

Tomáš Hoskovec Petr Sgall

Section I

The Prague tradition in retrospect

Prolegomena to the history of the Prague School of Linguistics

Josef Vachek Translated by Z. Kirschner

Introduction

3

.

External history of the Prague Linguistic School

5

.

The origin and maturation of the Prague Linguistic Theory

12

.

The personalities of the founding generation of the Prague School

21

.

Hard post-war decades of the Prague School (1950–1990)

37

.

The heritage of the Prague School in our domestic and in world linguistics

43

We have finished reading Vachek’s Prolegomena (afterword by Oldrˇich Leška)

52

Appendix I. Articles of the association ‘‘Prage Linguistic Circle’’ Appendix II. Lectures delivered at the Prague Linguistic Circle

55 60

References Editors’ note

71 80

Introduction A research worker who decides one day to write out a detailed history of the Prague School of Linguistics (henceforth PS) will be facing no easy task. It is not only phenomenalistic data that he must be concerned with in his work (though mere detecting them today, after more than six decades, encounters considerable difficulties of heuristic as well as purely technical nature), but, above all, he will be expected to incorporate the linguistic endeavours of PS into national as well as international contexts. Another objective will consist in determining the manifold factors both internal and external that exerted their influence on PS at the time of its coming into existence and in the course of further development of its concepts; and, naturally, in tracing the differentiation emerging within the PS; also the range of its impact at home as well as abroad must be considered, as it was felt from the



Josef Vachek

beginnings up to the present. There is no doubt that each of the above points taken by itself is important enough to require monographic treatment. Circumscribed time and space resources allotted to us to deal with this task have forced us to confine ourselves to outlining just prolegomena to a more extensive work. We wanted to limit ourselves to recording the events in the history of PS and the features that have appeared to us as really essential and characteristic. With respect to the restrictions mentioned above we haven’t been able to go into the history of partial problems discussed in PS, whether phonological, grammatical or other (in part we have touched upon them, at least in passing, in the monograph Vachek 1966a).1 We decided to concentrate on keeping track of the origins and maturation of what appears to us as essential contributions and specificity of PS: its conception of language as a functionally and structurally regarded systemic whole. Another intention of ours lay in initiating (at least in part and approximately) present-day readers into that unique working climate, the captivating companionability, which was so characteristic of PS in the times of gradual ripening of its conception. In doing so we are conscious that owing to this intention our report acquires a more memoir-like character than might be regarded as tolerable. In our opinion, however, our direct participation in at least a certain period of the history of PS (in its ‘‘classical’’ stage) obliges us to share this personal experience with those who came after us and those who are still to come. We firmly believe that the danger of subjectivism inevitably connected with a personal view of the given topic has considerably diminished today, when a certain lapse of time separates us from the history described, though the intensity of this experience has not decreased even after so many years. By all that has been said here the sources of our work have been specified. They were our own reminiscences and written records as well as the recollections of other, at that time still living witnesses and participants in the activities and events to be mentioned; further, of course, also reports in publications (mainly scientific, as is only natural) that were appearing both in our country and abroad. On this occasion it is not possible to forget the always readily offered and repeatedly given support represented for us in this work by consultations with the then still living members of the founding generation of the Prague Linguistic Circle. They were especially Professor Bohuslav Havránek, Professor Bohumil Trnka, and, from among our contemporaries and of the same generation background, our unforgettable friend Professor Vladimír Skalicˇka. It was they who would often lend me materials and documents otherwise hardly accessible at that time. For all that they deserve acknowledgment and sincere thanks. In concluding this introduction let us emphasize that it is solely the linguistic aspects of the Praguian theory that stand in the centre of our attention, its particular applications in the works of PS, and its significance for linguistics here

Prolegomena to the history of the Prague School

as well as abroad. That is also why we do not take a closer look at the very important part of the activity of PS which lies in the realm of aesthetics and literary science; nevertheless, we cannot leave it unmentioned here, be it only in a remark made in passing. Its more detailed evaluation would require a separate monograph conceived, of course, by a research worker well versed in both the disciplines involved. Prague, October 29, 1994.

.

External History of the Prague Linguistic School

Naturally, the name Prague School has not been given to the Prague linguists by themselves: it does not occur in the history of our own and world linguistics before the Praguian researchers attracted the attention of foreign experts by their works and principles. As far as could be ascertained, the term Prague School (henceforth PS) was first used in a prospectus in which Dutch phoneticians were inviting broader circles of research workers to their First International Congress of Phonetics that was to take place in Amsterdam in 1932, i.e., probably some time in the second half of 1931, after the Prague linguists had already scored their successes at conventions in The Hague, Geneve and, of course, at the Prague Congress of Slavonic Studies. The term école de Prague became quite common surprisingly promptly, so that in the mid-thirties it appeared in general and univocal use, even in international context. The real origins of PS, however, reach much deeper into the past. The beginnings and the prehistory of the Prague Linguistic Circle, the first organisational form of PS, were described in a very interesting and truthful way by its founder Vilém Mathesius in a lecture delivered at the festive session of the Circle in 1936, on the occasion of an anniversary — the first decade of its existence (see V. Mathesius 1936a). From this paper we learn that the first meeting of the Circle, then only a free association, took place at the Prague University Department of English (Mathesius was its director at that time, the Department having its seat at Nr 4, Veleslavínova Str.) on October 6, 1926. Five Prague members took part in the session (Bohuslav Havránek, Roman Jakobson, Vilém Mathesius, Jan Rypka and Bohumil Trnka); the only foreign participant, Dr. Henrik Becker from Leipzig, read a paper on ‘‘Der europäische Sprachgeist’’ (The European language spirit). The lecture was followed by a lively discussion and the participants agreed to meet at such sessions regularly. As shown by Mathesius, during the first academic year of its existence, in all, nine lecture sessions of the Circle took place, while in the following year the number went up to eleven. Papers were given not only by Prague lecturers





Josef Vachek

(besides the participants in the first session, who all contributed their lectures, it was also Vladimír Buben, Bohuslav Ilek, Jan Mukarˇovský and František Oberpfalcer), but also by distinguished foreign guests: the Russians Sergej Karcevskij, Nikolaj Sergeevicˇ Trubeckoj (Trubetzkoy), Petr Bogatyrev and Boris Tomaševskij, and the French scholars Léon Brun and Lucien Tesnière. However, the number of participants in the meetings, according to the then preserved attendance records, for a long time failed to exceed ten. This made it possible that for several years a remarkable practice was established to organise, in addition to the usual meetings held on the premises of the Department, also informal lecture talks invited to the households of individual members. This custom, which substantially strengthened friendship ties among the members of the Circle, at the same time aided in introducing into its activities a good mutual understanding indispensable for collective work — a feature very soon to become characteristic of the Circle. This custom was abandoned only in the first half of the thirties, after the number of members grew to such an extent that similar meetings were no longer technically possible. Becker’s lecture marks the beginning of the Circle’s history; however, it is necessary to pay heed to its prehistory as well, which can be traced yet much deeper back into the past. Mathesius mentions another earlier informal meeting of a handful of young linguists in which Havránek, Jakobson, Karcevskij and Trnka took part. At this meeting Mathesius presented his reflections on new currents and trends in linguistics (later published as Mathesius 1926). Nevertheless, for the oldest public presentation in which many of the later principles of the Circle had been proclaimed we have to go still further back to the past, as early as the year 1911. On February 6 of that year, at a meeting of the philosophical-historicalphilological class of the Royal Czech Society of Sciences, Mathesius delivered a lecture ‘‘O potenciálnosti jevu˚ jazykových’’ (On the potentiality of the phenomena of language), which was then printed soon afterwards, in the subsequent year (Mathesius 1911–1912). Even after many years Mathesius used to recollect with a bitter smile how his exposition at that meeting fell practically flat and failed to evoke any response, there being no discussion — the authority in chair, Professor Josef Zubatý, very politely bestowed his praise on the lecture, which he nevertheless judged ‘‘too subtle’’ for the audience then present. Obviously, Mathesius at that time, with his views in which he emphasized, among others, the necessity of a synchronic, ahistorical approach to language facts, remained in complete isolation. Only after some young researchers with understanding for his conception appeared within the Czech linguistic community could Mathesius envisage more clearly a formation of an organisational and working centre, which the Prague Circle proved to become later; Mathesius himself explicitly points out that it was his young colleagues R. Jakobson and B. Trnka with whom he first found understanding and support for his positions.

Prolegomena to the history of the Prague School

By coincidence it happened that the Circle made its debut abroad and not at home. An occasion for it was offered by the First International Congress of Linguists convened to The Hague for April 1928 and to whose organizers four members of the Circle (Jakobson, Karcevskij, Trubetzkoy and Mathesius) sent their individual answers to the principal question posed by the preparatory committee of the Congress: which is the most suitable method for a full description of language. The three Russian members of the Circle and Mathesius found that their own theses and the theses of the followers of the Genevese school, Charles Bally and Albert Sechehaye (their materials had also been submitted to the Congress in advance) agreed in a number of essential points. Therefore, still before the relevant plenary session took place, they arrived at a joint resolution proclaiming a programme of a new, synchronically oriented linguistic analysis. This programme was then unanimously approved by the Congress at its session on April 12, 1928; this was the first great international success of the new Prague linguistic conception. Encouraged by this success, the Prague linguists decided to work out (for the first international congress of Slavonic scholars which was to convene in Prague in October of the following year 1929 on the occasion of the centenary of the death of Josef Dobrovský) a detailed conception of linguistic analysis based on the principles of the Prague approach to language facts, an approach which they called structurally functional. The theses summarizing this conception were being elaborated for several months by a special committee of the Circle consisting of Havránek, Jakobson, Mathesius and Mukarˇovský in cooperation with some other members, above all Miloš Weingart. As Mathesius reminisces, the meetings of the committee used to take place at a small coffee-house Derby in the then Beˇlského Rd (nowadays Dukelských hrdinu˚) in Prague 7. The theses represented a collective work indeed, although their separate sections were prepared by individual members of the committee. The theses were submitted to the Congress in two versions, Czech and French; the Congress approved them and elected an international commission of ten linguists assigned to do the preparatory work in the analysis of Slavonic languages according to the principles presented by the Circle.2 The Congress of Slavonic Scholars was very important for the Prague linguists also in that the Circle’s very rich editorial activity was started in connection with it. On the occasion of the Congress the first two volumes appeared of a series that was later to become one of the most renowned in the world linguistics, Travaux du Cercle linguistique de Prague (henceforth TCLP). The first volume entitled Mélanges linguistiques contained in addition to the French version of the theses of the Circle (1929) altogether twelve treatises on general linguistic themes. The language of both the first volumes was predominantly French, which remained to serve as the permanent basic vehicle of TCLP, and to a lesser extent German; English was to be met in TCLP only much later. The second volume of the series contained





Josef Vachek

Jakobson’s French monograph on the phonological evolution of Russian (Jakobson 1929) as compared with similar developments in other Slavonic languages. Both the volumes met with extraordinary interest of the wide linguistic public, not only that concerned with Slavonic studies. Contacts with foreign research workers holding similar views were established, especially with the Polish and Russian followers of Jan Baudouin de Courtenay and with modern American linguistics represented, above all, by Edward Sapir (1922) and Leonard Bloomfield (1933).3 Such and similar contacts before long crystallized into a design to appear before the Second International Congress of Linguists (which was to take place in Geneva at the end of August 1931) with a detailed programme of research in phonology, i.e. in a functionally conceived study of speech sounds, and, of course, with an account of factual results achieved up to then in that particular sphere of work. To make it possible to discuss a number of relevant problems of methodology, the Circle dared to assume a responsibility which appears hardly credible today: an international phonological conference was invited to Prague for December 1930, in which event no less than fifteen foreign linguists from eight European countries took part. Among them were such very famous names as that of the Austrian psychologist Karl Bühler, Dutch linguists Jacob van Ginneken and Albert Willem de Groot, and also a Norwegian Celtic scholar Alf Sommerfelt; from Germany the Ukrainian psychologist, later on a renowned Comeniologue, Dmytro Cˇyževs´kyj, from Poland the Slavonic scholar Witold Doroszewski, the Polonists Kazimierz Nitsch and Stanisław Szober; and, of course, neither Karcevskij nor Trubetzkoy failed to appear. For four days the Czech as well as foreign participants were giving fundamental papers, followed by lively discussions. The papers and a summary of the discussions were then published in the fourth volume of the series Travaux du Cercle linguistique de Prague, still in time before the opening of the Genevese Congress (even this seems incredible to us nowadays). The same volume contained a proposal drawn up in French of a standardised phonological terminology and a brief outline of the principles which phonological transcription should comply with, both accompanied by illustrative examples (Cercle 1931a, b). Also the organisational success of the conference was important: it was agreed to found the International Phonological Association whose task would be phonological description of the greatest possible number of the languages of the world. Trubetzkoy was elected as chairman of the association, Jakobson became acting secretary and Mathesius treasurer. The Genevese Congress in August 1931 unanimously approved the principles of the Circle regarding the research in phonology and the International Phonological Association was equally unanimously recognized as an organisation affiliated to CIPL (Comité international permanent de linguistes, in charge of the organisa-

Prolegomena to the history of the Prague School

tion of international congresses of linguists). As is obvious, also the Genevese Congress meant another international success of the Prague functional structural linguistic conception, in the same way as later both the first two international congresses of phonetics (in Amsterdam in 1932 and in London in 1935), not to speak of a number of international meetings of minor scope or of more distant subjects of study; such was, e.g., the session of the linguistic section at the international congress of ethnologists and anthropologists in London in 1934. Moreover, an increasing need was felt for a more profound impact on the domestic linguistic events, although as early as 1929 Mathesius, Havránek and Trnka reported on the principles of PS at a convention of secondary school teachers of philosophy, philology and history in Prague. (This convention, however, failed to get the publicity given to the subsequent actions of the Circle.) An opportunity for a more notable action presented itself in 1930, when the Circle published, on the occasion of the 80th birthday of the Czechoslovak President Tomáš Garrigue Masaryk, two papers, delivered already in the preceding year by J. Mukarˇovský (about Masaryk as a stylist) and by R. Jakobson (on linguistic problems in Masaryk’s work); both the lectures appeared in one booklet (Cercle 1931c). The motivation of the lectures was, of course, not purely scientific: it was, at the same time, an expression of gratitude to a sponsor who by subscribing a considerable sum made possible the publication of the first two volumes of TCLP. A much more important entry into the still rather stagnant waters of the domestic linguistic life was implemented by the famous cycle of lectures on Standard Czech and on language culture, organized by the Circle at the beginning of 1932. This cycle laid a firm basis for a functional structural conception of language correctness and language culture in general. An external stimulus to organise the cycle, in which lectures were given by Mathesius, Havránek, Jakobson, Mukarˇovský and Weingart, was offered by the then extremely puristic practice of the journal Naše rˇecˇ (Our Language) represented above all by its editor-in-chief Jirˇí Haller. The lectures appeared in the same year in book form (Havránek and Weingart 1932), together with the ‘‘General principles of language culture’’ as a supplement, and were received by the cultural public with great sympathy (see, e.g., the commentary by F. X. Šalda in his Zápisník (Notebook) 1932, 1933). Even at the leftist cultural front they met with considerable sympathy, with the only exception of S. K. Neumann, who owing to lack of comprehension criticised them sharply in Peroutka’s Prˇítomnost (The Present) Nr. 46 (1932). It is well known that in the subsequent decades the principles of the Circle were to become a permanent starting point for the theoretical and practical care of the Czech language culture: in their spirit Pravidla cˇeského pravopisu (The Rules of the Czech Orthography) were reformed partially as early as 1941 and especially in 1957. In the same spirit also collective lexicographic works were conceived and





Josef Vachek

elaborated, in the first place Prˇírucˇní slovník jazyka cˇeského (A Shorter Dictionary of the Czech Language) in 9 volumes and later the Slovník spisovného jazyka cˇeského (A Dictionary of the Standard Czech Language) in 4 volumes published first in the years 1960–1971 (afterwards it appeared in 8 half-volumes), and, surpassing all, the outstanding collective Slovník spisovné cˇeštiny pro školu a verˇejnost (A Dictionary of Standard Czech for Schools and the General Public), elaborated under the guidance of František Daneš and Josef Filipec (Praha: Academia 1978). This dictionary was produced by the Lexicological and Lexicographical Department of the Institute for Czech Language under theoretical guidance and with methods developed by the Head of the Department, Josef Filipec, who was also the chief editor of the Dictionary. Moreover, still in the same decade, the new principles of the care of Czech language made their way into school grammars of Czech for the lower forms of secondary schools (the authors’ collective was again directed by B. Havránek). Their most valuable contribution can be seen, among others, in the consistently synchronic approach to Czech morphology, especially the classification of Czech verbs and in the regard to the functional stylistic differentiation of language means. It can be said, therefore, that in the course of just one decade of its existence the Circle attained a significant position in Czech cultural life. In this connection even a seemingly formal detail deserves to be mentioned: in the autumn of 1930 the Circle became a regularly organized and politically authorised association; until then it had been merely a free group. Now it had its official Statutes and a formally stated aim of its activities (see Appendix I). According to the Statutes, the aim consisted in the pursuit of linguistics on a functionally structural basis; well worth mentioning is the fact that here for the first time in the history of Czechoslovak linguistics an organisation was established which pragmatically delimited its domain of research, but also the method to be applied in this work. From this, however, no conclusion can be drawn that the Circle would censure members using other methods. For that matter, among the seventeen signatories of its Statutes we can find the names of, e.g., Oldrˇich Hujer, Václav Machek, Vladimír Buben and František Oberpfalcer, who never adhered to the principles of structurally functional conception in their work: Hujer was an enlightened Junggrammatiker (NeoGrammarian), Oberpfalcer-Jílek a refined connoisseur of the material which he, however, approached from eclectic rather than structuralist positions, and even V. Buben, a prominent Czech scholar in the field of Romance languages and literatures, as well as V. Machek, a comparative linguist, as further signatories of the Statutes could not be entirely classed with the followers of the functionally structural approach to language, although they undoubtedly sympathised with this point of view. Nevertheless, such an open profession of the Circle’s adherence to a definite research conception even so meant a new feature in our conditions; for a number of more conservative linguists this, at times, gave cause for reprimanding the alleged

Prolegomena to the history of the Prague School

sectarianism and intolerance. Such criticism did justice to the reality only in part — it was certainly necessary to make way for the new approach using conclusive arguments, which, however, as a rule did not exceed the limits of fair play. A proof of the successes attained by the Circle during the first decade of its work can also be seen in a great number of congratulations and acknowledgements which the Circle obtained on the occasion of the celebration of its first decade in 1936.4 The clear, convincing and fruitful conception kept attracting to the Circle young linguists entering upon scientific careers. As early as 1931 they read their linguistic papers at the Circle’s sessions; among them Josef Vachek (from the beginning of 1930 engaged as a student in secretarial work for the Circle), further, in the same year, L’udovít Novák, later Vladimír Skalicˇka, Pavel Trost, Karel Horálek, Jozef Ružicˇka and many others. For completeness’ sake it is necessary to add that information on the work done in PS was presented abroad in two issues of the Bulletin of the Association internationale pour les études phonologiques (International Phonological Association), above all information on the already finished works in the domain of phonology, and on those in preparation. The first issue appeared in 1932 in French, the second three years later, in English; no further issues were published afterwards. It was for the International Phonological Association that the Circle ensured the publication of a brochure by Trubetzkoy (1935) with an instruction concerning the phonological description of the world’s languages (it appeared in German in Brno; later it was translated into English by the Australian linguist Bluhme). It is, of course, well known that the members of the Circle frequently published in other than their own journals and periodicals; above all, it was the Cˇasopis pro moderní filologii (Journal for Modern Philology), Filologické listy (Philological Papers), Naše rˇecˇ (Our Language) as well as various University publications in Prague, Brno, Bratislava, etc. Ample response to the work of the Circle from abroad has already been mentioned: it was strong especially in places connected with a deep-rooted tradition drawing on Baudouin and Šcˇerba. We have also mentioned the favourable reception of the phonological work of PS in American linguistics, particularly by Sapir and Bloomfield, who in their works were approaching the main concept of phonology — the phoneme — in their own way. The same can be said, with minor differences only, about the prominent English phonetician Daniel Jones, who was coming close to that concept as a practical transcriptionist (J. Vachek wrote about him in the miscellany Charisteria (Cercle 1932) dedicated to Mathesius on his fiftieth anniversary). Also Dutch linguists were showing great interest in the Praguian activities (a festschrift published to honour the fiftieth anniversary of Trubetzkoy contains contributions by e.g., Hendrik Josephus Pos and Nikolaas van Wijk).5 Of the younger generation especially the Dutch Javanese scholar and general linguist Eugen M. Uhlenbeck must be mentioned, who during





Josef Vachek

the hard times of what was called ‘‘political normalisation’’ (détente of the early 1970s — editor’s note) provided invaluable help to our young linguists by granting short-term stays to them in the Dutch Institute of Research Workers, which he had founded himself especially for young scholars from behind the Iron Curtain. In a similar way substantial help to our young researchers was provided by Robert Auty, a British Slavonic scholar and Bohemicist, professor in Cambridge, London and later in Oxford, who, among other things, had a perfect command of Czech. From the same country came the help of William Haas, a general linguist from Manchester (a German from Northern Moravia by origin), who cooperated with the Circle’s scholars in those bad times; he shared a deep interest in the problems of written language with our scientists, an interest also shared by Philip A. Luelsdorff, an Anglicist from Regensburg, who was of great help especially as regards publication activities. In 1948 members of the Circle took an active part in the first post-war linguistic Congress in Paris; they had submitted their own theses in advance, which were then duly read and commented upon at a plenary session of the Congress, though not by their own authors and main speakers: Jakobson’s paper on the mutual relation of phonology and morphology was read in his absence by his colleague John Lotz, the English translation of the paper by B. Trnka, who could not be present at the Congress, was read by J. Vachek. Thus, though small in number, the PS delegation (consisting of three participants only, Havránek, Horálek and Vachek) did not appear as wholly insignificant. Otherwise, in that post-war period, the entire atmosphere, both ideological and political, was getting darker and more sinister. This development, naturally, could not remain without consequences for the PS and its Circle. There were, however, too many factors in play, whose interaction was so complicated as to require a special chapter (Chapter 4).

.

The origin and maturation of the Prague Linguistic Theory

Mathesius in his University lectures (and in their posthumous edition which we have prepared for press, cf. V. Mathesius 1961) very fittingly sketched the climate of the period in which the Prague linguistic theory came to be formed and, at the organizational level, the Prague Circle was founded. He pointed out therein that up to the beginning of the twenties of our century two main competing currents had existed side by side in the world linguistics. The first of them, genetically comparative, found its origins in the efforts of Franz Bopp and Rasmus Rask, and culminated in the Neo-Grammarian School, whose prominent representatives were Hermann Paul, Karl Friedrich Brugmann, etc. The second trend, according to Mathesius, can be denoted as analytically comparative and its beginnings reach

Prolegomena to the history of the Prague School

back to Wilhelm von Humboldt: among its most outstanding representatives, e.g., Heyman Steinthal, Franz Nikolaus Finck and Georg von der Gabelentz can be classed. The former current elaborated an exact scientific method, yet it often erred in atomistically isolating the facts examined, in lacking a sense for the structural patterning of languages and for their synchronic, ahistoric study in general, regarded by them as downright unscientific. The latter trend had a much deeper sense and feeling for synchronic specific features of the languages examined, yet it failed to produce methods that could compete with those of the former current in respect of exactness and scientific accuracy: the differences between individual languages were frequently accounted for by differences in national psychologies of the given language communities. In this way, of course, the problems in question were transferred from the domain of linguistics to the domain of psychology, a science at that time still unable to work its way to a scientific method of the desirable degree of exactness. Mathesius was well aware of both the strong and the weak points of the two linguistic approaches 6 and strove for a synthesis that would succeed in combining the merits of both the conceptions. He was conscious of the necessity of getting to know completely and reliably the language facts from which generalizing theoretical conclusions were to be drawn. This necessity led him already in the year 1911 to the requirement of a static (i.e., synchronic) analysis of language in addition to the dynamic (i.e., diachronic). Mathesius often stressed that only a synchronic analysis of a language can offer an exhausting and easily verifiable picture of a given language fact. Note that Mathesius’ emphasizing the necessity of synchronic study was not connected with rejecting the diachronic approach; this holds true, for that matter, of the activities of PS as a whole (today it may seem unnecessary to stress it, but as late as the second half of the thirties, and even after World War II there were linguists — even prominent ones — who blamed PS, and especially its phonological studies, for ahistoric approach.) An important circumstance here is the fact that Mathesius’ deep interest in phenomena of stylistic nature, attested already in his earliest works, proved clearly his essentially functional approach to language facts. On the other hand, it cannot be overlooked that in his beginnings Mathesius wrongly laid undue stress on the role of individuals in the language development and that he had a certain mistrust of what he then called ‘‘one-sided emphasizing of the social nature of speech’’. (This mistrust accounts for certain sympathies of his for Croce’s and Vossler’s theories at that time.) It is interesting that it was then the same ‘‘one-sided emphasizing of the social nature of speech’’ he held against Humboldt and his followers. It is also very well known that in this point Mathesius later substantially revised his initial opinions. As can be seen from the posthumous edition of his lectures (delivered at the turn of the twenties and thirties, cf. Concluding remarks of the editor to Mathesius 1961, pp. 249–52) he objected later (in the same way as the present PS) to another feature of





Josef Vachek

Humboldt’s conception — his emphasis on constant creation in language (conceived as energeia), which concealed from Humboldt the fact that language is essentially a system; it can hardly be doubted that it is its systemic character and, consequently, the existence of a certain norm resulting from it that constitute the necessary precondition for the social functioning of language. Similar objections could be later raised to Chomsky’s conception referring to Humboldt, which obscured the systemic nature of language from the American researcher in the same way as it had done previously from Humboldt.7 There can hardly be any doubts about the motives underlying the change in Mathesius’ views of the problem: it was obviously his close contact with the friends with whom he was organizing his linguistic Circle. It was especially Jakobson and Trubetzkoy who influenced Mathesius substantially; they had learned about the social impact of language phenomena from the Russian tradition of Baudouin and Šcˇerba, although, otherwise, as will be shown presently, Mathesius’ scientific type differed in a number of respects from that represented by these two friends of his. Nevertheless, it was obviously from them that Mathesius drew his knowledge in this respect, and, undoubtedly, it was their influence from which Mathesius’ later interest in the problems of phonology sprang; as regards their quantitative studies, he became a true pioneer not only at home but also in the world context. As far as the internal differentiation of PS is concerned, it can be safely stated that two markedly distinct tendencies were competing (which is a better term than rivalling) in it. On the one hand it was the Trubetzkoy-Jakobsonian trend distinguished by a grand vision of general systemic outlines, with a deep sense of what used to be newly called language modelling. On the other hand, there was the Mathesius-Havránek’s trend whose characteristic feature lay in refined tracing of inner systemic relationships, with special attention to the functional specialization of the means of expression of the given language system. Perhaps it could be said that in the characteristic epithet often used about PS and its approach to language facts, i.e., in the epithet ‘‘functionally structural’’, the Trubetzkoy-Jakobsonian trend emphasized rather the second member of this composite term, whereas Mathesius-Havránek’s view stressed more the first component. At the same time, as has been pointed out before, this difference in approaching the same facts can by no means be regarded as a contradiction: both the approaches complemented each other extremely well, and it was in the synthesis of both that PS displayed its strongest point, since this dual approach helped it to avoid impending one-sidedness. Nor can we say that the differentiation was due to national differences, i.e., differences between the ‘‘Russian’’ and the ‘‘Czech’’ approach. There were also Czech researchers to whom the Trubetzkoy–Jakobsonian approach appealed more strongly than that of Mathesius-Havránek, e.g., B. Trnka, J. M. Korˇínek, and, on the contrary, it is possible to find scholars on the

Prolegomena to the history of the Prague School

Russian side as well who were close to the Mathesius-Havránek’s position; this holds true, e.g., particularly about Leontij V. Kopeckij. At the same time this internal differentiation shows how unjust it was to blame PS for the assumed dogmatic sectarianism allegedly forcing all adherents to swear by the official canon of principles of the Prague Circle: it is obvious that acceptance of the functional structural approach as the point of departure provided sufficient space for the members of PS to assert their own specificity of opinions. We can even find, in addition to the two main trends within PS that were characterized above, also other members of the Circle who deviated from the official canon of principles still more profoundly — from the classical period of the Circle J.M. Korˇínek could be mentioned with his peculiar conception of langue, or some other aspects of language structure as well (see especially Korˇínek 1942). Nevertheless, the basic standpoint of all members of PS was concordant to such an extent that it made possible mutual fruitful cooperation and a unified collective procedure of all. The structural approach was being formed in clear-cut formulations opposed to the former Neo-Grammarian atomistic tendencies, the functional aspect made the needs of expression of a given language community into an important criterion in language analysis and evaluation. Simultaneously, the language means performing the purely communicative function (Bühler’s Darstellungsfunktion, at times also referred to as projective function in Czech writings) were carefully distinguished from the means used for signalling emotive function (appeal, or individually expressive function, i.e., serving the functions for which Bühler used the term Appellfunktion or Kundgabefunktion). These aspects were applied in the analysis of all language levels from the phonological through the morphological and lexical to the syntactical. By this respect to the functional differentiation of language means PS has always characteristically differed from other structuralistically oriented linguistic schools, especially from the Copenhagen glossematics and American descriptivism. In the same way PS can be favourably distinguished from these approaches in its recognition of the stylistic difference between the lexical layer synchronically domestic and that synchronically foreign (see especially Mathesius 1932b), a difference that, e.g., the American descriptivists came to notice only after World War II, and of whose existence the generative-transformational grammar began to be conscious still later (cf. C. C. Fries and K. L. Pike 1949, C. Rohrer 1967, on this topic also J. Vachek 1968c). It has already been mentioned here that in spite of laying great stress upon synchronic analysis, PS by no means objected to the application of diachronic methods in linguistics. On the contrary, it is emphasized from the very beginning that language changes can be understood more fully only from the functionally structural point of view, viz, that it is necessary to examine any language change — not only a phonetic one — from the angle of the whole system that has been





Josef Vachek

affected by it.8 Such an approach often shows that the examined language change largely has a therapeutical function, i.e., that it has been motivated by the need for preserving the endangered equilibrium of the language system, or, as the case may be, for recovering that equilibrium where it has been disturbed: what is meant by equilibrium here is the adequacy of language means for the purpose they are to serve. Already at its early stages PS often used to refer to this explanatory principle as teleological (for the first time this was formulated by Jakobson in his lecture in the Circle in January 1927; cf. Jakobson 1928, 1933). In this PS conception the study of language development thus gets transformed into a study of the development of the language system as a whole, a study of its structural problems and their solution. It can be said that also this nonatomistic and non-static, i.e., dynamic conception of the relationship between language synchrony and diachrony represents another very important specific feature of the Praguian linguistic theory, unparalleled in the contemporary linguistic world, at least as regards the consistency of this conception. It is undoubtedly true, as Jakobson points out (1963), that ‘‘what was regarded as a specific contribution of Prague to the development of modern linguistics appears to represent to a considerable extent a common denominator of a number of convergent trends in the scientific life of various European countries’’. (p. 482). As the common denominator of the work of such scholars as, e.g., the Dutch researchers Albert Willem de Groot and Hendrik Josephus Pos, the French linguist Émile Benveniste, the Norwegian Alf Sommerfelt, the Pole Jerzy Kuryłowicz, the Rumanian Alexandru Rosetti, the Hungarian Gyula Laziczius and the Russian Jevgenij Dmitrievicˇ Polivanov, Jakobson presents in his paper the very fact that all of them were striving to create a linguistic means-end model. Nevertheless, of course, at the same time it must be stated — as has been already said above — that in none of the mentioned countries was such a model applied so consistently and in such a notable way as in PS — which was particularly Jakobson’s merit. In his lecture delivered in 1931 at the Second International Congress of Linguists in Geneva, Mathesius regards as two main pioneers of new linguistics Jan Baudouin de Courtenay, with whom ‘‘an eminent role is played by the idea of function’’, and Ferdinand de Saussure, who impressed Mathesius above all by ‘‘the requirement of synchronic analysis of language and the idea of language system’’ (Mathesius 1932a). As regards Baudouin, Mathesius disagreed with ‘‘his having been deluded by the mystifying light of psychology, and attaching undue importance to the fact that speech keeps constantly changing’’. — Although marking out the two prominent scholars was essentially just, this by no means represents full enumeration of the factors that contributed to the change in linguistic views in our country, not only as regards the linguistic factors but also as far as other entire scientific branches are concerned.

Prolegomena to the history of the Prague School

In the same lecture Mathesius himself mentions the analytic-comparative methods of Henry Sweet, who — like Otto Jespersen — helped him to get over the one-sidedness of the official Neo-Grammarian linguistics that predominated at the time when Mathesius’ research profile was being formed. (It can hardly be a coincidence that the first item in Mathesius’ chronologically ordered bibliography was represented by nothing else than a review of Jespersen’s popular monograph Growth and Structure of the English Language.) Besides, Mathesius mentions two further opponents of the official Neo-Grammarian trend, the Romance scholar Hugo Schuchardt and the Czech Josef Zubatý. Although, as Mathesius rightly points out, neither of them went so far as to lay a ‘‘reliable foundation for a new period of linguistic research’’, even their role as precursors marking out a new era was certainly important. From among other branches of science the situation in linguistics at that period was certainly affected by the rise and influence of the Gestalt psychology and similar conceptions in musicology and theory of art, and equally, the abandonment of positivist philosophizing at that time: Mathesius often preached against ‘‘positivist narrow-mindedness’’, by which he understood adhering to phenomenalistic data and lack of courage to draw more general conclusions that could follow from them. This much had to be said concerning the question of external influences which helped to form the teachings of PS. We would particularly emphasize the fact that the work of eminent external authorities (and also of the Russian protagonists of PS, Trubetzkoy, Jakobson and Karcevskij) was able to assert itself in such a pronounced way in the Praguian context for no other reason than that it had found the ground well prepared by the preceding efforts of the local, Czech members of this school (here especially the names of Mathesius, Havránek and Trnka should be mentioned). Anyway, the very fact that a new linguistic school of functional structuralist orientation was able to originate in Prague and not, e.g., in Vienna, where otherwise particularly advantageous conditions would have existed for its rise, is noteworthy: both Trubetzkoy and Bühler (whose influence upon Trubetzkoy’s conception was beyond question) worked as university teachers there, and so were also the eminent comparatist Paul Kretschmer and the English scholar Karl Luick; in some points of their works both of them were inspired by Trubetzkoy’s conception, or, in some respects, even anticipated it (cf. Vachek 1933). If, in spite of all these circumstances, a new linguistic school failed to originate in Vienna, then it was probably due to the fact that the ground was not prepared here to such an extent as it was in Prague; this was due in part to the emphasis (mentioned above) laid by Mathesius on the necessity of synchronic study of languages, in part to the influence of Zubatý, whose non-conformist attitude towards the mechanizing Neo-Grammarianism was taken over by his disciples as well, among whom a prominent place belonged to no one else than B. Havránek. (It certainly exposes a very poor knowledge of





Josef Vachek

an otherwise prominent linguist to class Zubatý as a ‘‘narrow and orthodox NeoGrammarian’’, see Stankiewicz 1965.) Havránek shared with Mathesius not only his mistrust of the dominating Neo-Grammarian trend, but also the essentially identical functional standpoint: one of Havránek’s first articles deals with the question how Czech in the course of its development expressed the meaning conveyed by Latin adjectives ending in -bilis (Havránek 1929b); incidentally, an outline of the linguistic situation of the late twenties as seen by Mathesius in 1926 has much in common with Havránek’s view as given in his introductory chapter of the first volume of his extensive monograph on voice in the Slavonic verb, see Havránek (1928–1929). It was this functional approach to language that enabled Havránek to create an important functional theory of standard language, of language correctness and language culture in general, a theory the principles of which he outlined in his lecture on the tasks of standard language and its culture (Havránek 1932). According to this theory, accepted later by some prominent linguists abroad (e.g., Einar Haugen, Alf Sommerfelt, Paul L. Garvin), the correctness of a word or of a construction does not follow from historical criteria, but from functional adequacy of the given expression. Here Havránek together with Mathesius and Jakobson pointed out that any concern about foreign expressions (especially Germanisms) is totally unsubstantiated, as they have their legitimate place in the language norm, provided they fill a need in the language. Havránek then proved the importance of the functional approach by demonstrating in a number of examples how a sound form and a grammatical form in a given language can be profoundly influenced by functional needs of the given language community (cf. Havránek 1929a). For that matter, in this functional view any stylistic problems of a language appear as a formal differentiation of language performance given by functional needs of different kinds of language utterances; it was from this point of view that stylistic problems have always been analysed in PS (cf. Karel Hausenblas 1955, Lubomír Doležel 1964). From the preceding paragraph a further no less important characteristic property of PS can be seen, i.e., its keen interest in everyday life situations, the problems of which the Praguian functional structuralists want to assist in solving. They do not belong — unlike, e.g., the linguists of the glossematic school — to research workers who seclude themselves in purely theoretical scientific work, but they want to serve the language community and contribute to satisfying its needs of expression. In perfect agreement with this essential tendency of PS is its lively interest in questions of the language of poetry (which, of course, is not dealt with in detail in our outline), an interest that caused the lecture sessions of the Circle to be often frequented by prominent critics, writers and artists in general, who contributed to the Circle’s press platforms. Among them there were the novelist and play-writer Karel Cˇapek, who in his column greeted the publication of the

Prolegomena to the history of the Prague School

periodical Slovo a slovesnost (Word and Verbal Arts), the literary critic František Xaver Šalda, the author of an essay on poetical autostylization, and others, in particular Jaroslav Durych, Vladislav Vancˇura, Jindrˇich Honzl and Víteˇzslav Nezval. It was, of course, also the negative reaction of artists against the then purist practice of the journal Naše rˇecˇ (Our Language) aimed at the leading Czech writers, that gave the last, but not the only, impulse to the Prague Linguistic Circle to organize its series of lectures on Standard Czech and language culture. The sense for serving the language community, however, manifested itself also in the domain of language pedagogy and didactics. A mention has already been made of the textbook of Czech for junior secondary schools prepared by a collective headed by Bohuslav Havránek (Havránek-Kopeckij-Starý-Získal 1933–36). Besides, there appeared a series of textbooks and dictionaries of foreign languages, especially of English and Russian, in which the authors verified in practice the soundness of the new linguistic theory within one of the most sensitive testing fields, in language teaching. Such names as Bohumil Trnka (Potter-Trnka 1926–28), Josef Vachek (1946b), Ivan Poldauf (Osicˇka-Poldauf 1948), Kurt Roubícˇek (Kru˚ta-Roubícˇek 1941), Karel Hais (1938), most of them Mathesius’ disciples, can be adduced here; in the domain of Russian a particular mention must be made of an important representative of this section of the work of the Circle, Leontij V. Kopeckij, author of excellent dictionaries and textbooks of Russian (Kopeckij 1935) and the head of a collective that after World War II elaborated a large Russian- Czech dictionary. A more detailed list of publications of this kind would lead us too far, beyond the limits allotted to this outline. In this connection it should be noted that by these schoolbook publications the Circle laid foundations to a systematic research using a method which Mathesius himself had called analytically comparative (as distinct from the historically comparative approach) and for which in the contemporary world the term confrontational or contrastive method is used. In addition to the textbooks of English and Russian also two remarkable, though popularising, monographs appeared on the ground of PS, comparing more systematically in this contrastive way a foreign target language with Czech, the pupil’s mother tongue. This concerns, above all, the series of Mathesius’ broadcast lectures accompanying a radio course of English, which appeared in book form in the mid-thirties with a telling subtitle ‘‘What cannot be found in textbooks’’ (Mathesius 1936b). Following the example of Mathesius, a similar, also successful book on German was written five years later by the Prague secondary school teacher Jaroslav Nosil (1941), who was in constant touch with Mathesius when writing it. Also in the domain of Russian important works appeared in the field of language contrasting, above all owing to the care of Vladimír Barnet and a group of his collaborators. The importance of all this comes to the fore if we remind ourselves of the fact that western linguistics, especially that in the US, became aware of the possibility of





Josef Vachek

such a synchronic comparison of languages first in the sixth or seventh decade of our century and developed then an intense activity in this field (see, e.g. W. G. Moulton 1962, H. L. Kufner 1962, G. Nickel 1971), often without knowing that these problem areas had been treated methodologically and practically by PS three or more decades before. In his small booklet on English Mathesius sketched, be it only in a passing outline and with a predominantly popularising intent, a concrete example of what he theoretically postulated in Czech linguistics from the mid-twenties (and with which he acquainted the world linguistic public at The Hague Congress in 1928), i.e., a demonstration of linguistic characterology of a specific language. Mathesius himself explained the postulated concept as follows: ‘‘If the task of a descriptive grammar consists in presenting a complete inventory of all formal and functional elements that exist in a given language at a given stage of its development, linguistic characterology deals only with important and basic features of the given language at the given point in time, analyzes them on the basis of general linguistics and endeavours to find the relationships that exist between them.’’ At the same time, of course, attention should be paid not only to qualitative, but also to quantitative characteristic features of the language examined; it is known that, as far as quantitative study of language facts is concerned, Mathesius belonged to the first pioneers in our country, at least as regards facts of the sound level (Mathesius 1929, 1931a). From what has been said here it follows quite univocally that PS never intended to pursue science just for science sake (though its interest in general problems of linguistics was always lively and profound), but that PS as a rule endeavoured to draw as many practical consequences from theoretical knowledge as possible for the benefit of language users; in other words, through its scientific pursuits it wanted to serve the general language community. We can certainly venture the opinion that of all the structurally oriented linguistic trends, PS acquitted itself as by far the most sensitive to the needs and requirements of daily life, although it cannot be denied that some other similarly oriented groups manifested some understanding for such needs as well. What remains, however, beyond any doubt is the fact that nowhere else did this understanding attain so high a degree as in PS. For that matter already Bulygina (1964: 123) approvingly cites the statement by the German scholar H. Lüdtke (1959) that PS represents the only structuralist trend that ‘‘doesn’t take refuge in an ivory tower and doesn’t result in sterile formalism’’. It is certainly true that, e.g., the Yale descriptivist school gained recognition for the application of its conception of language to practical language instruction. Nevertheless, it is also true that a number of other topical problems of daily life remained constantly beyond its attention. This holds, e.g., for such important domains as standardization of language: it is symptomatic that even in the subsequent decades after the prestige of descriptivism was weakened, when interest in

Prolegomena to the history of the Prague School

these problems began to emerge, the main American linguist in this field proved to be an American scholar of Indian descent (Punya Sloka Ray 1963). Finally, it was only after World War II that in American linguistics a more systematic research in the domain of the language of poetry began — pushed through, especially by Roman Jakobson and by a former Prague member of the Circle, René Wellek. Moreover, as a prominent American dialectologist Raven I. McDavid pointed out, the Yale descriptivists came to deal with the domain of language instruction by a coincidence of external circumstances (by carrying out a government assignment in connection with the entrance of the US into World War II) rather than from their own initiative, as it was in the history of PS — there the interest in these problems represented an organic outcome of the whole linguistic conception from the very beginning, mainly in the approach of Jakobson and Mukarˇovský. McDavid pointed out, too, that the American tradition — at least in his time — failed to bring linguists and literary critics together to a mutual fruitful dialogue, which in the Prague cultural tradition was conducted for decades.9 So much had to be said about the origins and maturation of the Praguian linguistic theory and its specific features, by which it distinguished itself both from the preceding linguistic schools (especially from the Neo-Grammarians) and from other linguistic schools of structuralist orientation. After this general and factual view it will be useful to concentrate our attention on some concrete personalities of the founding generation of the Prague Circle — we shall see how each of them contributed to the creation of the generation’s profile by this or that particular feature. Although for technical reasons we have to confine ourselves to the founding generation of the Circle, it certainly will not do any harm to illustrate by further minor portraits the entire background in which PS originated and was gradually formed. In these portraits we base ourselves on our own personal reminiscences supplemented by recollections of other contemporaries with whom we have shared them and, last but not least, on the personalities’ own work.

.

The personalities of the founding generation of Prague School

It is only natural that the first personality we want to deal with in this outline will be the founder of PS himself, Vilém Mathesius (1882–1945). Although already in the preceding chapters many a thing was mentioned in passing or suggested, no less than a minor monograph would be necessary to let his scientific and human personage come out in its full plasticity. This task has been at least in part fulfilled by the memorial volume (edited by J. Vachek) Jazyk, kultura a slovesnost (Language, Culture and Verbal Arts), published in 1982 on the occasion of the centenary of Mathesius’ birthday (there a more detailed biography and a complete bibliography can be found; cf. Mathesius 1982).





Josef Vachek

First we should like to touch upon Mathesius’ moral pathos that emanated from his entire personality, which played such an important role in forming PS, and by which all his disciples without exception were affected. We want to give thought to the sources from which this pathos sprang: it resulted in part from his personal qualities — his pedagogical mastery, the depth and breadth of his scientific vision, his extraordinary industriousness and a deep sense of fair play towards everyone, in part from his hard personal fate, first caused by a serious eye disease, later, furthermore, by tuberculosis of the spine. This handicap, however, transposed his moral greatness to a still higher, heroic, level. The origins of his poor health Mathesius himself connected with his laborious efforts to provide the bare necessities of life during World War I: on Sundays he was obliged to get provisions for his family (since they had no relatives in the country) by trying to find better-off farmers willing to let a humble petitioner have some butter, flour or potatoes, not for money, of course, but for various valuables. The catastrophe occurred after the war, in 1924. By chance a report of an eyewitness was preserved showing how Mathesius himself heroically endured such blows. A miscellany prepared for the occasion of his 60th anniversary (completed and even set up for print, yet never published as the German occupation authorities severely curtailed scientific publications) contained among others also a personal remembrance by an intimate friend of Mathesius’ family, Professor Jaroslav Peklo, a phytopathologist. (Owing to some circumstances it was just this very contribution that failed to be published later, although all the other papers appeared elsewhere, partly in the Cˇasopis pro moderní filologii (Journal for Modern Philology), in part in the provisional revue Cˇeský cˇasopis filologický (Czech Journal of Philology) that at least temporarily was to fill the gap caused by the liquidation of the two periodicals, Cˇasopis pro moderní filologii and Listy filologické (Letters of Philology). In his recollections Professor Peklo describes how after having been alerted by an ominous phone message, in which the Mathesius family notified him that their father had lost his sight, he hurried to their flat near the then Letná water works. There he found that of the whole family only Mathesius himself remained calm, tried to comfort the others and to allay their fears. He was then, as his friend relates, in full control of the situation. In the same way he impressed us, his disciples, as well as his other contemporaries in the after years. Otokar Fischer, who belonged to the same generation as Mathesius and who had also spent his young years in the lowlands around Kolín, in his commemorative article on the occasion of Mathesius’ fiftieth birthday (1932) characterizes the honoured scientist as a rather classicist personality (while Fischer himself preferred to be taken as a romanticist). It can be said that to our generation, which came in touch with him in the second half of the twenties, Mathesius appeared to be a classicist par excellence. He always had full control of his emotions, never a stronger word passed through his lips and, at any time, he

Prolegomena to the history of the Prague School

was able to curb even the most righteous indignation he felt. He was absolute master of his time, and that is probably why he was able to give it out liberally to others who often turned to him asking for help, whether during their studies or after they finished them. All these classicist traits were obvious to anyone who got to know Mathesius even from afar. Few people, however, would have guessed what Mathesius himself confided to the author of these lines in a private letter from 1942, three years before he died: that he had to fight hard with himself to achieve his equanimity since he was of a temperament ‘‘more explosive than quiet’’. Also this admission eventually completes the portrait of Mathesius as a man of strong character, who by his will power was able to overcome even such obstacles that would wear down anyone else. Mathesius knew how to radiate the strength of his intellect and spirit from himself around — he was always full of optimism, which was not cheap but gained by striving with his enlightened faith and capability to view all, even his personal fate and everyday troubles, sub specie aeternitatis. In the obituary one of his disciples confessed ashamedly that in the hard times of the Nazi occupation we, the healthy ones, kept visiting the bedside of an incurably ill man to draw necessary doses of optimism, indispensable for us to be able to live, survive and go on fighting. Using the same strength of his spirit Mathesius succeeded in grouping around himself his friends and pupils. When owing to his serious disease he could no longer attend the lecture meetings of the Circle he used to invite regularly to his flat a group of young people he called ‘‘Little Circle’’ so as to have once a week a discussion with them on both linguistic and general cultural problems. It was also this very strength of spirit that had enabled him formerly to organize his Prague Linguistic Circle and to succeed as editor of a number of journals and miscellanies in associating round himself groups of people (of similar opinions, it is true, but rather dissimilar in personal temperament and style of work) and in inducing them to participate in fruitful work. Perhaps, he was the only personality at that time that was able to play such a unifying role: his merits did not consist only in having founded the Circle, but above all in that he succeeded in endowing it with a spirit of companionable friendship resulting in collective cooperation. If one of the beneficial activities of the Circle consisted in putting methods of collective work through, it is particularly Mathesius’ merit to have brought about that very atmosphere in which sociable cooperation thrived. It is only to be regretted that Mathesius’ work did not become better known in the world linguistics at an earlier date. It would have deserved it if for nothing else than for the really extraordinary importance of his treatise on the potentiality of the phenomena of language (Mathesius 1911–1912) — this remained without wider response, abroad mainly because it had appeared only in Czech and at





Josef Vachek

home because it could not meet with enough understanding at the beginning of the second decade of the century. When shortly after World War I Jakobson got to know this treatise, he told its author that in Moscow in 1911 he would have caused a veritable revolution in linguistics by its conclusions. And when finally in 1964–after more than half a century — the English translation of this lecture appeared (in the anthology J. Vachek 1964d), C. F. Hockett, a well-known American linguist, expressed his sincere admiration that such ideas could have been formulated at so early a period. (The Russian translation of Mathesius’ article appeared after the English version in Kondrašov’s anthology 1967.) Nevertheless, even so Mathesius gained international recognition still in his lifetime, especially by having been elected as the Czechoslovak representative in CIPL, the permanent international committee of linguists, which organizes international linguistic congresses. Thus he was recognized not only for his works in English studies, which were being published already from the first decade of our century and whose culmination was to become the posthumously published volume devoted to synchronic analysis of English (Mathesius 1961), but also for his works in Czech studies. These were concerned systematically with — also in this respect Mathesius was a pioneer — the living colloquial speech. There he mainly dealt with important problems of functional sentence perspective, which, only after his decease, became a major topic of our linguistics as well as in the international linguistic context, and with modern problems of language culture — especially in his call for stability of the Standard language (1932c). An invaluable bequest to Czech studies was made by his posthumous volume Cˇeština a obecný jazykozpyt (Czech Language and General Linguistics; 1947), not to mention his collection of essays Možnosti, které cˇekají (Chances that Lie Ahead; 1944). *

*

*

There is no doubt that in respect of international popularity Mathesius was at that time surpassed by another prominent representative of the Circle to whom we are going to devote our attention here, Nikolaj Sergeevicˇ Trubeckoj (Trubetzkoy, 1890–1938). He came from an old Russian aristocratic family from which a number of eminent politicians and scientists had descended. According to Jakobson, Trubetzkoy attended Moscow University from the age of 14, and he was seventeen when he started his publication activities, by that time not yet as a linguist but as an ethnologist. However, soon his interests centred on the field of linguistics, which prevailed to such an extent that by the time immediately preceding the Bolshevist Revolution he was preparing himself for associate professorship in comparative linguistics. The stormy period of the revolution and the subsequent civil war deprived Trubetzkoy not only of his reference library, but also of his home in his native country. After a transition period which he spent as

Prolegomena to the history of the Prague School

a professor in Rostov on the Don (in difficult conditions both personal and professional), he took refuge in Bulgaria, where he worked as a professor of Sophia University. At last in the year 1922 he was invited on Antoine Meillet’s recommendation to Vienna University as professor of Slavonic linguistics, to the chair once occupied by Miklošicˇ, Jagic´ and Vondrák. Here he undisputedly had much better working conditions than before, in Sophia. His working engagement became considerably more intensive owing to frequent contacts with Jakobson, who from the first half of the twenties lived in Czechoslovakia, and with the Prague linguistic background in general. What these contacts with this environment meant for Trubetzkoy can be illustrated by a passage from his letter sent to Mathesius on the occasion of the celebration of the 10th anniversary of the Circle in 1936. In an excerpt from it, as it was published in the periodical Slovo a slovesnost (1937: 63f.), we read: ‘‘Various stages of the development of the Circle which I have lived through with it come to my mind today — first the modest minor meetings at the chairman’s, then the heroic time of preparations for the first Slavonic Congress, the days of the Prague phonological conference never to be forgotten and many other beautiful days that I spent in the company of my Prague friends. All these recollections are connected in my mind with a special stimulating sensation, since at any contact with the Prague Circle I experienced a new upsurge of creative joy that in my lonely work far from Prague keeps sinking unwittingly again and again. This revival of and stimulus to intellectual creation represents the spirit inherent in our association and results, I insist, from the collective work of allied research workers sharing a common methodological approach and inspired by identical theoretical ideas.’’ Those years were probably the happiest period in Trubetzkoy’s scientific and pedagogical life, but they were already darkened by warning signals of his deteriorating health: Trubetzkoy suffered from angina pectoris, which was undoubtedly a toll for the hard years of privation and homelessness that his weakened organism had to pay. Those who remember the lectures in the Prague Circle, the meetings of which were taking place in a coffeehouse milieu, would long recall how at the beginnings of the sessions, still before Trubetzkoy arrived, they were asked by the organizers to abstain for this once from smoking because of Trubetzkoy’s heart decease. Understandably, the deteriorating political climate in the world (and especially in Austria) was by no means beneficial to his health. Under these circumstances his continuing scientific activities can be rated as the culmination of a supreme heroic effort. After the occupation of Austria by Hitler’s Germany Trubetzkoy clearly saw that his further existence at the nazified Vienna University was no more sustainable. The last possibility which, at least theoretically, remained, his emigration to the USA (where Trubetzkoy had friends and where he could go on with his research) was considered. However, it was too late: his flat was invaded and thoroughly searched by Gestapo and he himself was subjected to a merciless





Josef Vachek

interrogation procedure. This shocking experience resulted in a serious heart attack to which Trubetzkoy finally succumbed in hospital care, dictating almost to the last moments the remaining chapters of his famous Grundzüge der Phonologie. He died on June 25, 1938, when according to Jakobson’s information only some twenty pages were missing to make Trubetzkoy’s great life-work complete. Many valuable facts about Trubetzkoy’s life as well as about the development of his linguistic (especially phonological) views were revealed by at least partial publication of the many-years’ regular personal correspondence between Trubetzkoy and Jakobson. The letters written by Trubetzkoy to Jakobson were fortunately preserved thanks to the intrepidity of Professor B. Trnka, to whose care they had been entrusted by Jakobson, then on the run from the Nazis; Professor Trnka was able to give them back again to Jakobson in order after the war. The latter then published them in 1975 with Mouton as a valuable piece of evidence also for the early history of PS; unfortunately, Jakobson’s letters to Trubetzkoy were not preserved, but Jakobson gave brief accounts of their contents. Trubetzkoy’s heritage survives undiminished even after his decease. His thoroughly thought-out theory of phonological oppositions served Jakobson as the point of departure for his well-known theory of binary oppositions of distinctive features. The stimulating nature of the Grundzüge had as its consequence a number of re-editions, in the original German version as well as in translations (into French by Jean Cantineau 1949, Russian by A. A. Xolodovicˇ 1960 and, lastly, into English by C. A. M. Baltaxe 1969). In Vienna (and later on also in some other smaller Austrian cities) several international phonological conferences took place (after 1964); twenty-five years after his death, Vienna University honoured Trubetzkoy by fixing a commemorative plaque in the university columns, as has been the custom pertaining to the most illustrious professors of that institution. Trubetzkoy’s communication with Jakobson was not only continuous but also highly concordant in opinions as regards fundamental problems. Jakobson appears to have often inspired Trubetzkoy, who, now and then, brought Jakobson’s formulations to more perfect shape. They formed an excellent team together that used to participate in discussions jointly (as in The Hague case mentioned above). There is no doubt that in respect of practical organization it was Jakobson who played the leading role in the cooperation of the two scholars; on the other hand, Trubetzkoy synthesized Jakobson’s stimulating ideas in a most remarkable way: see, e.g., Trubetzkoy’s system of phonological oppositions. As is known, Jakobson formulated his theory of binary oppositions of distinctive features — at least in its rudimentary shape — still at the very close of Trubetzkoy’s life. The fact that this binaristic conception fails to be consistently accepted in Trubetzkoy’s Grundzüge does not necessarily involve, as has been inferred, e.g., by André Martinet, an essential contradiction between Trubetzkoy and Jakobson in ques-

Prolegomena to the history of the Prague School

tions of classification of phonological oppositions. It should be considered under what difficult conditions of life Trubetzkoy was bringing his manuscript to conclusion: he had absolutely no chance to subject it to a final revision and edition, which undoubtedly would have introduced desirable last corrections in more than one place. Anyway, it can be said that in essential matters the two researchers’ views concurred in a surprising way. The warm relationship of Jakobson to Trubetzkoy is shown by the fact that it was to him that Jakobson dedicated his Remarques (Jakobson 1929 = TCLP 2). In turn, as the Introduction supplied by the editors to the Grundzüge has revealed, in a planned yet never written preface Trubetzkoy intended to dedicate this work to Jakobson. In some points there is certainly no doubt that Jakobson induced Trubetzkoy to make some minor corrections in his formulations. This concerns, e.g., some of Trubetzkoy’s psychologizing definitions or, as the case may be, his psychologizing phraseology from the beginnings of phonological research, when Trubetzkoy was ridding himself of Baudouin traditions at a somewhat slower pace than Jakobson: the points in question were mainly the definitions of phoneme and morphoneme. Never, however, were the differences as serious as to indicate the existence of any fundamental contradiction between the two scientists. *

*

*

As is natural in the given context, the following lines will be devoted to the scientist whose name was repeatedly mentioned here in connection with Trubetzkoy’s work, Roman Osipovicˇ Jakobson (1896–1982). During his studies at Moscow University he got to know very well not only the traditions of the school of Jan Baudouin de Courtenay and Lev Vladimirovicˇ Šcˇerba, but also the subtle Neo-Grammarian methods of Filipp Fedorovicˇ Fortunatov and his school. From the very beginning he manifested a lively interest in problems of poetics and poetical language, an interest he did not relinquish even later, when he concentrated his main efforts on questions of structural grammar and phonology. By coincidence he came into his first closer contacts with Czech ambience through his well-known monograph on Czech verse as compared with Russian (Jakobson 1923), which appeared three years later also in its Czech version. He then became so familiar with the Czech environments and problems of Czech language, literature and culture in general that Arne Novák in his Prˇehledné deˇjiny literatury cˇeské (Outline of Czech Literature; 1936–38 = 1995:1617) did not hesitate to characterize him as a ‘‘Czechized Russian’’. From 1933 till the Nazi occupation he worked as professor of Russian and Slavonic studies at the University of Brno. After the country was occupied, in 1939, he fled to escape political and racial persecution and took refuge in Denmark, then in Norway and finally in the





Josef Vachek

United States, where he lived until his death. During the war he held for some time the post of professor of Czech language and literature at Columbia University in New York, from where at the beginning of the fifties he moved to Harvard University in Cambridge, Mass. as professor of Slavonic studies; at the same time, he was professor of general linguistics at the Massachusetts Institute of Technology. Our main interest will be centred on that period of Jakobson’s activities in which his relationship with the Czech ambience was very close and in which he was regarded as one of the most prominent members of PS. He was one of its founding members: his name appears among the names of the participants of the first meeting of the Circle, and it was he who at one of the first meetings (on January 13, 1927) presented his views on teleogically conceived phonology. From the very beginning he was vice-chairman of the Circle and there is no exaggeration in maintaining that he was its spiritus agens. This was especially true about his role at the international phonological conference in 1930, where he not only delivered three important papers, but also prepared the wording of the ‘‘Outline of Standardized Phonological Terminology and of the Principles of Phonological Transcription’’; moreover, he edited the minutes of the discussions, not to speak about his editorial share in publishing the fourth volume of TCLP, containing all the materials of the conference. In a similar way it would be possible to demonstrate his indefatigable initiative and efficiency by enumerating further activities, e.g., in connection with the cycle of lectures on Standard Czech and language culture, a number of longer and shorter articles published in Slovo a slovesnost, and his last editorial work, which had to be anonymous, viz the edition of Trubetzkoy’s Grundzüge published as the 7th volume of TCLP. It was probably Jakobson to whom we are indebted for the arrangement of Trubetzkoy’s bibliography (outwardly signed by B. Havránek) appended to the 8th volume of TCLP, in which treatises and essays devoted to Trubetzkoy’s memory were contained. Jakobson’s own article could not, of course, appear here. (Some of his papers succeeded in being eventually printed, mostly in the miscellany edited by Mathesius (1939–1940) Co daly naše zemeˇ Evropeˇ a lidstvu (What our Countries Gave to Europe and Mankind): they appeared under the name of Olaf Jensen — yet the initiated knew well who was hidden under this pseudonym.) Therefore, it is no exaggeration to say that if it had not been for Jakobson the Circle would have been deprived of many a characteristic feature and its activities as a whole would have been much less diverse. It can be even maintained that without Jakobson the Circle would not bear its name, which was to become a concept known all over the world: the name Prague Linguistic Circle was presumably chosen after the model of the Moscow Linguistic Circle, a progressive association of which Jakobson used to be a very agile member before he moved to Prague. Even after years Jakobson came back to this point and substantiated the choice by the intent to express by this name a nontraditional scientific endeavour striving for new, progressive ways (Jakobson 1965).

Prolegomena to the history of the Prague School

At the meetings of the Circle Jakobson belonged to the most regular and best informed debaters: his profound philosophical erudition (he was, e.g., an eminent connoisseur of Husserl’s phenomenology) enabled him prompt orientation in the problems in question and his clear rigorous logic qualified him to formulate his conclusions precisely and convincingly. In discussions with the adversaries of the Circle he was insuperable — his formulations were forceful as blows with a hammer and at the same time sharp as a razor. Nevertheless, he was completely devoid of any professional pride: at any time he was willing to discuss with anyone questions that fell within the domains of linguistics and poetics, being most obliging in this respect towards young adepts. Jakobson’s fervid interest in questions of verse, poetic language and poetics was closely connected with his constant contacts with poets, writers and artists in general who often attended the meetings of the Circle and contributed, as we already know, to its periodical. Jakobson thus represented, along with Jan Mukarˇovský, the main connecting link of the Circle with the artistic world. That is also why he very energetically took part in the cycle of lectures (often mentioned here before) on Standard Czech and language culture, where he subjected to scathing criticism the then puristic practice of Jirˇí Haller (at that time the editor of Naše rˇecˇ). The contacts with artists were naturally connected with Jakobson’s intensive social life: he liked to visit wine restaurants together with men of letters (as has been also recollected by such poets as Víteˇzslav Nezval and Jaroslav Seifert in their memoirs). It was obviously due to his influence that for long years it was a custom in the Circle to visit after finishing a lecture session some cosy wine bar ‘‘to continue the discussion’’. This is only seemingly a marginal fact that might be omitted in this outline: it was also through such informal meetings that friendly ties were strengthened, ties that served as one of the main guarantees of successful results of a work done collectively. However brief this outline may be, it must contain a mention of Jakobson’s (1936), as well as Frank Wollman’s (1936), dauntless denouncement of a monograph written in the spirit of aggressive German chauvinism by Konrad Bittner, senior lecturer of the Prague German University, entitled Deutsche und Tschechen I. Both articles were based on papers read at a lecture session of the Circle and they exposed the unscientific character of Bittner’s theses intended as ideological preparation of the Nazi occupation. It was this courageous deed that (in addition to the menace of racial persecution) forced Jakobson to leave Czechoslovak territory after the Nazi invasion as soon as possible. All these facts show in a very convincing manner how significant a role was played by Jakobson in the early PS. It would be possible to demonstrate in detail that a great majority of his theses (e.g., about the openness of the language system, about the therapeutic function of vowel changes, the need for finding general language laws, etc.) are valid till today not only within the confines of PS but on the international linguistic scene as well. It can be added here that even





Josef Vachek

Jakobson’s well-known theory of binary oppositions of distinctive features has its origins already in the Prague period of his activity — for the first time he dealt with this topic more systematically in one of his Prague lectures, at the meeting of the Circle on March 21, 1938, devoted to acoustic classification of consonants; Jakobson’s conclusions were then printed in full in French in the Proceedings of the 3rd International Congress of Phonetic Sciences, Ghent, 1938. Further development of his theses falls within the postwar period, when Jakobson worked outside the Czechoslovak domain. *

*

*

Another distinguished personality to be paid special attention is Bohuslav Havránek (1893–1978). He had entered the history of PS as one of the participants in the first meetings of the Circle, particularly by his very intensive cooperation in formulating the Prague theses from 1929 and, later, the general principles of language culture (1932). He substantially contributed to the success of the international Prague conference in 1930 and gave lectures at international congresses of Slavonic studies (beginning by the year 1929), where he always successfully promoted the functionally structural approach to the given language problems. Of great importance is his editorial work in the revue Slovo a slovesnost; here he originally collaborated with V. Mathesius and then, after Mathesius died, continued as its editor-in-chief. At the Prague University Faculty of Arts, where Havránek studied in addition to Czech also classical philology, his teachers were mainly Josef Zubatý and Jirˇí Polívka. The significance of Zubatý’s scepticism in the period of Neo-Grammarian hegemony has been mentioned above: among his pupils it was Havránek who belonged to the most perceptive. For Havránek’s further scientific development an important role was played by his assignment to the Office of the Dictionary of Czech, where he penetrated deep into the problems of Standard language and where he benefited greatly from discussions with the researchers then working on the Dictionary, especially Kvido Hodura and the untimely deceased Václav Ertl. Just as Ertl, the originator of the theory of ‘‘good authors’’, Havránek worked his way to the conception of language correctness given by function, not by historical purity, and, in the same way as Ertl, he was able to incorporate his theoretical views into practical schoolbooks. — Havránek was also an excellent dialectologist; his first lecture as senior lecturer was devoted to Czech dialectology. Havránek’s overview of Czech dialects as presented in the third volume of Cˇeskoslovenská vlastiveˇda (Czechoslovak National History and Geography) has not yet been surpassed, although a distance of more than six decades intervenes between its publication and our time. His two-volume monograph on verbal voice in Slavonic languages is regarded as a great contribution to the Slavonic as well as general

Prolegomena to the history of the Prague School

grammar. However, Havránek’s main contribution to Czech studies was made by his works devoted to the theory and practice of standard language. We have already mentioned Havránek’s fruitful discussions with Hodura and Ertl, which helped to form his functional approach to the problems in question. This approach can be traced already in his paper on Slavonic adjectives corresponding to the Latin type in -bilis; however, the study to be noted here above all, is Havránek’s extensive treatise based on his lecture on the tasks of standard language and on its culture (from the cycle organized by the Circle in 1932). This lecture laid firm foundations of the functional structuralist theory of language in respect of grammar as well as style. Havránek also showed how the function of standard language can influence the grammatical and phonological structure of the language; further, he drew a distinction between such important terms as ‘‘norm’’ and ‘‘codification’’, at a time when the Neo-Grammarian linguistics completely ignored this difference and when even progressive currents in western linguistics were not aware of the existence of these problems. Havránek’s collected works dealing with these problems were published in the sixties (Havránek 1963). Havránek’s interest in questions of bilingualism and linguistic areas (Sprachbund) must be regarded as equally modern. As an expert in Balkan languages he devoted special attention particularly to the Balkan language situation, having thus anticipated in this field by several decades the sociolinguistic research, so topical in the contemporary world. As an exacting but fair pedagogue, Havránek, who came from the family of a secondary school teacher, could unerringly pick young talented students; a long row of his pupils, Slavists, Bohemicists as well as specialists in Russian acquired university posts in Prague or Brno. He never denied any of his pupils advice and help, either during their studies, or after they finished them. Also his relationship towards the staff and employees of the Institute of Czech Language, the director of which he became in 1950, was always distinguished by humane kindness, although he never made concessions in his exacting demands on work. Moreover, we have not yet mentioned Havránek’s editorial participation in numerous dictionary projects beginning with Prˇírucˇní slovník jazyka cˇeského (A Shorter Dictionary of Czech Language) through Slovník spisovného jazyka cˇeského (A Dictionary of Standard Czech Language) up to the recent Slovník starocˇeský (A Dictionary of Old Czech); everywhere he joined in with pleasure and willingly placed his encyclopedic knowledge and many years’ experience at the disposal of all. In all these works Havránek’s contemporaries admired his amazing concentration on the given task and his precise matter-of-fact objectivity unaffected by transient moods, both being indispensable prerequisites of a successful scholar and teacher. In connection with the work on the Old-Czech Dictionary let us recall that Havránek succeeded in training the leader of the lexicographic collective of this dictionary, Igor Neˇmec; it was through the work of this scholar that the early idea





Josef Vachek

of the Prague theses from the year 1929 has materialized: the concept of a teleologically conceived subsystem of the Czech language as seen in the course of its development from the forteenth century up to the present time. Neˇmec has fulfilled this task, after preparing himself by a longer series of profound partial studies of the same teleological orientation; he thus established a model to be followed by the work of further researchers in a similar conception (cf. J. Vachek 1994a). Havránek was truly one of the few scholars who combined profound scientific erudition with pedagogical mastery. He demonstrated this quality at secondary school as well as at University both in Brno and Prague, where he lived up to his teacher’s reputation, though being hampered by numerous official functions: in Brno he was the Dean of the Faculty until the Czech universities were closed by the Nazis, and in Prague he worked for a number of years as the Rector of the then University of the Russian Language. In these positions he often dauntlessly helped good causes — e.g., it was his merit that the valuable libraries of individual departments of Brno University lived through the war and occupation without having suffered substantial losses and were able to start operating normally soon after liberation. As an experienced teacher Havránek was aware of the importance of a reliable bibliography — especially for young adepts of science. That is also why for decades he devoted much energy and effort to this task. After the war he iniciated a number of more or less extensive bibliographical publications accomplished under the guidance of his disciple Zdeneˇk Tyl or of others (e.g., František Daneš, Marie Teˇšitelová, etc.) It is also in connection with Havránek’s pedagogical activities that several treatises characterizing the language of major literary works appeared: as early as 1921 he noted the folk features of the language of the writer Alois Jirásek, in the novel Babicˇka (Grandmother) by Božena Neˇmcová and in the language of the new post-war literature; later he also examined the language of the romantic poet Karel Hynek Mácha. At the same time he was also an eminent specialist in older Czech and the language of the National revival period — all this perfectly qualified him to write his monumental ‘‘Vývoj spisovného jazyka cˇeského’’ (Development of Standard Czech) in the second series of the Cˇeskoslovenská vlastiveˇda (Czechoslovak National History and Geography; Havránek 1936). This work can be regarded as pioneering in its own way, namely in that it succeeded in joining the respect for the inner development of the language (regarded sometimes by some structuralists as an exclusively immanent matter) with attention given to social, economic and political context, local and temporal, within which the language developed. It is to be regretted that excess of other duties did not allow Havránek to prepare for print a new extended edition of this monograph. Havránek indisputably belongs to those members of PS who represent its veritable pillars of support — without his specific contribution involving a

Prolegomena to the history of the Prague School

specific touch the gamut of the working repertoire of PS would be considerably narrowed.10 *

*

*

Last but not least, it is necessary to mention here in more detail Bohumil Trnka (1895–1984), who acted as secretary of the Circle for long years and who was also one of those standing at its cradle. A native of Kletecˇná near Humpolec, he studied modern philology at Prague University. By his profound and wide interests he attracted the attention of his teacher Vilém Mathesius, who chose him as his research assistant when such assistance was approved by the Ministry of Education after Mathesius had been afflicted with his serious eye disease. (It is worth mentioning that Mathesius never had an ordinary assistant and never asked for one to be appointed to him. This he explained by pointing out that cooperation with junior research assistants and student assistants suited him better.) At that time Trnka had already gone through some remarkable activities outside the domain of linguistics: he dealt with problems of shorthand systems in great detail and himself was the author of an original system of Czech shorthand. Although this system had failed to win in an official competition, it offered its author ample opportunity to contemplate the problems of combinatorics of Czech speech sounds and the frequency of their occurrence. This all came very useful to Trnka when he started to work on questions of phonology, both qualitative and quantitative. In the field of English studies, however, Trnka began his research not by phonology, but by thorough analyses of syntax, which led him from the Old English period to that of Early New English (his study of the syntax of the Early New English verb appeared as the third volume of TCLP, cf. Trnka 1930). The contacts with Jakobson and Trubetzkoy soon directed his attention to phonology, where — certainly under the influence of his earlier investigations in stenography, but also following the example of Mathesius — he presented a very original and elaborate quantitatively conceived phonological analysis of Present Day English (Trnka 1935). From there his linguistic interests, inspired among others by Karel Skála-Rocher (also handicapped by eyesight affliction — Trnka helped him out in his literary work, too), turned partly to the study of homonymy, partly to special questions of phonology and grammar. Trnka went into the problems of phonological neutralisation, subjected to critical analysis Trubetzkoy’s systematics of phonological oppositions, and, above all, elaborated his own systematics of the morphological language level resting on exactly ascertainable morphological oppositions of a given language (Trnka 1958). Trnka takes up a very prominent position also in the field of historical phonology. Already one of his first contributions compares the phonological system of Old English with that of New English (Trnka 1929a); later he dealt with





Josef Vachek

the phonological situation in (Proto)Germanic, and lastly in Late Middle English and Early New English. Here, as we are convinced, Trnka definitively solved an old mystery of the English historical phonology, viz the motivation of a large complex of changes referred to as the ‘‘Great Vowel Shift’’ (Early New English shift of vowels). The initial impulse that had set this whole complex in motion Trnka correctly ascribed to changed relations between individual components of the short and long subsystem within the system of Middle English vowel phonemes (Trnka 1959, Vachek 1959). An important event may be seen in the foundation and subsequent lasting guidance by Trnka of the working group for functional linguistics attached to the Kruh moderních filologu˚ (Circle of Modern Philologists), whose members used to meet regularly in Trnka’s study and developed the principles of the Prague Linguistic Circle, then already disintegrated. From this group, by the way, originated the well-known programmatic declaration on Prague structural linguistics that Trnka’s group sent to Moscow, where it was published in the same year in Voprosy jazykoznanija and, a year later, in the Prague revue Philologica Pragensia (B. Trnka, J. Vachek et al.1957). In the context of the then unclear world situation such a declaration can be appreciated as a bold act indeed. Trnka explicated his overall conception of English grammar in his university textbook published in three parts (Trnka 1953–1956); in the same form his lucid explanations of Old English and Middle English literature appeared; he also wrote a profound treatise on Shakespeare’s philosophy. The sphere of his interests was very wide and Trnka, as a rule, reached very deep, aiming at crucial questions of the given theory. Although his conclusions were not always generally accepted (e.g., his objections to the motivation of changes at one language level by the structural situation at another level, his rejection of lexis as one of the language levels), it is beyond any doubt that his conception represents a fully original feature, a specific variant of the theory of PS, much appreciated abroad as well. Trnka acquired his reputation of a foremost pioneer in the world context by his interest in the quantitative aspects of language phenomena. He gained recognition after the postwar Paris Linguistic Congress appointed him Secretary of the International Commission for Quantitative Linguistics attached to the Permanent International Committee of Linguists charged with organizing linguistic congresses. Another of Trnka’s merits has already been mentioned: the preservation of the set of Trubetzkoy’s letters to Jakobson, which he kept concealed during the whole war at Jakobson’s request (see Jakobson 1975). Trnka had two important pupils, Jirˇí Nosek (who compiled a bibliography of Trnka’s works in PSE 11, 1955, 89–93) and Libuše Dušková, professor of English language at the Prague University Faculty of Arts. Nosek succeeded in obtaining the position of associate professor in time, but he had to wait for full professorship until the Communist régime in Czechoslovakia broke down in November 1989; he

Prolegomena to the history of the Prague School

had been denied the nomination earlier for ‘‘ideological’’ reasons. On similar grounds the academic career of Dušková was delayed, although she was the author of a number of valuable treatises and, last but not least, the main author of the excellent Mluvnice soucˇasné anglicˇtiny na pozadí cˇeštiny (A Grammar of Contemporary English against the Background of Czech; 1988). For a long time she kept writing her works as an ordinary research worker in the Cabinet of the Academy of Sciences, as political authorities prevented her even from being appointed associate professor — only after November 1989 she received her satisfaction at last. *

*

*

From among the prominent members of the Circle we shall mention, even though only briefly, the best known ones, starting by the non-Czech members. The first to be mentioned was the professor of Indo-European linguistics at the Prague German University, Friedrich Slotty. He was a convinced socialist who did not yield even to an intensive Nazi pressure and remained a member of the Circle all the same. After the war he voluntarily chose the East German State and worked there as a university professor until his death in 1964. Although a great friend of Czechs, he had a poor command of Czech (writing and lecturing in the Circle, but in German). Two other foreign scholars who regularly attended the lecture sessions of the Circle were two Ukrainian friends, professors of the provisional Ukrainian University in Prague, Agenor Artymovycˇ, a classical philologist, and Vasyl Simovycˇ, a Ukrainist. The former entered the history of PS by his lecture showing important autonomous features of written language (Artymovycˇ 1932). This topic had been dealt with in PS already before (Trubetzkoy, L’udovít Novák and Josef Vachek); Artymovycˇ, however, belonged to its pioneers. Artymovycˇ died in Prague in 1935 (a warm obituary by Jakobson himself appeared in SaS). Simovycˇ moved to Lviv, where he later died. Chronologically, two other Russian linguists belong to the first generation of PS, Sergej Karcevskij and Leontij Vasiljevicˇ Kopeckij. The former, well versed in the Saussurean theses, was teaching in the second half of the twenties at the Prague Russian (Exiled) Secondary School and communicated closely with Jakobson and Trubetzkoy, with whom he conceived the well-known theses for The Hague Congress in 1928. He wrote two valuable contributions to the first and fourth volumes of TCLP; later he returned to Geneva, where he lived as director of the Institute of Slavonic Studies until the end of his life. Kopeckij lived in Prague from the twenties as associate professor of Russian at the High School of Commerce, after the war as professor of Charles University. He was an eminent specialist in the didactics and methodology of foreign languages, in which domains he applied the principles of PS.





Josef Vachek

From the Czech members of PS it was Karel Horálek, a Slavonic scholar, Vladimír Skalicˇka, a general linguist, the author of a structuralist typology, and, further, a prominent Germanist and Baltic scholar Pavel Trost, who all became well known in the subsequent generation. An important role, of course, was played by Havránek’s Czech and Moravian pupils, e.g., František Daneš from Prague; from among the Brno scholars the following must be named: Adolf Kellner, an excellent dialectologist, and, from the younger generation, Miloš Dokulil, Jaroslav Bauer, a Slavonic scholar, and Arnošt Lamprecht, a Bohemicist. To them also Miroslav Komárek from Olomouc can be added, who in recent times successfully confronted the teachings of PS with the objections raised to the Prague approach by post-structuralists such as Jacques Derrida. Also experts in other domains, often famous abroad nowadays, have learned much from PS, too (e.g., Anglicist Jan Firbas and Greek and Mycenian scholar Antonín Bartoneˇk). The Statutes of the Circle were signed by some further prominent linguists who thus became founding members of the Circle as an officially recognized association, yet they never used the functionally structural methods in their work (although occasionally manifesting considerable sympathies for them). It was, above all, the professor of comparative Indo-European linguistics Oldrˇich Hujer (1880–1942), an adherent of the Neo-Grammarian current, but at the same time an eminent methodologist and an excellent teacher of a number of linguists from Prague and Brno (for some years he used to give lectures there, too). One of his pupils, Václav Machek (1894–1965), became associate professor at Brno University; he had signed the Statutes as well and lectured in the Circle on the subject of his thesis submitted to obtain associate professorship. In some points he came very close to the principles of the Circle, however, he never accepted them in full. The third of these signatories was the associate professor, later full professor of Charles University František Oberpfalcer-Jílek, a researcher of encyclopedic knowledge, but of eclectic methodological conception. It was he who brought Sapir’s (1925) famous article on ‘‘sound patterns’’ to the attention of the Circle, but he did not deal with problems of phonology more profoundly. Lastly, let us recall as another signatory Vladimír Buben, professor of French, then teaching in Bratislava, later in Brno and Prague. Among the founding members of PLC also a signatory of its Statutes Jan Rypka (1886–1968), then associate professor of Turkish and Modern Iranian literature, later professor of Charles University must be mentioned. His main interest concerned the literature of these languages (History of Iranian literature, written under his guidance, appeared in English and Rypka was awarded distinctions for it also from the Iranian side). A special position, already among the members of the first generation of the Circle, was taken by Josef M. Korˇínek (1899–1945), a pupil of Josef Zubatý and Oldrˇich Hujer at Prague University, from the mid-thirties professor of compara-

Prolegomena to the history of the Prague School

tive linguistics at the University of Bratislava. His death after the war prevented his transfer to Charles University. He specialized particularly in problems of expressive function of language (his main work Studie z oblasti onomatopoje (Studies in Onomatopoeia) appeared in Prague in 1934); here he arrived at many significant conclusions. Korˇínek’s work as an academic teacher in Bratislava has been highly appreciated by his pupils. They dedicated to his memory the entire first volume of Recueil linguistique de Bratislava (1948).11

. Hard post-war decades of the Prague School (–) The undeniable successes of the Prague School of Linguistics and its systematic effort to go on working out the principles inherited from the founding generation must be appreciated still more if we consider the complicated situation in which its members were obliged to work after World War II. Let us state only that two of the leading members of the founding generation (presented in more detail in the preceding chapter) could not take part in the postwar work any more since they did not live to see the end of the war: Trubetzkoy left his collaborators at its very beginning, and the founder of the Circle himself, Vilém Mathesius, died shortly before its termination (on the same day as F. D. Roosevelt, April 12, 1945). No less a loss for the postwar Prague Linguistic Circle was incurred by the fact that it could not rely upon the participation of its Vice-President Roman Jakobson, who was forced to leave Czechoslovakia in a hurry due to the occupation. He temporarily stayed in Denmark, then in Norway, and after its occupation he took refuge with his wife in Sweden, from where he succeeded in getting to the US. For some time he worked at a university establishment founded for scholars who had to flee the countries governed by Nazi Germany, then for some time he lectured at the Department of Czech Language and Literature at the University of New York (Columbia University). After the war he would have preferred to return to Czechoslovakia, where he would have been accepted by the Circle with enthusiasm. However, on the inquiry of the presidium of the Circle, the Ambassador of the USSR in Prague, Zorin, advised ‘‘waiting with it’’. Soon it was to appear that the disapproval concerning Jakobson’s arrival in Prague had deeper roots. In the Russian postwar linguistics important positions had been occupied by stadialist conceptions of language, building on the theory of the Caucasiologist Nikolaj J. Marr; his conception under the term ‘‘new teaching on language’’ claimed to be the only one ideologically acceptable, antagonistic towards all ‘‘bourgeois linguistic trends’’, consequently towards the structuralism of the Praguian type as well. The first signal of ideologically motivated criticism of structuralism was given in an article by the Soviet linguist Nikolaj S. Cˇemodanov as early as 1947. The Circle responded to it in a lecture by Vladimír Skalicˇka (Skalicˇka 1947–1948) ‘‘The





Josef Vachek

Copenhagen structuralism and the Prague School’’ (later published in Slovo a slovesnost), which pointed out significant differences between the conception of Louis Hjelmslev (which separates language reality from both the material realization and the meaning reality of the outer world, reducing thus language to an algebra of mutual relationships between its components) and the functional structuralist conception of PS, which is respectful of both the material realities, by which mainly phonic realization of language utterances is meant, and the functional sign nature of the language system with respect to external reality. Unfortunately, this argumentation soon proved to be a weak defence against the more and more powerful assaults of Marrism. Already by the end of 1947 and again a year later arguments by Alexander V. Isacˇenko could be heard at the meetings of the Circle, which showed a shift towards Marrist positions. A visible symbol of the installation of the ‘‘new teaching on language’’ as the official linguistic credo in our lands was presented in a lecture by a prominent advocate of Marrism, Marr’s disciple Ivan Ivanovicˇ Mešcˇaninov; the wording of this article duly appeared at the turn of the years 1948 and 1949 in Slovo a slovesnost (Mešcˇaninov 1949) — the desirable orientation of linguistic work in this country had been clearly marked out in it in the sense of Marrist stadialism. Fortunately, however, already in summer 1950 an essential reversal took place at the Soviet linguistic front. It happened so that Stalin (allegedly owing to the intervention by A. S. Cˇikobava) let himself be persuaded that an administrative measure was required against the Marrist vulgarizations, which compromised Russian linguistics in the international scientific context. This intervention naturally brought the Marrist era to an end in our country as well. In this situation something like reinstatement of the functional structural conception represented by PS could be expected. Curiously enough, a conclusion followed that was equally strange as nonsensical (obviously inspired by the Prague political sphere) resulting almost in practical identification of the teachings of PS with Marrism. As an external manifestation of that blatant disregard for facts, the prominent Party weekly Tvorba (years ago given at the disposal of the Communist Party by F. X. Šalda) started criticising (in a series of papers appearing regularly) various theses of PS, attacking especially Roman Jakobson. Although later, owing particularly to Havránek, the ideologically aggressive argumentation was substantially moderated, practical consequences of this campaign persisted in many respects. One of them involved actual dissolution of the Prague Linguistic Circle, ‘‘replaced’’ partly by Jazykoveˇdné sdružení (Linguistic Association) for Bohemist, Slavonic and general linguistic studies, partly by Kruh moderních filologu˚ (Circle of Modern Philologists), dealing with both linguistics and literature in the field of languages other than Slavonic. An important event was the establishment, within the Circle of Modern Philologists, of the Group for Functional Linguistics, thanks to the brave Trnka, who presided over it for years, having assembled a number of younger

Prolegomena to the history of the Prague School

researchers working fully in the spirit of the principles of the abolished Circle. Trnka’s group entered the history of our linguistics in 1957 by participating in the Soviet discussion on linguistic structuralism. Its report (Trnka, Vachek et al. 1957) subsequently appeared in the Soviet revue Voprosy jazykoznanija and was received with sympathies. This was indeed a bold step; the English translation of the theses of the report was then published in a new revue Philologica Pragensia (which had until then appeared as a Supplement to Cˇasopis pro moderní filologii). Another negative consequence of the confusing campaign of the post-Marrist period was the fate of the once renowned series of TCLP. From this series after the 7th volume (Trubetzkoy’s Grundzüge der Phonologie 1939) yet the 8th volume (Études phonologiques . . . 1939) appeared, originally intended to honour the fiftieth anniversary of Trubetzkoy’s birth, after his death published to pay homage to his memory. As the ninth volume Trubetzkoy’s Altkirchenslavische Grammatik was being prepared; however, its publication was hampered by an incredible number of obstacles. First, it seemed inappropriate to issue it in German; therefore, Trubetzkoy’s daughter translated it into French and in this version it was completely set up for print. Nevertheless, no one bold enough could be found in the given situation to support the publication, so that the type had to be broken up. At last the book was published by the Viennese Academic Publishing House in its original German version in 1954. The harm caused thus to our institutions, not only in terms of economy, but, above all, in respect of moral values, cannot be adequately appraised . . . It was largely Havránek’s merit that the subsequent confused post-Marrist period calmed down; in the sixties (1965) he complied with the call for ‘‘pure Marxist linguistics’’ by tentatively formulating the principles of such a theory — more or less in the spirit of good Praguian traditions (SaS 26, 1965:101–4). Also the author of a particularly sharp attack against the Jakobsonian conception of Prague structuralism (SaS 13, 1951:1–11) regretted his one-time rashness (SaS 17, 1956) and at his workplace at the Faculty of Mathematics and Physics organized a team, in its own way implementing the classical objectives of functional structuralism, building its own conception of a stratificational grammar, in some respects close to the conception of the American linguist Sydney M. Lamb. His changed point of view was formulated in detail in a number of papers and publications of his own as well as in those of his team (E. Hajicˇová and others). In the interests of truth it should be added, however, that Jakobson did not respond positively to the change of the author’s opinions and persisted in his implacable attitude towards the repentant author permanently. On the other hand, it is also necessary to state that very unjust reproofs have also been (and still are) directed at Havránek (e.g., in the third part of Pameˇti (Memoirs) of Václav Cˇerný). However, as a matter of fact, it was Havránek, who, valuing his colleagues above all according to their honest scientific work, was always ready to help effectively many of those who failed to





Josef Vachek

conform to the norms and ideas of the Party, to overcome their difficulties. Therefore, it was only he who could succeed in straightening out the post-Marrist confusion, which process, of course, took some time. Very convincing support for our above appreciation of Havránek can be found with an otherwise severe and not always just critic of the situation in our linguistics of the Marrist and post-Marrist period, viz Pavel Novák (1991). It is a proof contained in a footnote only (No. 14, p. 189), yet a very significant one. It says: ‘‘Soon the Memoirs of V. Cˇerný will be generally accessible (they appeared in 1992, note by J.V.) with very harsh formulations concerning the role of Professor B. Havránek after the coup d’état in 1948. All the same, the knowledgeable and sensitive Professor P. Trost, in his concluding speech at the meeting to honour his own seventieth birthday (in 1977) was able to state about Havránek: It was he to whom I owe most — not only myself but the majority of us, not only individuals, but our academic discipline and its very essence after it survived Marrism and anti-Marrism.’’ Among other things, it was also this approaching calm in the sphere of our linguistics that facilitated the publication of my Dictionaire de linguistique de l’École de Prague by the Dutch Publishing House Spectrum (Utrecht-Antwerpen 1960), again thanks to the approval of Havránek; the edition was successful even in Russia so that it appeared in Russian in Moscow. It might be concluded that the Russian translation of the terminological dictionary of PS was favourably accepted in our own as well as in Russian circles (where already in the past the Praguian theses were received with approbation by such scholars as V.A. Zvegincev, N.A. Kondrašov, M.M. Guxman, T.V. Bulygina, etc.). Especially the positive evaluation by the Estonian school of Tartu must be mentioned here, represented particularly by Jurij Mixailovicˇ Lotman (1922–1993), whose monograph Struktura xudožestvennogo teksta (Structure of artistic text; 1970) marked out with sympathies the work of Mukarˇovský. Thanks to the calming down at the ideological and political front it was possible to resurrect the Prague series of Travaux (though under the changed title Travaux linguistiques de Prague). There appeared four volumes of this revived series (1964, 1966, 1968 and 1971), the first two under my editorship. Perhaps it will not be regarded as immodest to recall that the permission to renew this series was obtained through the present author’s colleague, the technician Josef Bartolomeˇj Slavík, a personal friend of the corresponding member of the Czechoslovak Academy of Sciences Bacˇkovský, then in chair of the editorial commission of the Presidium of the Academy of Sciences, who gave his consent at Slavík’s intercession. It is only fair to mention that it was my colleague from the Institute of Czech Language Zdeneˇk Hlavsa who regularly participated in the editorial work for the new Travaux. This general loosening of the ideological and political atmosphere doubtless facilitated my holiday stay at Bloomington, Ind. in 1964, at the invitation by

Prolegomena to the history of the Prague School

Thomas A. Sebeok, general linguist at the local University, to give two series of lectures. There also two books resulting from my stay were published: an anthology Prague School Reader in Linguistics 1964 and The Linguistic School of Prague 1966, both by the Indiana University Press. Instead of the royalties I asked for 300 copies of each of the books to distribute them among our linguists from the PS and to present them to the libraries, Departments of English and of general linguistics at all Czechoslovak universities. It is beyond any doubt that all this could not have materialized without the clearing up of the overall atmosphere at that time; this also concerns my stay at the famous Dutch University of Leiden as visiting professor of English, which took place in the unforgettable years 1968–69. The very dates show that during that time again ominous clouds started to gather over PS and its fate, which nearly prevented me from assuming my position in Leiden. In the end all came off well, in spite of the fact that the so-called ‘‘fraternal help’’ against the supposed counter-revolution was then organized (Soviet military invasion into Czechoslovakia in August 1968 — editor’s note). I succeeded in concluding the assignment in summer 1969 (and was able to invite my family to Benelux and to Britain for the saved-up salary), but hard times lay ahead of me and of the Prague linguistic school after my return . . . Bad times came also for the Institute of Czech Language, the members of which accepted the events connected with the Prague Spring with great sympathies; which, of course, could not remain unpunished by the ideologically ‘‘rectified’’ Presidium of the Academy. Therefore, another spell of ideological screening took place, which, on the whole, fortunately passed without substantial losses; even so, reprisals were taken against Alexander Stich, the scientific secretary of the Institute during the Prague Spring, not rehabilitated before November 1989 (since then professor of Czech studies), and Jitka Štindlová, who owing to premature death did not live to see her satisfaction. At the head of the Institute Havránek was replaced by František Daneš, whom Havránek himself recommended as his successor already before (although Daneš was not a member of the Communist Party); as for Havránek, he later worked for the Institute as a consultant only. However, Daneš did not remain in the position of director for a long time since he failed to comply with the order to make his adult son — a technician — come back from the German Federal Republic. Before Daneš could return home from this commission, he found himself deprived of the directorship; he got to know it only ex post facto. With the management of the Institute Miloš Dokulil was provisionally charged, who acted in a very serious and humane way in his function, although a hard and energetic policy was expected from him. Thus a need arose to replace him with someone more acceptable for the presidium. Also in other respects the situation grew more severe. The Travaux linguistiques de Prague, recently revived, for whose further volume a miscellany specialized in sociolinguistics was prepared under Daneš’s editorship, were





Josef Vachek

liquidated again. The editorial commission of the presidium of the Academy denied the permission to publish the miscellany although this was already prepared and although it would be very advantageous: the famous Amsterdam publishing house John Benjamins wanted to participate as coeditor. The John Benjamins house had published our volume Praguiana in 1983. The sociolinguistic volume could be published only in 1987, with a changed editorial staff, in which Daneš’s name was not allowed to appear any more (as editors Jan Chloupek and Jirˇí Nekvapil were named). The Dutch publishing house was interested in the subsequent volume of Praguiana as well, yet not even that was approved by the editorial commission of the Presidium. At that time the Institute as well as the advisory board of linguistics of the Academy were directed by a confirmed enemy of PS Jan Petr, who pledged himself to ‘‘rid our linguistics of any remnants of non-Marxist conception’’. For a short period Petr was preceded in his post by Dokulil’s successor Karel Horálek, a man of great knowledge and scope of vision, but mentally unbalanced, who from a once compact collective soon made a set of people mutually distrustful and united only by their aversion to the director. When after Horálek Jan Petr was appointed director of the Institute, further changes involved abolition of some departments with good traditions, other departments were joined together and some eminent researchers were dismissed from the Institute. After the new Travaux ceased to exist, we were making an effort to revive them: since it was obvious that their main adversary, among other things, resented particularly the name Travaux in their title, Daneš and I suggested a change in the title of the series, viz Opera linguistica Bohemica. This proposal was refused again under the flimsy pretext that the Academy was not allowed to found a new series of publications — in spite of the fact that in our proposal we had explicitly emphasized that the series in question did not represent a new one, but a series already existing under another name . . . That was also why, for a number of years, the research workers of PS were entirely dependent on the series published by the Faculties of Philosophy of the Prague, Brno and, in part, Olomouc Universities — which, of course, offered very slight possibilities, not to mention the fact that the intervals between submitting the manuscripts and their publication kept increasing. The aversion to the term Travaux in the title of the series also manifested itself in the fact that the lectures on functional sentence perspective organized by the Institute in Mariánské Lázneˇ in 1970 could not be printed until four years later by the publishing house Academia (Daneš 1974), and, of course, under a title in which the term Travaux was not allowed to appear. It should be stated that in this situation publishers abroad helped us a lot (first the house Mouton, later mainly the above mentioned John Benjamins), but if the publication in question failed to appear in co-edition with the Academy participating — and such cases were more and more frequent — the books published abroad remained practically inaccessible to our readers.

Prolegomena to the history of the Prague School

Nevertheless, new books could be prepared just ‘‘for oneself ’’ or ‘‘into the drawer’’ as it used to be referred to, or, as the case might be, in the form of ‘‘samizdat’’. E.g., Sgall’s group published in this way a manuscript festschrift to honor the anniversary of Professor Hajicˇová); some works appeared owing to the help of our Viennese friends in their publication series. After the November ‘‘Velvet Revolution’’ of 1989 the Prague Linguistic Circle was resurrected, while already before November, on the premises of the Technical Museum there was a series of lectures initiated by both the linguists, above all Oldrˇich Leška, and literary scientists, pupils of Felix Vodicˇka, led by Miroslav Procházka, given to prepare that event. At the inaugural meeting on February 15, 1990 Havránek’s and Jakobson’s former disciple Miloš Dokulil was unanimously elected chairman; however, he resigned his post before long for health reasons. The present chairman comes from among the initiators of the renewal of the Circle; after Dokulil the above mentioned O. Leška was elected. The opening stages of this resurrection were not devoid of some tragic chords: in January 1991 Vladimír Skalicˇka, one of the members of the Old Guard of the Circle, passed away after a long disease; another component of this chord was the voluntary departure from the world in December 1989 of Jan Petr, a staunch adversary of PS, who failed to cope with the new situation both political and scientific; Petr was a gifted linguist — had he been able to face his failure, he might have contributed some useful work in the future. The restoration of PS, the revival of its organizational structure may have even outwardly marked the correctness of our firm conviction that the most important achievements of PS remain a lasting contribution to our own as well as to world linguistics, not merely a glorious but bygone chapter of its history; the correctness of this conviction of ours we try to prove in the following chapter of our Prolegomena. Hopefully, we needn’t refrain from a certain, however cautious, optimism concerning the future generations of our linguists, in the belief that PS will represent a lasting source of inspiration for their own work, too.12

.

The heritage of the Prague School in our domestic and in world linguistics

Although a more detailed history of PS in the last two or three decades should stay beyond the framework of the present Prolegomena (mainly because no satisfactory lapse of time has passed to separate us from the events that took place in them), it is possible, or even imperative, to raise the question what the classical period of PS, which we have been keeping track of here, has left as a lasting heritage to our contemporary linguistics and has been unquestionably further developed, and, of course, what on the contrary has become outdated. To answer this question no





Josef Vachek

great historical distance is necessary since the point is rather the confrontation of two scientific states, two historical stages, not the evaluation (be it positive or negative) of the processes and motives which have brought the situation in our linguistics from the previous to the ensuing stage. If consistently thought out, these questions can be summarized in one that can be taken as, so to say, the common denominator of all the partial questions: is it possible to speak about the existence of PS at all, today? Isn’t it rather a once and for all closed, though glorious, epoch of Czechoslovak linguistics? This question is posed not only by linguists abroad (e.g., Robert Austerlitz 1964, even Tat’jana V. Bulygina 1964) but also by our linguists (in discussions such a sceptical view on the existence of PS today has been voiced by Jaromír Beˇlicˇ and Karel Hausenblas). Such doubts are partly due to the dissolution of the organisation of the Prague Linguistic Circle as the original centre and promoter of the principles of PS, partly to the changed situation in the world linguistics as a whole. On the other hand, however, it can be rightly maintained that some ideas and currents both in linguistics and elsewhere, are kept alive by the preserved vitality of their principles in spite of the changes in the overall situation in other respects. This criterion can be safely applied also to solve our question concerning the very existence of PS at present. If we avail ourselves of the above criterion, we find that the two basic theses, i.e., the respect to the structure of the language system and that to the function of concrete utterances materialized on the basis of this system, have become a more or less self-evident prerequisite of all linguistic research nowadays. Nowhere in serious linguistics today is an inquiry into isolated, atomized language facts conducted, but more or less systematically the relationships with other components of the language system, or, at least, of the pertinent partial subsystem, are considered. Likewise we do not find any serious linguistic approach today that would disregard the basic language function, i.e., the communicative function, by proclaiming that the semantic aspect of language units should be left aside. On the contrary, at the present time even in the U. S. linguistics (which in the period of dominating descriptivism as well as in the first phase of the generative transformational grammar refused to take semantics into consideration as a matter of principle), an increasing effort is made to develop a systematic semantic analysis performed with the aid of semantic components that would be analogous to the distinctive features in phonology; a similar effort supported by the experience of modern logic can be found elsewhere, too (cf. e.g. Geoffrey N. Leech 1969). Also the principle of PS according to which language is to be examined with the aid of synchronistic as well as diachronistic methods, is slowly but surely gaining ground in the world linguistics. This principle must be paid attention, at least in a passing mention, in what follows. The emphasis put by PS on the need for synchronistic analysis, formulated by Mathesius already in 1911, never implied a denial of the appropriateness and

Prolegomena to the history of the Prague School

usefulness of diachronistic analysis, though even serious scholars could be found, who had ascribed this false thesis to PS in earnest, e.g., the Prague Germanist comparative scholar Josef Janko and, after the war, the Brno Bohemicist František Trávnícˇek. As we have pointed out elsewhere (Vachek 1964a), the real dividing line separating the functional structuralistic linguistics from the older NeoGrammarian linguistics is not drawn between the diachronistic and synchronistic approach to language, but between the linguistics which essentially atomizes and the linguistics which approaches language from the positions of structuralist standpoints. A number of Prague publications, mostly by Jakobson and Trnka, bear evidence that since 1929 it has always been so, the functionally structuralist approach being applied to questions of language history as well. This approach has never diappeared in PS, although the works of synchronistic orientation have always considerably prevailed over the works examining the problems of language history. This is understandable since the teachings of PS reflected a reaction against the orientation of traditional Neo-Grammarian linguistics, which regarded diachronistic analysis of language as the only scientifically acceptable one. It should be noted that respect to language history disappeared for a long time from the linguistics in the USA, both from Bloomfieldian descriptivism and from generative-transformational grammar; only later the transformationalists started to pay at least partial attention to it (cf. Vachek 1970a). Various other factors are connected with the process of gradually bridging the gap between diachrony and synchrony, which need to be considered here. First of all, there is the question of the motivation of language changes and language development in general. In the second half of the twenties Jakobson formulated his principle of therapeutic changes (mentioned already above) which serve to restore or save the endangered or impaired balance of the language system, the balance of the system being understood as its reliable functioning in the given language community. This principle was applied by the members of PS to concrete examples from the history of individual languages (see, e.g., Trnka’s above mentioned explanation of the causes of the late Middle English vowel shift). This was highly appreciated even by some linguists standing outside PS; thus, e.g., Alphonse Juilland (1967) regarded it as ‘‘a primary explanatory principle of language development’’. It is interesting that Jakobson himself in the time after World War II refrained from using that explanatory principle directly in his works, although he never renounced it explicitly. The point here is essentially the principle of immanent language development which with some adherents of PS became, so to say, a battle-cry in the late twenties and early thirties. It is well worth noticing that Jakobson himself was not an advocate of an absolute, radical immanentism, even at that time. He stated (Jakobson 1929: 14) that ‘‘the action of a language system is not confined to reactions to blows received from without, and to the cure of blows suffered’’, but





Josef Vachek

that ‘‘in the course of its development language reacts to its own internal problems as well’’, especially those of style — here it can be seen how important a role in PS was played particularly by the respect to the function of language utterances. In another place Jakobson admits (p. 95 f.) that the discovery of immanent laws of language development, though allowing characterization of any change of the system, at the same time fails to explain how fast the change will set in and in which direction the development is going to proceed, if there are more evolutionary possibilities. The solution follows, according to Jakobson, either from the mutual relationship between the phonological level and other language levels, or from the mutual relationship between the language system and other systems ‘‘of social and geographical order’’. All these systems, according to Jakobson, in their mutual relations constitute a system characterized by its own structural laws. Jakobson, however, adds that ‘‘heteronomous explanation of the development of a phonological system cannot replace immanent explanation, but only supplement it’’. Nevertheless, the very fact that Jakobson took heteronomous factors into consideration and was thus able to avoid radical immanentism (probably ideologically controversial anyway) is very important. It should be pointed out that the development of opinions concerning this question (as can be traced in PS) is connected with further expounding and working out the principles of this moderate immanentism, as they were in nuce given, yet not explicated, in Jakobson’s formulation quoted above. It is generally accepted today that regard to the balance of the system undoubtedly occupies an important place in the development of language, but it plays a supervising role rather than an initiating one. In other words, external factors indisputably intervene in the development of languages — such factors being facts of cultural, social as well as political and economic life (but also the laws of the mechanics of speech organs, etc.) — yet they function in this way to the extent as not to endanger the basic task of language, i.e., smooth fulfilling of the communicative function (cf. Vachek 1962). It is certainly no coincidence that we find this principle — though also in nuce only — suggested by another prominent member of PS, B. Havránek. In his contribution to the discussion at the Prague phonological conference as early as 1930 he pointed out that after all ‘‘it is inner causes that can decide the question why some foreign influences are effective while others remain without any effect’’ (cf. TCLP 4: 304). Today, of course, this thesis of Havránek’s has been refined and substantiated by a number of concrete examples, especially from the history of English (cf. Vachek 1961a). Another principle of PS dating back to pre-war times to be theoretically elaborated at the present time and concretely verified in individual languages concerns the so-called openness of the language system. This systemic openness followed indirectly from Jakobson’s original conception of teleologically conceived phonology: if the function of language changes consists above all in restoring the

Prolegomena to the history of the Prague School

balance of the system, it follows that no language system can ever be in a state of perfect balance, which is because language development is a general phenomenon (a language universal), which cannot be absent in any living language system. From this it necessarily follows that any language system is an open system and that in it, along with a relatively firm (‘‘closed’’) system core (centre), a less firm (‘‘open’’) language periphery can be found. Peripheral language phenomena are characterized either by being less firmly interconnected with the general systemic relationships dominating the given language system (if we want to use André Martinet’s (1955) term, they are less ‘‘integrated’’ in it), or by being distinguished from the elements of the system centre by their relatively very low functional load, or by their insignificant frequency of occurrence in the concrete utterances of the given language. This differentiation between the central and peripheral elements of the system has become a common feature in the post-war Czech linguistics maintaining the heritage of PS. How great an importance has been attributed to this differentiation can be seen from the fact that the central theme of the second volume of the renewed series of Travaux which appeared in 1966 was formulated as ‘‘Problems of the Centre and Periphery of the System of Language’’. The volume (TLP 2) brought a number of fundamental theoretical papers as well as concrete evidence clarifying the differentiation examined. On the other hand, on the part of Prague scholars it has been pointed out that Western linguistic currents very often fail to respect this important difference and that this lack of differentiation between the centre and periphery in their theories conduces to considerably less convincing conclusions (see, e.g., the criticism of Noam Chomsky’s arguments by Vachek 1964c). From the openness of the system of language it then follows that the character of any language system is dynamic, i.e. non-static at all times; also this property of the system of language was briefly stated in the prewar PS. So, e.g., Jakobson already 60 years ago (Jakobson 1931) wrote: ‘‘It would be a serious mistake to regard statics and synchrony as synonyms. Any static section is fictitious: it is only an auxiliary scientific procedure, not a special kind of existence. Perception of a film can be considered not only diachronically, but also synchronically: yet the synchronic aspect of the film is not identical with an isolated frame taken from this film — the perception of movement is present in the synchronic aspect as well. And so it is with language.’’ Nevertheless, this correct, though only figuratively sketched statement was later thoroughly reflected upon and again verified with the aid of concrete material from individual languages (especially for the phonological level, cf. J. Vachek 1968a and further literature listed there). In one of the quotations from Jakobson in the preceding paragraphs an interesting point must be noted. It is maintained there that if there are more possibilities of development, the solution can follow from the mutual relationship between the phonological and other language levels. This is again a very conse-





Josef Vachek

quential statement, although — as far as we were able to verify — it was never confirmed by adducing concrete evidence in the early period of PS. Here again the contemporary Czech and Slovak linguistics works out the impulses from the past times: Rudolf Krajcˇovicˇ (1957) explains some important changes in Primitive Slavonic on the basis of morphology; J. Vachek then brings examples from the history of English illustrating the motivation of phonological changes by the situation at the morphological and lexical level (Vachek 1961a); in a similar way the preservation of the Czech phoneme /rˇ /, whose functional load is relatively low, can be accounted for by the situation in the Czech system of morphology (Vachek 1963), etc., etc. It can hardly be doubted that the main principles proclaimed by PS in the late twenties and early thirties have not ceased to be valid after those sixty and more years that separate us from their original formulation, but that they essentially prove to be right even today as a referential frame for functionally structural analysis of language. Moreover, it appears that at the present time these principles can be worked out more consistently than could have been done by the generation which formulated them for the first time. From all this it should naturally follow that PS is living up to its continually applied principles even today, in the representatives of the new generation, which is still connected with the founding generation through the remembrances of older members who were its younger contemporaries. It is, of course, natural that in the course of those 60 and more years that have elapsed since the original principles of the Circle were formulated, certain modifications were bound to take place. Nevertheless, if we consider the rapid development through which the world linguistics has passed only since the end of World War II, i.e., during a much shorter period, e.g., in the USA (where the leading positions were subsequently occupied by at least two linguistic schools with diametrically different basic approaches to the examined facts), then we can be rightly surprised by how relatively few and, on the whole, insignificant modifications had to be introduced into the teachings of PS in the past four decades, to make their ideas do justice to the contemporary atmosphere in the world linguistics. Perhaps, the most important of these modifications involves the change in viewing the relation of linguistics to other sciences. While at the time of the origins of the Circle an endeavour prevailed to delimit linguistics precisely and strictly against the rest of the sciences and thus secure its autonomy, nowadays, on the contrary, in the world linguistics the emphasis is increasingly laid on what connects different disciplines with one another. The existence of a number of disciplines, sometimes referred to under the term ‘‘hyphenated linguistics’’, e.g., sociolinguistics, psycholinguistics, the study of paralinguistic phenomena, supplies a convincing proof of it. The cause of this changed attitude of the linguists was explained in a pertinent way by Jakobson himself in his Bucharest lecture on linguistics and the adjacent sciences (Jakobson 1969). He convincingly pointed out

Prolegomena to the history of the Prague School

that at the end of the first quarter of our century the objective of linguistics was to defend its right to be an independent scientific discipline. Therefore it was emphasized that it was not only its right but, what is more, its duty to examine facts that belonged to its domain of interest by its own methods and that it must not expect that its problems could be solved by some other scientific discipline. Therefore in the beginnings of PS such a great effort was made to eliminate from phonological theory the remnants of psychologism of the Baudouin type, still occasionally found in Trubetzkoy’s work, and even in early Jakobson as well. At the present time, however, the situation is completely different: the status of linguistics as an independent science has, according to Jakobson, been already won through and guaranteed beyond any doubts, and also purely linguistic procedures have been developed for the description of language facts and their scientific analysis. Thus linguistics can without fear confront its own scientific methods with those of related scientific disciplines, and, of course, on occasion, use the results of these methods to check the outcome of its own effort. The selfevident principle that linguistics must solve its own problems by means of its own methods and procedures, not by methods and procedures of other scientific disciplines naturally continues to hold. Taken all in all, a more detailed examination of the situation at the turn of the twenties and thirties will show that the procedure recognized as correct at present was not completely unknown to the then PS. It has been already mentioned here that the structuralist approach to language facts learned many a thing from the approach of Gestalt psychology, the musicologistic analysis of music forms, etc. Some other keen observations are known as well, which endeavoured to elucidate a linguistic problem with the aid of an analogous parallel taken from the domain of some other discipline — cf., e.g., Jakobson’s fitting comparison of mutual relationships between phonetics and phonology with the relationships which can be ascertained between the technology of goods and economics, or between numismatics and financial science. In all such cases the tertium comparationis is given by the contradiction between mere description and profound examination of a system of values. In the later development of PS an attempt can be observed at a more consistent and systematic engagement of linguistics and its problems in pansophist context, an engagement of such a scope that Jakobson attempted to suggest in Bucharest: there he, e.g., in a new way pointed to unsuspected analogies, which thanks to modern biology begin to stand out between the behaviour of linguistic and biological structures (Jakobson 1969: 100–3). For that matter, some time ago a British linguist and the founder of the London linguistic school, John Rupert Firth, looked on language as a means which makes it possible for a human individual as well as for a collective to survive; coping with situations in life with the aid of language he compared, by no means unfittingly, to the biological process of basal metabolism (Firth 1957).





Josef Vachek

It is clear, of course, that the changing general atmosphere in science and particularly in linguistics, above all in the last decades, could not pass unnoticed in PS and that PS had to come to grips with it. In many cases, however, it was only necessary to resume some component of its own tradition and to develop it thoroughly and consistently. Thus, e.g., facing the world’s emphasis on the necessity to analyze language phenomena from the quantitative aspect, too, PS was able to meet this trend successfully by building on the foundations laid decades before by V. Mathesius, B. Trnka and their pupil Jirˇí Krámský — how successfully the followers of PS conduct themselves can be seen from a number of volumes of the new series Prague Studies in Mathematical Linguistics (since 1966), in which the problems of quantitative linguistics occupy a very important place indeed (though not to the exclusion of other points). In the same series (as well as in the new Travaux linguistiques de Prague) also contributions inspired by another very powerful current in the world linguistics can be found, the conception of the generative transformational grammar. Even to this current and its conception the Czechoslovak linguists building on the foundations laid by the former PS had to do justice in their own way. The representative of this trend in the works of the contemporary Prague linguists is Petr Sgall, a one-time severe critic of the Prague structuralist conception, who, nevertheless, in his more recent works positively appraises many of its results and strives in a remarkable way for a synthesis of new methodological advances of the generative transformational approach with some fundamental theses of PS; thus, e.g., a very important position in his conception is taken by the fact that there exist different language levels — in this respect his theory in its own way resembles the so-called stratificational conception of grammar as it had been worked out by a Yale linguist Sydney M. Lamb (1966) — and, of course, by an important methodological principle of the so-called functional sentence perspective. The generative approach to language is combined in an original and fruitful way with the Prague tradition of dependency grammar by a prominent Czech Bohemicist František Daneš (1964), who, following the example of Mathesius (already 1942!), deals with problems of the structure of texts (Daneš 1970,1994) and of the semantic analysis of verbal constructions. Other researchers, who in the field of mathematical linguistics and in the effort to achieve its maximum formalisation presented interesting contributions, are Pavel Novák and Ján Horecký. All that has been said here, naturally, should by no means be understood in such a way that in PS the principles of the generative transformational approach are accepted without any reservations — critical comments were voiced especially in the field of generative phonology (Vachek 1964a, 1970a). The contemporary linguists who develop in a creative way the heritage of PS of the twenties and thirties do not leave unnoticed even other important trends and currents that can be found in today’s world linguistics. This holds true about, e.g., sociolinguistic research, which is pursued promisingly in both Russian and

Prolegomena to the history of the Prague School

American linguistics, and which in more than one respect begins to discover the themes that were highly topical in PS already in the period between the two world wars (thus, e.g., the problems connected with the multiform stylistic differentiation of a given language, questions of external intervention into language processes, the so-called ‘‘language planning’’, including even normalisation and codification of standard language, problems of mutual relations between external and internal factors in the development of language, etc., etc.). Outwardly this continuing interest of contemporary Prague linguists in these questions materialized in the preparation for press and later publication of a volume of sociolinguistic studies (Chloupek and Nekvapil 1987). We have been able to touch upon at least some more important zones in which the present Prague linguists in a remarkable way try to further develop a number of fruitful principles that up to the present day represent a valuable heritage of the early PS. A more detailed analysis of the work in each of these fields would, of course, require a special monograph (and obviously not a single one). However, in our opinion even from this passing view a sufficiently convincing conclusion can be drawn that even at the present time, after more than six decades from their origins, the main principles of PS remain a basis of a great bearing capacity, no less than an inspiring point of departure for further revealing inquiry both in the field of concrete language research and in the domain of general language studies. It seems to us, therefore, that the question we posed at the very beginning of this chapter, whether we can even today speak about PS, or whether it represents but one of the chapters of the history of linguistics, can be answered quite univocally: so far as the principles of this school survive and acquit themselves — which they undoubtedly do — we can, and what is more, we must speak about the continuing life of the school which formulated them for the first time. We can do so the more plausibly because what we witness is not a mere survival of a few isolated theses of PS, but a continuously developing rich life of all the essential component parts of the whole functional structuralist conception of language. Therefore, we may declare with confidence that even in further development of the world linguistics this conception is able to play a significant role in domestic as well as world context. To what extent PS will really play such a role depends on those who represent it today and will represent it tomorrow, who will become bearers of its ideas and principles.

Notes . A brief introduction to the history and theoretical issues of PS at all language levels is given also there and in T. V. Bulygina (1964) and R. Jakobson (1963). See also P. L. Garvin (1964), where, among other things, basic information on the theory of the language of





Josef Vachek

poetry is contained. . The French version of the Prague theses was published in TCLP 1 (1929), 3–25. The original Czech text can be found in PCˇSL 1 (1969), 35–65. . After the war fruitful contacts with PS were also maintained from the American side by a researcher in Indian studies Morris Swadesh, and a younger general linguist, the late Paul L. Garvin (died 15/5/1994), who propagated the principles of PS, later also through his well-known anthology (Garvin 1964). . A summary is contained in Slovo a slovesnost (Word and Verbal Arts, SaS) 3, 1937, pp. 62–4. . In Holland the work of PS was accepted with great sympathies, in general (see J. Vachek 1968b). . See Mathesius’ very lucid formulation already from the year 1911. .

Cf. N. Chomsky and M. Halle (1968); critical comment on it by J. Vachek (1970a).

. On this topic see recently J. Vachek (1994a). . See R. I. McDavid’s contribution to the dicussion concerning Vachek’s paper at the Bucharest Congress (Actes du Xe Congrès international des linguistes II., 236f.). . A comprehensive assessment of Havránek’s importance was presented in 1954 by K. Horálek; a detailed bibliography of Havránek’s writings compiled by M. Tylová appeared in the miscellany Slavica Pragensia 34, 1990 arranged on the occasion of the centenary of Havránek’s birth. . Korˇínek’s obituary was written in very warm words by J. Ru˚žicˇka in RLB 1948. . Inspirative themes for the perspectives of further development of the Prague Linguistic Circle were suggested by its chairman O. Leška in 1992 in CˇMF (Journal for Modern Philology) 74 (the works of the late professors V. Skalicˇka and B. Trnka are valued very highly there).

We have finished reading Vachek’s Prolegomena The literature on the Prague Linguistic Circle comprises today an impressive number of items; it includes monographic treatises as well as partial studies on members of the Circle or on the problems treated by them. The texts of the basic works representing the school are quite easily available in several miscellanies and in the selected works of their authors, which, too, allow the reader to form an idea of how the contributions of the individual scholars integrated into the whole. It would of course be easy to point out what remains to be done or what ought to be corrected to any degree — but these are points that would be made only by a rather well-informed inquirer. Jindrˇich Toman in his recently published The Magic of a Common Language. Jakobson, Mathesius, Trubetzkoy, and the Prague Linguistic Circle (Cambridge, Mass. — London 1995) has partly covered, and also

Prolegomena to the history of the Prague School

pointed out the hitherto most sensitive deficit, i.e. the inadequate presentation of the Prague Linguistic Circle within the context of the period in which it was established and developed its activities: the author presents a number of archival documents with attached brief introductory notes, thus clarifying a number of matters, but even more valuable are the question marks that spring up intrusively on the margins. Toman’s book brings us to Vachek’s Prolegomena. Indeed, we are aware of how little we learn from the actual people involved of how they themselves perceived the times in which they lived. It was Vilém Mathesius who probably revealed the most in his memories, partly published in Jazyk, kultura a slovesnost (Language, Culture and Verbal Arts; 1983), and in his articles, collected in Možnosti, které cˇekají (Chances that Lie Ahead; 1944). In his talks with his wife Krystyna Pomorska (Jakobson and Pomorska 1980) Jakobson, of course, writes first about himself. Still, a glance into the name index reveals a surprisingly curt treatment of Prague and his Prague colleagues. Were it not for the perfect artistic memory of Milada Soucˇková (The Prague Linguistic Circle: A Collage in L. Mateˇjka´s collection Sound, Sign and Meaning. Quinquagenary of the Prague Linguistic Circle, 1976:1–5), we would never be able to imagine the atmosphere of the Circle’s debates at the café Louvre. In the Prolegomena, Professor Vachek was not writing reminiscences of the Circle and he certainly did not attempt to draw genre, let alone portrait studies — incidentally, that was not his greatest strength. The text is nonetheless valuable because Circle matters are being described by somebody who from the very start had linked his scholarly life to it, and had lived in the Circle also at times when the debating evenings were but a shadow of their former selves — which however did not divest them of their realness. In the introductory note the author apologises, quite unnecessarily, for any subjectivism in the text; in my view, the book is specially charming due to the audible note of the Professor’s intonation, permeating the written text of the occasional dialogues in which he speaks of the issues with a personal keenness and characteristic distribution of objective accents. Vachek presents a relatively uniform image of the Circle; personal participation covers the ruggedness of the relief caused by the flow of time and by the diversity of the individual dispositions and tempers of the protagonists: he is moving around in his own world, in the friendly and industrious atmosphere among his teachers and colleagues, accompanied by the echoes of the achievements of the pioneering days when he had been present still as a student. It is only natural that Vachek’s peers do not reach the titanic dimensions of the founding fathers, although he does realize quite correctly that it had been especially the first generation of pupils thanks to whom the Circle survived, as a living organism, the worst times until its restoration in the beginning of 1990. And so, by coincidence, linguistic stars of the first order, like Vladimír Skalicˇka and L’udovít Novák, are given but fleeting notice, although it had been precisely these ‘‘cheeky lads’’ — as





Josef Vachek

they were reportedly called by the Circle (in many ways unlike the Circle’s founders and yet also their faithful spiritual sons) — who were the best proof of the source of the force that, again and again, revived the Circle. The core of the book is composed of five profiles: of V. Mathesius, N. S. Trubetzkoy, R. Jakobson, B. Havránek and B. Trnka. The author describes the issues of the day as could be experienced only by somebody deeply immersed in them, somebody who felt their pressing urgency as co-creator — that is without distancing himself by resorting to the medium of the book. Hence perhaps a certain apodictic aspect of the formulations: the questions are the same that Vachek posed to himself and wrestled with. He follows the development of Prague phonology along the road marked out by Jakobson when the latter was still in Prague (the decomposition of the phoneme into acoustic distinctive features) but — like Jakobson — he never exceeds the limits of the school. It is only natural that Vachek’s description reflects the times when the author selflessly promoted the Circle — and also courageously defended it; in his defensive he stressed the advantages of the ways in which the Prague position differed from other streams in modern linguistics. Only Vachek could make that generous gesture of distributing among our linguists hundreds of complimentary copies of his books acquainting foreign readers with the Circle. In his Prolegomena, Vachek restricted his topic to the linguistic component of the Circle; this may be a reasonable constraint as long as we do not lose sight of the fact that the Circle managed to incorporate linguistic issues into a broader semiotic framework and excluded a mechanistic understanding of this connection. If the Circle was at one time ideologically unacceptable, it was in particular because of this programmatic feature. Vachek’s book of reminiscences Vzpomínky cˇeského anglisty (Memoirs of a Czech Anglicist; H&H, Prague 1994) — containing an apposite portrait of Vachek by Libuše Dušková — has thus been joined by the non-commemorative, yet deeply personal Prolegomena. This is the last manuscript which Professor Vachek, with his typical meticulousness, prepared for print. He died on March 31, 1996; the Prague Linguistic Circle paid tribute to its Honorary Chairman at a commemorative meeting held in the beginning of May. Oldrˇich Leška

Appendix I

Articles of the Association Prague Linguistic Circle Pražský lingvistický kroužek Cercle linguistique de Prague Prager sprachwissenschaftliche Vereinigung Article 1 The mission of the association Prague Linguistic Circle is to apply the functionally structural method for the advancement of linguistic research. Article 2 The means of achieving this mission are: i. regular members’ meetings with lectures and debates; ii. administrative members’ meetings; iii. public meetings with the aim of spreading interest in general linguistic issues; iv. establishment of commissions for collective processing of scientific issues; v. publication of scholarly works; vi. establishment and maintenance of a specialised library and archive. Article 3 The resources needed to implement these goals shall be acquired from: i. annual membership dues paid both by regular and corresponding members; ii. the yield from the sale of the publications of the Circle; iii. subsidies from relevant scientific associations and foundations; iv. gifts and bequests. Article 4 The Prague Linguistic Circle consists of: i. regular members; ii. corresponding members.



Josef Vachek

Article 5 The first regular members are the founders of the Prague Linguistic Circle, whose names are listed below. Any scholar who, as a guest invited by the Committee, gives a lecture and has been accepted at the recommendation of the Committee as a regular member by a twothirds majority of the members present at an administrative meeting whose agenda explicitly includes the proposal at issue, shall also become a regular member. Any scholar whose works are in compliance with the goals of the Circle, yet who cannot for serious reasons attend Circle meetings, may be elected corresponding member at the recommendation of the Committee, by a two-thirds majority of the members present at an administrative meeting. Article 6 Regular members have the right to: i. attend lectures, debates and administrative meetings; ii. receive one free copy of the publications of the Circle; iii. elect the Committee (chairman, deputy-chairman, bursar and secretary), two auditors and members of the commission for the collective treatment of scientific issues; iv. introduce guests to lecture meetings of the Circle and recommend scholars to the Committee for regular or corresponding membership. Article 7 Regular members have the duty to: i. attend lectures, debates and administrative meetings, and participate in all the scholarly activities of the Circle; ii. pay membership dues set by the administrative members’ meeting for the current year. Article 8 The rights and duties of the corresponding members shall be the same as those of the regular members, with the exception of the duty to attend meetings other than meetings of the Committee and of the right to elect officers of the Circle or to be elected officers of the Circle. Article 9 A member shall cease to be a member of the Circle if: i. he informs the Committee in writing of his intention to leave the Circle; ii. for more than one year and without due excuse he does not meet any of the obligations of membership; iii. he is expelled from the Circle. Any member of the Circle shall have the right to propose to the Committee the expulsion of a member whose conduct contravenes the mission of the Circle. Such a proposal shall be voted on by an administrative members’ meeting under the same conditions as those applying to the acceptance of new members.

Prolegomena to the history of the Prague School

Article 10 The seat of the Circle is Prague. Article 11 The administration of the Prague Linguistic Circle shall be governed by a four-member Committee, also called the ‘‘Presidium’’, consisting of a chairman, deputy-chairman, secretary and bursar. The Committee shall meet at least once a month and shall constitute a quorum if three members are present. It shall decide by a simple majority of votes; in the event of an equal number of votes, the proposal supported by the chairman shall be adopted. The Committee may invite to its meetings officers and chairmen of research commissions for the collective treatment of scientific issues to report on the results of the work of the commissions. Such invited members of the Circle shall attend in an advisory capacity. Article 12 The election of the Committee shall take place at a general meeting convened every third year in January or, if necessary, at an extraordinary general meeting convened at any time on the basis of a decision by the Presidium or at the request of at least a two-thirds majority of regular members of the Circle. Besides the Committee, the general meeting elects two auditors, also for a term of three years. (Art. 14, 16 & 17.) Members are notified of the convention of the general meeting by an ordinary or open letter not later than ten days in advance. The general meeting decides (with the exception of the provisions of Art. 16 & 17) by a simple majority of members present if the meeting is attended by at least one-half of all regular members. Should the meeting not constitute a quorum, a new general meeting is convened half an hour later, which shall decide irrespective of the number of members present. Article 13 Administrative members’ meetings are convened at least twice a year. They discuss the internal matters of the circle. (Art. 5, Art. 6 para 3, Art 7 para 2, Art. 9.) At members’ meetings, with the exception of acceptance and expulsion of members, as well as at consultation meetings of research commissions, decisions about matters assigned to these Circle bodies shall be adopted by a simple majority of members present, irrespective of their number. In the event of an equal number of votes the decisive vote shall be that of the chairman of the meeting. Article 14 Disputes that may arise within the Circle shall be decided by a majority of votes of a threemember arbitration panel elected by the general meeting. Article 15 The Circle is represented externally by its chairman and in the absence of the latter by the deputy-chairman or secretary. All written documents of the Circle are signed by the





Josef Vachek

chairman (or deputy-chairman or secretary), financial matters also by the bursar. Members shall be informed of Circle announcements in writing. Article 16 The Circle shall cease to exist if, at the recommendation of the Committee, two-thirds of the members present at an extraordinary general meeting convened for such a purpose decide so. The assets of the Circle shall in that case fall to the public and university libraries in Prague. Article 17 Any alterations of the Articles must be adopted by two-thirds of the votes of the members present at a general meeting. B. Trnka, V. Mathesius, P. Bogatyrev, O. Hujer, J. Mukarˇovský, J. Rypka, V. Buben, R. Jakobson, F. Oberpfalcer, Fr. Slotty, O. Fischer, L. Kopeckij, E. Rippl, M. Weingart, B. Havránek, V. Machek, G. Ružicˇic´ *

*

*

Notes These Articles were approved by the relevant authorities in an official letter on 23rd October, 1930, ref. 565919- ai 1930/19A3200/2 ai 1930. Two names must be added to those of the fifteen founding members who signed the Draft Articles, the Finno-Ugric scholar Pavel Bujnák and the Slavonic scholar Josef Kurz, who at the time of the signing of the Articles were not present in Prague. The administrative members’ meeting of the Prague Linguistic Circle held on 16th January, 1931, accepted these two scholars, too, as founding members. Besides the institution of members, there was also the institution of guests of the Circle (there had been more than fifty by the time of the tenth jubilee of the Circle); for a small fee, a guest would be invited to all lecture meetings but would not enjoy membership rights. The founding members’ meeting took place on 1st December, 1930, and elected the first Committee of the Circle (it in fact merely approved the officers of the hitherto existing free group). The first chairman, as we all know, was from the beginning until his death Vilém Mathesius, the deputy-chairman Roman Jakobson, the secretary Bohumil Trnka; the bursary was first entrusted to Jan Rypka who, however, had to resign this office due to a large workload and the office was then taken over by Jan Mukarˇovský. The composition of the Presidium of the Circle then remained unchanged until the end of the war when Bohuslav Havránek became chairman after the deceased Mathesius. The deputy-chairman post remained unoccupied after the departure of Jakobson from Czechoslovakia (this was obviously a protest gesture against his forced departure); in case of need, the secretary, B. Trnka, deputised for the chairman. Trnka, too, was secretary from the beginning until the disintegration of the Circle.

Prolegomena to the history of the Prague School

The chairmen of the working commissions used to be invited regularly to Committee meetings: Bohuslav Havránek on behalf of the phonological section, Jan Mukarˇovský for the section for research into poetic language, Friedrich Slotty for the bibliographical section, and Miloš Weingart for the section for research into colloquial Czech. The report on the activities of the Circle in 1936 lists a further section — orthoepic, convener František Trávnícˇek (later a harsh critic of the Prague School). It ensues from the report that the original phonological section and the section for research into colloquial Czech merged into the commission for the description of the languages of Czechoslovakia — the convener was B. Havránek. During the whole life of the Circle four members resigned membership (J. V. Sedlák, M. Weingart. E. Rippl and F. Liewehr), nobody was expelled. The lecture meetings took place first in Mathesius’ office on the second floor in Veleslavínova street, where the English seminar was located until 1929; when the new building was opened the meetings were moved to the first floor, to Mathesius’ office (no. 126), and later to the English seminar classroom (no. 128), where the famous phonological conference of 1930 took place as well as the series of lectures on Standard Czech and language culture in 1932. When the Faculty started demanding payment for rent of the rooms, the lecture meetings started being held at Prague cafés (usually at the café Louvre, later the premises of the Prague Language School, and after that at the Zlatá husa café on Wenceslas Square). Towards the end of the war, the Circle found refuge at the Municipal Library on Mariánské Square; after the end of the war the meetings started being held once again at the Faculty of Arts, on what is today Jan Palach Square. Committee meetings used to be held at the café SIA (today’s Academic café on The Curies’ Square). The book series of the Circle, which the Circle used to publish itself before the war, i.e. TCLP and Studie Pražského lingvistického kroužku (Papers of the Prague Linguistic Circle), were sold through the Union of Czechoslovak Mathematicians and Physicists (in Czechoslovakia) and Otto Harrassowitz (abroad). Slovo a slovesnost, the journal of the Prague Linguistic Circle, was published by Melantrich. It survived the disintegration of the Circle and continues to be published as the journal of the Institute of Czech Language by the Academia Publishing House. The same Publishing House also published all four volumes of TLP. A short final note on the German name of the Circle. The name as proposed by F. Slotty is that printed in the sub-heading of the Articles. However, it has never become widely accepted, and the name ‘‘Prager linguistischer Zirkel’’ is more common.



Appendix II Lectures delivered at the Prague Linguistic Circle The list has been compiled on the basis of records in the ‘‘Report on the activity of the Circle in the first ten years’’ (until June 1936) and on the basis of reports and excerpts published regularly in SaS (until the end of 1948). Unfortunately, no reports on Circle activities, or attendance sheets (with the exception of the very last one) have been preserved from the 1949 to 1952 period. That is why those years are documented in our list only very sparsely. A very valuable source is L. Mateˇjka’s list of lectures (Mateˇjka 1976: 607–22). The report on the decade of the Circle makes a precise distinction between lectures presented in Czech or Slovak (here given in English translation) and in other languages (here given as such), while outlines and excerpts in Slovo a slovesnost are usually presented only in the Czech version — also in the case of lectures given originally in a foreign language. The exact wording can only be guessed. Nevertheless, in spite of all these shortcomings the list presented below is not without interest and documents very convincingly the broad range of themes treated by the Circle. Unfortunately, it has not been possible to find out the exact date of the last meeting of the Circle (it was probably in the autumn of 1952, the year when the Czechoslovak Academy of Sciences started its activities). 1926 Oct. 6 Nov. 3 Dec. 2

H. Becker B. Havránek B. Ilek

Der europäische Sprachgeist The grammatical category of passive voice Language culture in some new Russian publications

1927 Jan. 13 Feb. 3 Mar. 3 Apr. 7 May 12 June 2

R. Jakobson B. Trnka F. Oberpfalcer J. Mukarˇovský S. Karcevskij V. Mathesius

Oct. 2 Nov. 2 Dec. 2

J. Rypka L. Tesnière F. Oberpfalcer

The concept of the sound law and the teleological criterion Semasiology and its significance for linguistics Bally’s Le langage et la vie On motoric processes in poetry Sur les rapports entre le langage et la pensée On linguistic characterology with illustration from modern English On the interpenetration of languages in the Islamic world Duel et géographie linguistique Psychoanalysis and linguistics

Prolegomena to the history of the Prague School

1928 Jan. 13 V. Buben Feb. 3 J. Mukarˇovský Feb. 6 N. Trubetzkoy Feb. 7 B. Tomaševskij Feb. 14 R. Jakobson (reporting) Mar. 9 L. Brun Mar. 30 J. Mukarˇovský May 11 P. Bogatyrev June 1

R. Jakobson

Oct. 5 Nov. 9 Dec. 14 Dec. 16

B. Trnka G. Vinokur V. Mathesius J. Tynjanov

Dec. 18 N. Trubetzkoy 1929 Jan. 11 Feb. 8 Mar. 1 & Apr. 9 May 7 May 10 May 31 June 28 Nov. 8 Dec. 16

French argot and the colloquial speech of Paris Sound repetition in Mácha’s Máj and its relation to other elements of the poem Alphabet und Lautsystem La nouvelle école d’histoire littéraire en Russie. A draft of the phonological theses for The Hague Congress Rémarques faites en traduisant du tchèque en français Three solutions to the subjectivity of epic poetry in Máj, Alfred and Satanela K voprosu ob ètnologicˇeskoj geografii [Toward the question of ethnological geography] New works on the role of Church Slavonic elements in Russian On the essence of verbal aspect Linguistik und Philologie On the phonological system of modern English Problema literaturnoj èvoljucii [The problem of literary evolution] Vergleich der Vokalsysteme

O. Fischer On Brˇezina’s rhymes J. V. Sedlák Literary history and literary science Kopeckij, Ober- Discussion of an international artificial language pfalcer, Trnka, and Jakobson (reporting) The Congress of Slavonic philologists: discussion of the program B. Trnka Synchronic and diachronic linguistics P. Savickij Les problèmes de la géographie linguistique du point de vue du géographe The Congress of Slavonic philologists: discussion of the Theses O. Fischer On Kollár’s sonnets V. Mathesius Notes on Czech word order

1930 Jan. 27 D. Cˇyževs´kyj Feb. 20 V. Vážný Mar. 11 F. Slotty

Phonologie und Psychologie Slovak linguistic atlas Die syntaktische Funktion der Wortart





Josef Vachek

Mar. 25 R. Jakobson J. Mukarˇovský (Formal public session) Apr. 7 V. Machek June 22 G. Becking Sept. 15 R. Jakobson Sept. 23 Oct. 16 Nov. 3 Nov. 17 Nov. 20 Nov. 24 Dec. 1

N. Trubetzkoy O. Fischer R. Jakobson M. Seeman M. Weingart L. Kopeckij V. Mathesius

1931 Feb. 13 Feb. 17 Mar. 5 Mar. 23

J. Mukarˇovský J. Vachek J. Mukarˇovský L’. Novák

Mar. 26 V. Jirát Apr. 10 May 4 May 15 May 22 June 5 June 18 Oct. 29 Dec. 14

O. Fischer V. Simovycˇ G. Ružicˇic´ J. Frcˇek V. Jirát J. Vachek L. Kopeckij

1932 Jan. 4 O. Fischer Jan. 12 V. Mathesius Jan. 25 B. Havránek R. Jakobson Feb. 8 M. Weingart J. Mukarˇovský

Linguistic questions in the work of Masaryk Masaryk as a stylist

On the coining of expressive terms Schallanalyse Toward a phonological characterization of the Eurasian league Phonologie und Dialektologie Erben’s Kytice On polytony On the significance of phoniatry Leoš Janácˇek as a phonetician On the use of phonology in teaching On the problem of capacity and combinatory potential of phonological units

The semantics of Brˇezina’s poetic language (I) On the phonological nature of diphthongs The semantics of Brˇezina’s poetic language (II) The history of Standard Slovak from a phonological point of view and the problem of the reform of Czechoslovak orthography Mácha’s rhyme Discussion of the language of instruction in Sub-Carpathian Ukraine Konrad Wallenrod Adjective inflection in Ukrainian Measuring the melody in Czech words and sentences The vertical series of consonantal disturbances Rhyme in the work of Viktor Dyk Notes on combinatory structural analysis in phonology Recent linguistic discussion in the USSR

Echo of Russian Songs On the requirement of stability in the standard language On the tasks of the standard language and its culture Concerning the problem of purism On the cultivation of the sound aspect of language Poetry and the standard language

The papers read on Jan. 12 and 25 and Feb. 8 were delivered in public sessions as a cycle entitled Standard Czech and language culture.

Prolegomena to the history of the Prague School

Feb. 24 Feb. 25 Mar. 14 Mar. 15 Apr. 2 May 6

A. Artymovycˇ L’. Novák M. Rostohar N. Trubetzkoy V. Mathesius B. Trnka R. Jakobson Nov. 30 G. Becking Dec. 6 A. Bém 1933 Jan. 9 D. Cˇyževs´kyj Jan. 17 O. Zich Feb. 1 S. Pus¸cariu Feb. 13 Mar. 13 Apr. 4 May 24 May 27 Sept. 29 Nov. 27 Dec. 4 Dec. 18

J. Mukarˇovský K. Kožešník J. M. Korˇínek V. Jirát R. Jakobson O. Jespersen B. Bareš H. Siebenschein A. Artymovycˇ

1934 Jan. 8 Jan. 15 Jan. 29 Feb. 21 Mar. 12 Mar. 19 Apr. 23 Apr. 28 May 7 May 14 May 28 June 4 June 25 Oct. 5

V. Skalicˇka B. Trnka A. Bém V. Vážný R. Wellek F. Slotty D. Cˇyževs´kyj J. L. Fischer E. Otto A. Artymovycˇ B. Havránek F. Ružicˇic´ P. Bogatyrev

Oct 6

N. Trubetzkoy S. Karcevskij Oct. 15 G. Seidel Nov. 5 A. Isacˇenko

The written language On the historical phonology of Romance vocalism The psychological foundations of phonetics Dostojevskijs humoristische Erzählungen Loan words from a synchronic point of view On structural morphology Morphology of the Russian verb Musikwissenschaft und Phonologie Proper names in Dostoevskij

Zur Geschichte der russischen Philosophie The tempo of poetic language Phonologisches und Phonetisches auf dem Gebiete der rumänischen Sprache Milota Zdirad Polák Formal analysis of morals in La Fontaine’s fables Notes on the phonic realization of interjectional structures Aural and visual spheres in the images of Platen’s lyrics On Slavonic clitics Über Verners Gesetz German sound shift and the substratum theory Problems of translation (concerning Fischer’s Vilém Busch) Phonologie und Schrift

Towards a typology of external form On the phonological system of English On the question of literary influence Purism in Slovakia Šklovskij’s theory of prose Zur Theorie des Nebensatzes Eine neue Sprachphilosophie System of meanings Hauptfragen der Sprachwissenschaft Die Potentialität der Sprache Some problems of Czech dialectology On syllabic [r] Problems of structural ethnography Reports on congresses: Isacˇenko on the International Anthropological Congress in London; Mukarˇovský and Havránek on the International Slavonic Congress in Warsaw. Struktur des altkirchenslavischen Alphabets Réforme de l’orthographe russe Zum Problem der Sprache H. Heines Phonologie der slovenischen Mundarten





Josef Vachek

Nov. 19 B. Havránek Dec. 10 A. Bém & R. Wellek

Preliminaries to a structural history of Standard Czech Discussion of the methodological problems encountered in J. Mukarˇovský’s work on M. Z. Polák

1935 Jan. 4 Jan. 14 Feb. 4 Feb. 25 Mar. 18 Mar. 25 Mar. 30 Apr. 8

Interrogative intonation Musikwerk als Zeichen Concerning questions of phonological oppositions From the history of Middle Slovak Die Barockdichtung Tasks of linguistic pedagogy The science of verbal arts and its relation to adjacent sciences Víteˇzslav Hálek

Apr. 15 Apr. 29 May 6 May 13 May 20 June 3

June 4

S. Petrˇík G. Becking V. Skalicˇka L’. Novák D. Cˇyževs´kyj S. Hessen F. Wollman J.Mukarˇovský (public session) V. Brøndal R. Jakobson E. Utitz F. Trávnícˇek R. Carnap R. Jakobson, J. Mukarˇovský (reporting) J. Arnaudiès

P. Trost Oct. 21 J. Rypka Nov. 2 B. Trnka & J. Mukarˇovský B. Havránek R. Jakobson Nov. 4 R. Jakobson B. Trnka Nov. 18 E. Husserl Dec. 2 L. Silberstein Dec. 16 O. Kraus Dec. 17 B. Havránek & R. Jakobson

1936 Jan. 13 J. Cˇada Z. Vancˇura Jan. 27 P. Eisner & A. S. Magr (reporting) Feb. 3 A. Bém

Etymologie et problème des lois phonétiques Poetry of the Hussite period Sprache als Kultur Questions of normative pronunciation of Czech Über die logische Syntax Methodological discussion of Grund’s book Karel Jaromír Erben. Quelques remarques sur l’expression des relations de causalité en français Probleme des Sprachtabu A scholar’s impressions of Iran The Second International Congress of Phonetic Sciences Research trip to the USSR Research trip to Bulgaria and Rumania To the memory of A. Artymovicˇ General laws of clustering of phonemes Phänomenologie der Sprache Die tschechische philosophische Terminologie Reistische Sprachbetrachtung in ihrem Verhältnisse zur Logistik und Phänomenologie The critical and scholarly methods of Professor Weingart

Problems in the language of economists The language of economists Retrospect of the first year of Slovo a slovesnost The method of detailed observation in literary science

Prolegomena to the history of the Prague School

Feb. 22 Feb. 24 Feb. 29 Mar. 9 Mar. 16

B. Václavek V. Chaloupecký A. Mráz J. M. Korˇínek J. Arnaudiès

Mar. 23 Mar. 30 Apr. 6 Apr. 7 Apr. 27 May 11 May 18 May 25 June 4 June 22 June 23

B. Trnka S. Lyer P. Trost V. Brøndal C. Brailoiu A. Novák A. Frinta L. Landgrebe O. Fischer J. Hrabák Vl. Helfert E. Otto

Oct. 19 J. M. Korˇínek R. Jakobson & B. Trnka Nov. 3 V. Mathesius

Czech folklike songs Church Slavonic tradition in Bohemia (I.) L. Tolstoj and Slovaks Some remarks on the concepts of ‘‘langue’’ and ‘‘parole’’ Les idées de David García sur la proposition et le symbolisme logique The application of phonological principles in stenography Pause and its function in the Romance languages Structural analysis of Céline’s novel Voyage au bout de la nuit La structure des systèmes vocaliques Sur la doîna roumaine Style of the Czech literary criticism between 1870 and 1890 Jan Blahoslav: the oldest linguistic theory of Czech Feldbegriffe in Sprachwissenschaft und Sprachphilosophie The symbol of mothers in Goethe’s Faust A comparison of Old Czech and Old Polish verse On the question of musicality in poetry Grammatik und Stilistik vom Standpunkt der Wissenschaft und des Bildungsprozesses In memory of A. Meillet The International Congress of Linguists in Copenhagen Decennial Celebration of the Prague Linguistic Circle Ten years of the Prague Linguistic Circle (Introductory lecture)

Salutatory addresses : Professor Dr. A. Salacˇ (Dean of Charles University) Professor Dr. M. Murko (Slavonic Institute) Professor Dr. J. Rypka (Oriental Institute) Professor Dr. J. Cˇada (College of Business Trade) Professor Dr. E. Utitz (Prague Philosophical Circle) Lecturer Dr. K. Krejcˇí (Literary historical society) Dr. L’. Novák (The Matice slovenská society) gen. K. Herˇman (on behalf of the guests) Answer to the salutatory addresses : V. Mathesius, B. Havránek, R. Jakobson Salutatory communications : Professor Dr. J. Schrijnen (Nijmegen) (Permanent International Linguistic Committee) Professor Dr. P. Kretschmer (Vienna) (Indogermanische Gesellschaft) Professor Dr.V. Brøndal (Copenhagen) (Copenhagen Linguistic Circle) Professor Dr. E. Husserl (Freiburg) Professor Dr. L. Tesnière (Strassburg) Professor Dr. A. Sommerfelt (Oslo) Professor Dr. V. Simovycˇ (Lvov)





Josef Vachek

Professor Dr. S. Pus¸cariu (Cluj) V. Vancˇura (Union of Czech writers) L. Novomeský (Union of Slovak writers) and others Nov. 16 F. X. Šalda Nov. 23 E. Rippl Dec. 7 R. Jakobson & F. Wollman (reporting) Dec. 14 N. Trubetzkoy 1937 Jan. 18 L. Silberstein Feb. 3 Feb. 8 Mar. 8 Mar. 22 Mar. 24 Apr. 26 May 10 May 31 Oct. 1 Oct. 10 Oct. 25 Oct. 26 Nov. 8 Dec. 6

J. Horák R. Jakobson É. Benveniste F. Trávnícˇek V. Brøndal J. Mukarˇovský K. H. Menges P. Eisner R. Wellek E. Utitz L. Hjelmslev V. Skalicˇka F. Lexa F. Slotty

1938 Jan. 1 Feb. 14 Feb. 28 Mar. 21 Apr. 4 Apr. 5

A. Hartl A. Florovskij A. W. de Groot R. Jakobson J. M. Korˇínek J. Ružicˇka

Apr. 23 Apr. 25 May 30 May 31 June 27

K. Sandfeld N. van Wijk F. Trávnícˇek O. Broch B. Trnka

Nov. 7

J. Mukarˇovský

Mácha’s prose Humor in popular Czech Discussion of the book Deutsche und Tschechen by K. Bittner

Gedanken über das Indogermanenproblem

Historical terminology, the sociology of knowledge, and semantics Puškin and Mickiewicz On Puškin’s symbolics L’expression linguistique de la quantité The bases of Czechoslovak word order Le concept et les systèmes de cas en grammaire In memory of F. X. Šalda The languages of Turkestan The birth of poetic translation The development of English literary scholarship Towards a linguistic aesthetics Forme et substance de la langue Notes on the case system On the development of Egyptian Methods of deciphering Etruscan

The linguistic situation in Sub-Carpathian Ukraine Czech features in Old Russian literature Das Wesen der Wortarten Acoustic classification of consonants On the relation between sounds and word semantics Notes on the declension of the adjective in Indo-European with particular reference to its development in the Germanic and Balto-Slavonic languages Problems of linguistic interference Problèmes phonologiques et extra-phonologiques Genitive of negation in Czech Bedeutungsunterschied durch Intonationsunterschied Notes on the theory of combinatory variants and neutralization Concerning the semantic structuring of Mácha’s work

Prolegomena to the history of the Prague School

Nov. 23 F. Trávnícˇek Dec. 19 G. Porru 1939 Jan. 16 J. Vachek Feb. 27 J. Ružicˇka Mar. 6

G. Devoto

Mar. 13 M. Pavlovic´ May 22 June 12 June 19 Oct. 23 Nov. 13

J. Mukarˇovský J. Pru˚š ek P. Bogatyrev J. Mukarˇovský M. Gessman

Dec. 18 B. Trnka 1940 Jan. 15 J. Mukarˇovský Mar. 4 J. Vašica Mar. 18 Apr. 1 Apr. 22 May 27 June 24 July 1 Sep. 16 Oct. 7 Oct. 21 Nov. 19 Dec. 2

V. Skalicˇka J. Vachek J. Korˇínek J. Heidenreich J. M. Korˇínek J. Honzl J. Mukarˇovský J. Kurz V. Skalicˇka M. Seeman S. Lyer

Dec. 4 B. Havránek Dec. 16 A. V. Florovskij

1941 Jan. 13 Feb. 3 Feb. 17 Feb. 24 Mar. 3 Mar. 24

F. Žilka J. Mukarˇovský B. Trnka Cˇ. Loukotka F. Vodicˇka J. Mukarˇovský

Passive voice in Czech Italienische Phonologie

Remarks on some problems of the written language Phonological analysis of the primitive Indo-European gutturals Le système des occlusives indoeuropéennes à la lumière de la géographie linguistique et de la phonologie On the question of the affricates [cˇ], [dž], in Macedonian dialects Semantic structure and compositional basis of Cˇapek’s prose On the semantic structure of Chinese narratives Czech and Slovak folk theatre On Vancˇura’s semantic structure The phonological reconstruction of the Indo-European consonant system Some new publications on phonology

Problems in the aesthetics of language Liturgical character of the Kiev Leaflets: an attempt at a new solution Problems in morphology Morphological problems in Czech Monophonematic words On tradition and traditionalism in the works of Arne Novák On the substitution of languages in different styles The stage as a complex of signs Dialogue and monologue Chapters on Old Church Slavonic demonstrative pronouns Development of Czech declension Special means of voice formation The controversy about Catalanian in the light of comparative phonology Basic problems of style and stylistics The Czech Bible in the history of East Slavonic culture and literature

On methods of translating Scriptures Between literature and visual art The individual and the social in style On South American languages Methodological notes on the reception of a literary work On the poet





Josef Vachek

Mar. 31 Apr. 7 Apr. 21 Apr. 28 May 5 May 12 May 26

L. Rieger V. Buben Z. Kalista K. Horálek J. Veltruský J. Frey B. Trnka

June 19 B. Trnka

1942 Feb. 9 I. Poldauf Feb. 23 K. Horálek Mar. 9 B. Trnka Mar. 30 Apr. 13 Apr. 27 May 11

A. Frinta L. Rieger J. Mukarˇovský P. Poucha

B. Trnka & B. Havránek Sep. 21 J. Nováková Oct. 5 B. Havránek

Semantic analysis of philosophical texts The Slavonic influence on the Rumanian verb Czech legends in the Baroque era The place of Bulgarian in the area of the Balkan Sprachbund Dramatic text and the theater Verbal training of pre-school children Pronunciation of foreign words, especially proper names, in Czech On the analysis of children’s language from a phonological point of view (reporting on R. Jakobson’s Kindersprache, Aphasie und allgemeine Lautgesetze, written in exile)

Morphology of verbal aspect in Czech Analysis of Old Czech verse and the nature of stress Development and archaisms in the spoken and written language Foreign influences on language On the relation between phenomenology and structuralism Proverbs as a constituent of a text Phonology of Tocharian and Old Indian

Sep. 7

On the scholarly work of Professor Vilém Mathesius

Nov. 2 V. Skalicˇka Nov. 30 J. Mukarˇovský

Quantity in the Czech tonic verse Theses of the Prague Linguistic Circle in relation to the Shorter Dictionary of the Czech Language and to the Rules of Czech Orthography Ends and means of morphological typology The place of the aesthetic function among the other functions

1943 Jan. 11 J. Heidenreich Jan. 15 J. Mukarˇovský Feb. 17 V. Polák Mar. 17 May 5 May 26 June 7 June 23 Oct. 25 Nov. 22

B. Trnka J. Nováková J. Mukarˇovský V. Skalicˇka B. Trnka J. Nováková J. Slavík & J. Vachek Dec. 22 K. Horálek

Thematic structure in a work of poetry The concept of the development in art The problem of Indo-European protolanguage from the point of view of language interference Frequency and number in phonology Consonantal clusters in Czech verse Intentionality and unintentionality in art Typology of Bantu languages System and tense in linguistics Czech verse under the influence of quantity based meter The acoustics of the auditorium in theatrical praxis Greek model of the Old Church Slavonic translation of the Gospels

Prolegomena to the history of the Prague School

1944 Jan. 26 Feb. 21 Mar. 13 May 31 July 7

J. Honzl F. Vodicˇka V. Skalicˇka J. Beˇlicˇ A. Jedlicˇka

1945 Jan. 9 Feb. 12 Mar. 26 Apr. 30

J. Polišenský V. Skalicˇka J. Nováková J. Mukarˇovský & B. Trnka June 27 J. Mukarˇovský Oct. 9 J. Beˇlehrádek Oct. 15 J. Beˇlehrádek Dec. 10 J. Mukarˇovský 1946 Jan. 1 J. Nováková Jan. 28 A. Sychra Mar. 4 F. Vodicˇka Mar. 26 S. Urban´czyk

Apr. 7 June 3 June 11 Nov. 4 Nov. 25

A. V. Isacˇenko E. Pauliny J. Nováková K. Zeman V. Skalicˇka

Sign and signal in facial expression Jungman’s role in the development of Czech prose On the grammatical structure of Chinese Extra-linguistic factors in the development of dialects Jungman’s role in the creation of Czech terminology of literary scholarship and linguistics

Linguistic data and the work of historians Linguistic diversity Czech quantity based hexameter Professor Vilém Mathesius died The sentence in Vancˇura’s prose Structure and holism Holism and linguistics Structuralism

Czech syllabo-tonic hexameter Music and the word in the folk song The era of Jungmann in the development and periodization of Czech literature Geneza spółgłosek [s], [z], [c´], [n´] w polskich spieszczeniach [Genesis of the consonants [s], [z], [c´], [n´] in Polish terms of endearment] On the conative function of language Tense and Aktionsart in the Slovak verb Indian meters in the work of J. Jungmann and his school Distribution of Czech consonants Analogy and anomaly

1947 Feb. 17. L. Cejp Structural features of business correspondence Mar. 3 J. Honzl On the philosophy of sign in the theater Mar. 10 J. Vachek English vowel system according to American linguists Mar. 31 V. Polák Notes on the West European Sprachbund Apr. 18 S. C. Ross The Phoneme Apr. 21 S. C. Ross Philologica mathematica Session held conjointly with Charles University Institute of Mathematics May 5 May 19 June 23 Oct. 6

L.Cejp L. Hjelmslev E. J. Simmons A. Sychra

Art and journalism in the work of H. G. Wells Langues de différents degrés The study of modern languages in the United States The material of music from the phonological point of view





Josef Vachek

Oct. 20 Oct. 27 Nov. 17 Dec. 1

V. Skalicˇka G. Devoto K. Horálek M. Mayenowa

Dec. 8

A. V. Isacˇenko

1948 Jan. 19 L. Cejp Feb. 27 A. Belic´ I. Lekov Mar. 1 V. Skalicˇka Apr. 26 J. Nováková May 24 I. Poldauf Oct. 11 J. Vachek & K. Horálek (reporting) Dec. 6 A. V. Isacˇenko Dec. 13 J. Pru˚šek

Copenhagen structuralism and the Prague school Contributions à la problématique des cas On disputed questions of Proto-Slavic phonology An attempt at distinguishing poetic language from other types of language Subject matter and limits of synchronic linguistics

The Cambridge school On the principles of studying standard languages Regularity in the development of Slavonic languages Concerning case Latin of the Silver Age Atemporality as a grammatical category of the Czech verb Reports on the Sixth International Linguistic Congress in Paris Morphology, syntax and phraseology On the aspect of the Chinese verb

1949 Nov. 14 I. I. Mešcˇaninov The role of N. J. Marr in the development of Soviet linguistics No further notes of sessions preserved 1950 Nov. 20 K. Horálek Concerning Indo-European No further notes of sessions preserved 1951 No notes of sessions preserved 1952 Jan. 24 A. V. Isacˇenko Feb. 6 J. Beˇlicˇ Mar. 10 K. Horálek Mar. 31 J. Šedivý May 12 V. Skalicˇka No further sessions held

Indo-European and Slavonic kinship terms in the light of Marxist linguistics Czech language as national language On the so called motivatedness in language Friedrich Engels’ Fränkischer Dialekt and some problems of Germanistic studies On dialects

Prolegomena to the history of the Prague School

References Abbreviations used in the references BSE CˇMF CˇOJ

Brno Studies in English. Brno. Cˇasopis pro moderní filologii. [Journal for Modern Philology]. Praha. Cˇeština a obecný jazykozpyt. [Czech Language and General Linguistics]. Praha. See Mathesius 1947. NRˇ Naše rˇecˇ. [Our language]. Praha. PCˇSL Prameny cˇeské a slovenské lingvistiky, rˇada cˇeská. [Sources of Czech and Slovak Linguistics, Czech series] Praha. See Vachek (ed.) 1970, 1972. PLK Pražský lingvistický kroužek. [Prague Linguistic Circle]. PP Philologica Pragensia. Praha. PSE Prague Studies in English. Praha. PSML Prague Studies in Mathematical Linguistics. Praha. PŠ Pražská škola jazykoveˇdná [Prague School of Linguistics]. RLB Recueil linguistique de Bratislava. Bratislava. SaS Slovo a slovesnost. [Word and Verbal Arts]. Praha. SPFFBU Sborník prací filosofické fakulty brneˇnské university. Brno. TCLP Travaux du Cercle linguistique de Prague. Praha. See Cercle . . . 1929, 1931; Jakobson 1929, Trnka 1931. TLP Travaux linguistiques de Prague. Praha. See Vachek (ed.) 1964, 1966; Isacˇenko (ed.) 1968. Artymovicˇ, A. 1932. ‘‘Pysana mova’’. In Naukovyj Zbirnyk Vys. ukrajin. ped. instytutu v Prazi 2,1–8. Austerlitz, R. 1964. Review of J. Vachek, 1964d. Word 20, 458–65. Baudouin de Courtenay, J. 1895. Versuch einer Theorie phonetischer Alternationen. Strassburg. Beˇlicˇ, J. 1956. Sedm kapitol o spisovné cˇeštineˇ. [Seven Chapters on Standard Czech]. Praha. Bloomfield, L. 1933. Language. New York. Bühler, K. 1931. ‘‘Phonetik und Phonologie’’. In TCLP 4, 22–52. Bulygina, T. V. 1964. ‘‘Pražskaja lingvisticˇeskaja škola’’. In M. M. Guxman and V. J. Jarceva, 1964, 46–126. Cercle linguistique de Prague 1929. ‘‘Thèses présentées au Premier congrès des philologues slaves’’. In TCLP 1, 3–29. (Czech version in PCˇSL 1, 1970) Cercle linguistique de Prague 1931a. ‘‘Principes de transcription phonologique’’. In TCLP 4, 323–65. Cercle linguistique de Prague 1931b. ‘‘Projet de terminologie phonologique standardisée’’. In TCLP 4, 309–23. Cercle linguistique de Prague 1931c. ‘‘Masaryk a rˇecˇ’’. [Masaryk and speech]. Praha. Cercle linguistique de Prague 1932. Charisteria Guilelmo Mathesio quinquagenario a discipulis et Circuli linguistici Pragensis sodalibus oblata. Praha. Cercle linguistique de Prague 1948. ‘‘Thèses’’. In Actes du Sixième congrès international de linguistes, 135–6 ff.





Josef Vachek

Chloupek, J. and Nekvapil, J. (eds) 1987. Reader in Czech Sociolinguistics. Amsterdam: Benjamins. Chomsky, N. A. 1964. Current Trends in Linguistics. Den Haag. Chomsky, N. A. 1968. Language and Mind. New York. Chomsky, N. A. and Halle, M. 1968. The Sound Pattern of English. New York — Evanston — London. Cˇyževs´kyj, D . 1931. ‘‘Phonologie und Psychologie’’. In TCLP 4, 3–21. Daneš, F. 1956. Intonace a veˇta ve spisovné cˇeštineˇ. [Intonation and the Sentence in Standard Czech]. Praha. Daneš, F. 1964. ‘‘A three level approach to syntax’’. In TLP 1, 225–40. Daneš, F. 1966. ‘‘The relation of centre and periphery as a language universal’’. In TLP 2, 9–21. Daneš, F. 1970. ‘‘Zur linguistischen Analyse der Textstruktur’’. In Folia Linguistica 4, 72–8. Daneš, F. (ed.) 1974. Papers on Functional Sentence Perspective. Praha: Academia. Daneš, F. 1994. ‘‘Odstavec jako centrální jednotka tematicko-kompozicˇní výstavby textu’’. [The paragraph as the central unit of the thematic and compositional structure of text]. SaS 35, 1–17. Daneš, F. and Vachek, J .1966. ‘‘Prague Studies in Structural Grammar Today’’. In TLP 1, 21–32. Daneš, F. and Filipec, J. 1978. Slovník spisovné cˇeštiny pro školu a verˇejnost. [A dictionary of Standard Czech for Schools and the General Public]. Praha: Academia. (2nd ed. 1994) Dokulil, M. 1962. Teorie slov v cˇeštineˇ 1: Teorie odvozování slov. [Word Formation in Czech 1: Theory of Derivation]. Praha: Academia. Dokulil, M. and Daneš, F. 1958. ‘‘K tzv. významové a mluvnické výstavbeˇ veˇty’’. [On the socalled semantic and grammatical structure of the sentence]. In O veˇdeckém poznání soudobých jazyku˚, 231–46. Praha. Doležel, L. 1964. ‘‘Vers la stylistique structurale’’. In TLP 1, 257–66. Firbas, J. 1962. ‘‘Ze srovnávacích studií slovosledných’’. [From comparative word order studies]. SaS 23, 161–74. Firth, J. R. 1957. ‘‘Synopsis of linguistic theory, 1930–1955’’. Studies in Linguistic Analysis 1930–1955. Oxford. Fischer, O. 1932. ‘‘Vzpomínka’’. [A remembrance]. In Cercle 1932, 3–4. Praha. Fries, C. C. and Pike, K. L. 1949. ‘‘Coexistent phonemic systems’’. Language 25, 29–50. Garvin, P. L. (ed.) 1964. A Prague School Reader in Esthetics, Literary Structure, and Style. Washington, D. C. Guxman, M. M. 1964. ‘‘Istoricˇeskije i metodologicˇeskije osnovy strukturalizma’’. In: Guxman, M. M. and Jarceva, V. N. 1964, 5–45. Moskva. Guxman, M. M. and Jarceva, V. N. 1964. Osnovnyje napravlenija strukturalizma. Moskva. Hais, K. 1938. Anglicky v 30 lekcích. [English in 30 lessons]. Praha. Hamm, J. (ed.) 1967. Phonologie der Gegenwart. Graz — Wien — Köln. Hausenblas, K . 1955. ‘‘K základním pojmu˚m jazykové stylistiky’’. [On basic concepts of linguistic stylistics]. SaS 16, 1–15) Havránek, B. 1928–1929. Genera verbi v jazycích slovanských I.–II. [Genera Verbi in Slavonic Languages]. Praha.

Prolegomena to the history of the Prague School

Havránek, B. 1929a. ‘‘Influence de la fonction de la langue littéraire sur la structure phonologique et grammaticale du tchèque littéraire’’ In TCLP 1, 108–20. (Czech version in B. Havránek 1963, 19–30) Havránek, B. 1929b. ‘‘Prˇíspeˇvek k tvorˇení slov ve spisovných jazycích slovanských (Adjektiva s významem latinských adjektiv na -bilis)’’. [A contribution to word formation in Standard Slavonic languages (Adjectives with the meaning of Latin adjectives in -bilis)]. Slavia 7, 766–84. Havránek, B. 1932. ‘‘Úkoly spisovného jazyka a jeho kultura’’. [The tasks of the standard language and its culture]. In: B. Havránek and M. Weingart (eds) 1932, 32–84. — Partly translated in P. L. Garvin (ed.) 1964, 3–16. Havránek, B. 1933. ‘‘Zur phonologischen Geographie. Das Vokalsystem des balkanischen Sprachbundes’’. In Archives Néderlandaises de la phonétique experimentale 8–9, 28–34. Havránek, B . 1936. ‘‘Vývoj spisovného jazyka cˇeského’’. [The development of standard Czech]. In Cˇeskoslovenská vlastiveˇda II. — Spisovný jazyk cˇeský a slovenský (Czechoslovak National History and Geography II — Standard Czech and Slovak Language). Praha. Havránek, B. 1962. ‘‘Aktuální metodologické problémy marxistické jazykoveˇdy’’. [Current methodological problems of Marxist linguistics]. In Problémy marxistické jazykoveˇdy, 9–19. Praha. Havránek, B . 1963. Studie o spisovném jazyce. [Studies of Standard Language]. Praha. Havránek, B. 1965. ‘‘Stav cˇeské lingvistiky v jubilejním dvacátém roce’’. [The state of Czech linguistics in the anniversary year 1965]. SaS 26, 101–4. Havránek, B. 1970. ‘‘Principi praške lingvisticˇke škole’’. In Zbornik za filologiju i lingvistiku, XLII/I, 7–17, Novi Sad. Havránek, B. and Daneš, F. 1965. ‘‘Slovo a slovesnost a cˇeská lingvistika v novém období’’. [The journal Slovo a Slovesnost and Czech linguistics in the new era]. SaS 26, 310–17. Havránek, B., Horálek. K., Skalicˇka. V. and Trost, P. 1958. Réponses aux questions linguistiques au IVe Congrès international des slavistes. Moskva, 50–3. Havránek, B., Kopeckij, L., Starý, E. and Získal, A. 1933–1936. Cvicˇebnice jazyka cˇeského pro první (druhou, trˇetí, cˇtvrtou) trˇídu strˇedních škol. [A Textbook of the Czech Language for the First (Second, Third, Fourth) Form of Secondary Schools]. Praha. Havránek, B. and Mukarˇovský, J. (eds) 1942. Cˇtení o jazyce a poesii. [Studies in Language and Poetry]. Praha: Družstevní práce. Havránek, B. and Weingart, M. (eds) 1932. Spisovná cˇeština a jazyková kultura [Standard Czech and Language Culture]. Praha: Melantrich. Horálek, K. 1948. ‘‘La fonction de la ‘structure des fonctions’ de la langue’’. In RLB I, 39–43. — Reprinted in J. Vachek 1964d, 421–5. Horálek, K. 1954. ‘‘Prˇehled Havránkovy veˇdecké cˇinnosti’’. [A survery of Havránek’s scholarly activities]. Studie a práce lingvistické I, 11–35. Horálek, K . 1964. ‘‘Les fonctions de la langue et de la parole’’. In TLP 1, 41–6. Isacˇenko, A. V. 1948. ‘‘O prizyvnoj funkcii jazyka’’. In RLB I, 45–57. (English version in J. Vachek 1964d, 88–97) Isacˇenko, A. V. (ed.) 1968. Études structurales dédiées aux Sixième Congrès des slavistes à Prague. Praha: Academia. (= TLP 3).





Josef Vachek

Jakobson, R. 1923. O ˇcešskom stixe, preimušcˇestvenno v sopostavlenii s russkim. Berlin. (Czech version Základy cˇeského verše. [The Principles of the Czech Verse]. Praha 1926) Jakobson, R. 1928. ‘‘Pojem hláskoslovného zákona a princip teleologický’’. [The concept of the sound law and the teleological principle]. CˇMF 14, 183–4. — English translation in Jakobson 1962a) Jakobson, R. 1929. Remarques sur l’évolution phonologique du russe comparée à celle des autres langues slaves. (= TCLP 2), Praha. — Reprinted in R. Jakobson 1962a, 7–116. Jakobson, R. 1931. ‘‘Prinzipien der historischen Phonologie’’. In TCLP 4, 247–67. (French version in R. Jakobson 1962a, 202–20) Jakobson, R. 1933. ‘‘La scuola linguistica di Praga’’. La Cultura 12, 633–41. Jakobson, R. 1936. ‘‘Usmeˇrneˇné názory na starocˇeskou kulturu’’. [Adjusted views on old Czech culture]. SaS 2, 207–31. Jakobson, R. 1938. ‘‘Observations sur le classement acoustique des consonnes’’. In Proceedings of the Third International Congress of Phonetic Sciences, Ghent, 1938, 34–41. — Reprinted in R. Jakobson 1962a, 272–9. Jakobson, R. 1939. ‘‘Nikolaj Sergeevicˇ Trubetzkoy’’. In Acta Linguistica I, 64–76. — Reprinted in T. Sebeok 1966, 526–42. Jakobson, R. 1956. ‘‘Serge Karcevski’’. Cahiers F. de Saussure 14, 9–13. — Reprinted in Sebeok 1966, 493–7. Jakobson, R. 1962a. Selected Writings I. Phonological Studies, Den Haag: Mouton. Jakobson, R. 1962b. ‘‘Retrospect’’. In Jakobson 1962a, 631–58. Jakobson, R. 1963. ‘‘Efforts towards a means-ends model of language in the inter-war period’’. In Trends in Modern Linguistics 104–8, Utrecht-Antwerpen. — Reprinted in J. Vachek 1964d, 481–5. Jakobson, R. 1965. ‘‘An example of migratory terms and institutional models (On the Fiftieth Anniversary of the Moscow Linguistic Circle)’’. In Omagiu lui A. Rosetti 427–31, Bucuresti. Jakobson, R. 1969. ‘‘Linguistics and the adjacent sciences’’. In Actes du Xe Congrès des linguistes 75–122, Bucuresti. Jakobson, R. (ed.) 1975. N. S. Trubetzkoy’s Letters and Notes. Den Haag: Mouton. Jakobson, R. and Pomorska, K. 1980. Dialogues. Paris: Flammarion. Jakobson, R., Karcevskij, S. and Trubetzkoy, N. S. 1928. ‘‘Propositions’’. In Actes du premier congrès international de linguistes à La Haye, 336. Leiden: A. W. Sijthoff — Reprinted in Jakobson 1962a, 3–6. Janko, J. 1928. ‘‘První mezinárodní sjezd lingvistický v Haagu’’. [The first international congress of linguists in The Hague]. CˇMF 14, 313–16. Jedlicˇka, A.1964. ‘‘Zur Prager Theorie der Schriftsprache’’. In TLP 1, 47–58. Juilland, A. 1967. ‘‘Perspectives du structuralisme évolutif ’’. Word 23, 350–61. Karcevskij, S. 1927. Système du verbe russe. Praha. Karcevskij, S. 1931. ‘‘Sur la phonologie de la phrase’’. In TCLP 4, 188–227. — Reprinted in J. Vachek 1964d, 206–51. Kondrašov, N. A. (ed.) 1967. Pražskij lingvisticˇeskij kružok. Sbornik statej. Moskva. Kopeckij, L. V. 1935. Ruština pro Cˇechy. [Russian for Czech].Praha. Kopecˇný, F. 1962. Slovesný vid v cˇeštineˇ. [Verbal Aspect in Czech]. Praha.

Prolegomena to the history of the Prague School

Korˇínek, J. M. 1934. Studie z oblasti onomatopoje. [Studies in Onomatopoeia]. Praha. Korˇínek, J. M. 1939. ‘‘Laut und Wortbedeutung’’. In TCLP 8, 58–65. — Reprinted in J. Vachek 1964d, 98–107. Korˇínek, J. M. 1942. ‘‘O jazykovém vývoji’’. [On language development]. CˇMF 28, 371–5. Krajcˇovicˇ, R. 1957. ‘‘Zmena g > y > h v západoslovanskej skupine’’. [The change g > (> h in West Slavonic]. Slavia 26, 341–57. Kru˚ta, J. and Roubícˇek, K. 1941. Anglicky od A do Z. [English from A to Z]. Praha. Kufner, H. L. 1962. The Grammatical Structure of English and German. Chicago & London. Lamb, S. M. 1966. Outline of Stratificational Grammar. Washington, D. C. Leech, G. N. 1969. Towards a Semantic Description of English. London. Leška, O. 1992a. ‘‘K lingvistické tematice Pražského lingvistického kroužku klasického období’’. [On linguistic subject matter of the Prague Linguistic Circle of the classical period]. CˇMF 74, 1–7. Leška, O. 1992b. ‘‘Pražská a ženevská fonologie’’. [The Praguian and the Genevese phonology]. CˇMF 74, 65–71. Lüdtke, H. 1959. ‘‘Das prosodische System des Urslawischen und seine Weiterentwicklung im Serbokroatischen’’. In Phonetica, suppl. ad vol.4, 125–56. Luelsdorff, P. A. (ed.) 1994. The Prague School of Structural and Functional Linguistics. Amsterdam: Benjamins. Machek, V. 1930. Studie o tvorˇení výrazu˚ expresívních. [A study of the formation of expressive words]. Praha. Martinet, A. 1955. Économie des changements phonétiques. Berne. Mateˇjka, L. (ed.) 1976. Sound, Sign and Meaning. A Quinquagenary of the Prague Linguistic Circle. Ann Arbor, Mich. Mathesius, V. 1911–1912. ‘‘O potenciálnosti jevu˚ jazykových’’. [On the potentiality of the pheomena of language]. In Veˇstník Královské cˇeské spolecˇnosti nauk 1911–1912. Trˇída historicko-filosoficko-filologická. — Reprinted in PCˇSL 1, 1970, 5–34; Mathesius 1982, 9–28. English translation in Vachek 1964d, 1–32. Mathesius, V. 1926. ‘‘New currents and tendencies in linguistic research’’. In MNHMA (Miscelany dedicated to J. Zubatý), 188–203. — Czech translation in PCˇSL 2, 5–17. Mathesius, V. 1927. ‘‘Lingvistická charakteristika a její místo v moderním jazykozpyteˇ’’. CˇMF 13, 35–40. — Reprinted in CˇOJ 9–16; English version ‘‘On linguistic characterology with illustrtion from modern English’’ in Actes du Premier congrès international des linguistes à la Haye 56–63 (1928); reprinted in Vachek 1964b, 59–67. Mathesius, V. 1929. ‘‘La structure phonologique du lexique du tchèque moderne’’. In TCLP I, 67–84. — Reprinted in Vachek 1964d, 156–76; Czech version in CˇOJ 59–86. Mathesius, V. 1931a. ‘‘Zum Problem der Belastungs- und Kombinationsfähigkeit der Phoneme’’. In TCLP 4, 148–52. — Reprinted in Vachek 1964d, 177–82. Mathesius, V. 1931b. ‘‘Die Persönlichkeit Josef Zubatýs’’. In Prager Rundschau I, 239–47. — Reprinted in Sebeok 1966, II, 77–86. Mathesius, V. 1932a. ‘‘La place de la linguistique fonctionelle et structurale dans le développement général des études linguistiques’’. CˇMF 18, 1–7. (Czech version in CˇOJ, 17–38)





Josef Vachek

Mathesius, V. 1932b. ‘‘Cizí slova ze stanoviska synchronického’’. [Loan words from the synchronic point of view]. CˇMF 18, 231–9. — German version reprinted in J. Vachek 1964d: 398–412; Czech version reprinted in CˇOJ, 96–109. Mathesius, V. 1932c. ‘‘O potrˇebeˇ stability ve spisovném jazyce’’. [On the need for stability in the standard language]. In Havránek and Weingart 1932, 14–31. — Reprinted in v CˇOJ, 415–35. Mathesius, V. 1936a. ‘‘Deset let Pražského lingvistického kroužku’’. [Ten years of the Prague Linguistic Circle]. SaS 2, 137–45. Mathesius, V. 1936b. Nebojte se anglicˇtiny! Praktický pru˚vodce jazykem. [Don’t be afraid of English. A practical guide to the language]. Praha. Mathesius, V. (ed.) 1939–1940. Co daly naše zemeˇ Evropeˇ a lidstvu. [What our Countries Gave to Europe and to Mankind]. Praha: Evropský literární klub. Mathesius, V. 1942. ‘‘Rˇecˇ a sloh’’. [Speech and style]. In Havránek and Mukarˇovský (eds) 1942, 11–102. Mathesius, V. 1944. Možnosti, které cˇekají. [Chances that lie ahead]. Praha. Mathesius, V. 1947. Cˇeština a obecný jazykozpyt. Soubor statí. [Czech Language and General Linguistics. Selected Papers]. Praha: Melantrich. Mathesius, V. 1961. Obsahový rozbor soucˇasné anglicˇtiny na základeˇ obecneˇ lingvistickém. [A Functional Analysis of Present Day English on a General Linguistic Basis], edited by J. Vachek. Praha: Nakladatelství Cˇeskoslovenské akademie veˇd. — English translation by L. Dušková, Den Haag — Paris: Mouton & Praha: Academia. Mathesius, V. 1982. Jazyk, kultura a slovesnost. [Language, Culture and Verbal Arts.] Praha: Odeon. Mešcˇaninov, I. I. 1949. ‘‘Nové ucˇení o jazyku v SSSR a jeho soucˇasné vývojové fáze’’. [The new teaching on language in the USSR and its present stages of development]. SaS II, 1–7. Moulton, W. G. 1962. The Sounds of English and German. Chicago & London. Mukarˇovský, J. 1932. ‘‘Jazyk spisovný a jazyk básnický’’. [Standard language and poetic language]. In Havránek-Weingart 1932, 123–56. — Partly translated in P. L. Garvin 1964, 17–30. Mukarˇovský, J. 1982. Studie z poetiky. [Studies from poetics]. Praha: Odeon. Neústupný, J. V. 1966. ‘‘On the Analysis of Linguistic Vagueness’’. In TLP 2, 39–51. Nickel, G. (ed.) 1971. Papers in Contrastive Linguistics. Cambridge. Nosil, J. 1941. Poznejte neˇmcˇinu. [Get to Know German]. Praha. Novák A. 1936–1938. Prˇehledné deˇjiny literatury cˇeské [Outline of Czech Literature]. Praha. — Reprint, 1995. Novák, P. 1991. ‘‘K poválecˇným osudu˚m cˇeské lingvistiky’’. [On the post-war situation of Czech linguistics]. SaS 52, 183–93. Osicˇka, A. and Poldauf, I. 1948. Velký anglicko-cˇeský slovník Unikum. [Large English-Czech dictionary Unikum]. Praha. Pauliny, E. 1943. Štruktúra slovenského slovesa. [The structure of the Slovak verb]. Bratislava. Pauliny, E. 1948. ‘‘La phrase et l’énonciation’’. In RLB I, 59–66. PCˇSL: Prameny cˇeské a slovenské lingvistiky: Rˇada cˇeská. 1–2. 1970–1972. PCˇSL 1. U základu˚ pražské jazykoveˇdné školy. [The Foundations of the Prague Linguistic School]. Praha. (= Vachek 1970)

Prolegomena to the history of the Prague School

PCˇSL 2. Z klasického období pražské školy 1925–1945. [From the Classical Period of the Prague School 1925–1945]. Praha. (= Vachek 1972) Pos, H. J. 1939. ‘‘Perspectives du structuralisme’’. In TCLP 8, 71–8. Potter, S. and Trnka, B. 1926–28: Ucˇebnice jazyka anglického pro strˇední školy I-III. [A textbook of English for secondary schools I-III]. Praha. Ray, P. S. 1963. Language Standardization. Studies in Prescriptive Linguistics. Den Haag. Rohrer, Chr. 1967. ‘‘Die Behandlung der französischen Nasalvokale in der generativen Phonologie’’. In J. Hamm 1967, 287–97. Ružicˇka, J. 1948. ‘‘Josef Miroslav Korˇínek’’. In RLB I, 5–11. Ružicˇka, J. 1956. Skladba neurcˇitku v slovenskom jazyke. [The syntax of the infinitive in Slovak]. Bratislava. Rypka, J. 1968. History of Iranian literature. Written in collaboration with Otakar Klíma and others. Dordrecht, D. Reidel. Sapir, E. 1922. Language. New York. Sapir, E. 1925. ‘‘Sound Patterns in Language’’. In Language 1, 37–51. Sebeok, Th. A. (ed.) 1966. Portraits of Linguists I-II. Bloomington & London. Sgall, P. 1964. ‘‘Zur Frage der Ebenen im Sprachsystem’’. In TLP 1, 95–106. Skalicˇka, V. 1937. ‘‘Promluva jako lingvistický pojem’’. [The utterance as a linguistic concept]. SaS 3, 163–6. Skalicˇka, V. 1941. Vývoj cˇeské deklinace. [The development of Czech declension]. Praha. (= SPLK 4). Skalicˇka, V. 1947–1948. ‘‘Kodanˇský strukturalismus a ‘pražská škola’’’. [The Copenhagen structuralism and the ‘Prague school’]. SaS 10, 1947–1948: 135–42. SPLK: Studie Pražského lingvistického kroužku. [Papers of the Prague Linguistic Circle]. 1–4. 1937–1941. Praha. SPLK 1, 1937. Josef Hrabák, Staropolský verš ve srovnání se starocˇeským. [Old Polish Verse in Comparison with Old Czech Verse]. SPLK 2, 1939. Bohuslav Indra, Havlícˇkovy práce o verši cˇeské lidové písneˇ. [Karel Havlícˇek Borovský’s Works on the Verse of Czech Folk Songs]. SPLK 3, 1941. Josef Hrabák, Smilova škola. [The Literary School of Smil Flaška z Pardubic]. SPLK 4, 1941. Vladimír Skalicˇka, Vývoj cˇeské deklinace. [Development of Czech Declension]. Stankiewicz, E. 1965. Review of J. Vachek 1964d. In The Slavic and East European Journal 9, 196. Šcˇerba, L. V. 1912. Russkije glasnyje v kacˇestvennom i kolicˇestvennom otnošenii. SanktPeterburg. TCLP: Travaux du Cercle linguistique de Prague 1–8. 1929–1939. TCLP 1, 1929a. Mélanges linguistiques dédiés au Premier congrès des philologues slaves. Praha. TCLP 2, 1929b. Remarques sur l’évolution phonologique du russe comparée à celle des autres langues slaves. Praha. (= Jakobson 1929) TCLP 3, 1930. On the Syntax of the English Verb from Caxton to Dryden. (= Trnka 1930) TCLP 4, 1931. Réunion phonologique internationale tenue à Prague (18–21/XII 1930). Praha. TCLP 5, 1935. Description phonologique du russe moderne. I–III. — Only part II (= TCLP 52) appeared, cf. Trubetzkoy 1935. TCLP 6, 1936. Études dédiées au Quatrième congrès de linguistes.





Josef Vachek

TCLP 7, 1939a. Grundzüge der Phonologie. (= Trubetzkoy 1939) TCLP 8, 1939b. Études phonologiques dédiées à la mémorie de N. S. Troubetzkoy. TLP: Travaux linguistiques de Prague 1–4. 1964–1971. TLP 1, 1964. L’école de Prague d’aujourd’hui. Praha: Éditions de l’Académie Tchécoslovaque des Sciences. (= Vachek 1964e) TLP 2, 1966. Les problèmes du centre et de la périphérie du système de la langue. Praha: Academia. (= Vachek 1966d) TLP 3, 1968. Études structurales dédiées aux Sixième Congrès des slavistes à Prague. Praha: Academia. (= Isacˇenko 1968) TLP 4, 1971. Études de phonologie, typologie et de la linguistique générale, ed. by F. Daneš. Praha: Academia. Trnka, B. 1928. ‘‘Pražský lingvistický kroužek’’. [The Prague Linguistic Circle]. CˇMF 14, 182–6. Trnka, B. 1929a. ‘‘Some Remarks on the Phonological Structure of English’’. In Xenia Pragensia 357–64, Praha. Trnka, B. 1929b. ‘‘Méthode de comparaison analytique et grammaire comparée historique’’. In TCLP I, 33–8. — Reprinted in J. Vachek 1964d, 68–73. Trnka, B. 1930. On the Syntax of the English Verb from Caxton to Dryden. (= TCLP 3) Trnka, B. 1935. A Phonological Analysis of Present Day Standard English. (= PSE 5) Trnka, B. 1946. ‘‘Vilém Mathesius’’. CˇMF 29, 3–13. (English version in Sebeok 1966, II, 474–89) Trnka, B. 1948a. ‘‘Jazykozpyt a myšlenková struktura doby’’. [Linguistics and the ideological structure of the period]. SaS 10, 53–70. — English translation in J. Vachek 1966a, 152–65. Trnka, B. 1948b. ‘‘From Germanic to English’’. In RLB I , 139–49. Trnka, B. 1953–1956. Rozbor nyneˇjší spisovné anglicˇtiny I-III. [An Analysis of Present-day Standard English]. Praha. Trnka, B. 1958. ‘‘Morfologické protiklady’’. [Morphological oppositions]. In O veˇdeckém poznání soudobých jazyku˚, 93–104. Praha. Trnka, B. 1959. ‘‘A Phonemic aspect of the great vowel shift’’. In Mélanges F. Mossé in memoriam, 440–3. Paris. Trnka, B. 1966. ‘‘The distribution of vowel length and its frequency in Czech’’. In PSML 1, 11–16. Trnka, B., Vachek, J. et al., 1957. ‘‘K diskussii po voprosam strukturalizma’’. In Voprosy jazykoznanija 1957, nr. 3, 44–52. (English version PP 1, 1958, 44–52) Trost, P. 1936. ‘‘Bemerkungen zum Sprachtabu’’. In TCLP 6, 288–94. Trubetzkoy, N. S. 1935. Anleitung zu phonologischen Beschreibungen. Brno. Trubetzkoy, N. S. 1939. Grundzüge der Phonologie. (= TCLP 7) Vachek, J. 1932. ‘‘Professor Daniel Jones and the Phoneme’’. In Cercle 1932, 28–33. Praha. Vachek, J. 1933. ‘‘Professor Karl Luick and problems of historical phonology’’. CˇMF 19, 273–92. Vachek, J. 1939. ‘‘Zum Problem der geschriebenen Sprache’’. In TCLP 8, 94–104. — Reprinted in J. Vachek 1964b, 441–52. Vachek, J. 1946a. ‘‘Za hrdinou ducha. K prvému výrocˇí skonu univ. prof. dr. Viléma

Prolegomena to the history of the Prague School

Mathesiuse’’. [In memory of a spiritual hero. On the first anniversary of the death of Professor Vilém Mathesius]. In Vyšehrad 1 (24): 13–15. Vachek, J. 1946b. Anglicky sveˇže a spolehliveˇ. [A Bright and Safe Road to English]. Praha . Vachek, J. 1959. ‘‘Notes on the quantitative correlation of vowels in the phonematic development of English’’. In Mélanges F. Mossé in memoriam, 444–56. Paris. Vachek, J. 1960. Dictionnaire de linguistique de l’Ecole de Prague. Utrecht-Antwerp. — Reprinted in 1966. Russian version Lingvisticˇeskij slovar’ pražskoj školy, Moskva 1964. Vachek, J. 1961a. ‘‘Some less familiar aspects of the analytical trend of English’’. In BSE 3, 9–78. Vachek, J. 1961b. ‘‘A propos de la terminologie linguistique et du système de concepts linguistiques de l’Ecole de Prague’’. In PP 4, 65–78. Vachek, J. 1962. ‘‘K otázce vlivu˚ vneˇjších cˇinitelu˚ na vývoj jazykového systému’’. [On the influence of external factors on the development of the system of language]. In Slavica Pragensia 4, 35–46. (English version in Lingua 11, 433–48.) Vachek, J. 1963. ‘‘The place of the sound [rˇ] in the structures of Slavonic languages’’. In SPFFBU A 11, 81–92. Vachek, J. 1964a. ‘‘Prague phonological studies today’’. In TLP 1, 7–20. Vachek, J. 1964b. ‘‘Notes on gender in modern English’’. In SPFFBU A 12, 189–94. Vachek, J. 1964c. ‘‘On some basic principles of ‘Classical’ Phonology’’. Zeitschrift für Phonetik . . . 17, 409–31. Vachek, J. (ed.) 1964d. A Prague School Reader in Linguistics. Bloomington. Vachek, J. (ed.) 1964e. L’école de Prague d’aujourd’hui. (= TLP 1) Vachek, J. 1965. ‘‘Slovo a slovesnost jako tribuna pražské školy jazykoveˇdné v letech 1935–1943’’. [The journal Slovo a slovesnost as the platform of the Prague School of Linguistics in the years 1935 to 1943]. SaS 26, 305–10. Vachek, J. 1966a. The Linguistic School of Prague. An lntroduction to its Theory and Practice. Bloomington & London. Vachek, J. 1966b. ‘‘On the integration of the peripheral elements into the system of language’’. In TLP 2, 23–37. Vachek, J. 1966c. ‘‘Soveˇtský sborník o jazykoveˇdném strukturalismu’’. [A Soviet miscellany on linguistic structuralism]. Review of M. M. Guxman and V. N. Jarceva 1964. SaS 27, 56–66. Vachek, J. (ed.) 1966d. Les problèmes du centre et de la périphérie du système de la langue. (=TLP 2) Vachek, J. 1968a. ‘‘O dynamickém pojetí fonologie, zvlášteˇ cˇeské’’. [On the dynamic conception of (mainly Czech) phonology]. SaS 29, 246–55. Vachek, J. 1968b. Dutch Linguists and the Prague Linguistic School. Leiden. Vachek, J. 1968c. ‘‘Phonology today’’. Linguistics 46, 77–98. Vachek, J. 1968d. Dynamika fonologického systému soucˇasné spisovné cˇeštiny. [The dynamics of the phonological system of Present-Day Standard Czech]. Praha. Vachek, J. 1970a. ‘‘Z nových cest generativní fonologie’’. [Generative phonology looking for new ways]. SaS 31, 111–24. Vachek, J. 1970b. ‘‘On some less known aspects of the early Prague linguistic school’’. In Actes du Xe Congrès international des linguistes II, Bucharest, 333–7.





Josef Vachek

Vachek, J. 1970c. ‘‘Peˇtadvacet let od skonu Viléma Mathesia’’. [Twenty-five years since the death of Vilém Mathesius]. CˇMF 52, 121–6. Vachek, J. (ed.) 1970. U základu˚ pražské jazykoveˇdné školy. [The Foundations of the Prague Linguistic School]. Praha. (= PCˇSL 1) Vachek, J. (ed.) 1972. Z klasického období pražské školy 1925–1945. [From Classical Period of the Prague School 1925–1945]. Praha. (= PCˇSL 2) Vachek, J. (ed.) 1983. Praguiana. Praha: Academia & Amsterdam: Benjamins. Vachek, J. (ed.) 1989. Written Language Revisited. Amsterdam: Benjamins. Vachek, J. 1994a. ‘‘K teleologickému pojetí zmeˇn v historii jazykového systému’’. [On the teleological approach to changes in the history of the language system]. NRˇ 77, 127–32. Vachek, J. 1994b. Vzpomínky cˇeského anglisty [Memoirs of a Czech Anglicist]; Praha: H&H. Wollman, F. 1936. ‘‘Literárneˇveˇdné metody v Bittneroveˇ knize Deutsche und Tschechen’’. [Methods of literary criticism in Bittner’s book]. SaS 2, 201–7. Zvegincev, V. A. 1964–65: Istorija jazykoznanija XIX i XX vekov v ocˇerkax i izvlecˇenijax, I–II. Moskva.

Editors’ note Habent suafata libelli. In the early 1990s, Professor Josef Vachek, the honorary chairman of the recently re-established Prague Linguistic Circle, offered the publishing house H&H two manuscripts for publication: Vzpomínky cˇeského anglisty (Memoirs of a Czech Anglicist) and Prolegomena k deˇjinám Pražské školy jazykoveˇdné (Prolegomena to the history of the Prague School of Linguistics). The latter was much more on Vachek’s mind, but the publisher, a newcomer to the Czech book market, preferred the Memoirs. Only when the book appeared in 1994 did H&H decide to publish the Prolegomena as well. Vachek’s introduction to it bears the date of October 29, 1994, yet it was only in summer of 1996, almost three months after Josef Vachek’s death (March 31, 1996), that the publisher asked Professor Oldrˇich Leška, the then chairman of the Prague Linguistic Circle, to write an afterword. In the same year, the Circle’s Committee made a thorough revision of the Appendices and Secondary sources. All the same, neither Oldrˇich Leška (d. August 9, 1997) lived to see the book published as the Prolegomena appeared only in 1999. Josef Vachek’s Prolegomena provide a specific perspective on the history of the Prague Linguistic Circle, one that present-day scholars working in the same linguistic tradition can no longer offer. The editors of the new Travaux series are therefore pleased to be able to present this historical document in an English translation to the international community of linguists. The whole body of Josef Vachek’s text was translated by Zdeneˇk Kirschner. The Afterword and Appendix I were translated by Nad’a Abdallaová, Appendix II and References were adapted by the editors. In this translation of Vachek’s Prolegomena, proper names from languages using non-Latin alphabets (mostly Russian and Ukrainian) are rendered in philological transcription with diacriticized letters, e.g. Dostoevskij, Cˇexov (instead of Dostoyevsky, Chekhov), which was the usage already in the pre-war Travaux du Cercle linguistique de Prague as well as in Roman Jakobson’s Selected Writings. The philological

Prolegomena to the history of the Prague School

transcription of the name of Roman Osipovicˇ Jakobson is identical to the usual spelling (when leaving out his patronymic), while the philological transcription of the name of Nikolaj Sergeevicˇ Trubeckoj is not. Since Trubeckoj used two canonical Latin alphabet spellings of his family name, viz. Trubetzkoy when writing in German and Troubetzkoy when writing in French, we render his name, as the only exception, rather in his own German spelling, than in the philological transcription.



Anton Marty’s philosophy of language

Oldrˇich Leška

0. Topicality of Anton Marty’s work Metaphysical philosophy was having a strong and long lasting impact on linguistic thought, esp. as far as problems of meaning were concerned. For Aristotle (De interpretatione) the content of language expressions (words, . . .) in different languages was common both to Greeks and to the rest of the linguistic world, i.e. to the Barbarians. The problem of linguistic relativism was nonexistent: the content was fixed, given and thus independent of the structure of a language. This provided an inexhaustible feeding ground for both parties to the dispute between anomalists and analogists which cannot be considered as resolved and thus a matter of a distant past (Seuren 1985). But this is not the line of thought to be followed here. A consequence of a metaphysical frame of mind is reification of meaning. It should be noted that questions like what does ‘sarcastic’ mean? are quite normal and do not presuppose a metaphysical frame of mind or a definite philosophical outlook whatsoever. We are dealing here with a case when the language user in his reply ‘sarcastic’ means . . . isolates the meaning as a result of a natural hypostatic attitude by which he verbalizes his own linguistic experience. Hypostasis is not exclusive to matters of meaning; we can meet it at whatever level of linguistic structure. Analytical philosophy (philosophy of everyday language) noticed quite aptly that the linguistic form may lead one astray (cf. questions like what is a peony? : : what is virtue?) by inducing the same understanding of the ontological status of the items in question. It is with Franz Brentano’s philosophy which was further developing in the work of his circle including A. Marty that an antimetaphysical turn came about. So that if we term the period in philosophy since Descartes as modern then Brentano’s school made the first decisive step along the path towards philosophical postmodernism. Thus one can see Marty’s work in a definite general perspective. As Professor of philosophy at the Prague German university Marty asembled here a circle of followers and thus strengthened the position of Brentanian philosophy; through mediation of Tomáš Garrigue Masaryk, another Brentanian,



Oldrˇich Leška

Professor at the Prague Czech university, Marty was instrumental in changing the outlook of Czech national philosophical community. Marty’s teachings have left an unmistakable trace in the theory and practice of the Prague Linguistic Circle. The memory of A. Marty has not faded away with time; interest in his ideas may be attested throughout the decades following his death up to the present reaching far beyond the limits of Europe now.

.

A few data concerning A. Marty’s life and work

Anton Marty was borne 1847 in Schwyz, Switzerland, into a family whose atmosphere combined with the young boy’s penchant led him to an early decision to become priest. Marty started his theological studies with success but he was never ordained. The reason why he did not feel fit to continue along the chosen path was the announcement of new doctrines of the Roman Catholic Church concerning the papal infallibility and the immaculate conception of Virgin Mary, two points he found unacceptable. At the same time he became acquainted with Franz Brentano and his philosophy. He studied under his guidance in Würzburg (1869) and the relationship between Brentano — teacher and Marty — student developed into a deep and lasting friendship. Although completely different in temperament, these two shared the same painful experience of a conflict between the inner voice of conscience and the clerical discipline which both of them resolved — quite independently — in the same way: as rebels. This was interpreted in some circles as an anticlerical gesture which made Brentano’s position all the more attractive. The basic lines of Marty’s language conception were formulated in his thesis on the origin of language (Marty 1875). After a brief stay in Czernowitz/Cernauti at a new founded Austrian university Marty accepted the proposed professorship at the Prague German university (1880) where he was active to his end (1914; Brentano — by 9 years senior of Marty — outlived his friend and died in Florence 1917). Because of scientific authority and rare features of character Marty served later as Dean of the Faculty of Arts and finally (1909) was elected Rector of the University. His philosophy of language was elaborated in detailed studies that were assembled for the most part and published by his Prague pupil and follower Oskar Kraus (Kraus 1916–1920). Marty was preparing a synthesis of his ideas in a voluminous work of which only the first book was published during his lifetime (Marty 1908); the rest remained in manuscript fragments or sketches. Thanks to Marty’s student Otto Funke we are in possession of a reconstruction of the whole (Marty 1940, 1950); Funke published some important parts of the text (Marty 1950). Luckily enough, Funke found and published the text of Marty’s lectures on philosophy of language held in the academic year 1904/5, which represent a résumé of Marty’s unfinished work (Marty 1940).

Anton Marty’s philosophy of language

The published first volume of Marty’s Untersuchungen found a ready though limited resonance. Karl Bühler wrote a detailed review (Bühler 1909) which was critical but it reflected a real understanding for the problems treated by Marty. On the other hand, Antoine Meillet (Meillet 1909) dismissed the book as rather irrelevant for understanding any concrete language. One does not find this very surprising: Meillet never had any deep rooted interest in theoretical problems of language. One must however concede that reading Marty could have been hardly a treat, esp. for Meillet. Marty’s texts are not easy to read; he follows the professorish style of exposing his ideas with lenghty digressions commenting on what his colleagues wrote on the subject so that his own line of thought is not always easy to follow. On the whole none of Marty’s published texts passed unnoticed; the appearence of some was followed by a lenghty polemic. He was very alert to what was going on in his field. Throughout his scientific life he was on the guard against the ideas of Wilhelm Wundt, Heyman Steinthal and Moritz Lazarus; on the other hand, he liked Karl Brugmann, Herman Paul, Ludwig Tobler, one of the finest syntacticians I have ever read.

.

What was characteristic of Brentano’s philosophy

For understanding Marty’s philosophy of language a few words on Brentano’s philosophy should be useful. Franz Brentano (1838–1917) started with a critical analysis of Aristotle’s philosophical system. On this basis he developed his phenomenal-descriptive psychology setting it in opposition to the then predominant line of thought in psychology which concentrated on the mechanistically conceived processual aspect of psychical phenomena (Brentano 1874). The underlying notion for Brentano’s philosophy is the inherent intentionality of human consciousness which makes it possible to abstract from the existence of the objective reality as an indispensable presupposition of a philosophical system. It is in the intentional act that reality constitutes itself; the objects one is conscious of may be independent of one’s consciousness (as, e.g. a tree ) or dependent on it (as in the case of a unicorn), both kinds of objects are real. A concrete intentional act should be distinguished from its content which is linguistically communicable and thus stripped of any subjective uniqueness. In this way of thinking, language is set as the guarantee of the intersubjective objectivity; it is clear why problems of language (of its character and structure) move — in the framework of Brentano’s system — into the focus of attention. It was natural that one had to concentrate on those features of language which were universal so that the dividing line between philosophy of language and linguistics, whose concern is all that is specific of a given language, could be clearly drawn. At the same time one had to pay attention to the distinction between what





Oldrˇich Leška

is present in the linguistic conscience of a native speaker as an active component of the linguistic structure and those components that have to be arrived at in the process of scientific analysis of language. This is connected with important problems of elementary psychological relevance of language and conditions of experiencing language, which are problems that cannot be passed over in silence even now. The approach to language must take into acount synchronic relationship (statics) on the one hand and diachronic relationship that obtain in the process of linguistic change on the other (dynamics) as different points of view. It is the synchronic relationship that are harbingers of linguistic change. It was characteristic of Brentano’s philosophical circle that its members who had taken important positions at Austrian and German universities were continuing to share the basic ideas of their teacher while applying them to particular domains of knowledge (cf. Wilhelm Dilthey, Carl Stumpf, Anton Marty, Alfred Kastil, Emil Utitz, Edmund Husserl). This fact explains why Brentano’s philosophy had become so influential (Brentano exerted his influence mainly through personal contact — most of his writings remained in manuscritp form; it is characteristic that a short stay in Italy achieved more in spreading his ideas than what he had written). At the same time this philosophy was not a closed system, it opened new vistas for philosophical thought as evidenced by the work of Edmund Husserl.

.

Marty’s functionalist and teleological conception of language

As we know, Marty introduced himself to the world of science by his thesis on the origin of language (Marty 1875). Usually one does not start reading books on this topic without a suspicion of diletantish speculation. Marty’s book however is not an attempt at substituting speculative phantasy for lack of facts. One can say that the title of the book was misleading because what the author was trying to do was to pinpoint the constitutive features of human language, to determine on what condition it is justified to call a communicative system — a natural language. Language emerged as a practical tool of communication and it preserved its instrumental character throughout its history. The inherent goal-orientedness of language provides a natural basis for its teleological conception. Language does not follows in the course of its development any definite blueprint. It is supposed to respond to the requirements of the communication; thus one cannot expect any systematicity in the changing make-up of language. Practical considerations decide whether a new addition based on individual linguistic creativity will be sanctioned by usage and thus included into the commonly accepted stock — or not. In this way, through language processes, the relative order characteristic of the synchronic coherence is maintained.

Anton Marty’s philosophy of language

If onomatopoeic and/or affective expressions played some role in the formation of language, this can be understood only in the sense that they served as the inner form of nominative expressions (cf. waw waw as an imitation of the bark of a dog and wawaw meaning and denoting a dog or the dog is barking in child language). The inner form does not arrest the hearer’s attention. It serves as transitory component in the internal structure of the expression leading to its meaning function as the case is, e.g. in metaphors (chick ‘a young girl’ ) or metonymic expressions (roof ‘house’, ‘home’ ), see below. If language is goal-oriented, instrumental in its character, it is quite natural that it should be differentiated according to its diverse functions. Thus it is erroneous to apply the same criteria to the poetic use of language as to its use in communication. It is possible to complement language when adjusting particular sets of its means to new branches of knowledge (language of scientific discourse); language may be trimmed and pruned to the needs of a smooth everyday linguistic intercourse, esp. to the tastes of the cultured social strata. Language may be subject to critical analysis from the logical point of view. Evaluative judgments on language matters from the ethical point of view may be found in what we call language politics in cases of language suppresion or aggresion. Texts, Marty says, have according to their function a specific tonality, in other words they are marked by a stylistic homogenity/cohesion.

. Marty’s philosophy of language Marty makes a distinction between a practical and a theoretical branch of philosophy of language. The problems making up the contents of the practical philosophy of language were indicated in the preceding paragraph. The theoretical branch of philosophy of language should not overlap with linguistics, i.e. with the science whose object are concrete historical languages, language groups or language types. Having as its empirical basis the descriptive/ phenomenal psychology, it should deal with general semasiology (allgemeine Bedeutungslehre). This is based on the following types of intentional acts as postulated by Franz Brentano: 1. representation acts / Vorstellungsakte 2. judgment acts / Urteilsakte 3. evaluative acts / wertende (heischende) Akte. These types form a hierarchy: evaluative acts presuppose judgment acts, judgment acts presuppose representation acts. In language they find correspondences in universal functional categories:





Oldrˇich Leška

1. representation suggestives (Vorstellungssuggestive, Ausdrücke) 2. judgment suggestives (Urteilssuggestive, Aussagen) 3. value suggestives (Emotive, wertende Aussagen). These universal functional categories are language nonspecific, i.e. nonspecific as far as their form/structure is concerned. Thus a representation suggestive (Ausdruck) may take a form of a word (cf. seamstress), of a combination of words (cf. sowing machine), of a sentence (cf. what you may call it as an indefinite filling pronoun). Against these universal functional categories Marty opposes the language specific Ausdrucksmethoden, i.e. ways of their structuring in different languages; it is linguistics that has to deal with these problems. Because of their universality the functional categories can serve as a basis for contrastive linguistics. A general semantic distinction between linguistic formations is the distinction between autosemantics and synsemantics. Autosemantics are those formations that can function as selfcontained utterances. All other cases that do not meet this condition are synsemantic; thus synsemantics are e.g. a sentence term or a clause: the hearers attention does not concentrate on them passing on to their complements until the possibilities of the construction frame are exhausted (cf. verbal forms and their complements). Nouns in an independent position are considered theoretical autosemantics, however not as a morphological category: abstracts like waiting, fondness are — according to Marty — synsemantics because they are nominalizations, i.e. nominal substitutions for more complex and explicit constructions and in this sense abreviations sui generis. Words like conjunctions (e.g. but) are not empty words; they are synsemantics like other synsemantic expressions; their alledged ‘‘semantic emptiness’’ results from the fact that synsemanticity is only another way of saying that complex formations are wholes/constructions whose components are structurally context bound; these constructions ties in a case like that of conjunctions are especially strong. In Marty’s terminology die konstruktive Form (the construction form) is a universal feature of language. At first sight Marty’s conception may look like a psychologistic one, but it is not so. Let us see what he says about meaning: the meaning (Bedeutung, Bedeutungsfunktion) of a linguistic expression (e.g. of a word) consists in its function to call forth / elicit in the mind a certain representation; thus the meaning is not the elicited representation itself. A word is understood by the speaker and the hearer if the content of the representations produced is identical with or essentially similar to the content of the representations which as a rule are elicited by this word; this could be paraphrased in a modernized — yet adequate — way as follows: the word is propely understood if it is used according to the rules of its current use. Besides meaning/meaning function (Bedeutungsfunktion) a word (an expression) has a naming function (Nennfunktion); from the situational context it is

Anton Marty’s philosophy of language

possible to infer what the word (the expression) is related to, what its referent is. In the system of Marty’s logic the referential relationship has a form of a judgment in which the expression is a predicate like in a formula A is p. This makes it possible to tell whether an utterance is true or false if the truth value does not consist in the form of the construction itself (cf. p ∨ — p ). The naming function makes interlinguistic understanding possible as manifested in translation. The term Nennfunktion (naming function) can be understood as a presupposition of particular naming acts in which the context of an intentional act by taking on a definite linguistic shape is subject to the rules of usage thus being stripped of its subjective individuality. In cases like honey (address), blacksmith, plumber the meaning has a sliced structure: at the dinner table when saying I like the parson’s nose one almost certainly does not refer to the face of a particular parson; one may be aware of the humorous aspect of the expression, but one does not dwell on the semantic elements immediately present in the expression, these are the inner form (die innere Form) of the meaning function. Marty’s innere Form is — besides the construction form — another universal feature of language. The term inner form was introduced into linguistics by Wilhelm von Humboldt in a different sense: thanks to language the world perception has a certain shape (form), in relation to it — language is its inner form.

.

Anton Marty and the Czech national setting

The position of the Brentanian philosophy was rather strong at German universities including the German university in Prague. The situation at the Czech university in Prague was different; the climate in Czech philosophy was determined by a long tradition of Herbartian philosophy; it was only with T. G. Masaryk, another of Brentano’s students, coming to Prague that the situation shows signs of a change. Masaryk’s participation in the political life was extraordinarily significant, perhaps even more effective than his university teaching. In his programmatic book Cˇeská otázka (Topical problems of Czech national life; Masaryk 1895), he demonstrated convincingly the grave shortcomings of the onesided historicism and proclaimed the necessity of a thorough knowledge of the Czech society in its current state including matters of language. However this was not Masaryk’s first attempt at rehabilitating synchrony and its laws as a legitimate object of scientific inquiry. He made his views concerning this problem known in his book Základové konkrétní logiky (An outline of a concrete logic; Masaryk 1885) with references to A. Marty. To form an idea of what the relationship of the Czech and German speaking communities was in Prague, a closer look at Marty in his ambiance at the begin-





Oldrˇich Leška

ning of twentieth century may be useful. Marty lived on Marienstraße (Mariastreet to honor the Empress Marie Anne) which is known as Opletalova since 1945. In Marty’s time it was not only a quiet, but an extremely pleasant street bordering a beautiful park. Marty used to lecture in the study of his appartment; after a lecture attended not only by his students but by some of his followers as well his audience would move to the Café Louvre on the then Ferdinand street (Národní trˇída now) next to the corner formed by Národní and Spálená streets (Národní 20). It was there that discussions of this Brentanian circle continued in a more relaxed and informal atmosphere to which the quiet appartment of A. Marty formed a natural transition. This was typical of the Brentanians: Brentano would spend hours with his students after a class taking them for long walks and then bringing them to his hospitable home for dinner. When one goes through memoirs of contemporaries to find out who belonged to Marty’s circle, one can find, e.g., the name of Franz Kafka and of others of the Jewish community, but thus far I have not come across any Czech name. But it was typical also of that period of time. Prague was no sleepy town any more: people would form discussion groups according to their social and cultural complexion; these would rival with each other, jealously guarding their identity. Things were changing during the First Czechoslovak Republic (i.e. between the Wars). The Brentanians formed a Franz Brentano Society (Franz Brentano Gesellschaft) which was active until 1939 when it had to transfer most of its archives to Oxford. In the 1930s, Prague became a safe haven, at least for a short while — for refugee intellectuals from Nazi Germany; everything acquired a specific flavour. At that time instead of the old Café Louvre, which disappeared in the prewar years, there appeared (1936) on the same place a modern café with the same old name — Louvre. It could be hardly a mere coincidence that the meetings of the Prague Linguistic Circle moved there (Soucˇková 1976). By this time, the landscape of Czech philosophy changed considerably. Wilhelm Dilthey (1833–1911) who may be considered as one of the Brentanians exerted a strong influence in Czech circles, e.g. on František Xaver Šalda (1867–1937), an outstanding figure in Czech cultural, esp. literary life. Phenomenology — a further step in the Brentanian direction — was taking roots; a phenomenologically oriented Philosophical Circle was formed here, one of E. Husserl’s students — Jan Patocˇka — became active at the Czech (Charles) University. The heritage of A. Marty moved at that time, as we shall see, into the focus of attention.

. A. Marty and V. Mathesius Vilém Mathesius (1882–1945) — as we know from his recollections — was not happy with what university course would offer him. His independent mind could

Anton Marty’s philosophy of language

not be let astray by attempts to underpin the Neogrammarian practice by a theory based on the traditional psychology (cf. Paul 1920) which could not — as convincingly shown by E. Husserl (Husserl 1913) — serve as a basis of any viable theory. Matheisus was intuitively feeling, since his student days, that the synchronic state of a language is a legitimate object of scientific analysis. The topics of his term papers clearly bear witness to this fact; nothing could have been more alien to the routine of the Neogrammarians than the young Mathesius’ deep interest in problems of word order in Czech, German and English — and more generally — in problems of syntax throughout his university studies. Mathesius enrolled in both Prague universities, Czech and German, because the German university could offer him more in French and esp. in English. Thus one is not surprised to find Mathesius’ name among those who attended A. Marty’s lectures on philosophy of language in the academic year 1904/1905, which was mentioned earlier. Mathesius undoubtedly knew about the meetings of the Brentanian circle at Café Louvre and this might be one of the reasons why in the 1920s he embraced Jakobson’s idea to form a discussion circle, which was to become known as the Prague Linguistic Circle. In Marty Mathesius — a young man with perspective of a scholarly career — found a sound notional basis for what he intuitively felt was opening new vistas in linguistics. Marty was not the only source of his theoretical inspirations. We know he was acquainted with Masaryk’s Základové konkrétní logiky that contained useful information on problems of methodology. He attended lectures of František Cˇáda (1865–1918) on the logic of science. As Mathesius tells in his memoirs he was very interested in new developments in natural history, esp. in the field of physiology of plants. All this proves a searching mind of a young man entering the scientific world. What must have attracted Mathesius in Marty’s theory? First of all, it was a sound basis for a non-apriori functional theory and description of language discriminating between synchrony and diachrony. Second, it was inherently content oriented; as we know, Marty’s philosophy of language was a general semasiology (allgemeine Bedeutungslehre). Because of its functional/instrumental orientation, this theory made it possible to rely on simple evidence in establishing differences and identities in meaning. Besides, it explains why the basic idea of phonology — the content relatedness — could be so readily adopted and implemented. As the functional categories were universal as opposed to their diverse formal structuring in different languages, it was possible to arrive at a better understanding of the factors organizing language into a structure-based and goal-oriented whole. Finally, Marty’s universal categories defined a framework for a contrastive analysis of different languages, no matter whether genetically related or unrelated. Thus a pendant to the historical comparative method was introduced into the linguistic arsenal. Marty’s distinction between autosemantics and synsemantics established a new hierarchy of levels of linguistic structure.





Oldrˇich Leška

It was easy for Mathesius to translate these general ideas into an effective system of descriptive functional grammar. When he was appointed head of a newly organized chair of English philology in 1908, he started his course on history of English with a then provocative introduction presenting an analysis of the contemporary English, both of its expression and content plane. It may be assumed that the internal structuring of the content plane analysis was basically that which we find in his paper O potenciálnosti jevu˚ jazykových (On the potentiality of the phenomena of language), read in the meeting of the philological section of the prestigious Czech Royal Society of Sciences and then published in the Veˇštník KCˇSN (Annals of the Czech Royal Society of Sciences; Mathesius 1911). Mathesius’ system of functional grammar contains two chapters: one dealing with devices of nomination (nominative/naming acts; functional onomatology), the other with syntagmatic procedures of sentence forming acts (functional syntax). This basic outline remained unchanged throughout Mathesius’ career; we have a good idea of what his course looked like in detail in its later version (Mathesius 1942). After what has been said about Marty’s universal categories (Vorstellungssuggestive, Ausdrücke; Urteilssuggestive, Aussagen) it was easy to recognize the source of Mathesius’ inspiration. Each of these chapters deals with the formal features of the respective functional units; in other words morphology in Mathesius’ conception is not a separate stratum in the model as in traditional grammar, it is rather a way of looking at things. One would say — in paraphrasing Marty — that morphology deals with the language specific Ausdrucksmethoden. Mathesius’ approach allows to see linguistic objects without the spectacles of traditional linguistic notions, which is exactly what he says: ‘‘I do not discover things in texts, I rather create them in the process of analysis’’. It was this non-apriori approach operating on living speech combined with contrastive method that led Mathesius to formulate the basic principles of the functional sentence perspective. In functional styles and functional types of language we recognize Mathesius’ deep rooted interest in these matters dating back to his highschool years, which was only strenghtened when he became familiar with Marty’s applied philosophy of language.

.

A. Marty and R. Jakobson, B. Trnka, J. Mukarˇovský

In the mid 1930s, when the Prague Linguistic Circle was reviewing its achievements at the close of the first decade of its existence (Mathesius 1936a) and was formulating a new program, problems of meaning and semiotics were considered pilot topics for the years to come. This was due to the changes in the overall climate which were touched upon in the proceeding section of this paper. Besides the Circle had to capitalize on its vast experience accumulated in dealing with avantgard art of the

Anton Marty’s philosophy of language

time and to bring in notional clarity. A witness to this development is the editorial published in the first issue of the journal Slovo a slovesnost (Word and verbal arts). A pansemiotic outlook of this editorial may lead to A. Marty as the most probably source of inspiration even if there were no references to his work. There was much more than just programmatic announcement. Starting with Roman Jakobson is by no means a random choice. He had read Marty (Marty 1908) before he came to Prague 1920. We know Jakobson’s unsatiable curiosity which could have led him in the right direction, to Marty’s work. But it could have been the Moscow philosopher Gustav Špet who recommended it to Jakobson. Špet, a propagator of Husserl’s ideas, held seminars on phenomenology at the Moscow University and Jakobson was one of his enthusiastic students. It is characteristic of this period that semiotic thinking finds its way into such well established disciplines as phonology which was soon to become known in its canonical version presented by Trubetzkoy in his Grundzüge (Trubetzkoy 1939). Jakobson, who must be considered — besides Trubetzkoy — one of the founders of the Praguian phonology, undertook a thorough revision of its basic notions from the semiotic point of view in two lectures delivered in Copenhague in 1939 after he was forced by circumstances to leave Czechoslovakia and was in search for a new home (Jakobson 1939). It is very interesting to see the superficial sign conception of language replaced by a rigorous analysis introducing semiotic thinking into linguistic professional practice; it is then that it becomes a powerful analytical tool. Jakobson makes de Saussure’s notion of a purely differential character of structural units a point of departure for his analysis of what phonology is. Thus far his position of 1939 is identical with what he announced at the beginning of his theses on phonology at the Congress in The Hague in 1928; in this sense, Jakobson was right by pointing to de Saussure as to the father of phonology. A phoneme is a specific kind of sign, different from other signs, because its purely distinctive (negative) character is its only semiotic value; phonemes are signs of signs, and as constituent components of the expression plane of a sign they are signs on signs as well. In passing from phonemes over to contextually conditioned sound units (variants), one is passing to signs of a higher order because these signs are not only distinctive (i.e. negative), but indexical as well (i.e. positive) in referring to their conditioning context. A phonology understood as pure phonemics is unable to differentiate properly these two kinds of signs as it deals with distinctive (negative) signs only. Stylistic sound features (Lautstylistik) were reinterpreted; they are analogons to morphemes as far as content is concerned (they have a positive content component as morphemes have); their expression plane however cannot be analyzed into expression constituents, i.e. phonemes (as it is the case with morphemes). Similarly, junctures and all of the repertory of phrase phonology may be interpreted





Oldrˇich Leška

as signs related to signs, however with a positive semiological value which makes them different from phonemes. This summary of Jakobson’s reinterpretation of basic notions of phonology shows how the semiotic approach cleared the ground; in this light, many discussions appear useless and misleading because they are not language oriented but aim at solving general ontological problems which do not relate to phonology only; those engaged in these discussions must be found guilty of trespassing: phonology is a linguistic discipline, ontology — dealing with modes of existence — belongs to philosophy. The accent put on the instrumental, action related character brought Jakobson close to J. R. Firth (Firth 1964) and C. Bally (Bally 1944), which could have triggered new developments in the Praguian phonology. However, as mentioned above, it was too late to think of an unhindered organic development of the Praguian phonology: those who used to assemble in the meeting room at the Café Louvre were not to meet again after the war. Jakobson dealt with the same topic again in his eventual host country — at the French École libre in New York where he found a position thanks to his friend Lévy Strauss (published under the title Six leçons sur les sons et le sens). Jakobson’s interest in semiotics and phonology was not epizodic. As a matter of fact it dated back to the time of Jakobson’s first appearances on scientific scene when the problem was approached in a way that combined youthful fascination by avantgard poetry with curiosity of a searching mind: were the possibilities of correlating sound expression with content in the poetic use of language really limitless as they seemed to be? It was Velemir Xlebnikov by whom Jakobson was attracted and who was for Jakobson, esp. on his recollections of old age a powerful source of inspiration (Jakobson 1921, Holenstein 1988). During his American years, Jakobson became acquainted with the writings of C. S. Peirce and found them very stimulating. He adopted Peirce’s view according to which any linguistic sign may be included in a synonymic series whose members relate to each other differing in their explicitness and precision; this series may be eventually extended. By this metalinguistic procedure the problem of meaning was considered resolved. There was another point in Peirce’s semiotics which was important — or rather welcome — to Jakobson. Peirce teaches that one may distinguish indexical sings based on contiguity (cf. come here), iconic signs motivated by similarity (cf. swalow tailed coat) and symbols which are conventional (cf. hat). This does not mean of course that a sign repertory consists of three separate compartments; these sign types should be understood rather as aspects that may be combined in a sign in different proportions. This was a very important hint for Jakobson who engaged throughout his scientific life in a polemic with de Saussure’s basic principles, one of them being the arbitrary character of the linguistic sign. It is clear that by expanding the realm of iconicity in language, conventionality (arbitrariness) should shrink. This is what Jakobson was doing in his reprise of an old theme he

Anton Marty’s philosophy of language

was busy with in Prague in the mid 1930s. At that time, Jakobson established a system of invariant meanings which organized the repertory of the Russian verbal forms (Jakobson 1932) and that of the Russian case forms (Jakobson 1936) into a hierarchical system of binary oppositions. Thus Jakobson’s structural morphology formed a parallel to (was isomorphic with) the phonological system which was established in the opening period of the Prague Linguistic Circle and which made the linguistic world speak of the Prague School. In America Jakobson (Jakobson 1956, 1958) changed his approach, esp. his treatment of verbal forms; but these differences are not of major interest in connection with our topic. What must however attract attention to Jakobson’s American variations on the Prague theme is his thorough analysis of the expression : : content relationships that one does not find in the Prague version. Iconicity is not limited to such simple cases as illustrated above; it obtains in paradigmatic series as well that exhibit a clear mirroring of content features in the sound, shape on the expression level. These cases are called by Peirce diagrams (diagrammatic iconicity). Thus in Russian, e.g. the peripheral cases D pl and I pl exhibit diagrammatic iconicity in correlating the content features directional plural with the distinctive feature nasality on the expression level. If one takes substantives, then all peripheral cases plural (I D L pl) exhibit desinence initial =a- , i.e. the feature compactness (in contradiction to the adjectives that have a desinence initial =iwhich is diffuse). The sound features are in these cases indices of the respective content feature. Diagrammatic iconocity restricts the arbitrariness of the linguistic sign indeed. At the same time, this should be understood as Jakobson’s attempt at proving the psychological relevance of the distinctive features, which played an important role in his polemic with another of de Saussure’s principles — that of linearity of the linguistic sign; since, e.g., a phoneme is according to Jakobson a bundle (and not a sequence) of distinctive features, then the structure of a sign is not linear in all of its constituent elements. It should be noted that in the same direction, i.e. towards proving the psychological relevance of distinctive features, there point Jakobson’s analysis of poetry (cf. the analysis of C. Baudelaire’s poem Les chats written in collaboration with Lévy Strauss): in poetry, because of the specific principle of structuring a poetic text (paradigmatic relationships are being projected onto the syntagmatic axis), the reader may experience linguistic structural elements that otherwise, in a nonpoetic use of language, would pass unnoticed, including both phonological and content (grammatical) distinctive features. The unity of the Prague School should not be understood in a doctrinaire sense: it was a natural result of striving towards a common goal along individual paths. This is clearly demonstrated by the responses to the spirit of the Circle’s new program. Quite significant is the reaction of V. Mathesius.





Oldrˇich Leška

When Vladimír Skalicˇka published his structural typology under a rather misleading title Zur ungarischen Grammatik (Skalicˇka 1935), Mathesius recognized the importance of Skalicˇka’s book and concluded his review by saying that the character of modern linguistics may be propely understood only when the functional approach is complemented by a structural analysis (Mathesius 1936b); this may be seen as an indication of a shift in his basic position. As we know, Mathesius used — in matters of meaning — to rely on simple evidence; later he recognized the importance of this problem (Mathesius 1942). Bohumil Trnka (1895–1984), one of the key figures in the Circle, was building up systemtically a stratificational model of laguage since the early 1930s. Since the early 1940s he was interested in what he termed experiencing language, i.e. in the process in which the linguistic structure becomes part of the activities of its user (Trnka 1948) who is its discoverer and caretaker. In Jakobson’s Festschrift published on the occasion of his 70th birthday, Trnka expounded his theory of meaning which can be characterized as a bridge between Marty’s conception of meaning and de Saussure’s valeurs (values) manifested in language use by significations (Marty: Bedeutungsfunktionen; Trnka: meaning functions). Exactly as a dollar is a dollar, a word is a word; Bühler (Bühler 1934) says that one cannot buy anything with a word whereas one can with a dollar. This is simply not true: one can acquire a piece of reality with a word, i.e. one can grasp it by a creative use of a word, which is made possible thanks to the potential openness of the word as a structural element to further concretizations. Above we have mentioned the vast experience, esp. with avantgard art that has accumulated in the Circle. In verbal art, the Russian formalism was influential, but it had became clear quite early that this conception with the notions like device, make-up had serious limitations. It calls forth an idea of blueprint which allows for variations within certain limits; it presents a poetic work as a static construction and could not account for its inherent dynamism. A real turning point in Prague semiotics is marked by Jan Mukarˇovský’s functional structuralism. Its principles were formulated in his studies Umeˇní jako semiologický fakt (Art as a semiotic fact; Mukarˇovský 1934), Estetická funkce, norma a hodnota jako sociální fakty (Aesthetic function, norm and value as social phenomena; Mukarˇovský 1936) and Místo estetické funkce mezi ostatními funkcemi (Aesthetic function in its position among other functions; Mukarˇovský 1942). Mukarˇovský gives a definition of the term function: facing the outside world which may be either immediate (nonsemiotic behavior, plain action — practice) or mediate (semiotic behavior — semiotic). A symbolic function is directed towards an object while the aesthetic function implies that a sign was not put to use as a tool, as an instrument. An aesthetic sign does not reflect a certain reality, does not relate to a certain reality. On the contrary — reality in its totality finds a reflection in this sign and the subject reads its image into it according to his internal

Anton Marty’s philosophy of language

complexion. Mukarˇovský’s system of pragmatic functions differs from the functions established earlier by Bühler in his Organonmodel (Bühler 1934) or the communicative functions set up by R. Jakobson (Jakobson 1962, 1972). There is a striking similarity between Mukarˇovský’s and Marty’s conception of a poetic work (Marty 1908). According to Marty, a poetic text is defined by lack of the referential dimension. Reference reaching out beyond the text would be in conflict with the constitutive selfcontained character of the thus defined space. A poetic text is for A. Marty a fiktive Rede. It is a global representation suggestive of a fictious world/space; this is why all linguistic entities — no matter what their character and size are — become components of the inner form of a thus structured whole. Linguistic material being relieved of its primary symbolic function can be perceived as representing pure value. The pliability of a poetic text/work and its multifarious character may be understood as conditioned by its very nature.

References Bally, C. 1944. Linguistique générale et linguistique française. Bern. Brentano, F. 1874. Psychologie vom empirischen Standpunkte. Reprint: Hamburg 1955. Bühler, K. 1909. (= Review of A. Marty, Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie l. Halle 1908). Göttingische gelehrte Anzeigen 171 (1909) 947–79. Bühler, K. 1934. Sprachtheorie: Die Darstellungsfunktion der Sprache. Jena: G. Fischer. Firth, J. R. 1964. Papers in Linguistics 1934–1951. Oxford: Oxford Univ. Press. Holenstein, E. (ed.) 1988: Roman Jakobson, Semiotik: Ausgewählte Texte 1919–1982. Frankfurt am Main: Suhrkamp. Husserl, E. 1913. Logische Untersuchungen. 2nd ed. Halle: Niemeyer. Jakobson, R. 1921. ‘‘Novejšaja russkaja poèzija’’. Praha 1921, — Reprinted in SW V, 299–354. — German version: Die neueste russische Poesie. Texte der russischen Formalisten II. Texte zur Theorie des Verses und der poetischen Sprache. Hg. von W.-D. Stempel. München 1972. Jakobson, R.1932. ‘‘Zur Struktur des russischen Verbums’’. Charisteria 1932, 74–84. (Reprinted in Vachek 1964, 347–59; SW II, 3–15). Jakobson, R. 1936. ‘‘Beitrag zur allgemeinen Kasuslehre (Gesamtbedeutungen der russischen Kasus).’’ TCLP 6, 240–88. — Reprinted in SW II, 23–71. Jakobson, R. 1939. ‘‘Zur Struktur des Phonemes’’. Zwei an der Universität von Kopenhagen in Mai 1939 gehaltene Vorträge. SW I, 280–310. — Reprinted under the title ‘‘Die eigenartige Zeichenstruktur des Phonems’’. See: Holenstein 1988. Jakobson, R. 1956. Shifters, Verbal Categories and the Russian Verb (=Russian Language Project, Dept. of Slavic Languages and Literatures). Cambridge, Mass: Harvard University. — Reprinted in SW II, 130–47. Jakobson, R. 1958. ‘‘Morfologicˇeskije nabljudenija nad slavjanskim sklonenijem (Sostav russkich padežnych form).’’ In American Contributions to the Fourth International





Oldrˇich Leška

Congress of Slavicists, Moscow, Sept. 1958, 127–56. The Hague: Mouton. — Reprinted in SW II., 154–83. Jakobson, R. 1961. ‘‘Linguistics and communication theory.’’ In Proceedings of Symposia in Applied Mathematics, vol. XII, Structure of Language and its mathematical Aspects. American Mathematical Society, 245–52. — Reprinted in SW II, 570–9, Jakobson, R. 1962. ‘‘Les Chats’’ de Charles Baudelaire. L’Homme 2: 5–21. — Reprinted in SW III. Jakobson, R. 1972. ‘‘Verbal communication’’. Scientific American 227/3, 73–80. Jakobson, R. 1976. Six leçons sur le son et le sens. Paris: Editions de Minuit. Kraus, O., Eisenmeier, J. and Kastil, A. (eds) 1916–1920: see Marty 1916–1920. Marty, A. 1875. Ursprung der Sprache. Göttingen. Marty, A. 1908. Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie. Halle. Marty, A. 1916–1920, Gesammelte Schriften I. Bd. 1, 2. II. Bd. 1, 2. Hg. von Oskar Kraus et al. Halle: Niemeyer. Marty, A. 1940. ‘‘Grundfragen der Sprachphilosophie’’. Eine Vorlesung A. Martys aus dem Jahre 1904. ‘‘Psyche und Sprachstruktur’’. Nachgelassene Schriften von Anton Marty, hg. von Otto Funke. Bern. Marty, A. 1950. Gesammelte Schriften II, ‘‘Satz und Wort’’. Hg. von Otto Funke. Bern. Erste Ausgabe 1925. Reichenberg i. Böhmen:Stiepel; Gesammelte Schriften III. ‘‘Über Wert und Methode einer allgemeinen beschriebenden Bedeutungslehre’’. Hg. von Otto Funke. Bern. Erste Ausgabe 1926. Reichenberg i. Böhmen: Stiepel. Masaryk, T. G. 1885. Základové konrétní logiky. [An outline of a concrete logic]. Masaryk, T. G. 1895. Cˇeská otázka. [Topical problems of Czech national life]. Mathesius, V. 1911. ‘‘O potenciálnosti jevu˚ jazykových’’ [On the potentiality of the phenomena of language]. In Veˇstník Královské cˇeské spolecˇnosti nauk: Trˇída filosoficko-historickojazykoveˇdná 2, 1–24. — Reprinted in Vachek 1970; Mathesius 1982, 9–28. English translation in Vachek 1964, 1–32. Mathesius, V. 1936a. ‘‘Deset let Pražského lingvistického kroužku’’ [Ten years of the Prague Linguistic Circle]. Slovo a slovesnost 2, 137–45. — English translation in Vachek 1966, 137–51. Mathesius, V. 1936b. ‘‘Pokus o theorii strukturální mluvnice’’ [An attempt at formulating a theory of structural grammar] (= Review of Skalicˇka 1935). Slovo a slovesnost 2, 47–54. Mathesius, V. 1942. ‘‘O soustavném rozboru gramatickém’’ [On ways of a systematic grammatical analysis of language]. Slovo a slovesnost 8, 88–92. — Reprinted in Mathesius 1947, 157–75; 1982, 50–9. Mathesius, V. 1947. Cˇeština a obecný jazykozpyt. Soubor statí. [Czech language and general linguistics: Collection of papers]. Praha: Melantrich. Mathesius, V. 1982. Jazyk, kultura a slovesnost. [Language, culture and verbal arts]. J. Vachek (ed.). Praha: Odeon. Meillet, A. 1909. (= Review of A. Marty, Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie 1. Halle 1908.) Bulletin de la Société de linguistique de Paris 16, No 57 (1909), 60–1. Mukarˇovský, J. 1934. ‘‘L’art comme fait sémiologique’’. In Actes du Huitième congrès

Anton Marty’s philosophy of language

international de philosophie à Prague 2–7 Septembre 1934, 1065–1072. Praha 1936. — Czech version ‘‘Umeˇní jako semiologický fakt’’ in Mukarˇovský 1966, 85–8. Mukarˇovský, J. 1936. Estetická funkce, norma a hodnota jako sociální fakty. [Aesthetic function, norm and value as social phenomena]. Praha: F. Borový. — Reprinted in Mukarˇovksý 1966, 17–54; 1971, 7–65. Mukarˇovský, J. 1942. ‘‘Místo estetické funkce mezi ostatními funkcemi’’. [Aesthetic function and its position among other functions] (A lecture delivered in the Circle on November 30, 1942). First published in Mukarˇovský 1966, 65–73. Mukarˇovský, J. 1966 [1971]. Studie z estetiky. Výbor estetických prací J. Mukarˇovského z let 1931 až 1948. [Studies in aesthetics: Selected papers 1931–1948]. K. Chvatík (ed.). Praha: Odeon 1966. 2nd ed. 1971. Paul, H. 1920. Prinzipien der Sprachgeschichte. Halle: Niemeyer. Seuren, A. M. 1985. Discourse Semantics. B. Blackwell (ed.). Oxford. Soucˇková, M. 1976. ‘‘The Prague Linguistic Circle: A Collage’’. In Sound, Sign and Meaning. L. Matejka (ed). Ann Arbor: Univ. of Michigan, 1–5. Skalicˇka, V. 1935. Zur ungarischen Grammatik (= Práce z veˇdeckých ústavu˚, 39). Praha : Filos. fak. University Karlovy. — Extract ‘‘Morphem und Sema’’ reprinted in Vachek 1964, 320–8. SW: The Selected Writings of Roman Jakobson. 1962–1988. The Hague, Paris & New York / Berlin, Amsterdam & New York: Mouton SW I: Phonological Studies (= Selected Writings I). 1st ed. 1962; 2nd, expanded ed. 1971. SW II: Word and Language (= Selected Writings II). 1971. SW III: Poetry of Grammar and Grammar of Poetry (=Selected Writings III) ed. by S. Rudy. 1981. SW V: On Verse, Its Masters and Explorers (= Selected Writings V). S. Rudy and M. Taylor (eds). 1979. TCLP: Travaux du Cercle linguistique de Prague. 1–8. 1929–1939. TCLP 6, 1936: Études dédiées au Quatrième congrès de linguistes. TCLP 7: see Trubetzkoy 1939. Trnka, B. 1948. ‘‘Jazykozpyt a myšlenková struktura doby’’ [Linguistics and the ideological Structure of the Period]. Slovo a slovesnost 10, 73–80. — English translation in Trnka 1982, 49–60. Trnka, B. 1982. Selected papers in structural linguistics: Contributions to English and general linguistics written in the years 1928–1978. Ed. by V. Fried. Berlin — New York — Amsterdam: Mouton. Trubetzkoy, N. S. 1939. Grundzüge der Phonologie (= TCLP 7). Praha. Vachek, J. ed. 1964. A Prague School Reader in Linguistics. Bloomington, Ind.: Indiana Univ. Press. Vachek, J. (ed) 1966. The Linguistic School of Prague: An introduction to its theory and practice. Bloomington, Ind.: Indiana Univ. Press. Vachek, J. (ed) 1970. U základu˚ pražské jazykoveˇdné školy. [Foundations of the Prague linguistic school]. (= Prameny cˇeské a slovenské lingvistiky: Rˇada cˇeská, 1). Praha: Academia.



Die Typologie des Ungarischen

Vladimír Skalicˇka Übersetzt und eingeleitet von Markus Giger Am Anfang der sprachtypologischen Forschungen von Vladimír Skalicˇka und damit der ›Prager Typologie‹ steht die Beschäftigung mit dem Ungarischen (Skalicˇka 1935). Dennoch kehrte Skalicˇka nur einmal ausführlich zur Frage der typologischen Stellung des Ungarischen zurück, im Aufsatz »A magyar nyelv typológiája«, welcher im Sammelband »A magyar nyelv története és rendszere« (Redaktion S. Imre, I. Szathmári) 1967 in Budapest erschien. In diesen dreißig Jahren hatte sich Skalicˇkas Typologiekonzeption weiterentwickelt und vervollkommnet; es mag hier genügen, an die Erweiterung der syntaktischen Komponenten in Skalicˇka (1942) oder die Einführung des introflexiven Typus in Skalicˇka (1946– 1948) zu erinnern. In die Entstehungszeit des hier übersetzten Aufsatzes fallen die Reformulierung des Prinzips vom Sprachtyp als Konstrukt (Skalicˇka 1966), die typologische Definition des Begriffs ›Inkorporation‹(Skalicˇka 1968a) sowie die Auseinandersetzung mit der Greenbergschen Wortfolgetypologie (Skalicˇka 1968b), welche alle in diesem Text eine wichtige Rolle spielen. Aus diesen Gründen schien es wünschbar, den vorliegenden Aufsatz in einer Übersetzung breiter zugänglich zu machen, wobei sich das Deutsche in Analogie zu Skalicˇkas weiteren Arbeiten über die Typologie finno-ugrischer Sprachen anbot (vgl. neben Skalicˇka 1935 zum Ungarischen und Finnischen auch Skalicˇka 1968c, 1968d, 1975). Der vorliegende Aufsatz ist — wie so mancher Text von Vladimír Skalicˇka — sehr knapp und dicht gehalten, so daß dem mit seiner Typologiekonzeption nicht vertrauten Leser möglicherweise manches nicht auf Anhieb verständlich sein wird. Ich möchte es deshalb nicht versäumen, auf einige neuere zusammenfassende Darstellungen zu verweisen (Skalicˇka 1979; Skalicˇka/Sgall 1994; Sgall 1995). Im Text wurden Übersetzungen der ungarischen Beispiele beigefügt (wo nötig auch Interlinearübersetzungen) sowie die Literaturangaben ergänzt. Alle Beifügungen stehen in []. Ein Verzeichnis der verwendeten Abkürzungen findet sich am Ende. Für einige Hinweise danke ich Lucie Hašová (Prag) und Zsuzsanna Kovács (Zürich). Markus Giger (Zürich und Prag)

In den letzten Jahren sind viele neue linguistische Theorien entstanden. Wenn auch ich heute eine neue vorbringe, kann ich mich nur dadurch rechtfertigen, daß diese Theorie, von der ich sprechen will, von verschiedenen Gesichtspunkten aus wichtig ist. Die Typologie ist kein gänzlich neuer Zweig der Linguistik — das ist bekannt. Offensichtlich ist aber auch, daß wir uns mit der Typologie anders beschäftigen müssen als bisher.

 Vladimír Skalicˇka

Bislang sind viele typologische Schulen entstanden. Aber diese typologischen Schulen hatten einen gemeinsamen Fehler: Die Typologen wollten immer die charakteristischen Eigenarten der Sprachen herausheben — und daraus konnte kein System entstehen. Noch im Band mit dem Titel »Universals of Language« [Greenberg 1966a] betonten die amerikanischen Linguisten, daß die Linguistik nur mit empirischer Methode arbeiten kann. Daraus würden sie folgern, daß die Aufgabe des Typologen das Sammeln der Eigenarten sei [vgl. Greenberg 1966b:xii, Hockett 1966: 1–5]. Von einem System kann in diesem Fall keine Rede sein. Aber die Lage ist nicht dermaßen aussichtslos. In der Linguistik ist auch die deduktive Methode möglich — und in der Typologie ebenfalls. Die Zahl der Eigenschaften von Sprachen ist nicht unendlich. So unterscheidet sich z. B. die Länge des Wortes in Einzelsprachen. Aber das Wort kann nicht weniger als ein Morphem enthalten, und es kann nicht länger sein als der Satz. Ein anderes Beispiel: Die Verbreitung der Anomalie (das sind etwa Synonymie und Homonymie) unterscheidet sich je nach Sprache: Im Latein ist sie größer als im Ungarischen, so gibt es z. B. im Latein viele Deklinationen, im Ungarischen dagegen nur eine. Diese Unterschiede haben Grenzen: Die Zahl der Unterschiede kann nicht größer sein als die Zahl der Wörter, und sie kann nicht kleiner sein als null (der letztere Fall ist im Volapük verwirklicht). Daraus ergibt sich, daß die Zahl der möglichen Sprachen nicht unendlich ist, aber doch um vieles größer als die Zahl der verwirklichten Sprachen. Demnach ist es die primäre Aufgabe des Typologen festzustellen, welchen Platz eine Einzelsprache unter sämtlichen möglichen Sprachen einnimmt. Die möglichen Sprachen sind nicht alle gleichermaßen wahrscheinlich. Es gibt solche, die gänzlich unwahrscheinlich sind (z. B. eine solche Sprache, in der jedes Wort eine besondere Deklination hätte), es gibt wahrscheinlichere und weniger wahrscheinliche. Die verwirklichten natürlichen Sprachen gehören unter die wahrscheinlicheren. Die Regeln, welche die wahrscheinlicheren Sprachen von den weniger wahrscheinlichen trennen, sind zahlreich und kompliziert. Der größte Teil dieser Regeln verbindet die sogenannten Typen miteinander. Deshalb ist die einfachste Beschreibung einer Sprache die Feststellung dessen, zu welchem Typus die Sprache gehört. Die hundertprozentige Verwirklichung mancher Typen ist nicht wahrscheinlich und in vielen Fällen unmöglich. Deshalb kann man in Einzelsprachen mehrere Typen beschreiben. Wenn ich im weiteren versuche, das Ungarische typologisch zu beschreiben, bin ich mir bewußt, daß diese Beschreibung nur skizzenhaft und oberflächlich ist, keineswegs endgültig. Ich möchte nur zeigen, auf welche Weise eine Sprache typologisch beschrieben werden kann. Ich muß festhalten, daß ich hier nur von der sogenannten Makrotypologie spreche, welche sämtliche Sprachen, d. h. die existierenden und nicht existierenden, die möglichen und unmöglichen, die verwirklichten und noch nicht verwirklichten behandelt, also nicht von der

Die Typologie des Ungarischen 

Mikrotypologie, welche sich nur auf miteinander verwandte Sprachen erstreckt, z. B. auf die finnougrischen Sprachen. Der Auffassung der herkömmlichen Typologie gemäß gehört das Ungarische unter die sogenannten agglutinierenden Sprachen. Das heißt nichts anderes als daß die Sprachen in der herkömmlichen Typologie in vier, fünf Gruppen aufgeteilt werden und daß das Ungarische zu einer dieser Gruppen, zur agglutinierenden Gruppe gehört. Den neueren Typologien gemäß ist für das Ungarische das Wortfolgeschema SOV charakteristisch (Greenberg [1966c: 109]) oder ein geringes Passiv (Šaumjan [vgl. Šaumjan 1971: 449–55]) usw. Meiner Meinung nach ist das Wichtigste in der Typologie der Typus. Deshalb beginnt die typologische Beschreibung mit den Typen. Im Ungarischen ist der agglutinierende Typus der wichtigste. Auch im Ungarischen ist dieser Typus nicht rein, da ja ein reiner Typus unmöglich ist. Am wichtigsten ist im agglutinierenden Typ das Verhältnis zwischen sogenannten Semantemen und formalen Elementen. Im reinen (nicht verwirklichten) Typus wäre die Verkettung der Morpheme folgende: —...¦—...¦—...¦—... Das heißt, daß das Wort eindeutig charakterisiert ist, daß sich in einem Wort ein und nur ein Semantem befindet, daß keine formalen Wörter existieren (Artikel, Postpositionen, Pronomina, Konjunktionen), und es gibt keine Komposita. Dieses Schema wird im Ungarischen in großem Maße realisiert. Das Ungarische hat viele wort- und formbildende Suffixe und wenig formale Wörter. Es hat viel mehr formbildende Suffixe als etwa das Latein; das Russische etwa hat wiederum mehr als das Finnische. In vielen Fällen entspricht ein ungarisches Suffix einer anderen Konstruktion in den übrigen Sprachen: barát-a-i-m-nak [Freund-poss-pl-1.sg-dat] — to my friends, amicis meis; vár-hat-lak [wartkönn-1.sg subj/2. sg obj]- ich kann dich erwarten; autó-z-t-am [Auto-suff-prät1.sg] — ich fuhr Auto. Damit hängt eine weitere Erscheinung zusammen: Es entwickelt sich kein großer Unterschied zwischen den wort- und formbildenden Suffixen. Im reinen agglutinierenden Typus, im »Konstrukt«, existiert dieser Unterschied überhaupt nicht. Wenn er im Ungarischen besteht, so heißt das, daß ein anderer Typus, und zwar der flektive Typus sich mit dem agglutinierenden Typus verbindet. Damit hängt wiederum eine weitere Eigenschaft zusammen: Die Wortarten unterscheiden sich nicht so deutlich voneinander wie in den übrigen Sprachen. Im agglutinierenden Konstrukt gibt es keine Wortarten. Unter den verwirklichten Sprachen gibt es solche — z. B. das Tscherkessische — , in denen die Wortarten nur schwer zu unterscheiden sind. Im Ungarischen finden sich Spuren eines Zustandes, in dem die Grenze zwischen Verb und Nomen undeutlicher war als heute; vgl. z. B. die sogenannten

 Vladimír Skalicˇka

‘igenévszók’ [»Verbonomina«] (les [›auflauern‹] und les [›Hinterhalt‹], nyom [›drücken‹] und nyom [›Spur‹] usw.). Die infiniten Verbformen, d. h. die Übergangswortarten (Infinitiv, Partizip, Adverbialpartizip), spielten früher eine viel wichtigere Rolle als jetzt. Dieses Phänomen ist sehr wichtig vom syntaktischen Standpunkt aus: Wenn die infiniten Verbformen viel wichtiger waren, dann spielten wiederum die Nebensätze eine kleinere Rolle als heute. Es ist offensichtlich, daß in dieser Frage die ungarische Sprache nicht dermaßen agglutinierend ist wie etwa die türkische. Aber es gibt im Ungarischen viele andere agglutinierende Erscheinungen: das nominale Prädikat, die geringe Kongruenz (a szép ház-nak [def schön Haus-dat ›dem schönen Haus‹], három ló [drei Pferd ›drei Pferde‹] usw.), die grammatikalisch gebundene Wortfolge (a szép ház [›das schöne Haus‹], a ház [›das Haus‹], a ház ello˝tt [›vor dem Haus‹]), die aus vollen Silben bestehenden Endungen (-ban [iness], -ból [elat], -hoz [allat], -nak [dat], -tok [2.pl] usw.), die Vokalharmonie usw. Die Wortbildung ist überwiegend agglutinierend: Suffixe sind in großem Maße gebräuchlich (vgl. auch solche Wörter wie vadászik [›jagen‹ (< vad ›Wild‹)], halászik [›fischen‹ (< hal ›Fisch‹)], kártyázik [›Karten spielen‹ (< kártya ›Karte‹)], autózik [›autofahren‹ (< autó ›Auto‹)], dohányzik [›rauchen‹ (< dohány ›Tabak‹)] usw.). Aus alldem wird offensichtlich, daß der agglutinierende Typus sehr wichtig ist im Ungarischen, aber daraus ist nicht alles erklärbar. Jetzt werden wir die Untersuchung der anderen Typen versuchen. Der flektive Typus ist gemäß der herkömmlichen Typologie charakteristisch für die indoeuropäischen Sprachen. Wenn wir nun diesen Typus im Ungarischen nachzuweisen versuchen, dann postulieren wir damit nicht etwa indoeuropäischen Einfluß, z. B. Einfluß der slavischen Sprachen. Wir wollen nur nachweisen, daß die ungarische Sprache mit Sicherheit in Teilen zum flektiven Typ gehört. Sehr charakteristisch ist für den flektiven Typ, daß der Unterschied zwischen wort- und formbildenden Suffixen wichtiger ist als jener zwischen Semantemen und formalen Elementen. Dieses Prinzip ist im Ungarischen im Detail verwirklicht. So gibt es z. B. mehr Alternationen zwischen dem Wortstamm und der Endung als zwischen der Wurzel und dem Wortbildungssuffix: vgl. szamár [Esel-sg] –szamar-ak [Esel-pl], tehén [Kuh-sg]–tehen-ek [Kuh-pl], kehely [Kelch-sg]–kelyhek [Kelch-pl], sarok [›Ecke‹]–sark-a [Ecke-poss ›seine Ecke‹] usw. Im agglutinierenden Typus kann das Suffix nach einer Endung stehen, z. B. türkisch ben-im-ki [ich-gen-da ›meines da‹]. Das ist unbekannt im flektiven Typus und ebenso im Ungarischen. Wie gesagt unterscheiden sich die Wortarten im Ungarischen voneinander, aber nicht so rein wie in anderen Sprachen. Z. B. unterscheidet sich das Nomen vom Verb: Die sogenannten ‘igenévszók’ [»Verbonomina«] sind nur in Spuren vorhanden, die infiniten Verbformen kommen nicht sehr oft vor. Das nominale Prädikat existiert im Ungarischen, aber seine

Die Typologie des Ungarischen 

Verbreitung ist geringer als in anderen Sprachen. Kongruenz tritt im Ungarischen selten auf, aber dennoch existiert sie, in erster Linie zwischen Subjekt und Prädikat. Sehr wichtig unter typologischem Gesichtspunkt ist die ungarische Wortfolge. Die grammatikalisch gebundene Wortfolge kommt nur in einigen Fällen vor (a ház elo˝tt [›vor dem Haus‹]). Viel wichtiger ist im Ungarischen ein anderer Wortfolgetyp, der den Akzent audrückt (jártam az iskolába [›ich ging zur Schule‹] gegenüber az iskolába jártam [›ich ging zur Schule‹]). Voraussetzung für diese Wortfolge ist, daß der syntaktische Zusammenhang der Wörter ganz klar wird. Dies ist für den flektiven Typus charakteristisch. Für den agglutinierenden Typus ist dagegen die grammatisch gebundene Wortfolge charakteristisch, welche in der Regel auch die grammatischen Verhältnisse ausdrückt. Diese Wortfolge ist in anderen Sprachen verwirklicht, z. B. im Türkischen oder im Mongolischen. Die Erscheinungen des flektiven Typus sind somit nicht häufig im Ungarischen. Es ist aber sehr wichtig, auf sie aufmerksam zu werden: Der Charakter der ungarischen Sprache besteht in einer eigenartigen Verbindung der agglutinierenden und flektierenden Elemente. Der isolierende Typus (den wir aus der englischen oder französischen Sprache kennen) ist nicht bedeutend im Ungarischen. Formwörter sind nicht häufig (es gibt den Artikel, zusammengesetzte Verbformen: fogok várni [›ich werde warten‹], vártam volna [›ich hätte gewartet‹]; es gibt Modalverben und Modalwörter: kell [›es ist nötig, man muß‹], szabad [›es ist erlaubt, man darf‹], szoktam [›ich pflege (zu tun)]). Eine weitere Erscheinung des isolierenden Typs, die Verwendung von Nebensätzen, hängt offensichtlich mit dem Vorkommen des flektiven Typus zusammen: Nebensätze bedeuten Abbau der infiniten Verbformen, das heißt deutlichere Unterscheidung der Wortarten. Besonders wichtig ist in der ungarischen Sprache der sogenannte polysynthetische Typus, das heißt ein solcher Typus, in dem kein Unterschied besteht zwischen semantischen und nicht-semantischen Morphemen. Der polysynthetische Typus wäre in reiner Form die bloße Verkettung von Semantemen. Dieser reine Typus ist wiederum unmöglich. Der polysynthetische Typus ist im Chinesischen und in anderen ostasiatischen Sprachen verwirklicht. Unter den europäischen Sprachen hat er sich im Ungarischen am stärksten entwickelt. In der ungarischen Wortbildung ist der Anteil des polysynthetischen Typus sehr stark. Die Komposition ist in allen finnougrischen Sprachen wichtig, aber im Ungarischen hat sie sich am weitesten entwickelt. Uns ist nur ein Gebiet der nominalen Wortbildung bekannt, wo die ungarische Sprache weniger Komposita hat als eine andere Sprache: Dieses Phänomen hängt offensichtlich mit dem agglutinierenden Charakter des Ungarischen zusammen: Es gibt Komposita — z. B. im Deutschen — , in welchen die Funktion des zweiten Gliedes sehr abstrakt ist — eigentlich eine Suffixfunktion. In solchen Fällen steht im Ungarischen ein

 Vladimír Skalicˇka

Suffix: dt. Kaufmann–ung. keresked-o˝ [< kereskedik ›handeln‹], dt. Bergmann– ung. bányá-sz [< bánya ›Bergwerk, Grube‹]. In denjenigen Sprachen, in welchen der polysynthetische Typus wichtig ist, aber keine dominierende Rolle spielt, wie z. B. in einigen amerikanischen und paläoasiatischen Sprachen, kommt noch eine andere Eigenschaft vor, die sogenannte Inkorporation. Syntaktisch zueinandergehörende Wörter verschmelzen oft zu einem Wort, so bilden z. B. das Objekt und das Prädikat, die Adverbialbestimmung und das Prädikat, das regierende und das regierte Wort zusammen ein neues Wort. Die genaue Interpretation ist oft nicht einfach, so ist z. B. im Giljakischen (Nivch) die ganze Sache strittig, und wahrscheinlich können wir nur von unvollkommener Inkorporation (Subinkorporation) sprechen. Meiner Ansicht nach sind Beispiele für diese Inkorporation auch in der ungarischen Sprache leicht zu finden. Das eine Beispiel sollen die Verbalpräfixe geben. Ich weiß sehr wohl, daß auch dies nicht eindeutig ist; oftmals ist das Verbalpräfix überhaupt nicht nennbar als Adverbialbestimmung: el-nézni [›ansehen‹], ki-vívni [›erkämpfen‹], meg-találni [›auffinden‹]. Oftmals hat das Verbalpräfix nicht Semantemcharakter, z. B. in megtalálni. Aber die Zahl der Verbalpräfixe, welche eher zu den Semantemen gehören, ist groß: össze- [›zusammen-‹], vissza- [›wieder-‹] usw. Es gibt noch solche wie hátra- [›zurück‹] (hátramarad [›zurückbleiben‹]), elo˝re- [›vor, voraus‹], tovább- [›weiter-‹] usw. Groß ist auch die Anzahl der Fälle, in denen das Verhältnis zwischen Verbalpräfix und Verb dem Verhältnis zwischen Adverbialbestimmung und Prädikat ähnlich ist, z. B. vissza-néz [›zurückschauen‹], hozzácsatlakozik [›sich anschließen‹], el-ment [›ging weg‹], be-ment [›ging hinein‹], kiment [›ging hinaus‹]. Infolge dieser Gründe sind die Verbalpräfixe mindestens z. T. Manifestationen des polysynthetischen Typs.[1] Ein Beispiel für die Subinkorporation sehe ich in der ungarischen Attributkonstruktion. Das bestimmende und das bestimmte Wort bilden zwar kein Wort, aber oftmals bilden sie eine einzige Betonungseinheit (a nagy asztal [›der große Tisch‹]) — ihre Verbindung ist also enger als diejenige von zwei zusammengehörigen Wörtern. Ich beabsichtigte, nur eine kurze Skizze der ungarischen Typologie zu geben. Ich hoffe, es gelang mir zu zeigen, daß die Typologie der ungarischen Sprache nicht einfach ist und daß eine eigentümliche Verflechtung der Typen charakteristisch für sie ist.

Bemerkung [1]

Vgl. dagegen Giger 1998: 93, wo für die Interpretation der ungarischen ablösbaren Verbalpräfixe als Manifestation des agglutinierenden Typs argumentiert wird.

Die Typologie des Ungarischen 

Abkürzungen allat dat def elat gen iness obj

Allativ Dativ definiter Artikel Elativ Genetiv Inessiv Objekt

pl poss prät sg subj suff

Plural Possessivaffix Präteritum Singular Subjekt Suffix

Literaturverzeichnis Giger, M. 1998. »Zur typologischen Stellung des Obersorbischen innerhalb der westslavischen Sprachen: Das Verbalsystem«. In Beiträge der Europäischen Slavistischen Linguistik (POLYSLAV) 1, M. Giger und B. Wiemer (Hrsg.), 83–96. München. (Die Welt der Slaven. Sammelbände/Sborniki 2) Greenberg, J. H. (ed). 1966a. Universals of Language. Cambridge, Mass.-London. (2. Auflage; 1. Auflage 1963) Greenberg, J. H. 1966b. »Introduction«. In Greenberg (1966a), xi-xiii. Greenberg, J. H. 1966c. »Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements«. In Greenberg (1966a), 73–113. Hockett, Ch. F. 1966. »The Problem of Universals in Language«. In Greenberg (1966a), 1–29. Šaumjan, S. 1971. Strukturale Linguistik. München. (Internationale Bibliothek für Allgemeine Linguistik 12) — Original: Šaumjan, S. K. 1965. Strukturnaja lingvistika. Moskva. Sgall, P. 1995. »Prague School Typology«. In Approaches to Language Typology, M. Shibatani und T. Bynon (eds), 49–84. Oxford. Skalicˇka, V. 1935. Zur ungarischen Grammatik. Praha. (Facultas Philosophica Universitatis Carolinae Pragensis. Práce z veˇdeckých ústavu˚ XXXIX) Skalicˇka, V. 1942. »Problém druhého slovesa«. Cˇeský cˇasopis filologický 1, 9–14. Skalicˇka, V. 1946–1948. »Sur le rôle de la flexion interne dans la langue«. Linguistica Slovaca IV-VI, 13–22. Skalicˇka, V. 1966. »Ein ‘typologisches Konstrukt’«. In Travaux linguistiques de Prague 2, 157–63. Skalicˇka, V. 1968a. »Die Inkorporation und ihre Rolle in der Typologie«. In Travaux linguistiques de Prague 3, 275–9. Skalicˇka, V. 1968b. »Problém syntaktické typologie«. Slavica Pragensia X, 137–42. Skalicˇka, V. 1968c. »Über die Typologie der finnisch-ugrischen Sprachen«. In Congressus internationalis Fenno-Ugristarum Helsinki 23.-28. VIII. 1965. Pars I. Acta Linguistica, 494–8. Helsinki. Skalicˇka, V. 1968d. »Zum Problem des Donausprachbundes«. Ural-Altaische Jahrbücher 40, 3–9.

 Vladimír Skalicˇka

Skalicˇka, V. 1975. »Über die Typologie des Estnischen«. In Congessus tertius internationalis Fenno-Ugristarum Tallinae habitus 17.-23. VIII. 1970. Pars I. Acta Linguistica, 369–73. Tallinn Skalicˇka, V. 1979. Typologische Studien. Mit einem Beitrag von Petr Sgall. Braunschweig/ Wiesbaden. (Schriften zur Linguistik 11) Skalicˇka, V., Sgall, P. 1994. »Praguian Typology of Languages«. In The Prague School of Structural and Functional Linguistics, P. Luelsdorff (ed), 333–57. Amsterdam/Philadelphia. (Linguistic and Literary Studies in Eastern Europe 41).

Section II

Grammar

Theoretical description of language as a basis of corpus annotation The case of Prague Dependency Treebank*

Eva Hajicˇová Prague

.

Introduction

The raised interest in corpus linguistics in the last two decades of the past century has opened new horizons for linguistic (and not only linguistic) research, but also brought along several fundamental issues for discussion. One of them is that of designing annotation (tagging) scenarios. This issue is topical esp. at the present stage when no problems of principle are connected with a mere compilation of very large corpora of texts (not that much of speech, though, which is still a serious problem, mostly of a technical and organizational nature). In the present contribution we would like (i) to argue that such scenarios should include an underlying level of syntactic annotation and (ii) to demonstrate on the example of the Prague Dependency Treebank (using a part of the texts gathered in the Czech National Corpus) that a design of such a scenario if based on a sound explicit linguistic theory is a feasible task. We present first a brief account of such a theory, viz. the Functional Generative Description (FGD, Section 2), followed by a sketch of the overall tagging scenario of the Prague Dependency Treebank (PDT, Section 3). The core of the paper is Section 4, in which the theoretical assumptions of FGD concerning the tectogrammatical (underlying) syntactic level are confronted with the properties of the tagging scheme (the tectogrammatical tree structures, TGTSs). Some conclusions are drawn in Section 5.



Eva Hajicˇová

.

Functional Generative Description

. General overview The annotation scheme on the deep structure level (TGTSs) is based on the theoretical framework of the Functional Generative Description (FGD), namely on its level of the tectogrammatical representations (for motivating discussions and for more details, see e.g. Sgall 1967 and 1992; Sgall et al. 1986; a formalization can be found in Petkevicˇ 1987 and 1995). It has been shown in which way the class of these representations can be specified on the basis of a small number of general principles accounting for the core of grammar and by specific rules for peripheral patterns. The two features of FGD distinguishing it from most of other theoretical frameworks are the dependency-based syntax and the systematic incorporation into the description of an account of information structure of the sentence (its topic–focus articulation, TFA). . Tectogrammatical level .. A general characterization of the tectogrammatical level of FGD The tectogrammatical level can be characterized as the level of linguistic (literal) meaning, i.e. as the structuring of the cognitive content proper to a particular language. On this level, the irregularities of the outer shape of sentences are absent (including synonymy and at least the prototypical cases of ambiguity) and it can thus serve as a useful interface between linguistics in the narrow sense (as the theory of language systems) on one side and, on the other, as a background for such interdisciplinary domains as that of semantic interpretation (logical analysis of language, reference assignment based on inferencing using contextual and other knowledge, further metaphorical and other figurative meanings), that of discourse analysis or text linguistics, and so on. A tectogrammatical representation (TR) of the sentence basically has the shape of a dependency tree. The edges of the tree denote the dependency relations and the nodes carry complex labels indicating their lexical and morphological values. No nonterminals and no nodes corresponding to function words (auxiliaries, prepositions, conjunctions, articles) are present in the tree. Counterparts of function words (and function morphemes) are parts of the complex symbols of the nodes. Instead of using the notion of phrase, we work with subtrees (i.e. the governor and its dependents, or all its subordinate nodes, where ‘‘subordinate’’ is the transitive closure of ‘‘dependent’’, so that ‘‘b is subordinated to c’’ means ‘‘b immediately or through mediation of other nodes depends on c’’). The left-toright order of the nodes of the dependency tree corresponds to the topic–focus articulation of the sentence.

The Prague Dependency Treebank

TRs meet the strongly restrictive condition of projectivity: a dependency tree is projective if for every three-element set of nodes a, b, c present in the tree, it holds that if a depends on c, and b is placed between a and c in the left-to-right order, then b is subordinated to c. The cases of non-projective constructions in the surface structure (which are strongly limited as for their types, though not as for their frequency) can be described by means of movement rules concerning morphemics (see Sgall 1997b; Hajicˇová 1998a and the writings quoted there). The orientation of the dependency relation (i.e. the determination of which of the pair of the nodes connected by an edge is the governor and which is the dependent) can be specified on the basis of the following operational criterion: the dependent node is that member of the pair that is syntactically omissible, if not in a lexically specified pair of words (as is the case with the endocentric syntagms), then at the level of word classes. Thus e.g. in ((very) slow) progress the syntactic potential of the heads prototypically is identical to that of the whole groups. In Jim met Sally nothing can be deleted, but we know from other cases that the verb is never deletable (without a specific context), whereas object can be absent e.g. with read, and subject (or, more precisely, its prototypical tectogrammatical counterpart, the Actor or Actor/Bearer) is absent e.g. with rain (the E. pronoun it is just a morphemic filler, having no semantic relevance, since no other option is present, and in the so-called pro-drop languages such as Czech or Latin no subject pronoun is present). As mentioned above, function words do not occupy specific positions in the syntactic structure of the sentence as represented by the TRs. This is substantiated by the fact that articles and prepositions are, as a rule, connected with nouns, auxiliary verbs and conjunctions with verbs, and they cannot be freely modified by other elements of the sentence. Thus it appears not to be appropriate to assign them the same status as to proper (autosemantic) lexical units. Their underlying counterparts thus should be differentiated from those of the autosemantic words and denoted by more economical means than separate nodes. Thus e.g. an embedded clause such as (We knew) that Jim arrived is represented by a subtree the head of which is labeled by the lemma of its verb with the functor PAT corresponding to the function of the conjunction ‘that’ in this sentence: arrive.ant(erior).indic.PAT. The labels of the nodes of the TR are complex symbols consisting of three parts: (i) a lexical part, (ii) a combination of symbols (called grammatemes) for values of grammatical categories such as number, tense, modality, etc., and (iii) symbols (called functors) denoting the kinds of syntactic dependency (the valency positions). The functors can equivalently be written as labels of edges (or, in a linear notation, as indices of parentheses). The TR of a sentence thus can be rendered by a string of complex symbols corresponding to lexical occurrences (of the autosemantic words), with every dependent included in a pair of parentheses.





Eva Hajicˇová

Along with dependency, the TRs include a specification of several further relations. One of these is the topic–focus articulation (TFA), expressed in the surface structure mainly by an interplay of word order and sentence prosody (esp. the position of intonation centre); in the TRs, TFA is represented by the left-to-right order of the nodes (denoting the so-called communicative dynamism, i.e. the underlying word order) and by an index attached to the verb to denote whether it is contextually bound or non-bound; the nodes to the left of their governor are contextually bound, those to the right are contextually nonbound (for a definition of topic and focus as based on these primitive notions, see Sgall et al. 1986). Other kinds of syntactic relations are those of coordination (conjunction, disjunction and others) and of apposition. Their interplay with dependency cannot be accounted for with full adequacy by trees if we do not want to neglect the difference between the binary dependency relations and the coordinated (and appositional) constructions, some of which may have an indefinite number of members. However, even a network with a greater number of dimensions, which in this sense can serve as the shape of a TR, can be formally described in the form of its one-to-one linearization (see Petkevicˇ 1995; Sgall 1997a), namely by a string of complex symbols with which a pair of parentheses surrounds every dependent item. The kinds of dependency relations are written as indices of parentheses (attached to the parenthesis that is placed on the side of the head). Also the kinds of coordination are indicated by such indices (on the righthand parenthesis). .. Valency as the core notion of TRs The core of syntax in FGD lies in the notion of valency, i.e. of sets of kinds of dependents (see esp. Panevová 1974 and 1980; Hajicˇová and Panevová 1984). Within the dependents, arguments or inner participants are differentiated from free modifications (circumstantials, adjuncts) on the basis of the following criteria: a. inner participants are bound to certain groups of verbs only; b. they occur at most once as dependent on a single verb token. Five types of inner participants of verbs are distinguished: Actor/Bearer (Jim runs, sits, sleeps . . ., the brook runs), Objective (to build a house; to destroy a house; to see a house; to address someone; to elect the chairman; to choose a spokesman), Addressee (to give Mary a book), Effect (to do sth. as chairman; to elect somebody the chairman; to choose him as chairman), Origin (to make a canoe out of a log). Valency is not restricted to verbs; among the inner participants of nouns there is e.g. Material (Partitive, two baskets of sth.) and Identity (the river Danube, the notion of operator). There is a rich repertoire of free modifications: mostly for verbs, there are several types of Temporal circumstantials (when, how many times, since when, till

The Prague Dependency Treebank

when, how long, for how long), Manner, Regard, Extent (he spent his money to the last penny), Norm (in accordance with), Criterion (according to), Substitution (instead of), Accompaniment (with someone), Means (Instrument), Difference (two inches taller), Benefit (for someone), Comparison (as bright as something; brighter than sth.), Locative, three types of Directional-1.from where, 2.which way, 3.where to, Condition, Cause, Aim (in order to, for the sake of), Concession (although), Result (so that); dependent mainly on nouns, there are e.g. Appurtenance (my table, Jim’s brother, Mary’s car), Restrictive (rich man), Descriptive (the Swedes, who are a Scandinavian nation). A participant or a free modification can be either obligatory or optional with a given head: participants are prototypically obligatory (e.g. Actor and Objective with the verb meet: Jim met Eve.), but they can also be optional (e.g. the Addressee with the verb to read: to read a book (to somebody)). Free modifications are prototypically optional, e.g. to be sitting (somewhere) (for a reason) (for some time), but they can also be obligatory (as e.g. Manner with the verb to behave: to behave badly, or Temporal. how_long with the verb to last: to last for a week, or Direction-to where with the verb to arrive: to arrive at Prague). To decide whether a complementation of the verb is obligatory (i.e. present in the underlying structure), although deletable in the surface structure, the so-called ‘dialogue test’ was formulated by Panevová (1974, 1980; see also Sgall et al. 1986). It is based on an assumption that the speaker is obliged to be able to add the information he deleted in his utterance (assuming that it is an information known to the hearer), if he is asked for it. Thus the dialogue in (1) is not coherent, since the speaker A should be able to answer the question posed by the hearer B. (1) A: Jerry arrives tomorrow. B: Where to? A: I don’t know. The dialogue test exemplified in (1) indicates that with the verb arrive the free modification of Direction-3.where_to (‘predictable’ for the hearer, i.e. known by the speaker and deleted in the surface shape of the sentence precisely because A believes it to be easily recoverable by B) is obligatory; on the contrary, the dialogue in (2) is coherent, which indicates that with the verb arrive the free modification of Cause is not obligatory. (2) A: Jerry arrives tomorrow. B: Why? A: I don’t know. Along with the information on the valency requirements of each lexical entry, there is also other grammatical information included in the valency frames, such as surface deletability (e.g. Directional-3 with to arrive is deletable, Objective with





Eva Hajicˇová

to meet is not: We met there is a case of reciprocity, rather than of deletion), markers denoting an optional or an obligatory controller (e.g. Actor is an obligatory controller with to try, an optional one with to decide; Addressee is an optional controller with to advise, to forbid), and the dependent’s ability to occupy certain syntactic positions (e.g. of Subject with Passivization, of a wh-element) or to constitute barriers for movement, and subcategorization conditions. .. Topic–Focus articulation in TRs TFA is characterized on the basis of two concepts (discussed in detail in Sgall et al. 1986; Hajicˇová 1993; Hajicˇová, Partee and Sgall 1998; and the writings quoted there): a.

Contextual boundness: Contextually bound (cb) items are the primary grammatical counterparts of expressions carrying so-called given information, and contextually non-bound (nb) items refer in prototypical cases to ‘‘new’’ information; primarily, nb items belong to the focus of the sentence and cb items constitute the topic of the sentence. b. Communicative dynamism (CD): In prototypical cases the scale of CD corresponds to the surface word order, but there are secondary cases, e.g. with the most dynamic item (focus proper, which carries the intonation center of the sentence, often a falling stress) occurring elsewhere than at the end of the sentence, or the verb occupying the second position in the uppermost subtree according to language specific rules (which are more or less obligatory in German, optional in Czech). It should be emphasized that for a proper interpetation of the given occurrence of the sentence in a given discourse it is necessary to take into consideration the sentence prosody and other specific aspects. In the TR’s, CD is indicated by the left-to-right order of the nodes of the tree, in which every cb dependent is placed to the left of its head and every nb item is placed to the right of its head. The values f, t, c of the grammateme TFA indicate whether the given item is nb (in the focus), or cb (in the topic), or a contrastive topic, respectively, as indicated in the examples (3) through (5) below; the questions in brackets indicate the context for that reading of the sentence which is considered in these examples. (3) Father came home. (What about your father?) (3¹) father.t come.t home.f (4) (It is) father (who) came home. (Who came home?) (4¹) home.t come.t father.f

The Prague Dependency Treebank

(5) Mary went home and Fred stayed at school. (What about Mary and Fred?) (5¹) Mary.c went.f home.f and Fred.c stayed.f at school.f

.

The scenario of the Prague Dependency Treebank

The annotation scheme of the PDT consists of the following three layers of tagging: 1. The morphemic layer, arrived at by an automatic procedure of POS tagging and by disambiguation of the rich inflectional system of Czech, contains disambiguated values of morphemic categories (see Hajicˇ and Hladká 1997, 1998; problems of this step are not discussed in the present paper). 2. Syntactic tags on the analytic layer (analytic tree structures, ATSs), encoding functions of individual word forms (including also e.g. punctuation marks) as they are rendered in the surface shape of the sentence (see Hajicˇ 1998; a manual has been prepared for the human annotators, see Bémová et al. 1997, and translated into English); at the present time, about 100 000 sentences from CNC are tagged on this layer. The layer of analytic syntax does not immediately correspond to a level substantiated by linguistic theory, although in some aspects it may be viewed as coming close to the level of ‘surface syntax’ as present in the earlier stages of FGD (see Sgall 1992 as for reasons to abandon this level). The main difference between ‘surface’ and the analytic layer is that every function word and punctuation mark gets a node of its own in the syntactic network. We have been led to the inclusion of the analytic layer into the tagging procedure by two reasons: (a) it makes it possible to work with a relatively large set of syntactically tagged sentences without much delay, (b) it is a useful intermediate step in the transduction of the surface shape of the sentence (as a string of words) into its underlying structure (as a dependency tree structure), and (c) it allows for a comparison of the results with the outputs of several tagging and parsing procedures which have been implemented for other languages in different research centres. 3. Syntactic tags on the tectogrammatical layer (TGTSs) capture the deep (underlying, tectogrammatical) structure of the given sentence, i.e. its dependency based syntactic structure proper (see Hajicˇová 2000). A significant part of the annotation procedure of both 2. and 3. is carried out automatically. The annotators involved in the intellectual part of the procedure have a software tool at their disposal that enables them to work with the graphic representation of the trees on both these layers, modifying the trees in several





Eva Hajicˇová

respects, esp. in what concerns adding or changing the complex labels of the nodes, or adding and suppressing nodes.

. Tagging on the Tectogrammatical layer . Tectogrammatical tree structures (TGTSs) A node of the TGTS represents an occurrence of an autosemantic (lexical, meaningful) word; the contribution of the synsemantic (auxiliary, functional) words to the tectogrammatical structure of the sentence is reflected in the indices to the autosemantic words to which the function words ‘‘belong’’ (i.e. auxiliary verbs and subordinating conjunctions to the verbs, prepositions to nouns, etc.); coordinating conjunctions remain as nodes of their own (similarly as in the ATSs). In cases of deletions in the surface shapes of sentences nodes for the deleted autosemantic words are added to the tree structure. Not only the direction of the dependency relation (dependent from the right — dependent from the left), but also the ordering of the sister nodes is specified in the TGTSs (in accordance with CD, see Section 2.3.3.(b) above). Each TGTS has thus the form of a projective dependency tree with the verb of the main clause as its root (to be more precise, the root of the TGTS is a special node identifying the sentence of which the given structure is the TGTS, and the node of the main verb is the only node incident to this identifier). In case of nominal ‘sentences’ (i.e. of constructions without a finite verb), three possibilities obtain: (i) the governing verb is added (in case of surface deletions, which is relatively rare), or (ii) a symbol for ‘empty verb’ (‘EV’) is added as the governor (e.g. Od našeho washingtonského zpravodaje ‘From our correspondent from Washington’, with the node for ‘correspondent’ depending on ‘EV’), or (iii) the governing nominal node acts as the governor (e.g. with author names).

. Characterization of the labels of the TGTS nodes Each label of a node in a TGTS consists in the following parts: 1. the lexical value proper of the word (represented in a preliminary way just with the usual graphemic form of the word, the ‘lemma’); 2. the values of the morphological grammatemes (corresponding primarily to the values of morphological categories such as modality, tense, aspect with verbs, gender and number with nouns, degree of comparison with adjectives); 3. the values of the attribute ‘functor’, corresponding to (underlying) syntactic

The Prague Dependency Treebank

functions (Actor, Objective, Means, Locative, etc.; in our examples, we write the values of functors in upper case letters); as a matter of fact, in case of doubts, the annotators have the possibility to indicate two different values for every functor; 4. the values of the attribute ‘syntactic grammateme’, corresponding to secondary aspects of syntactic functions and combined with some of the functors, leading thus to a more subtle (semantic) subcategorization of these syntactic relations that is rendered on the surface first of all by prepositions and cases of nouns; this concerns the functors with the meaning of location LOC, DIR-1, DIR-2 and DIR-3 (corresponding to the questions ‘where?’, ‘from where?’, ‘through which place?’ and ‘where to?’, respectively); thus e.g. LOC (expressed in Czech by several prepositions combined either with the locative (Loc) or with the instrumental (Instr) case of the noun) is subcategorized into na+Loc (‘on’: na stole ‘on the table’), v+Loc (‘in’), u+Loc (‘by’), nad+Instr (‘above’), pod+Instr (‘under’), prˇed+Instr (‘in front of ’), za+Instr (‘behind’), mezi+Instr (‘among’), mezi+Instr (‘between’), etc.; as for functors having a temporal meaning, a similar subcategorization is established with the functor TWHEN (with the grammatemes AFT ‘after’, BEF ‘before’, ON ‘on Monday’, ‘next year’); a positive or negative grammateme is attached to ACMP (‘with’ vs. ‘without’), REG (‘with regard’ vs. ‘without regard’) and BEN (‘for’ vs. ‘against’), etc.; 5. the values of a special grammateme capture the basic information about the topic–focus articulation (TFA) of the sentence. At the present stage, the tentative and preliminary inventory of the tectogrammatical labels for Czech comprises 10 attributes for morphological grammatemes (e.g. number, tense, aspect, degrees of comparison) and 47 values for the attributes of ‘functor’ and ‘syntactic grammateme’ (functor: Actor, Patient, Addressee, Locative, Means, etc.; syntactic grammateme: see point 4. above). In addition, there are specific attributes for the relations of coordination, apposition, for marking ‘restored nodes’ (i.e. such nodes that have been deleted in the surface shape of the sentence and restored in its tectogrammatical structure), and three attributes to account for the basic features of coreference (both intra- and intersententential). . An illustration The (preferred) ATS of sentence (6) is given in Fig. 1, its TGTS (on the reading ‘(I tell you about) Mary (that she) is carrying books to the library’) in Fig. 2 (with many simplifications): (6) Marie nese knihy do knihovny Mary is-carrying books to (a) library



 Eva Hajicˇová

o

nést.sim.proc.indic

o

o kniha.pl.Obj

Marie.Subj

o do.AuxP o knihovna.sg.Adv

Figure 1.

A simplified ATS of sentence (1), where ‘nést’ is the infinitive of ‘nese’, AuxP is the syntactic label for a preposition, and the other abbreviations correspond to morphological values (tense: simultaneous, aspect:processual, etc.) and to types of dependency on the level of ‘analytic’ syntax. o nést.sim.proc.indic.F o Marie.ACT.T

o kniha.pl.PAT.F

o knihovna.sg.DIR-3.F

Figure 2.

A simplified TGTS of sentence (1), with abbreviated symbols corresponding to the values of grammatemes and functors. The indices T and F stand for the values of the TFA attribute, namely ‘contextually bound’ and ‘non-bound’, respectively, . Confrontation of TR’s and TGTS’s .. Motivation of differences As has been stated above, TGTSs are based on the theoretical conception of TRs; this does not mean that the resulting tagged structures are ‘deteriorated’ by being biased to a specific theoretical framework. On the contrary: TGTSs have a theoretically sound and empirically tested basis with a perspicuous structure and other properties favourable for the possibility of comparison with other frameworks; this might be taken as an advantage, since it makes it possible to specify the properties of TGTSs in a precise and explicit way. Thus these structures will be a useful source of information also for those who work in other frameworks. However, there are some points in which TGTSs differ from the TRs; the differences have been motivated by an effort, first, to encode peripheral relations (other than pure dependency) in a straightforward way (see below 4.4.2) and, second, to preserve also those pieces of information from the surface shape of the sentence that might be of interest for future (mostly linguistic) research (Section 4.4.3).

The Prague Dependency Treebank

.. Coordination and apposition To specify TGTSs as two-dimensional trees, coordination and apposition are treated in a way that differs from their treatment in FGD: although coordinating conjuntions belong to function words, they retain their status as nodes (labeled as CONJ, DISJ, etc., with the lexical value of the conjunction) in the TGTSs; the same holds for the expressions denoting an apposition. In addition, the nodes for the words standing in the coordination (apposition) relation get a special index. Compare the bracketted shapes of the TGTSs (disregarding other than structural relations) for (7): in (7a), the interpretation is ‘(old men) and (women)’), while (7b) stands for the interpretation ‘(old men) and (old women)’, with an added node (marked by ELID) corresponding to the restrictive adjunct ‘old’, which has been deleted in the surface shape of the sentence. (7) starˇí muži a ženy old men and women a. (CONJ.a (starý) (muž.CO) (žena.CO)) b. (CONJ.a ((starý) muž.CO) ((starý.ELID) žena.CO)) This exception makes it technically possible to work with rooted trees, rather than with networks of more dimensions. .. Further specific differences New attributes for the existing nodes are being established, carrying information that might be interesting for the use of the tagged corpus for further research, be it a research aiming at an enrichment of the tectogrammatical description, as e.g. in 1., or a research of how language is used in communication, as in 2. The following issues belong here: 1. In order to capture lexical collocations as wholes the component parts of a collocation get a positive value of a newly introduced attribute PHRi, where i is the serial number of this collocation (a ‘phraseme’ or idiom) in the sentence. 2. Special attributes COREF, CORNUM and CORSNT are introduced (for the time being, only to nouns and pronouns) to capture at least some basic aspects of (esp. textual) coreferential relations. The values of these attributes can be characterized as follows: in COREF, the lemma of the antecedent is inserted; the attribute CORNUM contains the serial number of the antecedent if it occurs in the same sentence, or else, the value is NA (non-applicable); the attribute CORSNT obtains one of the following two values: PREV (if the antecedent occurs in the previous sentences), or else NA. In cases of grammatical coreference (such as with the ‘subjects’ of infinitives as complements of the so-called verbs of control), the attribute COREF of the ‘restored’ subject gets the lemma of the ‘controller’ as its value; the attribute



 Eva Hajicˇová

CORNUM then gets the serial number of the controller as its value and CORSNT gets the value NA, cf. (8): (8) Rodicˇe radili Jirkovi nechodit tam parents adviced George not-to-go there In the TGTS of (8), there will be a node added as an Actor of the verb nechodit, with Cor as its lexical value, and with Jirka in its COREF, 3 in its CORNUM and NA in its CORSTN. 3. With ‘restored’ nodes standing for elements deleted in the surface structure of the sentence but present in its underlying structure the attribute DEL gets the value ELID, if the ‘restored’ element stands alone, cf. e.g. the TGTS (7b) above, or else ELEX (‘expounded’ deletion), see the TGTS (9’) for (9), if the interpretation is ‘(very old men) and (very old women)’, i.e. with a single added node for the deleted restrictive adjunct ‘old’ marked as ELEX rather than adding also the node(s) dependent on it. (9) velmi starˇí muži a ženy very old men and women (9¹) (CONJ.a (((velmi) starý) muž.CO) ((starý.ELEX) žena.CO). 4. Parenthetical items in the sentence without a specific syntactic relation to one of its elements get the functor PAR ( see (10)), while a parenthetical item which exhibits a dependency relation to some element of the sentence obtains a regular functor (see (11), where CRIT stands for the functor of Criterion and ‘já’ is the lemma for ‘I’). (10) Jirka myslím.PAR prˇijde pozdeˇ George I-think will-come late (11) Jirka (podle mne) je talentovaný pianista George according to-me is talented pianist (11¹) (Jirka.ACT) (já.CRIT) být ((talentovaný.RSTR) pianista.PAT) 5. A special functor PREC is introduced to denote the syntactic function of those elements of the sentence (with the analytic function of a particle) that as a rule stand at the beginning of the sentence, have a more or less discoursive function of cohesion but do not connect clauses into complex sentences; there belong the particles tedy, tudíž ‘thus’, tj. ‘i.e.’, totiž ‘as a matter of fact’, etc. (e.g. He was ill. Thus he couldn’t come there.) 6. Direct speech is distinguished by an index DSP (‘direct speech’) attached to the root of the TGTS of the sentence enclosed in quotation marks; if more than a single sentence is in quotation marks, an index DSPP (‘part of direct speech’) is

The Prague Dependency Treebank 

attached to the root of the TGTS of the first and of the last sentence of such a direct speech. 7. Quoted word(s), if occurring in quotation marks (be they single or double) in the surface shape of the sentence, get the index QUOT, unless they constitute a sentence of its own; e.g. while the noun pleasure in They call it ‘‘pleasure’’ gets ‘QUOT’, the verb come in He told her: ‘‘Come back soon’’ gets the index DSP indicating direct speech.

.

Concluding remarks

. Issues for further discussion A complete tectogrammatical tagging of a large corpus is, of course, a very demanding task and it is no wonder then that the specifications we are now working with cannot cover all subtle oppositions that should be distinguished in the representation of the meaning of the sentence. However, we have found our task very stimulating and leading to new insights concerning issues some of which either (i) are technically complex and could not yet been entirely integrated into our apparatus (as e.g. a fully automatic handling of PP attachment and other cases of morpho-syntactic ambiguity), or (ii) have not yet been analyzed in any existing grammar or monograph of Czech. Let us briefly mention here some of the open questions that are still waiting for a monographic inquiry; some of them concern the theoretical framework, some are connected only with the decisions concerning tagging and its ambiguities. .. Issues concerning types of valency slots The issues concerning valency belong to the crucial ones when completing the tagging scenario. This is true both of the determination of valency frames for verbs as well as those of nouns. Some of the functors seem to cover more than one type of syntactic relations and thus might be more subtly differentiated; this concerns e.g. the Locative (cf. the difference between zranil se v lese ‘he injured himself in the forest’ and zranil se na ruce ‘he injured himself on his hand’, the latter being in some sense closer to an inner participant (argument). However, the question remains of how to classify the Locatives v kuchyni ‘in the kitchen’, jednání uvnitrˇ koalice ‘discussions within the coalition’. Similar problems concern the modification of Means (cf. the difference between psát rukou ‘to write with hand’, na stroji ‘on the typewriter’, tužkou ‘with a pencil’ and pohnout rukou ‘to move one’s hand’). Another such question concerns the relation between the functor Dir-3 and the so-called Intent (e.g. šla nakoupit ‘she went shopping’). A further question of

 Eva Hajicˇová

this kind is that of the difference between e.g. pojmenovat neˇjak ‘to give some name’ and pojmenovat po kom ‘to name (something, somebody) after somebody’; the latter example certainly is not just an instance of the functor Manner. In the valency frames of nouns, we work with the modifications of Restrictive (adjunct) and Identity. It is then an open question how to distinguish among such examples as pan N. ‘Mister N.’, poslanec N. ‘the deputy N.’, termín sloveso ‘the term verb’. The following criterion might be applied: with two adjacent (congruent, or non-declined) nouns the second noun functions as an Identity modifier if (a) it is non-declined (e.g. parníkem [Instr.] Hradcˇany [Nom.] ‘with the steamer Hradcˇany’) or (b) it can be put (without a change of meaning) into a genitive case (e.g. pojem subjekt/u ‘the notion (of) subject-Nom./Gen.’). In all other cases the first noun would then be classified as a Restrictive adjunct. However, even with this rule some intermediate cases remain: e.g. in the combinations of first name — family name, the family name may also be in genitive (esp. if the first name has a shape of a nickname: Jan Novák, but Honzík Novák/Nom.Sg. or Nováku˚/Gen.Pl.). .. Topic–Focus articulation One of the ‘burning issues’ in the TFA annotation is to make more precise the notion of contrastive topic. It is also necessary to make sure whether for a given language a distinction between contrastive and non-contrastive (parts of) focus is grammatically determined, and to pay much more attention to the study of the systemic ordering of kinds of dependents (‘canonical order’). The boundary line between the syntactic function of focus sensitive particles (rhematisers, focalizers) and those of other subclasses of Attitude adverbials has to be systematically studied; up to now we distinguish between RHEM (rhematizer), ETHD (ethical dative), INTF (intensifier), and ATT (attitudinal adjunct). The primary (prototypical) and secondary (marked) positions of overt focalizers (for a most recent treatment, see Hajicˇová, Partee and Sgall 1998) should be taken into account. It also should be considered whether some of these functors should not be reclassified as grammatemes (in accordance with an older proposal by P. Sgall). .. Coordination constructions and deletions It should be further investigated under which conditions a coordination construction is to be understood as a coordination of sentences (clauses) or of their parts (up to now we handle such examples as Sedlák a Bureš objevili virus L. ‘S. and B. discovered the virus L’ as a narrow coordination, although such a sentence does not exclude that each of the persons discovered the virus separately). Also other cases of deletions are still open for discussion. Up to now, we do not generally restore the governing verb of a whole sentence the deletion of which is registered in the analytic trees by means of an extra node labeled as ExD. In those cases where perhaps also some of the dependents of the restored node

The Prague Dependency Treebank

should be restored, we are not yet capable to specify under which conditions this restoration should take place (to avoid repeating what does not belong to the deleted position). .. Issues of the lexicon and word formation The present tagging scenario does not work with a composition of lexical meaning (degrees of hyponyms, etc.) from its parts or features, and we are aware that lexical semantics is a domain to be investigated. In the subdomain of word formation, up to now we have only worked with some of the most productive affixes and their roles (the verbal prefix of negation, some postverbal nouns and adjectives, postadjectival adverbs, possessive adjectives and pronouns). The boundary lines of some of these groups (and of many other) have not been drawn with full adequacy. For instance, the intransitive verbs derived by the ‘reflexive’ particle se, such as šírˇit se ‘expand’ are treated as specific lexical units; their relationship to the base forms is only to be found in the analytic trees; however, it is necessary to look for criteria distinguishing the set of such verbs. .. Coreference in discourse It has been mentioned above that only the elementary cases of textual anaphora are recorded. It will be necessary to look for a more complete application of the considerations of our previous research (Hajicˇová 1993), i.e. to work more systematically especially with the degrees of salience of the items contained in the stock of information shared by the speaker and (according to the speaker’s assumptions) the hearer(s). .. Graphic symbols Our treatment of dashes, quotes and quoted words, direct speech, and so on, as well as of the difference between a full stop and a semicolon as marking sentence boundaries of different strengths, or of the boundaries between paragraphs, is only preliminary. It has to be studied to which degree and in which ways the corresponding graphic symbols (or the prosodic phenomena they represent) contribute to the underlying structure of sentences (which also characterizes some aspects of the sentence as occupying certain positions in the discourse pattern).

. Conclusion If compared to the prevailing present-day trends in parsing and annotation, the present scenario of PDT is promising in going deeper in the sense of including much of semantically relevant phenomena. The aim of tagging under this approach is not only to check the grammatical structure of sentences (and their well-



 Eva Hajicˇová

formedness), making choice of the reading to be preferred, but also to provide an adequate input for the semantico-pragmatic interpretation of sentences and of their specification in what concerns their embedding in context. It will of course take some time before the part of PDT equipped with tectogrammatical tags is large enough to be of actual relevance either for practical applications or for further studies. The expected application of statistically based methods should lead to a more general and efficient shape of the procedure, but even then the tagging will contain many errors of most different kinds. However, we believe that the resource offered by the Prague Dependency Treebank will be soon useful for authors of future monographic inquiries into Czech grammatical and textual phenomena and their relationships to those of other languages: the rich empirical material gathered and analyzed in PDT, certainly makes a more systematic insight into the studied issues possible, and the related studies will in turn help us to remove the individual errors and to amend the procedures. In this way a means for a principled way of testing the descriptive framework of FGD can be gained. We are aware that many questions remain open, and some of them have been sketched above. These and many other puzzles are i.a. connected with the freedom of language (the speakers being free to decide for any deviation of the norm they only can think of) and make it necessary to look for descriptive methods adequate to account not only for the norm, but also for most different deviations (cf. Sgall, this volume).

Notes The research reported on in this paper (a preliminary form of which with the title ‘‘Dependency-Based Underlying-Structure Tagging of a Very Large Czech Corpus’’, was published in T. A. L. 2000) has been carried out within the project supported by the Czech Grant Agency 405–96-K214 and by the Czech Ministry of Education LN 00A 063.

References Bémová, A., Buránˇová, E., Hajicˇ, J., Kárník, J., Pajas, P., Panevová, J., Šteˇpánek, J. and Urešová, Z. 1997. Anotace na analytické rovineˇ: návod pro anotátory [Annotations on the analytic level: instructions for the annotators]. Technical Report ÚFAL TR-1997–03. Prague: Charles University. Translated into English and available on the web site http://ufal.mff.cuni.cz Hajicˇ, J. 1998. ‘‘Building a syntactically annotated corpus: The Prague Dependency Treebank’’. In Issues of Valency and Meaning. Studies in Honour of Jarmila Panevová E. Hajicˇová (ed), 106–32. Prague: Karolinum.

The Prague Dependency Treebank 

Hajicˇ, J. and Hladká, B. 1997. ‘‘Probabilistic and rule-based tagger of an inflective language — a comparison’’. In Proceedings of the Fifth Conference on Applied Natural Language Processing, 111–18. Washington, D. C. Hajicˇ, J. and Hladká, B. 1998. ‘‘Czech language processing — POS tagging’’. In Proceedings of the First International Conference on Language Resources & Evaluation. Granada, Spain. Hajicˇová, E. 1993. Issues of Sentence Structure and Discourse Patterns. Prague: Charles University. Hajicˇová, E. 1998. ‘‘Movement Rules Revisited’’. In Procesing of Dependency-Based Grammars, Proceedings from the Workshop, COLING/ACL, S. Kahane and A. Polguere (eds), 49–57. Montreal. Hajicˇová, E. 2000. ‘‘Dependency-Based Underlying-Structure Tagging of a Very Large Czech Corpus’’. Les grammaires de dépendance. Traitement Automatique des Langues 41(2000):57–8 Hajicˇová, E. and Panevová, J. 1984. ‘‘Valency (case) frames of verbs’’. In Sgall 1984: 147–88. Hajicˇová, E., Partee, B. and Sgall, P. 1998. Topic–Focus articulation, tripartite structures, and semantic content. Amsterdam:Kluwer Panevová, J. 1974. ‘‘On verbal frames in Functional Generative Description’’. Prague Bulletin of Mathematical Linguistics 22:3–40; 23(1975):17–52. Panevová, J. 1980. Formy a funkce ve stavbeˇ cˇeské veˇty. [Forms and Functions in the Structure of the Czech Sentence]. Prague: Academia. Petkevicˇ, V. 1987. ‘‘A New Dependency Based Specification of Underlying Representations of Sentences’’. Theoretical Linguistics 14:143–72. Petkevicˇ, V. 1995. ‘‘A New Formal Specification of Underlying Representations’’. Theoretical Linguistics 21:7–61. Sgall, P. 1967. Generativní popis jazyka a cˇeská deklinace. [Generative Description of Czech and Czech Declension.] Prague: Academia. Sgall, P. (ed) 1984. Contributions to Functional Syntax, Semantics and Language Comprehension. Amsterdam: Benjamins and Prague: Academia. Sgall, P. 1992. ‘‘Underlying Structure of Sentences and Its Relations to Semantics’’. Wiener Slawistischer Almanach. Sonderband 33, T. Reuther (ed), 273–82. Wien: Gesellschaft zur Förderung slawistischer Studien. Sgall, P. 1997a. ‘‘Valency and Underlying Structure. An Alternative View on Dependency’’. In Recent Trends in Meaning-Text Theory, L. Wanner (ed), 149–66. Amsterdam/Philadelphia: Benjamins. Sgall, P. 1997b. ‘‘On the Usefulness of Movement Rules’’. In Actes du 16e Congr s International des Linguistes (Paris 20–5 juillet 1997), B. Caron (ed). Oxford: Elsevier Sciences. Sgall, P. (this volume). The Freedom of Language. Sgall, P., Hajicˇová, E. and Panevová, J. 1986. The Meaning of the Sentence in Its Semantic and Pragmatic Aspects. J. L. Mey (ed). Dordrecht: Reidel — Prague: Academia.

‘‘Conditionals’’ in Hebrew and English: same or different?

Yishai Tobin Be’er Sheva ‘‘Every language provides a means to coin out its own substance.’’ (Bolinger 1971: xi)

.

Introduction

In this chapter I will present a functional, sign-oriented analysis of the conditionals if/whether in English and their so-called Hebrew equivalents im/lu/ilu/lule/ilule (all translated as ‘if ’). The analysis is based on the monosemic concepts of invariance, markedness, and distinctive feature theory. I will show that despite the fact that all these forms share the same FUNCTION, their MEANINGS are quite different and therefore it would be difficult to view them as being equivalent. However the iconic relationship between the FORMS and their meanings in each language system does seem to be similar: the more complex the form, the more specifically marked the meaning. The theoretical and methodological question to be asked therefore is: What are the bridges and interfaces of form, meaning, and function in and across human languages? The answers I will provide are based on aspects of the semiotic view of language developed by the Prague and PostPraguian sign-oriented functional schools of linguistics discussed in Andrews and Tobin (1995); Contini-Morava and Goldberg (1995); Contini-Morava and Tobin (2000); Klein-Andreu 1983; Reid and Otheguy (1998); Tobin (1988, 1989); and Waugh and Rudy (1991). Most typological studies are based on the methodological tenets of sentenceoriented linguistics and their accompanying sets of traditional and neotraditional syntactic categories which are usually compared and contrasted crosslinguistically based on their ‘‘equivalence’’ and ‘‘translatability’’ (Garcia et al. 1987). Certain semiotic or sign-oriented linguistic theories however question or reject many of these traditional categories in favor of the linguistic sign and its fundamental component of invariant meaning as the basis of their analyses of linguistic systems within and across languages (Andersen 1991; Klein 1976; Lattey 1980, 1989). In

 Yishai Tobin

this paper I will show how the sign-oriented principles of invariance, markedness, and distinctive feature theory can provide an alternative way of analyzing supposedly universal categories within and across languages to show how different they actually may be (Tobin 1992a,b; 1994; 1997b). Most typological studies of universal categories usually choose the most frequently used, prototypical token of that category to represent that category in a language (as may seen Traugott et al. 1986 and van der Auwera 1997). For example, if is the universal conditional discussed for English, si for Latin, and im would be their so-called equivalent in Hebrew. One must remember, however, that if is in opposition to at least one other (at least partially) synonymous conditional: whether in English (Tobin 1990a: ch. 6) and im is in opposition to at least four other conditionals: lu, ilu, lule, ilule in Hebrew (all of which are also translated as ‘if ’) (Tobin 1994: ch. 3). In this paper we will show that although these forms may share a similar semantic domain: THE PRESENTATION OF POSSIBILITIES; they are marked for very different semantic features and, therefore, may be less equivalent and translatable across languages than was previously considered. The semantic analyses of these conditionals in English and Hebrew will be based on the following sign-oriented hypotheses: a. the conditionals in English and Hebrew differ in their monosemic, core, invariant meanings, or signifié’s (e.g. Bolinger 1977; Saussure 1913/1959; Tobin 1990a, 1993); b. this difference in meaning is expressed by a marked distinctive semantic feature in the sense originally postulated by Roman Jakobson (1957) of the Prague School; c. this marked distinctive feature (Semantic Integrality) (Tobin 1990a, 1992c, 1994) is based on the perception of entities in continuous versus discontinuous space, time, or existence in the sense originally associated with the school of the Psychomechanics of Language (Guillaume 1945, 1971; Hirtle 1982, 1995); d. according to this analysis: if is unmarked in English and indicates a possibility or possibilities viewed either in continuous or discontinuous space, time, or existence, while whether is marked for Semantic Integrality and therefore indicates possibilities which can be potentially perceived individually in discontinuous space but must now be viewed collectively: i.e., as part of a set viewed in continuous space, time or existence; e. the Hebrew system, on the other hand, distinguishes between factive and non-factive versus contrary-to-fact negative (Kiparsky and Kiparsky 1971) possibilities which are part of a relative hierarchy of possibility of occurrence or nonoccurrence; f. both the English and the Hebrew conditionals form semantic systems in the

‘‘Conditionals’’ in Hebrew and English: same or different?

sense of the Columbia School (Contini-Morava 1989; Garcia 1975; Huffman 1996; Kirsner 1979; Reid 1991).

.

The English System

Both if and whether share certain syntactic, semantic, and pragmatic qualities and functions which have made most grammars and dictionaries label them as synonyms based on the false assumption that function equals meaning. They both are labelled as conjunctions used in conditional clauses, reported yes-no questions, and utterances expressing wishes, stipulations, conditions and suppositions which are often referred to as direct and indirect questions or interrogatives (inverted or otherwise). They both have a strong tendency to collocate with verb forms or tenses (e.g., unreal past, conditional, future, historical subjunctive, etc.) and verbs (want, wonder, doubt, decide, think, see, know, and the modals shall/should, will/would, etc.) which may be considered appropriate for the kinds of speech acts mentioned above. Very often when both forms are interchangeable in the same utterance, the choice of if versus whether is deemed to be a stylistic one: if, the shorter and more usual form, is generally considered to be the more colloquial, informal, or less literary of the two. This stylistic explanation for the choice of one form over another is certainly open to question particularly since both forms can and do appear simultaneously in the same sentence, paragraph, or passage within the same written or spoken discourse and neither serves as a style marker in the sense of Enkvist (1973). It is also pointed out that whether has a stronger tendency to introduce alternative conditions or possibilities particularly to what I have referred to as whether or not and whether X or Y kinds of possibilities (Tobin 1990a: ch. 6). This preference for more binary or dichotic alternatives for whether is further strengthened by the diachronic fact that the etymons of whether (Old High German hwedar, Old English hwaeder, Middle English hwether) meant: ‘which one of (the) two things’. The semantic feature relevant to the present analysis of English conditionals, Semantic Integrality, has been previously postulated as a semiotic feature of human cognition and perception. The marked feature Semantic Integrality is based on the assumption that there are two alternative ways of perceiving a plurality of entities in space, time, or existence either as discrete entities (a+b=a+b) or as potentially discrete entities perceived as part of a continuous set (a+b=[ab]) as illustrated in Figure 1. Semantic Integrality may be viewed as a marked distinctive feature that can be used to explain linguistic concepts both in the grammar and the lexicon in a unified and systematic way that reflects the human perception of entities in space, time, and existence. Indeed, there may very well be a universal semantic develop-



 Yishai Tobin

Alternate ways of perceiving entities in a spatio–temporal–existential cline a+b=a+b (discontinuous space, time, or existence] or a+b=[ab] (continuous space, time, or existence] SEMANTIC INTEGRALITY: THE PERCEPTION OF POTENTIALLY DISCONTINUOUS ENTITIES VIEWED IN A CONTINUOUS SET

Figure 1. Semantic Integrality

ment of linguistic forms on a spatio-temporal-existential cline going from the most concrete spatial messages to the more abstract temporal, to the most abstract existential kinds of messages: (a) existence in space or in a place (e.g., in the room), (b) existence in time (e.g., in the morning), (c) abstract existence (e.g., to be in trouble, to be in the in-group) (Tobin 1990a: ch. 3). This spatio-temporal-existential cline has been considered a cognitive universal (Givon 1979; Traugott 1978; Wierzbicka 1972, 1980). It has been applied to the historical development of language in general and in the development of pidgins and creoles in particular (Traugott 1975); various word classes such as copulas, prepositions, and particles (Aphek and Tobin 1988, 1989; Garcia, van Putte and Tobin 1987; Traugott 1975, 1978); first language acquisition (Traugott 1974; Clark 1973); the origin and development of idioms in general (Pike 1976) and the development of phrasal verb idioms in particular (Makkai 1972). These studies provide additional evidence for Kronasser’s Law (Kronasser 1952; Kovacs 1961) which claims that semantic change tends overwhelmingly to move from the concrete to the abstract. The distinctive feature of Semantic Integrality may provide a means to better understand the connection between human cognition and perception and language and how this connection may be realized differently in various language systems. Semantic Integrality has served as the basis for analyzing the following linguistic phenomena: systems of grammatical number in general and so-called ‘‘irregular’’ plurals, mass/count nouns, agreement in English; the dual number in Hebrew; the quantifiers many vs. much; each vs. every; and some vs. any in English and their so-called Hebrew equivalents rav/merube/harbe/mirba; kol; and exad, axadim/eize; the adverbs of comparison like vs. as in English and their so-called Hebrew equivalents kemo/ke-; the connectives also vs. too in English and their socalled Hebrew equivalents gam,/gam-ken/af; the restrictives only vs. just in English and their so-called Hebrew equivalents aval/ax/ela, rak/levad, ax-ve-rak; the deletives without vs. -less in English and their so-called Hebrew equivalents beli/lelo; the synthetic vs. analytic systems of comparatives and superlatives (the) more/ most X/-er/-est X in English and their so-called Hebrew equivalents yoter X/X yoter/ha-X be-yoter, ha-xi X, od X; the diminishers few vs. less in English and their

‘‘Conditionals’’ in Hebrew and English: same or different?

so-called Hebrew equivalents me’at/ketsat; possessive constructions in English and Hebrew; and contracted vs. non-contracted forms in English and Hebrew in Tobin (1990a, 1994, 1998a). All of the abovementioned analyses were based on the Jakobsonian notion of markedness as it is used to distinguish between opposed members of a lexical or grammatical system (Andrews 1990; Battistella 1990, 1996 (the last reviewed in Tobin 1998b)). The asymmetric value relationship of the concept of binary markedness developed by the Jakobsonian-van Schooneveldian School of linguistics may be summarized as follows for the distinctive feature of Semantic Integrality: a.

The unmarked (U) member of an opposition is neutral. It makes no claim with regard to Semantic Integrality and is less complex cognitively. It means: PERCEIVE ENTITIES IN CONTINUOUS OR DISCONTINUOUS SPACE, TIME, OR EXISTENCE. b. The marked (M) member of the opposition makes a specific claim regarding Semantic Integrality. It is more complex cognitively. It means: PERCEIVE POTENTIALLY DISCONTINUOUS ENTITIES IN CONTINUOUS SPACE, TIME, OR EXISTENCE. The unmarked member of an opposition which makes no specific claim concerning the absence or the presence of a distinctive semantic feature is usually more flexible and open-ended than the marked member of an opposition. The unmarked member of an opposition allows for any and all possibilities and is very often the first to be acquired and the one more frequently used although this does not have to be an absolute rule and depends on linguistic and situational contexts. The marked member of an opposition which does make a specific claim for the presence of a distinctive semantic feature is less flexible and usually more limited in its distribution because of its greater precision in meaning. It is often acquired later and generally may be used less frequently than the unmarked member of an opposition although this, too, is not necessarily an absolute rule and depends on linguistic and situational contexts. Therefore it should always be remembered that neither order of acquisition nor the rate of frequency is the sole criterion for choosing the markedness value. However the asymmetric markedness relationship implies that the marked members usually may be replaced by the unmarked members while the opposite is not necessarily true. The feature of Semantic Integrality which makes a claim for the perception of potentially discrete entities as a single continuous set has been chosen as the marked feature for the following reasons: (i) The process of first perceiving concrete or abstract entities as they appear discretely in discontinuous space, time, or existence is the fundamental perception of identification. The later perception of these potentially discrete entities as



 Yishai Tobin

forming part of an integral set occupying continuous space, time, or existence is a more complex cognitive task. This greater cognitive complexity underlies and justifies the choice of Semantic Integrality as being the marked feature synergetically. (ii) If the notion of isomorphism is taken seriously, it becomes evident that the notion of integrality or fusion functions on all the levels of language as well. (iii) In phonology there is an opposition between vowels and diphthongs, nongeminated and geminated consonant phonemes, and stops/fricatives and affricates forming an obstruent consonant class. In all of these cases the second or marked member of the opposition represents a more complex unit or sound nucleus composed of the merger, fusion, or integration of two vowels or consonants and may be marked for Integrality (Tobin 1997a). (iv) In morphology or syntax there are also simple, analytic, independent units versus more synthetic, agglutinative, or compound units composed of the fusion, merger, or integration of different bound or unbound morphemes or clause types, etc., both in word formation and in clause and sentence formation. In all of these cases, the fused, merged, or integrated units are perceived as being more complex or marked as opposed to the simple or more neutral counterparts. (v) There is usually some sort of iconic connection between the signals themselves and their markedness values (Tobin 1990a, 1994). The signs marked for the more complex or specific meanings are usually larger or more complex in their signal or form while the signs which are unmarked and are simpler and less complex in their meanings are usually simpler in their signals and forms as well. The English system of if/whether is marked for the semantic feature Semantic Integrality. This feature is connected to aspects of human cognition and perception. The marked distinctive feature of Semantic Integrality is based on the assumption that there are two alternative ways of viewing entities in space, time and existence either as discrete entities: (a+b=a+b) or as potentially discrete entities perceived as part of a continuous set: (a+b=[ab]). In the present analysis the form whether is marked for the feature Semantic Integrality and the form if is unmarked for this feature. In other words, whether presents possibilities as an integral set (perceived in continuous space, time, existence) from which one (or more) possibility (or possibilities) is chosen while if makes no claim as to whether a possibility (or possibilities) appears as part of a set or not. (1) I don’t know whether you have brothers and sisters. If you do, I should like very much to have their addresses. (Vonnegut 1963: 10) (2) Tell me whether you’re coming or not, and if so (*whether so), whether I should prepare a big meal or just order in.

‘‘Conditionals’’ in Hebrew and English: same or different?

(3) Tell me whether you’re a virgin (pregnant, alive/dead), and if you are, I promise not to laugh. (4) We’ll make a nursery if and when (*whether and when) it’s necessary. (5) Talk to me as if (*as whether) you loved me! (6) If (*whether) I get any bigger I won’t fit through the door. In all of the above examples there is an inherent choice between or among possibilities which is always conveyed by the marked form whether which does not have to appear in either the single. or potentially multiple possibilities signaled by the unmarked form if. In other words, ‘‘it always takes two to tango’’ if you used the marked form whether.

.

The Hebrew system

Hebrew appears to have a system which is quite different from the English system both in form and in meaning although they share a similar function of presenting alternative perceptions of possibilities. Hebrew distinguishes between factive and non-factive positive possibilities versus contrary-to-fact negative possibilities both of which are part of a relative hierarchy of possibility of occurrence or non-occurrence. Thus, while the English system classifies the alternative perception of possibilities as being either discrete or part of an integral set of possibilities, the Hebrew system classifies possibilities as either being factive/non-factive and positive or contrary-to-fact and negative in a hierarchical order of possibility of occurrence or non-occurrence. The Hebrew system is composed of: (a) two simple independent (morphologically positive or affirmative) morphemes (im/lu) which indicate high and low degrees of factive or non-factive possibilities of occurrence or non-occurrence respectively and (b) three complex independent and bound (morphologically) negative morphemes (i-/–le (lo) = ‘no’) which are added to lu (LOW POSSIBILITY) to convey a hierarchy of contrary-to-fact occurrences or non-occurrences. Not surprisingly, the more complex five-member Hebrew system is iconic and economic to make it mnemonically more efficient to acquire and remember. The small, independent, morphologically positive signs denote the widest range of possibilities (factive and non-factive) on a two member hierarchical system im (HIGH) and lu (LOW): a. im = HIGH (MORPHOLOGICAL) POSITIVE FACTIVE/NON-FACTIVE (7) im ata ba, tavi li sefer. (present) if you come you-will-bring to-me book. ‘If you are coming, bring me a book.’



 Yishai Tobin

(8) im ata tavo, tavi li sefer. (future) if you-will-come you-will-bring to-me book ‘If you will be coming, bring me a book.’ (9) im kvar bata, lama lo heveta li sefer. (past) if already came-you, why no brought-you to-me book ‘If you came already, why didn’t you bring me a book?’ (10) im hayita ba, hayita mevi li sefer. (compound) if you-were come, you-were bring to-me book ‘If you would come, you’d bring me a book.’ b. lu = LOW MORPHOLOGICAL POSITIVE FACTIVE/NON-FACTIVE (11) lu yehi . . . (future) if it-will-be ‘If it will be . . .’ (12) lu bata/hayita ba, hayita mevi li sefer. if came-you/were-you come, were-you bring to-me book ‘If you came, you would have brought me a book.

(past compound)

The larger, morphologically negative signs composed of various dependent and independent morphemes denote more specific semantically and pragmatically complex contrary-to-fact possibilities, on a three member hierarchical system. These iconically more complex signs are transparent: they are composed of the (morphologically) positive or affirmative sign lu–(LOW POSSIBILITY) plus one or both negative morphemes: the prefix i- ‘negative’ (e.g. efshar ‘possible’ i-efshar ‘impossible’), and/or the suffix le which is written (lamed-aleph) just like the independent negative morpheme lo–‘no, not’ in an iconic tripartite hierarchical order — (i-lu-LOW, lu-le-MID, i-lu-le-HIGH — the more complex the form and the more negative markers it has, the higher it is on the negative, contrary-to-fact three member hierarchy. c. ilu= LOW MORPHOLOGICAL NEGATIVE CONTRARY TO FACT (13) ilu bata/hayita ba, hayita mevi li sefer. (past/ if came-you/were-you come, were-you bring to-me book compound) ‘If you had/would have come, you would have brought me a book. d. lule = MID MORPHOLOGICAL NEGATIVE CONTRARY TO FACT (14) lule bata/hayita ba, hayita mevi li sefer. (past/ (not)-if came-you/were-you come, were-you bring to-me book comp.) ‘If you hadn’t/wouldn’t have come, you would have brought me a book.

‘‘Conditionals’’ in Hebrew and English: same or different? 

e. ilule = HIGH MORPHOLOGICAL NEGATIVE CONTRARY TO FACT (15) ilule bata/hayita ba, hayita mevi li sefer. (past/ (not)-if came-you/were-you come, were-you bring to-me book comp.) ‘If you hadn’t/wouldn’t have come, you would have brought me a book. The reader will note that all of these forms are translated as if and the subtle semantic distinctions between them may be lost in the Hebrew-English translation. By the same token, the semantic distinction between if and whether in English may not necessarily be transferable into Hebrew as well. Therefore it should be clear that there may not necessarily be true equivalence in these systems of conditionals.

. Summary and conclusions One of the most fundamental problems in linguistics in general and in the study of linguistic typology and language universals in particular is the distinction between what is extralinguistic versus what is linguistic, or, in other words: which universals are part of language per se and which are reflections of universal principles of cognition, perception, or communication reflected in diverse ways in different languages. Indeed, this question of language universals is one of the key issues dividing traditional and neotraditional sentence-oriented and Saussurian, semiotic, or sign-oriented approaches to language. One might expect Hebrew to have a similar system for such a basic and fundamental universal linguistic category as conditionals. Like the English system, the Hebrew system of conditionals revolves around a similar semantic domain: exhaustively classifying alternative perceptions of possibilities. However, unlike the English system, the Hebrew system classifies these alternative perceptions according to a totally different set of semantic criteria. The Hebrew system classifies the shared semantic domain of alternative ways of presenting possibilities according to a different set of cognitive and perceptual extralinguistic criteria. It should be remembered that all of these criteria for classifying conditionals are based on extralinguistic factors which are not always encoded in the same way in all languages. As in the case of English and Hebrew, different extralinguistic criteria for categorizing conditionals may or may not have explicit linguistic signs that signal them in these languages. Therefore languages differ in terms of which extralinguistic criteria are formally incorporated in their sign systems and which extralinguistic factors are left to human inference and knowledge of the world. The Hebrew system is unlike the English system in other ways as well. Although the English conditionals (if/whether) have a strong tendency to collocate

 Yishai Tobin

with the same verb tenses (e.g. unreal past, conditional, future, historical subjunctive), this is not the case for the Hebrew system: im freely occurs with all tenses (past, present, future, compound) and strongly favors the future and has even been inaccurately called the ‘‘future conditional’’ (Bahat and Ron 1976, 1984); lu prefers the past, can appear in the compound and the present, and usually avoids the future; ilu/lule/ilule appear exclusively in the past and compound tenses. Not surprisingly, the unmarked form if appears much more frequently than the marked form whether in spoken and written texts of various styles and registers (Tobin 1990a: ch. 6). In Hebrew, im is the most frequently used form; followed by ilu; lu and lule have a similar frequency of occurrence; and the most iconically complex form ilule has the lowest frequency of occurrence. This nonrandom distribution and collocation of conditionals and tenses and their relative frequencies (culled from a corpus of twelve texts of diverse styles and registers) is not arbitrary and has been discussed elsewhere (Tobin 1990b: 486–9, 1994: 114). No matter how we look at these systems, we cannot escape from the fact that the same categorical label of conditionals is divided quite differently in English and Hebrew in the number and type of signals, the semantic features underlying the meanings of the signs comprising these language specific systems, and in their distribution and collocation. From the semiotic view of language, it is difficult to view both systems as being equivalent even though all the Hebrew forms can be translated as if, and both systems perform the same communicative function of presenting alternative ways of perceiving possibilities. If, as Saussure has taught us, everything in language is in a system of oppositions, then we should take these unique language specific oppositions into account before we make universal typological claims about crosslinguistic language categories. Or as Anthony Fox (1990: 5) in his book The Structure of German (reviewed by Christopher Beedham 1992) summarizes the basic tenets of European or Saussurian structuralism very accurately when he claims: . . . a language is seen as a self-contained formal system, where each part derives significance from its relation to the other parts and to the whole; hence all the parts are mutually defining. From this it follows that each language must be described in its own terms and not in terms of an earlier stage of its history or in terms of categories derived from another language. (Beedham 1992: 638)

The very idea of comparing and contrasting languages is as difficult as it is intriguing. English is generally assumed to be richer in its lexicon and more impoverished in its morphology and syntax than Hebrew. Yet, both languages are efficient and self-contained systems of communication serving their respective speech communities. The communication needs of human beings and the semantic or pragmatic domains they deal with are extralinguistic phenomena. How these extralinguistic phenomena are encoded in language and languages is

‘‘Conditionals’’ in Hebrew and English: same or different? 

multivaried and diverse. In some instances the English system seems more complex, and in other cases, such as conditionals, the Hebrew provides a wider variety of forms and constructions. A system in one language may rely on lexical items while bound morphology or syntagmatic signals are used in the other, or, possibly a mixture of all the above can be found in either language. Semantic or pragmatic domains may overlap in the traditional and neotraditional categories assigned to the same language or across languages. Sometimes the Hebrew systems may be more transparent or iconic because the signals share the same consonantal roots while the connections between the members of a system may be less transparent in English. The larger number of lexical items in English may appear to present more subtle semantic distinctions than Hebrew which may make more finely-grained distinctions through morphological and syntactic means. However, it may be said that the speakers of each language have chosen alternative means of expression within the basic synergetic principle inherent in all languages: to achieve maximum communication with minimal effort. Each language community solves this problem and reaches a compromise between the communication and human factors in a unique way. One can only laud these crosslinguistic diversities and declare Vive la différence!

References Andersen, P. K. 1991. A New Look at the Passive. Frankfurt: Peter Lang. Andrews, E. 1990. Markedness theory: The union of asymmetry and semiosis in language. Durham, NC and London: Duke University Press. Andrews, E. and Tobin, Y.. 1996. Towards a Calculus of Meaning: Studies in Markedness, Distinctive Features and Deixis. Amsterdam: John Benjamins. Aphek, E. and Tobin, Y. 1988. Word Systems in Modern Hebrew: Implications and Applications. Leiden: E. J. Brill. Aphek, E. and Tobin, Y. 1989. The Semiotics of Fortune-Telling. Amsterdam and Philadelphia: John Benjamins. Auwera, J. van der. 1997. ‘‘Pragmatics in the last quarter century: The case of conditional perfection’’. Journal of Pragmatics. 27: 261–74. Bahat, J. and Ron M.. 1976. ve-dayek! (Be Precise!) Tel-Aviv: Hakibbutz Hameuchad. Bahat, J. and Ron M. 1984. dikduk ivri (Hebrew Grammar). Tel-Aviv: Hakibbutz Hameuchad. Battistella, E. 1990. Markedness: The Evaluative Superstructure of Language. Albany, NY: SUNY Press. Battistella, E. 1996. The Logic of Markedness. Oxford: Oxford University Press. Beedham, C. 1992. ‘‘School Grammar and University Linguistics: A review Article’’. The Modern Language Review 87(3): 636–51.

 Yishai Tobin

Bolinger, D. 1971. The Phrasal Verb in English. Cambridge MA: Harvard University Press. Bolinger, D. 1977. Meaning and Form. London and New York: Longman. Clark, H. 1973. ‘‘Space, Time, Semantics and the Child. In T. F. Moore (ed.). Cognitive Development and the Acquisition of Language, 18–33. New York: Academic Press. Contini-Morava, E. 1989. Discourse Pragmatics and Semantic Organization: The Case of Negation and Tense-Aspect with Special Reference to Swahili. Berlin and New York: Mouton de Gruyter. Contini-Morava, E. and Goldberg B. S. (eds). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. New York and Berlin: Mouton de Gruyter. Contini-Morava, E. and Tobin Y. (eds). 2000. Between Grammar and Lexicon. Amsterdam and Philadelphia: John Benjamins. Enkvist, N. 1973. Linguistic Stylistics. The Hague: Mouton. Garcia, E. 1975. The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North Holland. Garcia, E., van Putte, F., Yishai Tobin. 1987. ‘‘Cross Linguistic Equivalence, Translatability and Contrastive Analysis’’. Folia Linguistica 21: 373–405. Givon, T. 1979. On Understanding Grammar. New York: Academic Press. Guillaume, G. 1945. L’architectonique du temps dans les langues classiques. Copenhagen: Munksgaard. Guillaume, G. 1971. Lecons de linguistique, 1948–1949, Serie B: psycho-systematique du langage: principes, methodes et applications I. Quebec/Paris: Presses de l’Université Laval/ Klinsieck. Guillaume, G. 1995. Meaning, data, and testing hypothesis. In: Contini-Morava and Sussman Goldberg, 153–68. Hirtle, W. 1982. Number and Inner Space: A Study of Grammatical Number in English. Quebec: Presses de l’Université Laval. Hirtle, Walter. 1995. ‘‘Meaning, data, and testing hypotheses.’’ In E. Contini-Morava and B. S. Goldberg, 1995, 152–68. Huffman, A. 1996. The Categories of Grammar: French lui and le. Amsterdam and Philadelphia: John Benjamins. Jakobson, R. 1957. Shifters, Verbal Categories and the Russian Verb. Cambridge, MA.: Harvard University Press. Kiparsky, P and Kiparsky C. 1971. ‘‘Fact’’. In D. Steinberg and L. A. Jakobovits (eds). Semantics: An Interdisciplinary Reader in Philosophy, Linguistics, and Psychology, 345– 69. Cambridge: Cambridge University Press. Kirsner, R. 1979. The Problem of Presentative Sentences in Modern Dutch. Amsterdam: North Holland. Klein, F. 1976. ‘‘Same’’ vs. ‘‘Different’’ Crosslinguistically: The Articles in English and Spanish’’. Papers from the Twelfth Regional Meeting of the Chicago Linguistic Society (CLS) 12. 413–24. Klein-Andreu, F. 1983. Discourse Perspectives in Syntax. New York: Academic Press. Kovacs, F. 1961. ‘‘A propos d’une loi semantique’’. Acta Linguistica Hungarica II(3–4): 405–11. Kronasser, H. 1952. Handbuch der Semasiologie. Heidelberg: Carl Winter Universitatsverlag.

‘‘Conditionals’’ in Hebrew and English: same or different?

Lattey, E. 1980. ‘‘Grammatical systems across languages: A study of participation in English, German and Spanish’’. City University of New York Ph.D. dissertation. Lattey, E. 1989. ‘‘On the Role of Form in Going from Sign to Text and from Text to Understanding. In: Tobin 1989, 45–59. Makkai. A. 1972. Idiom Structure in English. The Hague: Mouton. Pike, K. 1967. Language in Relation to a Unified Theory of the Structure of Human Behavior. The Hague: Mouton. Reid, W. 1991. Verb Number in English: A Functional Explanation. London and New York: Longman. Reid, W. and Otheguy, R. (eds) In press. Signal, Meaning and Message: Perspectives on SignOriented Linguistics. Amsterdam and Philadelphia: John Benjamins. Saussure, F. de. 1913/1959. A Course in General Linguistics. New York: Philosophical Library. Tobin, Y. 1990a. Semiotics and Linguistics. London and New York: Longman. Tobin, Y. 1990b. ‘‘The Future Tense in Modern Hebrew: From Sign to Text. Folia Linguistica 24(3/4): 1–56. Tobin, Y. 1992a. ‘‘Translatability: An Index of Cross System Linguistic, Textual and Historical Compatibility.’’ In Geschichte, System, Literarische Ubersetzung, H. Kittel (ed.), 307–22. Berlin: Erich Schmidt. Tobin, Y. 1992b. ‘‘Contrastive Linguistics: A Sign-Oriented Approach.’’ In New Departures in Contrastive Linguistics, M. Markus and Ch. Mair (eds), 311–35. Innsbruck: Institut fur Anglistik der Universitat Innsbruck. Tobin, Y. 1992c. ‘‘Semantic Integrality: A Universal Semiotic Feature of Language and Perception’’. In M. Kefer and J. van der Auwera (eds). Meaning and Grammar: Crosslinguistic Perspectives, 347–66. Berlin and New York: Mouton de Gruyter. Tobin, Y. 1993. Aspect in the English verb: Process and Result in Language. London and New York: Longman. Tobin, Y. 1994. Invariance, Markedness and Distinctive Feature Analysis: A Contrastive Study of Sign Systems in English and Hebrew. Amsterdam and Philadelphia: John Benjamins. Tobin, Y. 1997a. Phonology as Human Behavior: Theoretical Implications and Clinical Applications. Durham, NC and London: Duke University Press. Tobin, Y. 1997b. ‘‘Same vs. Different Crosslinguistically: The Articles in English, Spanish and Hebrew’’. In: R. Hickey and S. Puppel (eds). Language History and Linguistic Modelling. Berlin and New York: Mouton de Gruyter. Tobin, Y. 1998b. ‘‘Review: The Logic of Markedness by Edwin Battistella’’. Language 74(4), 832–934. Tobin, Y. 2000. ‘‘The Dual Number in Hebrew: Grammar or Lexicon, or Both?.’’ In Contini-Morava and Tobin (2000), 87–129. Traugott, E. C. 1974. ‘‘Explorations in Linguistic Elaboration: Language Change, Language Acquisition, and the Genesis of Spatio-Temporal Terms. In: J. Andersen and C. Jones (eds). Historical linguistics I, 263–314. Amsterdam: North Holland. Traugott, E. C. 1975. ‘‘Spatial Expressions of Tense and Temporal Sequencing: A Contribution to the Study of Semantic Fields. Semiotica 15(3): 205–30.



 Yishai Tobin

Traugott, E. C. 1978. ‘‘On the Expression of Spatio-Temporal Relations in Language. In J. Greenberg et al. (eds). Universals of human language, Volume 3, Word structure, 369–400. Stanford: Stanford University Press. Traugott, E. C., ter Meulen. A., Reilly, J. S., Ferguson, C. A. (eds). 1986. On Conditionals. London: Cambridge University Press. Traugott, E. C. and Heine B. 1991. Approaches to Grammaticalization. Amsterdam and Philadelphia: John Benjamins. Vonnegut, K.. Cat’s Cradle. London: Penguin. Waugh, L. and Rudy S. (eds). New Vistas in Grammar: Invariance vs. Variation. Amsterdam and Philadelphia: John Benjamins. Wierzbicka, A. 1972. Semantic Primitives. Frankfurt: Athenaum. Wierzbicka, A. 1980. Lingua Mentalis. New York: Academic Press.

Sur la paradigmatisation du verbe indo-européen (Deuxième partie)*

Tomáš Hoskovec Prague et Brno La première partie du présent mémoire a paru dans le 3e volume de cette nouvelle série des TCLP. Nous étudions le verbe indo-européen non comme des transformations successives d’un système primitif unique, mais comme des constructions particulières, entreprises, de façon individuelle et mutuellement indépendante, à partir d’un inventaire commun de matériaux morphologiques et morphosyntaxiques, formes et fonctions. Dans diverses langues, les matériaux primitifs i.–e.s ont été arrangés selon divers principes et hiérarchies en divers systèmes, ce qui donne l’individualité des langues i.–e.nes anciennes. Dans la deuxième partie, nous décrivons la constitution des paradigmes verbaux latin et grec (chapitre 7), balte et slave (chapitre 8).

7. L’opposition d’aspect et la processualité: le cas du grec et du latin Le cadre théorique de nos réflexions, donné au chapitre 3, présume deux stades consécutifs de l’indo-européen, le premier à flexion dérivationnelle (f-d), le second à flexion paradigmatique (f-p). Au chapitre 4, nous avons conjecturé, pour le stade à f-d, deux classes lexico-sémantiques du verbe, celles d’extraverti et d’introverti, sur lesquelles, au chapitre 5, nous avons bâti, à l’aide de l’opposition d’actualité, les premières catégories paradigmatiques. Au chapitre 6, nous avons présenté un dérivé morphologique particulier: le verbe thématique. Primitivement, il restait en dehors des catégories d’actualité. Dans certaines langues, le verbe thématique s’est soumis aux catégories paradigmatisées d’actualité jusqu’à s’y incorporer (voie par laquelle a disparu le subjonctif védique), mais nous allons voir dans ce chapitre 7 que dans d’autres langues, il a supplanté lesdites catégories en donnant naissance à des paradigmes neufs. . En dernière conjecture utilisée dans le présent mémoire, nous présumons avec Mme Kurzová (1993: 118), quoiqu’en terminologie différente, qu’à l’intérieur de

 Tomáš Hoskovec

chacune des deux classes lexico-sémantiques du verbe indo-européen commun, à savoir à l’extraverti et à l’introverti, a existé une sous-division, également d’ordre lexico-sémantique. L’extraverti aurait distingué un complexe d’un évolutif, l’introverti, un dynamique d’un statique. Dans les deux cas, les sous-classes en résultant étaient de nature aspectuelle sensu latiore; elles précisaient sous quel aspect se présentait le procès du verbe, tout en influençant le choix de contextes syntaxiques dans lesquels le verbe pouvait apparaître. Nous les présumons disjonctives et non nécessairement couvrant l’ensemble de leurs classes respectives. Illustration: Intuitivement, les deux sous-divisions peuvent être illustrées de la façon suivante: à supposer que les verbes français ‘parvenir’ et ‘marcher’ soient extravertis (procès du verbe s’accomplissant à partir du sujet et hors de lui, cf. 4.4), le premier est complexe vu sa sémantique lexicale (procès global, déterminé, présenté dans sa totalité), le deuxième évolutif (procès non-global, nondéterminé, présenté dans son développement); à supposer que les verbes français ‘tourner’ et ‘savoir’ soient introvertis (procès du verbe s’accomplissant dans le sujet et/ou en l’entourant), le premier désigne un procès dynamique, le deuxième, statique. Chose importante, une pareille intuition se manifeste dans le choix des contextes sémantico-syntaxiques où les verbes en question sont employés, technique de vérification que nous empruntons à M. Jan Šabršula (1962/65, 1999). Aussi peut-on dire, en parlant d’un déplacement vers un but, d’une part, les alpinistes sont parvenus au refuge de quoi il s’ensuit que les alpinistes ont atteint leur but, et d’autre part, les alpinistes ont marché vers le (mais non *au) refuge ce qui n’implique point qu’ils aient atteint leur but, cf. les alpinistes ont marché vers le refuge sans y parvenir. De même, on peut dire il a (ou bien il s’est) tourné brusquement, subitement, soudainement, mais non *il a su brusquement, subitement, à moins de modifier la valeur lexicale de ‘savoir’ dans le sens de ‘venir à savoir’, de sorte que ces syntagmes équivalent à dire il a appris subitement que . . . ; il s’est soudainement rendu compte de . . . ; il a compris brusquement ce qui se passait. Digression méthodologique: L’appareil notionnel des classes et des sousclasses lexico-sémantiques du verbe i.–e. tel que nous venons de le présenter est une hypothèse. Nous recommandons au lecteur de prendre ce mot au sens étymologique grec: ßB`h,F4l ‘ce que l’on met dessous pour bâtir là-dessus’. Notre tâche ne se résumera pas à vérifier l’hypothèse adoptée; il n’y a aucun moyen d’aller chercher un indo-européen primitif en vue de le comparer à notre appareil notionnel. Nous sommes, certes, tenu de justifier le choix de l’hypothèse: il ne faut pas qu’elle contredise ce qui est connu des langues anciennes. Or, le travail propre du linguiste consiste à construire une compréhension de la langue . . . et à le faire en pleine conscience des bases sur lesquelles il bâtit. .. La coexistence du dynamique et du statique au sein de l’introverti n’aurait été possible que grâce à la grande neutralité des deux classes principales d’extra-

Sur la paradigmatisation du verbe indo-européen 

verti et d’introverti vis-à-vis du rapport entre le sujet syntaxique et l’agent sémantique, cf. 4.1 et 4.4. Tandis que les classes lexico-sémantiques d’extraverti et d’introverti furent distinguées morphologiquement par deux séries différentes de désinences, leurs sous-classes respectives, telles que nous les supposons, ne le furent pas; le complexe et l’évolutif se servaient des mêmes désinences extraverties, le dynamique et le statique, des introverties. Cette hypothèse s’appuie sur l’origine commune, bien visible en grec et en indo-iranien, des désinences de l’aoriste et du présent (qui à la voix active résultent du complexe et de l’évolutif respectivement) d’un côté, et des désinences du moyen et du parfait actif (qui résultent respectivement du dynamique et du statique) de l’autre. .. Les termes complexe et évolutif soulignent le caractère différentiel, nonoppositionnel desdites sous-classes de l’extraverti du stade à f-d. Cette différence pouvait revêtir, au stade à f-p, le caractère oppositionnel de perfectif (remplaçant le complexe) et de non-perfectif (remplaçant l’évolutif). De même, les termes dynamique et statique relèvent du caractère différentiel, non-oppositionnel des deux sous-classes de l’introverti au stade à f-d, tandis qu’au stade suivant à f-p, elles pouvaient se changer en sous-classes oppositionnelles de nonprogressif (à base du statique) et de progressif (à base du dynamique). Les classes différentielles d’extraverti et d’introverti changent elles aussi, au stade à f-p, en donnant naissance aux classes oppositionnelles d’actif (à base de l’extraverti) et d’inactif (à base de l’introverti). Or, toutes les différences d’ordre lexico-sémantique n’ont pas été érigées en oppositions morphologiques, toutes les oppositions morphologiques n’ont pas atteint le même niveau parmi les nouvelles catégories paradigmatiques. C’est précisément ce en quoi consiste l’individualité du procès de paradigmatisation, ce qui fait la différence entre les langues i.–e.nes individuelles et ce que nous allons observer, à ce chapitre, au sujet des langues latine et grecque. .. Il est caractéristique des systèmes paradigmatiques latin et grec de morphologie verbale qu’ils «complètent» le verbe en tant qu’unité lexicale. Cela veut dire qu’ils pourvoient un verbe issu de la sous-classe lexico-sémantique d’extraverti évolutif (paradigmatisé en non-perfectif) d’un homologue extraverti complexe (en guise de perfectif), voire d’un introverti dynamique ou statique (paradigmatisés respectivement en inactif progressif ou non-progressif). La même chose arrive mutatis mutandis aux verbes issus des autres sous-classes. Il ne sera point notre objectif de reconstruire ici les sous-classes d’origine des verbes latins et grecs, mais de mettre en lumière le procès de paradigmatisation en latin et en grec. Exemples: Au sein d’une unité lexicale, il est logique de prendre pour primitif le thème qui est morphologiquement le moins chargé. Considérons le grec §-$0-< à côté de $"-\-<-T et $V-F6-T pour ‘se mettre en marche’. Du fait que les deux thèmes de présent sont marqués par rapport à celui d’aoriste on peut

 Tomáš Hoskovec

conclure que l’unité lexicale primitive fut jadis un extraverti complexe (→ actif perfectif) qui servit de base à la dérivation secondaire de deux extravertis évolutifs (→ actifs non-perfectifs), l’un à -<-j ,/@-, l’autre à -F6- ,/@. Par contre, la paire 8X(-T et §-8,>-"=§-8,(-F-" pour ‘ramasser’ fait penser que le verbe fut primitivement un extraverti évolutif (→ actif non-perfectif), pourvu ultérieurement d’un actif perfectif à -F-, ce qui vaut également pour le latin dı¯c-i-t à côté de dı¯c-s-i-t = dı¯xit pour ‘dire’. Le thème non-marqué de uı¯d-ı¯ à côté de uid-e-o¯ pour ‘voir’ admet pour explication que le verbe latin (actuellement actif nonperfectif) est d’origine introvertie. La comparaison de linqu-it à lı¯qu-it pour ‘rester’ mène à la même conclusion. C’est également à partir du thème morphologiquement le moins chargé, donc primitif, que sont produits les adjectifs verbaux en -to-, cf. $"-J`H dérivant d’un ancien extraverti complexe, 8,6-J`H dérivant d’un ancien extraverti évolutif (de même que le¯c-tus, dic-tus) et lic-tus, uı¯-sus < *uı¯s-sus < *uı¯d-tos dérivant d’anciens introvertis. Dans une paire latine comme tang-o¯ :: tetig-ı¯ les deux membres paraissent marqués, mais nous avons le témoignage des formes archaïques tag-a-m, at-tig-a¯-s (au lieu de tangam, attinga¯s du latin classique) pour ‘toucher’ qui prouvent le caractère secondaire et du thème de présent et de la réduplication au thème de parfait, ce dont témoigne également le participe tac-tus. .. En indo-européen commun, les quatre sous-classes lexico-sémantiques du verbe furent différentielles, non-oppositionnelles. Là-dessus, le latin et le grec bâtirent des catégories morphologiques oppositionnelles. Mme Kurzová (1993) montre que le premier pas vers la structuration latine et grecque consiste en l’établissement du présent thématique qui a remplacé à la fois l’ancien «présent athématique» (actualisé nvnc) et l’ancien «injonctif» (sans actualisation), cf. 5.2 pour le contraste de MA¯-M-I à MA¯-M. La confusion des deux membres oppositionnels a ensuite obligé le système verbal, ajoute-t-elle, à créer des subjonctifs, moyens neufs, pour exprimer la non-actualité du procès de verbe. En latin aussi bien qu’en grec, le présent thématique s’est placé au-dessus de la dichotomie primitive d’extraverti et d’introverti. Cependant, les catégories paradigmatiques des deux langues se sont réarrangées par différents systèmes d’oppositions (cf. 7.2 et 7.3). Les oppositions étaient d’ordre aspectuel et c’est sur elles que se sont superposés, comme une deuxième couche, les systèmes temporels latin et grec, également différents entre eux (cf. 7.5). . Le latin a systématisé la différence entre le complexe (perfectif) et l’évolutif (non-perfectif) en opposition de perfectivité. Issue du domaine de l’ancien extraverti, l’opposition de perfectivité a été prolongée dans le domaine de l’ancien introverti. C’est sur elle que sont bâtis les deux sous-systèmes du verbe latin, le perfectum (sur le perfectif) et l’infectum (sur le non-perfectif), qui se manifestent par deux thèmes verbaux différents.

Sur la paradigmatisation du verbe indo-européen 

.. Lors de sa paradigmatisation, le latin prend, en premier pas, l’axe qui séparait, originellement, le complexe de l’évolutif au sein de l’extraverti, et l’ayant réévalué en opposition de perfectif (← complexe) : : non-perfectif (← évolutif), l’applique à la classe d’introverti, marquant ainsi les introvertis statiques comme perfectifs et les dynamiques comme non-perfectifs. En deuxième pas, la langue identifie l’ancien introverti statique à l’ancien extraverti complexe (de quoi va naître le «parfait») et relie l’ancien introverti dynamique (→ «présent médiopassif») à l’ancien extraverti évolutif (→ «présent actif»). L’identification de l’ancien introverti statique à l’ancien extraverti complexe s’est faite, à base de l’opposition de perfectivité, sous l’aspect perfectif que les deux sous-classes primitives ont reçu en commun, et à force de concevoir tout état (la notion d’un verbe d’état étant originaire de l’introverti statique) comme résultant d’un procès achevé (donc englobé, déterminé; cette vision du verbe étant originaire de l’extraverti complexe). L’union des deux anciennes sousclasses donne le «perfectum» latin, une catégorie primitive, marquée par un thème particulier et par des désinences particulières, à partir de laquelle se développera ensuite un sous-système de catégories morphologiques du verbe latin. La double origine de cette catégorie de base se voit bien dans les deux sens du parfait latin uixit, à savoir ‘(il) a vécu [un nombre d’années]’ (extraverti complexe) et ‘(il) est perdu [=(il) a perdu sa vie]’ (introverti statique). De l’optique concevant un état comme résultat d’un procès achevé témoignent, à leur tour, les perfectiua tantum tel que odi ‘je hais’ [perfectif], par rapport à amo ‘j’aime’ [nonperfectif], ou memini ‘je retiens présent à la mémoire’ [perfectif], par rapport à scio ‘je sais’ [non-perfectif]. Le rapprochement des anciennes sous-classes d’introverti dynamique et d’extraverti évolutif s’est fait, lui aussi, à base de l’opposition de perfectivité, cette fois-ci sous l’aspect non-perfectif, mais il n’y a pas eu d’autre véhicule. Ensemble, elles constituent la catégorie primitive d’«infectum», base d’un autre système de catégories morphologiques verbales. L’infectum et le perfectum sont les deux premières classes oppositionnelles du verbe latin. Elles ne sont pas symétriques: le perfectum a homogénéisé ses deux sous-ensembles originels, l’infectum ne l’a pas fait. Ces derniers seront opposés l’un à l’autre par un nouveau trait sémantique, activité, issu de la différence primitive et originelle d’extraverti et d’introverti. L’ancien extraverti évolutif sera marqué actif, l’ancien introverti dynamique, inactif. L’opposition d’actif :: inactif restera limitée à l’infectum (cf. 7.2.3). .. La forme prototypique de l’infectum était le présent thématique actif. En vertu de sa valeur aspectuelle non-perfective (conception non-globale, nondéterminée du procès de verbe), le présent thématique servait de contrepartie, dans le contexte oppositionnel d’infectum : : perfectum, aux deux anciennes sousclasses ayant revêtu l’aspect perfectif, c’est-à-dire aux anciens extravertis com-

 Tomáš Hoskovec

plexes (uenio à ueni pour ‘venir’) et aux anciens introvertis statiques (linquo à liqui pour ‘laisser, rester’; uideo à uidi pour ‘voir’). La dérivation du présent thématique n’était pourtant pas univoque, les résultats revêtant parfois des traits sémantiques particuliers, tels que «verbe d’état», «verbe ingressif», «verbe inchoatif», tous opposés à un «verbe évolutif» neutre. Plus d’un verbe de perfectum (pf) se voit ainsi confronté à deux verbes d’infectum (if) différents, cf. sedi (pf) ‘je me suis assis’ → sedeo (if: verbe d’état) ‘je suis, me trouve assis’ & sido (if: évolutif) ‘je m’assois’. Exemples: Parmi d’autres cas d’asymétrie entre l’infectum et le perfectum, citons scio (if: évolutif) ‘je sais’ & scisco (if: inchoatif) ‘j’apprends’, opposés les deux à l’unique sciui (pf) ‘j’ai su/appris’; seneo (if: verbe d’état) ‘je suis faible comme un vieillard’ & senesco (if: inchoatif) ‘je deviens vieux’, opposés les deux à l’unique senui (pf) ‘j’ai vieilli’. Observation: Et le présent thématique (if) et le parfait (pf) sont donc habilités à exprimer un état. À défaut d’opposition aspectuelle, le présent thématique, développé originairement au-dehors de la différence primitive d’extraverti :: introverti, peut aller jusqu’à éliminer ce dernier sous sa forme paradigmatisée de perfectum. En témoignent encore des verbes comme gaudeo ‘je me réjouis’, soleo ‘je suis habitué’ dont les parfaits archaïques gauisi, solui ont disparu pour être remplacés, ultérieurement (sous forme d’opposition temporelle), par les tournures périphrastiques gauisus sum ‘je me suis réjoui’ et solitus sum ‘j’ai pris habitude’. .. Limité à l’infectum seulement, l’ancien introverti dynamique, classé inactif, s’oppose en latin à l’ancien extraverti évolutif, classé actif. Il n’exprime pas un aspect du procès du verbe, mais signale un rapport syntaxique particulier, à savoir une «disparition» de l’agent humain (ou conçu comme tel) du procès de verbe, disparu parce que devenu vague ou général, cf. sedetur ‘on est assis’, uideor ‘on me voit’. Le rôle paradigmatique de l’inactif latin consiste à produire des formes déagentives à partir des verbes imperfectifs de base (actifs). C’est pourquoi nous appelons le «passif» latin déagentif. Des traces de l’ancien inactif dynamique, verbes sans contrepartie active, restent notoirement visibles dans les «déponents», e.g. sequor ‘je (pour)suis’, (g)nascor ‘je nais’, morior ‘je meurs’. Marque de marginalité, l’inactif latin développe des désinences particulières ayant recours à l’élément -r qui, dans l’inventaire primitif i.–e., se tenait à part des désinences extraverties et introverties (voir remarque en 4.3). Remarque: Il n’y a jamais eu de forme synthétique d’inactif au sein du perfectum latin, la forme périphrastique étant secondaire. Le «passif» périphrastique, ajouté au perfectum plus tard par analogie, fut créé sur la base de syntagmes statiques. La paradigmatisation du perfectum eut donc un nouveau recours à l’identification d’une action accomplie à l’état en résultant (cf. 7.2.1): l’inactif (déagentif) périphrastique désigne un état en le présentant comme le résultat

Sur la paradigmatisation du verbe indo-européen 

d’une action achevée pour référer ainsi à l’action même. L’opposition d’actif : : inactif au sein du perfectum est donc issue du même schéma sémantique duquel était née l’opposition d’infectum :: perfectum. Rappel: L’ajout que présente, au perfectum latin, l’inactif périphrastique est lié à la paradigmatisation de l’adjectif verbal en -to (à allomorphe -so) en participe dit «passif» (participium perfecti passiui), paradigmatisation qui n’a pas eu lieu en grec. .. Pour expliquer le système des formes verbales en latin nous avons introduit l’opposition de perfectivité et, à l’aide de celle-ci, décrit les deux sous-systèmes oppositionnels du verbe latin, à savoir l’infectum et le perfectum. Ces deux soussystèmes sont caractérisés par deux thèmes verbaux différents. Or pour fléchir un verbe latin, il faut en connaître encore un troisième. Nous avons vu en 7.2.2 que le thème d’infectum peut être morphologiquement marqué par rapport à celui de perfectum, et nous allons voir en 7.4.2 que le cas inverse (thème de perfectum marqué morphologiquement par rapport à celui d’infectum) est également possible. Le troisième thème, celui de supin (il vaudrait mieux l’appeler «thème de participe passif», parce que c’est là que se manifeste son importance morphologique) qui échappe à la dichotomie aspectuelle d’infectum et de perfectum, peut être morphologiquement le moins marqué par rapport aux deux précédents, comme nous l’avons vu aux exemples 7.1.3 avec tac-t-um opposé à tang-i-t (if) et à tetig-i-t (pf). . Le grec emploie deux oppositions: celle de perfectivité (issue du contraste complexe vs. évolutif, le premier membre devenant marqué perfectif) et celle de non-progressivité (issue du contraste dynamique vs. statique, le deuxième membre devenant marqué non-progressif). Or du croisement des deux oppositions ne résultent pas quatre catégories (paradigmatisant ainsi les quatre sous-classes i.–e.nes primitives), mais seulement trois. Ceci est dû au fait que, contrairement au cas du latin, le perfectif reste dans le domaine de l’ancien extraverti et, par conséquent, tout l’ancien introverti se voit qualifié de non-perfectif; en même temps, le nonprogressif restant limité à l’ancien introverti, tout l’ancien extraverti est qualifié de progressif. Au sein du progressif non-perfectif, la langue garde l’ancien contraste extraverti vs. introverti qu’elle paradigmatise en opposition d’activité. .. L’ancien introverti statique (→ «parfait») se trouve opposé, d’un côté, à l’ancien extraverti complexe (→ «aoriste») en vertu de sa non-perfectivité qu’il partage avec l’ancien extraverti évolutif (→ «présent»). Or, le même introverti statique diffère de l’ancien extraverti évolutif en vertu non de son inactivité, mais de sa non-progressivité (procès statique opposé à procès en cours), laquelle nonprogressivité l’oppose aussi bien à l’ancien introverti dynamique (→ «moyen») et, naturellement, à l’ancien extraverti complexe (→ «aoriste»). La progressivité réunit donc l’ancien introverti dynamique (→ «moyen») à la fois à l’ancien extraverti complexe (→ «aoriste») et à l’ancien extraverti évolutif (→ «présent»).

 Tomáš Hoskovec

La non-perfectivité, à son tour, réunit l’ancien extraverti évolutif (→ «présent») aux anciens introvertis statique et dynamique (→ «parfait» et «moyen»). Tout compte fait, le non-perfectif et le progressif sont deux catégories sémantiquement très hétérogènes. Par contre, le perfectif contient seulement l’ancien extraverti complexe (→ «aoriste») et le non-progressif contient seulement l’ancien introverti statique (→ «parfait»), ce qui en fait deux catégories sémantiquement bien homogènes. Leur homogénéité permet de les grammaticaliser comme deux aspects marqués. Ce qui reste, c’est-à-dire le progressif non-perfectif (contenant et l’ancien extraverti évolutif et l’ancien introverti dynamique) servira aux deux précédents d’aspect non-marqué. Dans ce qui suivra, le progressif non-perfectif sera appelé processuel. La répartition du verbe grec en trois classes oppositionnelles d’aspect (perfectif, non-progressif et processuel), exprimées par trois thèmes différents, servira de base au paradigme catégoriel du verbe grec. Progressivement, un ancien introverti statique, devenu non-progressif, tel que 8X-8@4B" (le redoublement 8,- est secondaire),1 sera complété à la fois d’un processuel 8,\BT et d’un perfectif §-84B@<, toujours au sens de ‘rester, laisser’. Le complètement se fera mutatis mutandis pour des verbes originaires des autres sous-classes lexico-sémantiques primitives, mais il sera très lent et laissera grand nombre de verbes «défectifs». .. Au sein du processuel (=progressif non-perfectif) se rencontrent deux sousclasses lexico-sémantiques primitives, celles d’extraverti évolutif et d’introverti dynamique. Elles échappent aux deux oppositions paradigmatisées (perfectivité et non-progressivité) et ne peuvent être opposées qu’à l’aide de la différence primitive extraverti vs. introverti, élevée en opposition paradigmatique actif : : inactif. En résultera un processuel actif (→ «présent actif») et un processuel inactif (→ «présent moyen»). Ce dernier développera ensuite des désinences particulières, différentes de celles du présent thématique. Remarque: Nous interprétons les désinences du moyen grec comme résultant de l’adjonction de la série introvertie à la série extravertie (4.3), cf. 1ère sg.prim -mai < -m-H2 e-i (désinence extravertie + désinence introvertie + affixe d’actualisation), 3e sg.prim -toi < -t-o-i (désinence extravertie + désinence introvertie + affixe d’actualisation), 3e sg.sec -to < -t-o (désinence extravertie + désinence introvertie). Aux désinences primaires, l’attique généralise ultérieurement le vocalisme -a-, cf. -sai, -tai, -ntai. La restitution de -o- est due à Martín Sánchez Ruipérez (1952). D’origine non-perfective, l’opposition d’activité s’élargira au perfectif, donnant naissance aux aoristes actif et moyen. Cependant, au sein du perfectif, l’opposition morphologique d’activité est une couche accessoire qui n’a aucunement influencé tous les aoristes «au sens passif et aux désinences actives». Au sein du non-progressif («parfait»), l’opposition morphologique d’actif et d’inactif ne s’est pleinement développée qu’à l’époque hellénique, ce qui a rendu

Sur la paradigmatisation du verbe indo-européen

le non-progressif (1o) parfaitement parallèle au processuel et au perfectif, (2o) superflu parce que peu différencié par rapport aux deux précédents. En fait, la ‘‘riche floraison des formes du parfait en annonce dès le IIIe siècle [avant J.C.] la prochaine disparition’’, comme le dit Pierre Chantraine (1926: 254). .. L’inactif (moyen) grec partage avec l’actif le trait sémantique d’agentivité (dû apparemment à son appartenance au progressif), tandis que l’inactif (passif) latin est déagentif (la progressivité faisant défaut au système oppositionnel latin). Comparons les inactifs linquor ‘on m’abandonne’, dicor ‘on dit à mon sujet’ du côté latin, et les inactifs 8,\B@:"4 ‘je reste en arrière’, *,\6




Tomáš Hoskovec

matique de présent actif. Au fait, le processuel grec s’est généralisé sous forme thématique pure, 8,\BT, au détriment de la forme 84:BV", intel-lexi — §-8,>", rexi — ê-D,>". Originellement, l’élargissement -s- fut un moyen dérivationnel spécifique de certains verbes extravertis, et en effet, la dérivation grecque s’est arrêtée là, les verbes FJX(T — §FJ,>", 8X(T — §8,>", ÏDX(T — êD,>" n’ayant jamais développé de parfait (inactif non-progressif). L’ancien extraverti évolutif (→ actif non-perfectif) fero — nXDT ‘je porte’ n’a même pas fait dériver de perfectif, le latin ayant recours à tuli, le grec à ³<,(6@ §FB,4D" (présent @B,\DT < FB,D-j-T) ‘je semai’ et latin spar-s-ı¯ (présent spar-g-o¯) ‘je dispersai’, cf. lat. par-s-ı¯ (allomorphe de pe-per-c-ı¯, présent par-c-o¯) ‘j’épargnai’.

Sur la paradigmatisation du verbe indo-européen

.. La richesse dérivationnelle du verbe grec dépasse de loin les besoins des paradigmes catégoriels. Bien des racines font dériver plusieurs aoristes (perfectifs) qui sont différenciés sémantiquement, cf. ¦(,ÃD"4 ‘réveiller’ — ¦(DXFh"4 ‘s’éveiller’ — ¦(,Dh−<"4 ‘être (mis) en état de veille’. Inversement, d’une racine donnant un seul perfectif grec, tel §-FP-@-<, peuvent dériver plusieurs processuels comme §P-T ‘j’ai’ et ÇFP-T ‘je (re)tiens’ (pour le latin, rappelons les exemples pareils donnés en 7.2.2). Remarque: La dérivation radicale n’était pas vouée uniquement à compléter les catégories paradigmatisées. Elle permettait également de doter un perfectif d’un autre perfectif, cf. §-FJ0-< ‘je me plaçai debout’ → §-FJ0-F(") ‘je levai, dressai’; §-nL¯ -< ‘je naquis’ → §-nL-F(") ‘j’engendrai’ (à comparer les participes substantivés Ò nbl ‘fils’ = ‘celui qui est né’, Ò nbF"H ‘père’=‘celui qui a fait naître’). Les divers aoristes dérivant de la même racine diffèrent le plus souvent en valence verbale, différence qu’il ne faut pas confondre avec celle de diathèse. L’opposition de diathèse externe :: interne rend paradigmatique l’ancienne différence sémantique entre l’angle de vue présentant le procès du verbe comme dirigé hors du sujet (→ externe) vs. celui présentant le procès du verbe comme s’accomplissant sans quitter le sujet (→ interne), cf. 4.4. Si, ceteris paribus, deux formes verbales sont différenciées par leur valence,3 elles peuvent partager la même diathèse. Aussi préférons-nous expliquer le «passif» grec en -(h)0- hors de la diathèse, dans un large contexte d’activités morphologiques à la fois lexicales et grammaticales (cf. Hoskovec 1995). . Les oppositions aspectuelles sensu latiore (7.1) employées au début de la paradigmatisation — celle de perfectivité en latin (7.2), celles de perfectivité et de non-progressivité en grec (7.3) — concernaient à la fois l’ancien extraverti et introverti. Dans les systèmes paradigmatisés du latin et du grec, l’inactif — que ce fût en guise de déagentif ou de moyen — s’est détaché des autres oppositions qui concernent depuis lors l’aspectualité sensu stricto du procès du verbe (perfectif :: non-perfectif en latin, perfectif :: processuel :: non-progressif en grec). L’aspectualité latine repose sur deux thèmes verbaux (ceux d’infectum et de perfectum), la grecque sur trois (ceux d’aoriste, de présent et de parfait). Regardons dans ce contexte la catégorie du temps. .. Appuyé sur l’étude de l’usage des temps du verbe en latin classique, M. Guy Serbat (1976) critique avec justesse le tableau parfaitement symétrique, présenté par Antoine Meillet (cf. Antoine Meillet et Joseph Vendryes 1924: §§387–94), de deux séries temporelles parallèles, basées sur deux aspects, à savoir uidebam (imperfectum) — uideo (praesens) — uidebo (futurum) au sein de l’infectum et uideram (plusquamperfectum) — uidi (perfectum) — uidero (futurum exactum) au sein du perfectum. Cependant, il faut concéder le constat que les deux séries sont un fait de morphologie dérivationnelle: deux thèmes (ceux d’infectum et de



 Tomáš Hoskovec

perfectum, cf. uı˘d-e-o¯ et uı¯d-ı¯) développent chacun un temps marqué de passé (à valeur ‘‘ne-plus-actuel’’) et un temps marqué de futur (à valeur ‘‘ne-pas-encoreactuel’’). Les temps ajoutés sont tous morphologiquement homogènes, sauf pour le futur d’infectum où nous rencontrons dic-s-o, dic-s-is (non paradigmatisé), dicam, dic-es (paradigmatisé), dic-a-b-o, dic-a-b-is (paradigmatisé).4 L’opposition d’infectum et de perfectum était primordiale pour la paradigmatisation du verbe latin. Or, en vertu du principe sémantique alors appliqué, un état atteint (perfectum) a été identifié à une action achevée y aboutissant. Une fois introduites les deux oppositions d’actualité (à savoir, ‘‘ne-plus’’ et ‘‘pas-encore’’), le parfait (perfectum) se situe forcément, par rapport au présent thématique (infectum), comme un temps ne-plus-actuel vis-à-vis d’un temps neutre quant à l’actualité. Par ce biais, les deux séries temporelles se rencontrent et les temps du perfectum doivent dorénavant délimiter leur position respective par rapport à ceux infectum. Le latin classique a réarrangé l’ensemble de ses formes d’après le schéma bien connu, le latin vulgaire l’a réduit. L’opposition aspectuelle qui avait provoqué la collision des deux séries temporelles, conçues primitivement comme indépendantes, s’en est en conséquence ternie. Rappel: Le schéma des temps en latin classique est grossièrement le suivant: présent = temps neutre, implicitement à nvnc; parfait = passé pur et simple par rapport à nvnc; imparfait = passé concomitant un point de référence situé avant nvnc; plus-que-parfait = passé précédent un point de référence situé avant nvnc; futur = temps simple après nvnc; futur antérieur = temps futur précédent un point de référence situé après nvnc. ... Le temps et l’aspect du verbe sont deux catégories indépendantes. Nous voyons un excellent argument pour leur indépendance précisément dans la facilité avec laquelle ils s’associent et se dissocient, comme nous venons de le décrire pour le latin. Rappelons-nous encore une fois le védique et l’avestique où l’aoriste et l’imparfait ne sont que deux formes de prétérit, la deuxième (imparfait) désignant surtout un passé «révolu», «bien éloigné» (cf. la remarque 1.1). Il n’y a aucune raison pour penser que ces deux langues aient perdu une opposition aspectuelle primitive entre leurs aoristes et imparfaits. Une opposition aspectuelle peut aussi bien ne jamais s’être développée, c’est-à-dire que la différence entre le complexe et l’évolutif, postulée au stade à f-d pour le verbe actif (cf. 7.1), n’aurait jamais été paradigmatisée dans les langues indo-iraniennes anciennes (cf. 5.4). Ajoutons-y que l’«aoriste» et l’«imparfait» des vieilles langues slaves (cf. ci-bas 8.3.1) indiquaient respectivement un passé simple et un passé concomitant et n’ont été évalués quant à l’aspect qu’ultérieurement et seulement dans certaines des langues slaves. ... L’élan du système primitif de temps du latin (deux séries parallèles ante– nvnc–post) se reflète aussi dans le redoublement des subjonctifs, fait paradigma-

Sur la paradigmatisation du verbe indo-européen

tique qui n’a pas d’homologue ailleurs. Chacun des sous-systèmes (infectum & perfectum) en dispose d’un pour marquer la non-réalité nvnc et d’un autre pour marquer la non-réalité ante. Suite à l’affaiblissement des deux aspects, une confusion assez grande règne dans l’usage des subjonctifs du présent et du parfait, tandis que ceux de l’imparfait et du plus-que-parfait se sont transformés en irréel présent et passé respectivement. .. Le système tripartite aspectuel du verbe grec, opposant §84B@< (perfectif) à 8,\BT (processuel) d’un côté et à 8X8@4B" (non-progressif) de l’autre, n’a pas servi de support à des séries temporelles systématisées: l’aspect perfectif ne dispose que d’un seul membre (aoriste), donc point de relation temporelle; l’aspect processuel connaît un présent et un prétérit (imparfait); l’aspect non-progressif (parfait) est le seul à avoir développé, secondairement, un temps d’avant (plusquamperfectum) et un temps d’après (futurum exactum), les deux temps ajoutés étant très marginaux. Contrairement au latin, le temps du verbe grec n’est distingué qu’à l’indicatif. ... La langue garde l’ancien préfixe e- et s’en sert, en l’accompagnant des désinences secondaires (qui sont au fait morphologiquement superflues), pour marquer un procès révolu. Chose importante, le marquage du procès révolu, ‘‘actuel-au-temps-jadis’’, ne prend pas en considération la tripartition aspectuelle. Aussi l’aoriste se trouve-t-il marqué (à l’indicatif) sans avoir à son côté un temps non-marqué du même aspect. Contrairement à l’état védique et/ou avestique, l’opposition de l’aoriste à l’imparfait (les deux prétérits étant marqués ‘‘ne-plusactuel’’) est de nature aspectuelle. Ce sera le parfait, primitivement non marqué quant à l’actualité, qui perdra progressivement sa propre valeur d’aspect au profit de l’aoriste et finira, en grec moderne, absorbé par ce dernier, évolution décrite par Pierre Chantraine (1926). ... Le futur grec reste en dehors du système des thèmes aspectuels ici traités et ne représente, jusqu’à l’époque attique classique, qu’une facette modale du présent. La dérivation du futur est probablement originaire d’une dérivation de variantes intentionnelles/volontatives des verbes pour exprimer désir, décision, intention d’exécuter leur procès respectif. Les verbes intentionnels sont de diathèse interne, cf. §8B@:"4 ‘j’espère’, BD@hL:@Ø:"4 ‘j’aspire’, *4"J,\<@:"4 ‘je m’efforce’, etc. Aussi trouve-t-on à côté d’un présent actif (diathèse externe) •6@bT ‘j’entends’ un futur moyen (diathèse interne) •6@bF@:"4 ‘je vais/veux entendre, j’entendrai’, bien qu’au présent, la diathèse externe •6@bT ‘j’entends’ s’oppose à l’interne •6@b@:"4 ‘j’obéis’. Exemples: Prenons deux anciens extravertis complexes (→ actifs perfectifs) §8"$@< et §$0< qui font dériver, chacun, plusieurs non-perfectifs (cf. 7.4.3), dont un à modalité spéciale qui deviendra le futur de l’unité lexicale. De la racine



 Tomáš Hoskovec

*lagw-, c’est-à-dire *leH2-gw- (cf. §-8"$-@-<), dérivent 8":$V" et un inactif non-progressif *X*,4P", tandis que le latin dico ‘je dis’ se «contente» d’un seul dixi; de même, à un ancien introverti statique comme liqui ‘je suis resté’ il suffit d’opposer, en latin, un seul non-perfectif linquo, tandis que le grec (8X)8@4B" nécessite à la fois un processuel 8,\BT et un perfectif §84B@<. Le surcroît de la dérivation grecque est le mieux visible aux verbes secondaires qui sont parfaitement réguliers: à partir du grec n48ä < n48XT ‘j’aime’ sont produits régulièrement deux dérivés, ¦n\80F" et B,n\806", tandis que le latin amo a un seul amaui. Nous ne pouvons que souscrire à la thèse de M. I. A.Perel’muter (1977) montrant comment plusieurs traits morphologiques que la tradition tenait pour particulièrement «archaïques» dans le sens ‘remontant à un indo-européen primitif commun’ sont issus d’«innovations», c’est-à-dire de constructions particulièrement et proprement grecques.

8. Systématisation de classes lexico-sémantiques: le verbe balte et slave À propos des langues baltes et slaves on dit d’habitude que contrairement à leur déclinaison, restée bien fidèle à l’état hypothétique indo-européen, leur verbe s’en serait considérablement éloigné à force d’avoir subi de trop nombreuses innovations. Il est vrai que nous n’y trouvons point de catégories morphologiques mentionnées aux chapitres 5 et 7. Cependant, leur manque est moins dû à de prétendues réfections ultérieures qu’à une absence totale des processus de paradigmatisation qui sont survenus en grec et/ou en latin. Bien que nous rencontrions, dans les langues baltes et slaves, des dérivations morphologiques parallèles à celles connues dans les langues classiques, rien ne nous autorise à conclure qu’il y a jamais eu, pour le verbe balte et slave, des paradigmes morphologiques à la «gréco-latine».

Sur la paradigmatisation du verbe indo-européen

8.1 Commençons par le lituanien qui ne cesse d’impressionner les linguistes par sa richesse en moyens «archaïques» de dérivation de thèmes verbaux, tels que apophonie de racine, suffixes et élargissements radicaux. Or ce qui rend cette langue intrigante, c’est la nature de ses moyens morphologiques qui sont analogues, mais non homologues à ceux que l’on peut rencontrer dans d’autres langues i.–e.nes anciennes. Il est de grande importance d’étudier les thèmes du verbe lituanien dans le contexte structural du lituanien. Exemples: Pour la racine lituanienne Mi/e/aN- (dont les voyelles correspondent à l’apophonie i.–e.ne Mø/e/oN-) nous avons, différenciés par le degré, les thèmes simples min-i ‘(on) mentionne, rappelle’, men-a ‘(on) se rappelle’ et man-o ‘(on) pense’, à quoi s’ajoute encore, à thème élargi, ma˛sto = man-st-o ‘(on) réfléchit’. La racine lituanienne (non i.–e.ne!) De¯ /e˘- prend les élargissements -d- et -st-, cf. de-d-a ‘(on) met, pose’ et de˙-st-o ‘(on) expose, démontre, enseigne’; la racine STa¯- peut être élargie par -v-, cf. sto-v-i ‘(on) se tient debout’ vs. sto-j-a ‘(on) s’arrête’; la racine AU- peut être élargie par -n-, cf. au-n-a ‘(on) revêt’ vs. av-i < *au-i ‘(on) porte, est vêtu’. La racine ŽiN- forme un radical à elle-même, cf. su-žin-o ‘(on) apprend’, à l’élargissement -st-, cf. paži˛sta < pa-žin-st-a ‘(on) connaît’, au suffixe -o-, cf. pa-žin-o-j-a ‘(on) fait connaissance’. Convention: Le verbe fini lituanien sera donné toujours à la non-personne qui couvre toutes les formes verbales françaises aux sujets ‘‘il’’/‘‘elle’’, ‘‘ils’’/‘‘elles’’, ‘‘il’’ (impersonnel, cf. il vente) et ‘‘on’’. Les traductions françaises des exemples lituaniens seront citées aux sujets ‘‘il’’ ou ‘‘on’’. Nous n’irons pas à la recherche des morphes qui, dans diverses langues i.–e.nes correspondent (ou prétendent correspondre) aux moyens dérivationnels du verbe lituanien. Ce qui nous intéresse, ce sont les principes selon lesquels ce dernier a été paradigmatisé. Nous nous limiterons aux deux thèmes sur lesquels sont bâtis, respectivement, les indicatifs présent et passé (= le prétérit). Sauf pour l’«optatif», marginal et lié par sa forme au présent (dans le paradigme catégoriel du verbe lituanien, il ne présente qu’une variante de l’impératif), les autres catégories du verbe fini (futur, conditionnel, impératif) sont dérivées de façon uniforme à partir du thème de l’infinitif qui est étroitement lié à celui du prétérit. Ici, elles ne feront pas objet de nos réflexions. Au cours du présent chapitre nous esquisserons, dans un contexte comparé balto-slave, un tableau montrant que le choix du paradigme formel (c’est-à-dire le choix de la classe de conjugaison, cf. 3.2) est de pertinence lexico-sémantique au présent (8.1.1) et relève, au prétérit, d’un plus vaste domaine de dérivation lexicale (8.3.2). L’auteur du présent mémoire développe son interprétation systématique de la morphologie du verbe lituanien dans une monographie particulière (1997, manuscrite). .. La distinction primitive, à la fois lexico-sémantique et morphologique, du verbe i.–e. en extraverti et introverti (4.2) n’existe plus telle quelle en lituanien,



 Tomáš Hoskovec

mais cette langue a consciemment recréé — même à plusieurs reprises — une catégorie verbale lexico-sémantique qui, morphologiquement marquée, correspond aux anciens introvertis par le sens ‘‘procès non destiné à s’accomplir hors sujet’’ (4.4). Ce sont les verbes non-intentionnels, tous marqués par leur appartenance à la conjugaison en -a, aux élargissements radicaux -n- et -st. La signification des deux élargissements est identique et leur distribution suit mécaniquement la nature longue (→ -st postposé) ou brève (→ -n- intercalé) du radical verbal. Tout comme les anciens introvertis, les verbes non-intentionnels lituaniens sont bien marqués sur le plan morphologique, mais contrairement à ces derniers, ils le sont non seulement aux désinences (marque accessoire), mais surtout aux radicaux (marque première). Par conséquent, les verbes non-intentionnels lituaniens aux radicaux marqués sont dérivables sur les racines non marquées (leur dérivation étant le plus souvent accompagnée d’une apophonie quantitative proprement balte qui, pourtant, est fonctionnellement redondante). La langue moderne connaît un grand nombre de paires verbales (deux unités lexicales dérivant de la même racine) dont un membre est marqué comme non-intentionnel, cf. le verbe de base keit-
ia > keicˇia ‘(on) change = provoque un changement’ et le verbe marqué non-intentionnel ki-n-t-a ‘(on) change = subit un changement’, dérivant les deux de la racine Kei/iT-. Parfois on rencontre même un verbe de base, tel que trauk-i
a ‘(on) étend, tire [par les deux bouts]’, auquel sont opposés deux verbes non-intentionnels, cf. tru¯k-st-a ‘(il) se déchire, craque [à force d’être trop tendu]’ & tru-n-k-a ‘(il) dure [s’étend dans le temps]’, tous les trois radicaux dérivant d’une racine commune TRau/u¯ /uK-. Convention: Le signe i marque un élément morphonologique particulier qui se manifeste, au niveau phonologique, soit comme le trait distinctif de palatalisation de la consonne précédente, soit, si précédé d’une voyelle, comme la consonne j. L’orthographe lituanienne rend la consonne palatalisée, en position devant une voyelle antérieure, par le graphème i, cf. traukia; la palatalisation des dentales t et d peut aboutir à des semi-fricatives cˇ et
ˇ , respectivement, cf. keicˇia, leidžia < leid-i
a ‘(on) permet’. .. En i.–e. commun, l’extraverti et l’introverti furent deux classes lexicosémantiques non-oppositionnelles, distinguées par des moyens morphologiques. En lituanien moderne, le non-intentionnel ne participe à aucun système binaire. Il représente une classe lexico-sémantique parmi d’autres qui sont, elles aussi, distinguées (marquées) morphologiquement, tels que verbe d’état (marqué par la désinence -i) ou verbe intensif à désinence -o (qui comprend des fréquentatifs/itératifs à radical élargi par -st- aussi bien que des effectifs/causatifs à radical élargi en -d-). C’est pourquoi nous concevons toutes les classes verbales comme différentielles. Voyons à titre d’exemple la triple dérivation, à partir d’une racine commune Siau/iuT-, d’un verbe de base siaut-i
a > siaucˇia ‘(on) se met en rage/

Sur la paradigmatisation du verbe indo-européen 

colère’, d’un verbe d’état siaut-i ‘(on) est en rage’ et d’un non-intentionnel siu-nt-a ‘(on) devient fou’. Le verbe d’état et le non-intentionnel appartiennent à deux classes différentielles bien visibles, mais le verbe de base lui-même est positivement marqué: il porte la désinence -i
a qui, opposée à la désinence -a, caractérise les radicaux participant à l’apophonie quantitative. Exemples: À partir de la racine lituanienne TiNG- dérivent un verbe d’état ting-i ‘(il) est paresseux, évite tout travail’ et un verbe non-intentionnel ting-st-a ‘(il) ne fait rien, s’abandonne à la paresse’; à partir de la racine Mı¯ /iL- un verbe d’état myl-i ‘(il) aime, est amoureux de quelqu’un’ et un non-intentionnel pamil-st-a ‘(il) tombe amoureux de quelqu’un’; à partir de ŠaL- un verbe causatif/effectif šal-d-o ‘(il) refroidit’ et un non-intentionnel ša˛l-a < ša-n-l-a ‘(il) gèle, fait/a très froid’. Dans les trois paires, les deux membres sont morphologiquement marqués. La première paire (tingi–tingsta) représentant deux verbes intransitifs, la deuxième (myli–pamilsta), deux transitifs, on voit bien que la classe lexico-sémantique des non-intentionnels est en déhors de la division syntaxique des verbes en transitifs et intransitifs. Il n’y a pas de classe «intentionnelle» en lituanien. La notion d’«intentionnel» n’existe que par négation, tel «non-non-intentionnel», et n’offre aucune analogie avec l’ancien extraverti indo-européen. Même ceci rime parfaitement avec l’évolution paradigmatique des langues i.–e.nes où le médio-passif (qui, si jamais il existe, est issu de l’introverti) devient la voix positivement marquée par opposition à l’actif, devenu la voix de base, non-marquée, neutre. .. Pour la dérivation des verbes non-intentionnels en lituanien moderne, les moyens primordiaux sont les élargissements radicaux et la classe de conjugaison. L’apophonie de la racine peut être absente. Nous voyons en elle un moyen dérivationnel plus ancien qui servait avec toute probabilité le même but. Un troisième moyen dérivationnel produisant la classe lexico-sémantique de nonintentionnel est identifiable à la nouvelle conjugaison athématique dont nous parlerons en 8.4.1.1. Ces trois sortes de moyens dérivationnels relèvent de diverses étapes de l’évolution de la langue. Il en découle qu’après la perte des désinences introverties, conséquence d’une généralisation du verbe thématique, la classe primitive i.–e.ne d’introverti a été à plusieurs reprises reconstruite lors de la paradigmatisation du lituanien. . Le verbe slave ne connaît plus rien de l’ancienne division en extraverti et introverti, mais l’ancienne sous-division de l’extraverti en complexe vs. évolutif (cf. 7.1) y a été systématisée en deux classes oppositionnelles, à savoir perfectif :: imperfectif (conception globale & déterminée :: non-globale & non-déterminée du procès de verbe).5 note: Il serait plus adéquat de dire, en français, «non-perfectif» au lieu d’«imperfectif», comme le propose Jens Holt (1943: 65). Le non-perfectif est le

 Tomáš Hoskovec

membre non-marqué de l’opposition et c’est l’aspect qui apparaît en tout cas de neutralisation. Tout comme l’ancienne division i.–e.ne, présumée au stade à f-d, la nouvelle division slave n’est marquée par aucun morphème spécifique, mais conformément au stade à f-p, les langues slaves fournissent presque à chaque verbe imperfectif un perfectif (très souvent, elles en fournissent plusieurs qui sont différenciés sémantiquement) et uice uersa. Rares sont les verbes slaves qui n’ont pas de partenaire de l’autre aspect et rares sont ceux qui restent ambigus. Dans ce sens-là, la situation est bien parallèle à ce que nous avons dit à propos du verbe latin et grec dans les parties 7.2 et 7.3. Également parallèle à l’évolution du latin et du grec, l’opposition slave de perfectivité n’a pu avoir lieu qu’après la généralisation du présent thématique comme l’unique série de désinences qui avait fait disparaître la distinction formelle d’extraverti et d’introverti postulée pour l’i.-e. commun. Remarque: Les langues slaves n’attestent que quelques cinq verbes athématiques, parfaitement allomorphes aux thématiques. Leurs désinences n’apportent aucun témoignage sur la présence de la classe lexico-sémantique d’extraverti. Non plus que la forme veˇd-eˇ, allomorphe de la 1ère sg athématique veˇmь ‘je sais’, ne prouve la survivance, dans les langues slaves, de l’introverti i.–e. primitif.6 .. En expliquant la genèse du système aspectuel slave, M. Igor Neˇmec (1958) s’appuie essentiellement sur la catégorie d’ingressivité qui aurait été à l’origine du perfectif en tant qu’amorce du procès (état/action) désigné par la racine. Ce fut le suffixe -n-, pris dans l’inventaire dérivationnel de l’indo-européen, qui servit de porteur principal à l’ingressivité en slave. Exemples: Au début il y aurait eu, en slave, une dérivation lexicale analogue au grec :V2,, :"J,Ã< ‘sache!, savoir’ (procès de base) → :"-<-2V-
Sur la paradigmatisation du verbe indo-européen

... En grec moderne, le suffixe nasal s’est répandu de façon considérable, de nouveaux présents en -<-T ayant remplacé bien des anciens, cf. *\


 Tomáš Hoskovec

et «prétérit marqué» respectivement (en leur opposant encore un «prétérit périphrastique» en guise du parfait à participe en -l). Lier dès le début l’aoriste slave à l’ancien perfectif i.–e. et l’imparfait à l’ancien imperfectif serait un courtcircuit de pensées qui manquerait de justification à la fois en morphologie formelle et en analyse sémantique de textes et en vieux-slave (dorénavant, «vieuxmacédonien» selon notre convention) et en vieux-tchèque, comme l’ont bien démontré Bohuslav Havránek (1939) et Antonín Dostál (1954). Par leurs origines, les temps morphologiques du verbe slave étaient indépendants de l’aspect qui s’était constitué au sein d’un présent thématique plutôt atemporel. Ce ne fut qu’ultérieurement qu’en certaines langues slaves, dont le tchèque, l’aspect a pris une valeur temporelle (l’imperfectif signifiant le présent, le perfectif exclusivement le futur), tandis qu’il en reste indépendant en d’autres, dont le bulgare. ... Nous sommes loin d’identifier l’aoriste slave à un aoriste «gréco-indien» érigé en indo-européen. Sans nier la correspondance formelle entre slave by (et en vieux-macédonien et en vieux-tchèque) ‘(il) fut’ (à allomorphe by-stъ en vieuxmacédonien seulement) d’un côté, et sanskrit á-bhu¯-t, grec §-nØ de l’autre, nous voulons souligner que les aoristes grecs et indiens sont issus des paradigmatisations particulières à ces deux langues et que rien ne nous autorise à postuler le même développement paradigmatique pour les langues slaves (ni pour les baltes). ... En tant que paradigme formel, l’aoriste slave est très hétérogène. Ses désinences ont été homogénéisées partiellement à l’aide de l’élément sigmatique. Mais si nous voulons comprendre la nature secondaire de l’aoriste slave, plutôt que d’observer ses désinences, il faut prêter attention à ses radicaux. Pour la majorité absolue des racines, ils reposent sur un des suffixes vocaliques suivants: -n-o˛ /ú-, -eˇ-, -a-, -i-, -ov-a- qui appartiennent tous à l’inventaire balto-slave de dérivation lexicale du verbe. Tous ces suffixes servent, dans les langues slaves et à la dérivation grammaticale, en produisant des thèmes secondaires de prétérit marqués, suffixés, et à la dérivation lexicale, en produisant des verbes secondaires dont les deux thèmes — celui de présent et celui de prétérit — sont conjointement suffixés. Remarque: Par sa dérivation morphologique, l’aoriste slave est étroitement lié à l’infinitif. Or, aux étapes anciennes, l’infinitif a vécu un changement profond de sa valeur syntaxique en remplaçant le participe dans son rôle de prédicat condensé. Le remplacement du participe par l’infinitif est une tendance générale slave.9 Cependant, dans le passé attesté des langues slaves individuelles, nous rencontrons divers cas de vacillation entre l’infinitif, moyen neuf, et le participe, moyen ancien. Vu que le participe est produit à partir du thème de présent, la vacillation originelle entre l’infinitif et le participe ne reflète-t-elle pas, sur le plan syntaxique, que l’infinitif fut au début, sur le plan morphologique, moins établi, moins systématisé (donc dérivationnellement défectif) par rapport au participe? En

Sur la paradigmatisation du verbe indo-européen 

conséquence, la morphologie hétéroclite de l’aoriste slave vis-à-vis de l’indicatif présent ne reflète-t-elle qu’un seul aspect d’un processus morphosyntaxique plus large au cours duquel le thème d’aoriste-infinitif gagna son indépendance par rapport au thème de présent-participe? .. En lituanien, on voit encore mieux la nature secondaire du prétérit: il se sert des deux désinences à base vocalique longue (-o et -e˙) dont la première exprime également un présent. La désinence -o caractérise les intensifs qui, aux radicaux particulièrement marqués, peuvent être spécifiés comme des fréquentatifs/itératifs ou des effectifs/causatifs, cf. 8.1.2. Nous revenons au statut des deux désinences plus bas aux paragraphes 8.4.4 et 8.4.5. Exemple: De la racine ŽiN- dérivent — à part le présent (préfixé) nonintentionnel (élargi en -st-) paži˛sta < pa-žin-st-a ‘(on) (re)connaît’ — deux formes marquées par la désinence -o: un présent intensif/itératif žin-o ‘(on) connaît/sait’ [motivé par ‘reconnaît à maintes reprises’?] et un prétérit pa-žin-o ‘(on) connaissait/savait’ qui, au fait, représente le même intensif/itératif que žin-o à ceci près que son action est contextuellement — par opposition à paži˛sta — située dans le passé. ... Beaucoup de thèmes du prétérit lituanien sont formés par les suffixes -oet -e˙- à l’aide desquels sont également faits, de façon productive, des verbes secondaires. Les suffixes -o- et -e˙- (que nous nous appliquons à différencier des désinences -o et -e˙ pour d’importantes raisons prosodiques, voir plus bas en 8.4.4) correspondent aux suffixes -a- et -eˇ- de l’inventaire slave mentionné au paragraphe 8.3.1.2. Les autres suffixes slaves du même inventaire trouvent également leurs correspondants lituaniens, mais ceux-là servent, en lituanien, uniquement à faire dériver de nouvelles unités lexicales, non à produire des thèmes particuliers de prétérit. Par conséquent, les prétérits suffixaux lituaniens gardent une motivation sémantique que l’on ne peut plus repérer même dans les plus vieilles couches attestées des langues slaves. Exemples: Par le suffixe -e˙- sont marqués les verbes secondaires qui expriment le port et/ou le maintien de la qualité désignée par le nom de base, cf. sen-as ‘vieux’ — sen-e˙-j-a ‘(on) est/devient manifestement vieux=tombe en désuétude’ [opposé à non-intentionnel sen-st-a ‘(on) vieillit, s’invétère’]. Or, à partir du présent kalb-a ‘(on) parle’ le même suffixe -e˙- donne le prétérit kalb-e˙-j-o ‘(on) parlait’ [motivé par ‘(on) persévère en état de parler, s’adonne à parler’?]. Chose importante, les verbes non-suffixés au thème (primaire) de présent et suffixés par -e˙- au thème (secondaire) de prétérit manifestent nettement un des deux traits sémantiques: (1) énonciation de la parole, cf. kalb-a → kalb-e˙-j-o pour ‘parler’, šnek-a → šnek-e˙-j-o ‘causer, bavarder’, žad-a → žad-e˙-j-o ‘donner sa parole, promettre’ [apparenté à žod-is ‘mot’]; (2) mouvement continu, cf. jud-a → jud-e˙-j-o pour ‘se mouvoir’, tek-a → tek-e˙-j-o ‘couler’, varv-a → varv-e˙-j-o ‘découler/tomber en gouttes’. Et c’est précisément le suffixe -e˙- qui produit, de

 Tomáš Hoskovec

façon uniforme, les prétérits de tous les verbes d’état (qui sont marqué au présent par la désinence -i), cf. stov-i → stov-e˙-j-o pour ‘se tenir debout’, gul-i → gul-e˙-j-o ‘être allongé/couché’, av-i → av-e˙-j-o ‘être chaussé/vêtu [de qc.]’. Le suffixe -o- produit des dénominatifs à partir des thèmes en -o (=a¯), tel galv-a (< *galv-a¯) ‘(la) tête’ → galv-o-j-a ‘(on) pense’ (prétérit: galv-o-j-o). Or si jamais un verbe dont le présent prend la désinence -o a un prétérit suffixé, le suffixe est infailliblement -o-, cf. bij-o ‘(on) a peur’ → bij-o-j-o ‘(on) avait peur’, iešk-o ‘(on) cherche’ → iešk-o-j-o ‘(on) cherchait’. Le suffixe -o- produit également, souvent précédé de divers élargissements consonantiques et/ou de l’élément palatalisant qui en précisent la signification, des verbes déverbatifs, tel neš-a ‘(on) porte’ → neš-i
-o-j-a (prétérit: neš-i
-o-j-o) ‘(on) porte sans cesse çà et là’, ger-i
a ‘(on) boit’ → ger-i
-o-j-a ‘(on) boit souvent et en petites quantités’, ‘(on) «buvote»’ [d’après le modèle (il) pleut → (il) pleuvote]. Illustrons la double action du suffixe -o- à la racine ŽiN- qui s’en sert pour faire dériver et le radical du prétérit žin-o-j-o ‘(on) a su’, opposé au présent žin-o, et celui du verbe secondaire pa-žino-j-a ‘(on) fait connaissance’ (prétérit: pa-žin-o-j-o). . Abordons à la fin la structure des désinences du verbe balte et slave. Contenant chacune une série de désinences à segment vocalique initial, les conjugaisons du présent sont traditionnellement classées en «thématiques» (le segment vocalique correspond à e/o, e.g. v.–m. nes-o˛, nes-o˛-tъ, nes-e-mъ, v.–t. nes-u, nes-ú, nese-my, lit. neš-u, neš-a, neš-a-me) et «semi-thématiques» (le segment vocalique consiste en d’autres voyelles, e.g. v.–m. proš-o˛, pros-e˛-tъ, pros-i-mъ, v.–t. proš-u, pros-ie, pros-í-my, lit. praš-a-u, praš-o, praš-o-me), cf. Stang (1942: 21, 98–9). Les désinences athématiques sont marginales et sans motivation aucune, excepté une phase historique du lituanien. .. Le paradigme du verbe lituanien distingue deux personnes, locuteur et auditeur, à trois variantes de nombre (dont le duel, très marginal dans l’usage, mais régulier et productif dans le système), et une non-personne sans aucune différenciation de nombre, ce qui se déroule dans un contexte syntaxique particulier (cf. Hoskovec 1994, 1997). Aux désinences dites «thématiques» et «semi-thématiques» du présent et du prétérit on peut distinguer très nettement une base vocalique et un exposant de congruence syntaxique. Cet exposant a une forme particulière pour chacune des deux personnes aux trois nombres, et une forme zéro pour la non-personne. Tous les exposants disposent d’un allomorphe allongé (à voyelle longue ou à diphtongue) et sont capables de prendre l’accent à la syllabe précédente. La nonpersonne a pour désinence la base vocalique pure, sans exposant, qui ne s’allonge point et ne prend jamais d’accent. Les désinences athématiques lituaniennes observent les mêmes principes: elles n’ont certes pas de base vocalique, mais marquent bien l’opposition des deux

Sur la paradigmatisation du verbe indo-européen 

personnes à la non-personne. Les désinences -mi et -si de la 1ère sg et de la 2e sg s’allongent en -mie et -sie respectivement et attirent l’accent tout comme les désinences -me, -te, -va, -ta du pluriel et du duel (aux allomorphes -me˙, -te˙, -vo, -to respectivement). La non-personne des verbes athématiques est à la désinence -ti qui, non plus que la base vocalique pure, ne s’allonge ni ne prend d’accent. ... Il est certain que parmi les verbes athématiques du vieux-lituanien on rencontre des homologues modèles d’anciens verbes i.–e.s extravertis (marqués ‘‘actualisés’’, cf. 5.1), tels que dúomi ‘je donne’, demì ‘je mets’, eimì ‘je vais’, esmì ‘je suis’. Cependant, le système du lituanien est assez particulier et il ne serait pas judicieux de le rapprocher à la hâte d’un système gréco-indien (cf. 8.4.2). Aussi ne voulons-nous pas rattacher l’allomorphe -mie (équivalant au prussien -mai) à la désinence grecque moyenne -:"4 et préférons l’expliquer historiquement à l’intérieur du système du lituanien, comme le fait Jonas Kazlauskas (1968: 294–5). Remarque: Au XVIIème et au début du XVIIIème siècle survint, pour disparaître aussitôt, une vague grande et inattendue de nouveaux verbes athématiques. Elle eut lieu en lituanien, mais point en letton (le prussien n’étant plus attesté à l’époque). Christian Stang (1962) l’explique comme un rudiment de système grammatical marquant les verbes intransitifs et/ou d’état. La plupart de ces nouveaux verbes athématiques, parmi lesquels il y avait aussi des transitifs, pourraient être classés non-intentionnels au sens illustré en 8.1.2 et nous y voyons effectivement un essai de restituer une classe lexico-sémantique proche de l’ancien introverti. ... En soulignant l’individualité de l’ancien système athématique du verbe lituanien, nous voulons relever encore un autre aspect pertinent pour la compréhension historique: que la désinence athématique -ti de la non-personne lituanienne corresponde si exactement à celle de la personne ‘‘autrui’’, telle que nous l’avons postulée pour le stade f-d au paragraphe 4.3, et marquée en outre ‘‘actualisé’’, n’implique aucunement qu’il devait y avoir, en lituanien pré-historique ou bien en balte primitif, une personne secondaire ‘‘autrui’’ + ‘‘autrui’’ à désinence -nti. La personne primitive ‘‘autrui’’ a pu toujours rester sans spécification de nombre.10 En outre, suite à ce qui a été dit aux paragraphes 4.3 et 6.1, nul n’est obligé de présumer que le verbe thématique balte ait perdu les désinences de la 3e personne de tous les nombres: il peut n’en avoir jamais eu. Fait historique indéniable, nous n’en avons aucune trace. Rappel: Que nous soyons bien compris: nous ne postulons pas d’indoeuropéen primitif, langue tout faite dont les langues i.–e.nes anciennes attestées ne seraient que des modifications ultérieures; nous étudions comment sont constitués les paradigmes verbaux des langues i.–e.nes anciennes prises individuellement, l’i.-e. primitif n’étant pour nous qu’un inventaire de moyens linguistiques. Au stade à flexion dérivationnelle (f-d), le verbe à voyelle thématique et le verbe à exposant de congruence syntaxique sont deux dérivés du même niveau, de la même

 Tomáš Hoskovec

complexité morphologique. La voyelle thématique et l’exposant de congruence syntaxique sont deux composantes qui entrent, au stade à flexion paradigmatique (f-p), dans la désinence du verbe «thématique». Elles y entrent, dans diverses langues, en divers ordres, accompagnées ou non d’autres éléments (cf. chapitres 5, 6, 7). Quant au verbe balte, aussi loin que nous voyons, nous n’avons que le verbe thématique à 3e personne sans exposant de congruence. Il serait peu utile de se demander ce qu’il pouvait y avoir avant. Essayons plutôt de mieux voir ce qui est arrivé après. ... Dans ce qui nous reste du vieux-prussien, nous retrouvons des vestiges d’un faible essai d’introduire un exposant neuf à la 3e personne de la conjugaison thématique. C’est le suffixe -ts, originellement, le pronom *tas devenu enclitique, dont il serait plutôt risqué, vu la pénurie de matériaux, de porter un jugement quant à la spécificité en nombre et en genre. Une pareille formation n’est attestée dans aucune autre langue balte. Observation: Le Troisième catéchisme vieux-prussien contient, à une seule page, deux passages parallèles où le même verbe imma ‘il prit’ dans la même structure syntaxique et sémantique est accompagné (i) d’un sujet pronominal tonique imma tans ʃtan geitin ‘Nam er das Brodt’ (III 752) et (ii) d’une «désinence» en guise de suffixe enclitique immats dijgi ʃtan Kelkin ‘nam Er auch den Kelch’ (III 757). L’exposant -ts apparaît surtout là où le verbe est séparé de son sujet par d’autres membres syntaxiques de la phrase, voire par d’autres propositions, comme en témoigne, dans le Premier catéchisme, la différence entre Nuʃon Rekis Chriʃtus bela . . . ‘Vnʃer herr Chriʃtus ʃprach . . .’ (I 119) d’un côté (le verbe bela, sans exposant, suit immédiatement après son sujet), et Nuʃen rickis Jeʃus Chriʃtus . . . bha belats . . . ‘Vnʃer herr Jeʃus Chriʃtus . . . vnd ʃprach . . .’ (I 133 . . . 7) de l’autre (le verbe belats, avec exposant, est séparé de son sujet par un récit de quatre lignes). .. Au chapitre 6, nous avons souligné qu’au niveau à f-d, la dérivation du verbe thématique (à l’aide de l’élément e/o) et celle du verbe congruent (à l’aide des éléments -m, -s, -t, parmi d’autres) étaient deux activités morphologiques indépendantes. Dans les langues i.–e.nes individuelles, c’est-à-dire au niveau à f-p, les désinences de conjugaison pouvaient combiner les éléments porteurs des deux activités susmentionnées, mais elles pouvaient se servir aussi bien d’autres. Avant d’identifier un élément morphologique i.–e. commun dans une langue i.–e.ne particulière, il faut scruter minutieusement son mode d’existence. Exemples: Nous concédons volontiers qu’une forme toute faite a pu être reprise par plusieurs langues, e.g. grec ,É:4 & vieux-lituanien eimì, vieil-indien ásmi & vieux-macédonien jesmь. Or, un constat pareil ne dit rien sur un parallélisme éventuel des conjugaisons athématiques grecque et balte, indienne et slave. En dépit de l’identité formelle qui règne entre ,É:4 et eimì, la désinence lituanienne -mi doit

Sur la paradigmatisation du verbe indo-européen 

être tenue allomorphe de -mie, voire allomorphe secondaire de celle-ci. En effet, qui veut expliquer la prosodie de eimì par la loi de Saussure — Fortunatov doit partir de eimíe, forme attestée dans le vieux-lituanien devant des enclitiques, e.g. eimiegi ‘je vais donc’. Rien de pareil en grec. Quelle est donc la valeur heuristique du parallélisme ,É:4 ↔ eimì ? La belle et impressionnante correspondance indo-balte ásmi ↔ esmì, ási ↔ esì, ásti ↔ e˜ sti est ternie par une dissymétrie désagréable: tandis que la 3e sg passe identique, la 1ère et la 2e personne, à l’intérieur du système lituanien, portent deux désinences non-identifiables aux indiennes, à savoir -mie et -sie, qui ne sont abrégées que secondairement, histoire des syllabes finales oxytones accentuées. De même, les correspondances phoniques indo-slaves a ↔ je, s ↔ s, m ↔ m, i ↔ ь ne s’accordent ni avec celle de asti ↔ jestъ (de quoi nous parlerons abondamment au paragraphe 8.4.3), ni avec celle de asi ↔ jesi. Si nous prenons la correspondence asmi ↔ jesmь pour régulière, celle de asi ↔ jesi est troublante: pourquoi -mi ↔ -mь, mais -si ↔ -si? Si nous argumentons que la désinence athématique vieille-macédonienne -si restitue la voyelle i grâce à l’analogie avec la désinence thématique -ši (semi-thématique, pour Stang 1942:214), comment expliquer la version tchèque jsi, étant donné qu’au sein du système tchèque il n’y a jamais eu d’exposant de la 2esg autre que le segment -š pur et simple, sans voyelle? Il faudrait donc conclure qu’en dépit des ressemblances frappantes entre les formes athématiques grecques et indiennes d’un côté, baltes et slaves de l’autre, les paradigmes respectifs sont incomparables: dans le système indien, la conjugaison athématique est la conjugaison tout court, la conjugaison thématique n’en étant qu’une variante marquée; dans le système grec, la conjugaison athématique est un sous-système bien développé et nettement distinct de la conjugaison thématique; dans le système lituanien, la conjugaison athématique est une innovation récente, un essai abandonné au profit de la thématique, conjugaison modèle; dans les systèmes slaves, il n’y a pas de conjugaison athématique, quelques cinq verbes, au maximum, ayant des désinences «bizarres». .. À la lumière des paragraphes précédents 8.4.1 et 8.4.2, il est très instructif de comparer la conjugaison thématique de deux vieilles langues slaves, tchèque et macédonienne. Tandis qu’en vieux-tchèque la 3esg(pl) n’ajoute aucun exposant au segment thématique, c’est-à-dire à la base vocalique de la désinence, cf. ved-e ‘il conduit’ (ved-ú ‘ils conduisent’), en vieux-macédonien on trouve systématiquement un -tъ additif, cf. ved-e-tъ (ved-o˛-tъ). L’exposant -tъ est identique au N.sg.masc du pronom démonstratif ‘lui/celui-ci’ (cf. 8.4.1.3) et son vocalisme à yer dur (ъ) exclut toute parenté avec le -ti «indo-européen», c’est-à-dire gréco-indien, qui aurait donné -tь à yer mou (ь). En témoigne le vieux-macédonien lui-même par sa forme da-m-ь ‘je donne’ dont la désinence en -ь correspond à celle en -i de sanskrit da-da¯-m-i et grec *\-*T-:-4. La linguistique comparée traditionnelle qui voulait établir coûte que coûte un parallélisme indo-slave, tenu pour continuité

 Tomáš Hoskovec

évolutive, se vit forcer à mettre en jeu un processus obscur de «durcissement» de -ti i.–e. (cf. Wondrak 1908: 135, Meillet 1934: §§345–6, Stang 1942: 216–20). ... Avant de commenter les solutions proposées, nous voulons relever que la linguistique historique et comparée slave ne s’est jamais sérieusement arrêtée sur un point pourtant cardinal: Pourquoi le primat est-il donné aux formes vieillesmacédoniennes? Y a-t-il d’autres raisons que la tradition et la terminologie érigeant le vieux-macédonien en vieux-slave? Certes, le vieux-tchèque (traces isolées à partir du Xème siècle, tradition littéraire suivie à partir du XIIIème) est plus «jeune» par rapport au vieux-macédonien qui était déjà au IXème siècle une langue cultivée et qui fut travaillé à cette époque-là en une langue littéraire dont la portée devait dépasser de loin le territoire macédonien (les documents les plus anciens étant du Xème siècle). Le décalage des deux langues offre un temps suffisant pour que certaines formes linguistiques disparaissent complètement. Mais en étudiant les correspondances phoniques entre le v.–m. et le v.–t., nous ne trouvons pas un cas pareil de «perte» d’une consonne finale, et à plus forte raison, d’une syllabe finale: la différence v.–t. ved-e (ved-ú) vs. v.–m. ved-e-tъ (ved-o˛-tъ) n’a pas d’analogie et, par conséquent, doit être tenue phénomène purement morphologique, non phonologique. Ajoutons qu’il y a, dans les documents v.–m.s, des «exceptions», où le -tъ désinentiel manque, mais il n’y a pas un cas v.–t. où un -t(ъ) désinentiel se soit «exceptionnellement» conservé. Il ne faudrait pas négliger, non plus, l’apport d’un troisième membrum comparationis, à savoir la langue des petits Fragments de Freising du Xème siècle, documentant un parler slave du territoire slovène, probablement contemporain de l’élévation du vieux-macédonien en langue littéraire. Limitée qu’elle est, la langue des Fragments observe clairement le modèle v.–t. ved-e (ved-ú), non le v.–m. ved-e-tъ (ved-o˛-tъ). À l’intérieur des langues slaves anciennes, il n’y a donc aucun moyen de distinguer entre «perte» et «introduction» d’un élément désinentiel, le choix étant fait d’après ce que l’on imagine modèles i.–e.s primitifs. Dans notre approche, nous ne postulons pas de paradigmes i.–e.s communs, tout en nous bornant à étudier la structure et le fonctionnement des paradigmes verbaux au sein des langues i.–e.nes individuelles. Commentaires: Présumer un processus de «durcissement» de -ti i.–e. en -tъ «slave» implique plusieurs difficultés. Il faudrait (1o) lui faire succéder, dans les parlers slaves orientaux, un «ramollissement» inverse du segment -tъ en -tь — processus ayant lieu uniquement dans les désinences de la 3e personne et marquant systématiquement les textes slavons déjà les plus anciens de la rédaction russe (à savoir, ceux du XIe et du XIIe siècle) qui distinguent encore bien nettement le yer dur (ъ) du yer mou (ь), tandis que dans les écrits russes ultérieurs règne une confusion progressive entre les deux signes. Audit ramollissement slave oriental devrait suivre (2o) un «redurcissement» partiel qui, à l’époque moderne,

Sur la paradigmatisation du verbe indo-européen 

différencie le [grand]-russe нес-y-т à désinence dure (cf. нес-y-тъ selon l’orthographe prérévolutionnaire) de l’ukrainien нес-y-ть et du biélorusse няс-y-ць à désinence molle. Il est certes possible de tenir la forme slave orientale ancienne нес-y-ть pour primitive — par rapport à la vieille-macédonienne nes-o˛-tъ → et reflétant ainsi le segment -ti présumé i.–e. commun (cf. Meillet 1934: 320, Stang 1942: 215), de sorte qu’on n’ait à admettre qu’un processus de durcissement en vieux-macédonien, ayant une analogie dans le durcissement relativement récent des désinences [grand]-russes. Encore faudrait-il expliquer la distribution du segment -tь/-tъ dans les désinences des langues slaves orientales: en [grand]-russe il marque, sous l’aspect dur, à la fois la 3esg нес-e-t(ъ) et la 3epl нес-y-т(ъ), en biélorusse, de même qu’en ukrainien, il marque, sous l’aspect mou, la 3epl de tous les verbes, e.g. няс-y-ць, mais à la 3esg une classe seulement des verbes, e.g. няс-e vs. нoc-i-ць. La linguistique comparée connaît encore d’autres essais d’expliquer l’évolution du segment v.–m. -tъ, dont les plus importants partent des désinences «moyennes». M. Adolf Erhart (1988: 121–34) en a élaboré une théorie bien cohérente et donnant pleine satisfaction phonologique, qui cependant, a pour nous l’inconvénient de présupposer comme i.–e.nes primitives des désinences particulières de l’inactif latin et grec que nous tenons pour créations individuelles desdites langues (cf. 7.2.3 et 7.3.2). ... Vu que les langues slaves occidentales ne connaissent point d’élément -t aux désinences de la 3e personne (sauf éventuellement pour la 3esg jest, allomorphe à je, mais le verbe être constitue, dans beaucoup de langues, un paradigme à part, peu motivé par le reste de la conjugaison), et que certaines des langues slaves méridionales (slovène, serbe, croate, y compris la langue des Fragments de Freising) n’en connaissent rien non plus, tandis que d’autres (bulgare, macédonien moderne) l’utilisent au pluriel seulement, force est de reconnaître que la nonoccurrence de l’élément -t dans les désinences de la 3e personne du verbe slave est plus simple à expliquer par la non-introduction de cet élément à l’état originel du slave primitif que par sa disparition ultérieure. La disparition, si jamais il y en a eu lieu, ne pourrait pas s’expliquer comme un processus phonique (elle n’a touché, dans aucune langue slave, ni le -tъ final des participes, cf. prokle˛tъ ‘damné’, slovo˛tъ ‘renommé’, ni le -tь final des numéraux, cf. pe˛tь ‘cinq’, dese˛tь ‘dix’, ni celui des substantifs, cf. po˛tь ‘chemin’), mais seulement comme un processus morphologique concernant les paradigmes formels de conjugaison.11 ... L’idée de l’introduction (au lieu de la disparition) de l’exposant -t(ъ/ь) dans le système de conjugaison des langues slaves individuelles paraît beaucoup moins radicale si l’on met en corrélation la répartition de l’exposant -t(ъ/ь) à la 3esg/pl et celle de l’exposant -m(ь) à la 1èresg. Les langues slaves qui n’ont pas recours à l’exposant -m(ь) développent l’exposant -t(ъ/ь), à l’instar du vieux-

 Tomáš Hoskovec

macédonien et des langues slaves orientales, et uice uersa, comme dans les langues slaves occidentales aussi bien qu’en slovène, croate et serbe. Cependant, rien n’empêche d’autres langues slaves, dont le bulgare et le macédonien moderne, d’utiliser les deux à la fois. Le rôle des morphes -t(ъ/ь) et -m(ь) est à portée slave commune: ils servent de marques morphologiques positives de la 3e et 1ère personne respectivement. Remarque: Le vieux-tchèque présente un système de conjugaison slave où les deux marques sont absentes et c’est l’opposition de longueur qui en tient lieu, étant parfaitement homologue au segment vieux-macédonien -tъ, cf. v.–t. 1èresg ved-u :: 3epl ved-ú ≡ v.–m. 1èresg ved-o˛ :: 3epl ved-o˛-tъ. Les deux marques ont en outre ceci de commun qu’elles sont absentes du paradigme d’aoriste, où il n’y a pas de danger d’homonymie interne, cf. v.–t. 1èresg ved :: 3epl ved-u, v.–m. 1èresg ved-ъ :: 3epl ved-o˛. Cependant, dans les deux langues apparaît une homonymie externe entre la 1èresg du présent et la 3epl de l’aoriste.12 ... Ayant ébranlé, ci-dessus, l’identité i.–e.ne de l’exposant de la 3esg dans le v.–i. bhar-a-ti ‘(il) porte’ et v.–m. ber-e-tъ ‘(il) prend’, nous pouvons la restituer par une voie différente: au niveau à f-d, l’exposant -t d’un verbe extraverti congruent à la personne ‘‘autrui’’, peut être identifié, hypothèse acceptable, à la racine td’un «nom» déictique qui survit dans beaucoup de langues i.–e.nes, y compris dans les slaves, cf. v.–m. tъ ‘celui-là’. Au niveau à f-p, l’exposant -t apparaît dans la désinence verbale vieille-indienne -ti, cf. bhar-a-ti, et son analogue déictique, devenu pronom, se trouve incorporé, secondairement, dans certaines langues slaves à la forme verbale, cf. v.–m. ber-e-tъ à côté de v.–t. ber-e. Nous tenons cette «théorie», tout à fait plausible, pour un jeu intellectuel qui ne sert qu’à illustrer la pluralité des reconstructions i.–e.nes parmi lesquelles nous n’avons pas de base pour faire notre choix. L’approche méthodologique que nous avons adoptée dans le présent mémoire nous permet une autre vision du slave «primitif», slave «commun». C’était une langue en devenir, vivant à sa façon le passage i.–e. commun du stade à f-d au stade à f-p. Elle ressentait un vif besoin morphologique de combattre l’homonymie au sein de ses paradigmes. Nous ne pouvons rien dire sur l’origine d’une telle homonymie: à la lumière de la linguistique comparée traditionnelle, elle résultait de l’écroulement d’anciens paradigmes i.–e.s communs, à une lumière alternative, elle pouvait aussi bien refléter une évolution encore inachevée à partir du stade i.–e. préparadigmatique (les deux raisons ne s’excluent pas). Le verbe slave doit envisager des problèmes particulièrement graves: quelques mots isolés mis à part, il n’a qu’une conjugaison à base désinentielle vocalique (thématique ou semi-thématique) et il manque d’exposants de congruence syntaxique (c’est-à-dire, d’exposants de personne grammaticale). Le slave commun cherche de tous côtés . . . et trouve de tous côtés. En résultera la

Sur la paradigmatisation du verbe indo-européen

pluralité des langues slaves, pluralité dont l’évolution s’est déroulée en grande partie à l’époque historique: – à la personne ‘‘nous’’ (=‘‘moi’’ + ‘‘autrui’’/‘‘toi’’) sont attestés plusieurs exposants (morphes -mъ, -my, -mo) qui ne sont pas réductibles l’un à l’autre et dont le consonantisme témoigne d’une inspiration par le pronom ‘nous’; – à la personne ‘‘moi’’, on voit très nettement l’expansion du morphe -m. Au début de l’histoire du vieux-tchèque, le morphe -m ne se rencontre qu’auprès de quelques cinq verbes irréguliers (athématiques). Réagissant à une nouvelle homonymie, issue de la contraction de certains thèmes verbaux, la langue déploie ce morphe-là dans une partie considérable de ses paradigmes. Parent le plus proche du tchèque, le slovaque généralise l’exposant -m dans toute sa conjugaison. Or, le même exposant -m s’est généralisé aussi dans une langue qui n’a jamais connu de contraction, telle que le bulgare. Chose qui mérite d’être dite, le consonantisme de l’exposant est identique à celui des cas obliques du pronom ‘moi’; – aux personnes ‘‘autrui’’ et/ou ‘‘autrui’’ + ‘‘autrui’’, nous rencontrons l’exposant -t(ъ/ь). Il existe dans les peu nombreux verbes irréguliers slaves et son consonantisme reproduit celui du pronom démonstratif ‘celui-ci’. La dualité de son vocalisme peut refléter une pluralité de contextes influençant . . . ... Passons en revue, à la fin, le paradigme formel du vieux-macédonien. L’exposant -tъ, quoique formellement identique à un pronom au masculin singulier,13 s’y est généralisé (i) à toute la 3esg du présent thématique sans égard pour le nombre et le genre du sujet;14 (ii) à toute la 3epl du présent thématique, cf. v.–m. ved-o˛-tъ vs. v.–t. ved-ú ‘ils conduisent’; (iii) à la 3e personne des exemples très peu nombreux de la conjugaison irregulière, dite athématique, cf. v.–m. jes-tъ ‘il est’, das-tъ ‘il donne’, formes opposées à la fois à tchèque je, dá et à vieux-lit. e˜ sti, dúos-ti, pour le singulier, et v.–m. so˛-tъ ‘ils sont’, dade˛-tъ ‘ils donnent’, formes opposées à v.–t. sú, dadie, pour le pluriel. Vu la grande extension du morphe v.–m. -tъ dans les paradigmes formels du présent, nous n’allons pas interpréter les variantes en -(s)tъ de certains aoristes qui ne sont attestées qu’en vieux-macédonien — cf. umreˇ-tъ (v.–m. seulement) à côté de umreˇ (et v.–m. et v.–t.) ‘il mourut’, na-cˇ˛e-tъ (v.–m. seulement) à côté de nacˇ˛e /na-cˇe (et v.–m. et v.–t.) ‘il commença’ — en tant que vestiges d’un aoriste nonsigmatique i.–e. comme le fait Christian Stang (1942:65ss), et reprend M. Herman Kølln (1969: 19ss). Nous tenons le -tъ pour secondaire, telle particularité de la conjugaison du v.–m. où cet exposant se systématisa dans le paradigme formel du présent et pénétra partiellement dans celui de l’aoriste. Comme l’aoriste slave ne distinguait pas formellement la 2esg de la 3esg, l’exposant v.–m. -(s)tъ put se trouver aux désinences des deux personnes. Donc en dépit de sa «jeunesse» relative, ce ne serait pas le vieux-tchèque qui aurait perdu des désinences, mais le vieux-macédonien qui en aurait créé de neuves.



 Tomáš Hoskovec

Remarque: C’est également par innovation que le vieux-macédonien différencie formellement la 2edu de la 3edu, tandis que le vieux-tchèque ne le fait pas. .. Quittons les exposants de congruence syntaxique, c’est-à-dire les exposants de personne, pour les bases vocaliques des désinences. En lituanien, elles sont -a-, -i
a-, -i- (bases brèves) et -o-, -e˙- (bases longues). Aux trois bases brèves lituaniennes correspondent très exactement les bases slaves -e-, -i
e-, -i- (cf. mьr-o˛ & mьr-e-ši pour -e-, bor-jo˛ & bor-je-ši pour -i
e-, zьr-jo˛ & zьr-i-ši pour -i-), or les deux bases longues lituaniennes ne trouvent rien de correspondant dans les divers systèmes de conjugaison slave. Certes, à l’intérieur des présents slaves deˇl-a-je-ši, deˇl-a-je-te et um-eˇ-je-ši, um-eˇje-te on peut isoler les segments -a- et -eˇ-, mais ce sont des suffixes dérivationnels auxquels s’ajoute la base brève -i
e- suivie des exposants -ši et -te, tandis qu’aux bases lituaniennes -o- et -e˙- les exposants de personne s’ajoutent immédiatement, cf. mók-a-i, mók-o-te & mók-e-i, mók-e˙-te (devant les exposants vocaliques -i de la 2esg et -u de la 1èresg, les bases s’abrègent pour faire ensemble une diphtongue). Outre les bases vocaliques -o- et -e˙-, le lituanien connaît, lui aussi, les suffixes dérivationnels -o- et -e˙- qu’il ne faut pas confondre avec les premières: 1o les exposants vocaliques -u et -i s’ajoutent aux suffixes -o- et -e˙- à l’aide d’un -j- épenthétique; 2o aux suffixes -o- et -e˙- s’ajoutent, toujours à l’aide d’un -j- épenthétique, les bases vocaliques -a- et -o-; 30 les suffixes -o- et -e˙- reçoivent et attirent l’accent du mot, ce que les bases vocaliques ne font jamais (sauf si, suivies de l’exposant de la 1èresg ou de la 2esg, elles font des désinences à l’état diphtongué, cf. rãš-o → raš-aı˜, rãš-e˙ → raš-eı˜ pour ‘écrire’). Exemples: Comparons à ce sujet sáug-a-i, sáug-o-te (-o- = base de désinence, ne prend jamais d’accent) à côté de saug-ó-j-a-i, saug-ó-j-o-te (-o- = suffixe dérivationnel, prend l’accent), mók-e-i, mók-e˙-te (-e˙- = base de désinence) à côté de mok-e˙´ -j-a-i, mok-e˙´ -j-o-te (-e˙- = suffixe dérivationnel). Il semble que la différence dans la nature morphologique de la base de désinence et du suffixe de dérivation échappe aux linguistes qui cherchent à établir une préhistoire du verbe balte et slave. Aussi Stang (1942: 143) en affirmant ‘‘Das a¯ . . . tritt also nicht als Präsenssuffix, sondern als Verbalsuffix auf,’’ identifie-t-il les voyelles o qui suivent après le thème žin- l’une dans le présent žìno et l’autre dans le prétérit žinójo. Or, le suffixe et la base de désinence ont chacun une sémantique à part (cf. 8.3.2, 8.4.5) et un comportement morphologique particuler. Remarque: En letton, on ne rencontre que deux bases vocaliques, une longue, -a¯- (alternant à la non-personne avec -a), et une brève, -a- (qui, à la non-personne, ne se réalise même pas). La base longue -a¯- correspond à lituanien -o-, tandis que la brève a deux correspondances lituaniennes: elle équivaut à lit. -apour les verbes lettons qui ne palatalisent pas la consonne finale de leur thème (sauf éventuellement à la 2esg), cf. 2esg na¯c-∅, 2epl na¯k-a-t, non-pers. na¯k à côté de l’infinitif na¯k-t ‘venir’; elle équivaut à lit. -i
a- pour les verbes lettons qui palatali-

Sur la paradigmatisation du verbe indo-européen 

sent systématiquement la consonne finale de leur thème, cf. 2e sg brauc-∅, 2epl brauc-a-t, non-pers. brauc à côté de l’infinitif brauk-t ‘aller (en voiture)’. En outre, le letton connaît une conjugaison particulière qui correspond à la lituanienne à base -i-. Elle est à base -a-, ne palatalise curieusement pas la finale du thème et se manifeste par un exposant non-zéro, à savoir -i, à la 2esg. Elle a donc 2esg prot-i, 2epl prot-a-t, non-pers. prot, inf. prast < *prat-t ‘savoir’. L’exposant -i l’oppose aux deux autres conjugaisons lettonnes à base -a- (correspondant à lit. -a- ou -i
a-), cf. 2esg sa¯c-∅, 2epl sa¯k-a-t, non-pers. sa¯k, inf. sa¯k-t ‘commencer’; en même temps, la base brève l’oppose à la conjugaison à base longue qui, elle aussi, a l’exposant non-zéro -i à la 2esg, cf. 2esg sak-i, 2epl sak-a¯-t, non-pers. sak-a, inf. saca¯t ‘dire’. La conjugaison lituanienne à base -e˙- n’a point d’homologue en letton. .. La différence morphonologique entre les désinences verbales lituaniennes et slaves mise à part, il reste encore une autre différence importante. Les désinences lituaniennes (que l’on peut parfaitement caractériser par leurs bases vocaliques) sont de haute spécificité lexico-sémantique ce qui n’est pas du tout le cas des désinences slaves (sans parler de leurs bases vocaliques). Exemples: Observons ici le rôle des bases vocaliques dans les oppositions morphologiques suivantes: mók-a ‘(on) sait’ (base -a- non-marquée) & mók-o ‘(on) fait que l’on sache = (on) apprend/enseigne’ (base -o- marquant les «effectif/causatifs»); saug-ùs (adjectif) ‘sauf, sûr’ & sáug-o ‘(on) fait que l’on soit sauf, sûr = (on) protège/défend’ (base -o- marquant les «effectif/causatifs»); gùl-a ‘(on se) couche’ (base -a- non-marquée) & gùl-i ‘(on) est couché’ (base -imarquant un «état»); au˜-n-a ‘(on) chausse, met sur soi [des chaussures]’ (base -anon-marquée) & ãv-i ‘(on) chausse toujours, porte sur soi d’habitude [des chaussures]’ (base -i- marquant un «état»); etc., etc. Nous ne croyons pas que le lituanien ait conservé à merveille un état primitif balto-slave (sinon indo-européen) qui aurait disparu même dans les langues slaves les plus anciennement attestées. Nous croyons que le lituanien a développé de façon systématique un aspect qui est resté plutôt négligé par la majorité des langues flexionnelles anciennes: les différents paradigmes formels (cf. 3.2) peuvent être utilisés comme classeurs lexico-sémantiques. On peut constater une évolution pareille dans le développement de la déclinaison lituanienne. . Il est certes possible de parvenir, par reconstruction interne, à un balte et à un slave primitifs.15 Ils diffèrent du vieil-indien, du grec ou du germanique primitif par beaucoup de traits particuliers (par exemple, par un système proprement balte et par un autre proprement slave d’oppositions phonologiques de voyelles). Admettons donc que le balte et le slave primitifs (constitués les deux au stade à f-p) aient eu aussi des systèmes particuliers de conjugaison paradigmatique, systèmes bien différents de ceux de conjugaison hittite, indienne, grecque, latine et autre (constitués eux aussi au stade à f-p). Le balte et le slave primitifs ne connaissent de

 Tomáš Hoskovec

paradigme systématisé qu’un présent thématique sans valeur temporelle. À part cela, ils attestent, à la périphérie de leurs systèmes respectifs, des formes isolées ressemblant à d’anciens verbes athématiques d’origine extravertie (laquelle ressemblance est néanmoins fallacieuse, cf. 8.4.1, 8.4.2). Il serait vain et futile d’y chercher des reflets d’un système à l’indienne: et le slave primitif et le vieil-indien représentent pour nous d’ultimes étapes accessibles en remontant par divers chemins dans le temps et rien ne nous autorise à supposer que l’une des langues ait jamais exercé de l’influence sur l’autre.16 En plus, de pareilles recherches de reflets lointains nous détournent d’autres pistes d’investigation qui, se trouvant à l’intérieur des langues baltes et slaves où elles ont sur quoi s’appuyer, sont beaucoup plus sûres et, nous en sommes persuadé, beaucoup plus intéressantes.

9. Conclusion Dans les analyses et exemples ci-exposés, nous avons pu entrevoir l’ampleur de la dérivation qui jadis enrichissait le verbe i.–e. primitif en produisant, à partir des racines déjà existantes, de nouvelles nuances de signification. Les moyens dérivationnels qui y servaient étaient de haute spécificité sémantique et d’application restreinte. Lors de leur constitution (individualisation), les langues i.–e.nes ont réarrangé ces dérivés-là en catégories grammaticales à force de mettre en relief, pour chaque dérivé, certains traits distinctifs et reléguer les traits sémantiques primitifs au second plan. Aussi les formes des dérivés ont-elles acquis leur signification moins par une marque formelle quelconque que par le fait de se trouver opposées mutuellement au sein d’un système clos, et les moyens dérivationnels ont-ils perdu leur spécificité sémantique, les uns devenant des allomorphes qui servaient la même catégorie, les autres s’étendant aux racines auxquelles ils n’étaient pas applicables auparavant. C’est ainsi que se sont constitués les paradigmes grammaticaux dont la structure est spécifique dans chaque langue. Il s’en suit que l’occurrence, dans la langue A, d’une forme qui, dans la langue B, est typique pour une catégorie grammaticale ne garantit pas la présence de cette catégorie dans A. Vu la portée du processus de paradigmatisation, vu la pluralité et l’individualité des voies par lesquelles des catégories verbales de base pouvaient prendre structure dans les systèmes grammaticaux des langues indo-européennes anciennes, la question «Quelles étaient les catégories verbales de l’indo-européen commun?» nous semble mal posée. Rappel des arguments: Appuyés sur une distinction typologique entre une flexion dérivationnelle (f-d) et une flexion paradigmatique (f-p), introduite au chapitre 3, nous n’avons présumé, pour le verbe indo-européen primitif, que deux classes différentielles lexico-sémantiques, celles d’extraverti et d’introverti (4.2), marquées morphologiquement et sous-divisées respectivement en complexe vs.

Sur la paradigmatisation du verbe indo-européen

évolutif d’un côté et en dynamique vs. statique de l’autre (7.1), sous-divisions différentielles lexico-sémantiques sans marques morphologiques aucunes. Lors de la constitution des langues indo-européennes historiques (processus de paradigmatisation que nous concevons comme des passages strictement individuels du stade à f-d au stade à f-p), cette double division primitive a pu disparaître totalement (5.2.1), se reconstituer en d’autres classes lexico-sémantiques — soit différentielles (8.1), soit oppositionnelles (8.2) — ou bien se réarranger en catégories grammaticales oppositionnelles (7.1), le réarrangement donnant naissance à des paradigmes particuliers comprenant des catégories particulières telles que le processuel, le déagentif, le moyen (7.2 & 7.3). À cela se sont ajoutées, en deuxième rang, les catégories d’actualité (5.2) et d’antériorité/postériorité (7.5, 8.3). Nous nous sommes concentrés sur les thèmes verbaux et les classes de conjugaison en tant que paradigmes formels (3.2). Quant aux désinences, nous en avons postulé un inventaire minimum au stade à f-d (4.3), tout en soulignant l’individualité de leurs arrangements paradigmatiques en systèmes de désinences dans les langues i.–e.nes particulières au stade à f-p (5.1, 6.1, 8.4).

Notes * Pour la rédaction finale de la deuxième partie du présent mémoire, l’auteur a bénéficié, en 1997, d’une bourse de recherche de la Fondation Andrew W. Mellon, assignée à la Maison des Sciences de l’Homme à Paris. . Le redoublement fut à l’origine un moyen non-systématisé d’emphase et il est effectivement attesté aux trois thèmes du verbe grec (aoriste, présent, parfait). Ce ne fut que plus tard qu’il devint, systématisé, la marque du parfait, tout comme l’augment, systématisé lui me aussi, devint la marque du passé. M Kurzová (1993: 140) fait remarquer que dans les formes grecques à redoublement consonantique se répercutent les deux particules d’actualisation: i au présent, e au parfait, cf. *\*T:-4 ‘je donne’ vs. ¦-*,*T´ 6,4< ‘j’avais donné’, (\(<,J"4 ‘(il) naît’ vs. (,(X<0J"4/(X(@<, ‘(il) est né’. Ajoutons à son observation que les aoristes à redoublement reprennent la voyelle radicale et par conséquent restent en dehors de la distribution de i et de e, cf. –("(, ‘(en)mène!’, §<,(6, ‘porte!’. . «Agent», opposé à «affecté»/«récipiendaire», signifie ici acteur et/ou porteur du procès du verbe sans égard pour son évaluation d’après une sémantique cognitive (cf. 4.2.2 dans la première partie). «Agent périphrastique» est un argument verbal à statut différent. L’auteur a me développé sa conception d’agent de verbe, en élargissant celle de M Jarmila Panevová (1980), dans sa thèse de doctorat exempla Litvanica (1994). Un manuscript annexe non publié, exemplvm Latino-Graecvm (1995), discute, sous cet angle de vue, les faits latins et grecs. . Pour suivre nos arguments, il suffit ici d’imaginer la valence de verbe (cadre de compléments verbaux obligatoires et facultatifs) dans le sens que lui donne Lucien Tesnière (1959). Cependant, nos analyses sont faites à partir de la théorie élaborée dans l’école de M. Petr Sgall (1984, 1986).



 Tomáš Hoskovec

. Contrairement à l’avis répandu qui voit dans l’imparfait uidebam une forme analytique h me à auxiliaire *b u¯-, l’auteur souscrit à l’explication donnée par M Kurzová (1993: 187–9) h qui identifie l’élément -b- à l’élargissement i.–e. -d -, également présent dans le «passif» grec en -h0-, dans les prétérits «faibles» germaniques, dans le praeteritum frequentatiuum h lituanien. Émile Benveniste (1935=1984: 189) caractérise l’élargissement -d - comme exprimant l’état, et plus particulièrement, l’état achevé. Cette signification originelle de l’élement constitutif de l’imparfait nous semble convenir au principe sémantique qui reliait, pendant toute la paradigmatisation du latin, l’état avec une action achevée dont il résulte, principe que nous avons vu reposer également à l’origine des paradigmes latins de perfectum et de passif périphrastique (cf. 7.2.1 et 7.2.3). Les voyelles -a¯- et -e¯- qui précèdent le suffixe -b- font partie de l’inventaire i.–e. commun de morphologie verbale, comme l’a expliqué M. Adolf Erhart (1961, 1983). Nous n’interprétons pas non plus le plus-queparfait latin uideram par *uisus eram, mais proposons la reconstruction uideram < *uid-i-sam qui est structurellement parallèle à l’imparfait uid-e-b-am: la désinence propre -am est la même dans les deux formes, et à l’élargissement -b- de l’imparfait correspond le -s- du plus-que-parfait. Sur le même élargissement -s-, précédé d’un -i-, se sont appuyés d’autres paradigmes dérivés au perfectum, tels que uidero . . ., uiderim . . . dont nous interprétons les désinences comme structurellement parallèles à celles de faxo . . ., faxim . . . < fac-so . . ., fac-sim . . ., cette dernière interprétation étant empruntée à M. Jay H. Jasanoff (1987). Analogiquement, au sein de l’infectum, l’élargissement -b- a servi, telle marque de thème secondaire, à constituer un paradigme formel de futur. . Il n’est pas lieu et il n’est non plus notre intention de donner ici une théorie sémanme tique de l’aspect du verbe slave. À l’instar de M Barbara H. Partee (1999) qui met en corrélation l’aspect du verbe et les déterminants du nom, nous recommandons d’approcher l’aspect verbal en tant qu’un cas particulier de détermination. Toute langue est capable d’exprimer un procès comme déterminé. Certaines, dont le tchèque ou le grec (ancien et moderne), à force de viser dans le domaine du verbe, développent des systèmes d’aspect verbal, certaines autres, telle que le français ou l’anglais, à force de viser dans le domaine du nom, développent des systèmes de déterminants nominaux, tandis que d’autres encore, par exemple le latin ou le lituanien, sans développer un système grammatical particulier, se servent à ce but de moyens provenant de divers domaines. Ceci dit, il ne faut pas oublier que toute théorie sémantique est méta-linguistique à force de corréler les signes de langue avec la réalité externe à la langue. Une théorie sémantique de l’aspect verbal n’acquitte donc pas le linguiste du devoir de présenter une théorie linguistique de l’aspect, en corrélant les signes de langue entre eux. Sans pouvoir nous mettre à cette tâche, nous profitons de l’occasion pour témoigner des sympathies que suscite en nous l’approche méthodologique de Jens Holt (1943). . Il est facile de rapprocher le slave veˇd-eˇ du grec @É*", vieil-indien ve¯d-a. Cependant, la linguistique traditionnelle n’a jamais trouvé d’explication persuasive de la désinence -eˇ dans le proto-indo-européen (une revue critique des hypothèses avancées à ce sujet est donnée par Josef Miloslav Korˇínek 1948). Quelle que soit son origine, la désinence -eˇ doit être une création morphologique proprement slave. Rappelons que l’allomorphe veˇdeˇ est inconnu des textes les plus anciens en vieux-slave, c’est-à-dire en vieux-macédonien littéraire (il n’est fréquent que dans le Codex Suprasliensis, le chiffre exacte en est donné dans l’œuvre

Sur la paradigmatisation du verbe indo-européen 

posthume de Nikolaj Sergeevicˇ Trubeckoj 1954: 160); par contre, il était assez répandu en ème vieux-tchèque du XIV siècle, c’est-à-dire à l’époque où se préparaient déjà de grands ème changements morphologiques, survenus en tchèque au XV . . Il est traditionnel de tenir i-sk-o˛, i-sk-a-ti ‘chercher’ pour le seul verbe slave à radical en -sk- indo-européen, cf. Stang (1942: 21). Or il n’est pas légitime de présenter isk-o˛ isk-a-ti comme un thème slave en -sk-. Pour dire cela, il faudrait prouver que -sk- est un élargissement dans le système morphologique d’une langue slave quelconque, c’est-à-dire (i) qu’il commute avec un autre élargissement auprès de la même racine i- et (ii) qu’il s’applique également à une autre racine. Pour (i) on pourrait opposer i-sk-o˛ à i-d-o˛ auprès d’une racine i- signifiant un déplacement à pied (cf. i-ti ‘aller à pied’ opposé à ja-ti ‘se déplacer à cheval, en voiture’), mais comme aucun verbe ne prouve (ii), le rapprochement (i) reste hypothétique. . Le groupe -sk- n’est point stable en lituanien. Il se range dans l’ordre sifflantegutturale devant une voyelle, mais dans l’ordre inverse gutturale-sifflante devant une consonne, cf. dre˙sk-e˙ (prétérit) & dre˙ks-ti (infinitif) pour ‘gratter’. . Un tableau synoptique en est donné par Bohuslav Havránek (1927: 156–9), dans le domaine balte, un processus parallèle a été étudié par Pavel Trost (1958), la concurrence du participe et de l’infinitif en vieux-macédonien (vieux-slave) a été décrite par M. Radoslav Vecˇerka (1961: 65–9). . Il y a deux arguments qui sont en jeu ici: 10 non-différenciation du nombre, 20 noncoexistence des désinences -ti et -nti. L’idée du grand archaïsme de la non-différenciation e du nombre à la 3 personne a été émise pour la première fois par M. Vladimir Toporov (1961); une discussion typologique importante, assez sceptique, en donne Pavel Trost (1981). Ayant refusé la première, M. Adolf Erhart (1987) s’est servi de la deuxième idée, admettant la désinence -nt (sans i) en fonction «primaire», pour donner une explication nes phonologique satisfaisante de la disparition, par rapport aux autres langues i.–e. , des e exposants personnels à la 3 personne du verbe balte. . André Vaillant évite le processus de durcissement/ramollissement des yers finals en postulant un processus non moins ad hoc de «restauration des consonnes finales». ‘‘On e n’en connaît qu’un cas sûr, dans les désinences de 3 personne du singulier et du pluriel du présent, qui, étant dissyllabiques, subissent une réduction en slave comme en baltique . . .’’ dit-il (Vaillant 1950: 207). La sûreté de son processus ne s’appuie, hélas! que sur l’hypostasie e d’un verbe balto-slave primitif ayant des désinences de 3 personne en -t(i). Circonstance pertinente, l’argument de «restauration des consonnes finales» ayant été présenté dans le tome I (Phonétique), il n’est ni repris ni même rappelé dans le tome III (Le verbe), où l’auteur se contente de décrire la distribution de -t désinentiel, en remarquant (Vaillant 1966: 19): ‘‘Le système du slave s’explique par le système plus riche de l’indo-européen qu’il continue. Quant au système des désinences verbales de l’indo-européen, . . . il ne s’agit plus de l’expliquer, mais on peut relever certaines de ses particularités.’’ . M. Miroslav Komárek (1969: 36) voit dans la longueur de la désinence vieille-tchèque e -ú (3 pl.ind.praes) une trace de l’élément -tь/-tъ présumé slave commmun. Il établit un changement v.–m. jazъ > v.–t. já ‘moi’, v.–m. vedo˛tъ > v.–t. vedú ‘[ils] conduisent’ où la longueur serait provoquée par la chute de la dernière syllabe. Même si nous admettions le

 Tomáš Hoskovec

processus qu’il postule, cela ne serait aucune preuve de l’existence de l’élément -tь/-tъ dans un vieux-tchèque préhistorique. Or, l’argument de M. Komárek ne nous paraît pas convainquant: 10 Le savant met en parallèle une paire existante et une paire non-existante: le vieux-tchèque connaît effectivement les deux formes jaz et já, mais seulement la forme vedú. 20 La graphie du vieux-tchèque nous permet de lire non seulement jaz à voyelle brève, mais aussi jáz à voyelle longue. 30 Le mécanisme de ce processus devrait conduire également à v.–m. vedetъ > v.–t. *vedé ‘[il] conduit’, ce qui n’est pas le cas. L’auteur mentionne (p. 85) e des cas d’allongement secondaire de la désinence de la 3 sg en tchèque, qu’il essaie d’expliquer par son mécanisme (à savoir v.–m. vedetъ > v.–t. *vedé), mais il finit par avouer lui-même que ces cas-là ont plutôt d’autres raisons. Somme toute, la distribution de la longueur en tchèque est un problème très complexe qui manque toujours d’explication globale et unifiante, couvrant tous les faits attestés; on ne peut pas en tirer des arguments pour des faits non-attestés. Quelle que soit l’«origine» de la longueur dans la désinence vieille-tchèque ved-ú, sa «raison» dans le système de la conjugaison tchèque est de prévenir ère une homonymie malcommode au sein du paradigme de présent, en différenciant 1 sg vede u :: 3 pl ved-ú. Dans ce sens, elle remplit le même rôle que le segment vieux-macédonien -tъ ère e qui oppose 1 sg ved-o˛ :: 3 pl ved-o˛-tъ. . Nous remontons jusqu’à Filip Fedorovicˇ Fortunatov (1908) qui fut le premier à dériver l’exposant -tъ à partir du pronom *tås, et à P’otr Savvicˇ Kuznecov (1956). Néanmoins, l’identité formelle de l’élément désinentiel -tъ et du pronom tъ frappa d’autres contemporains de Fortunatov, dont Václav (Wenzel) Vondrák (Wondrak 1908: 135): ‘‘Schon frühzeitig trat hier eine Verhärtung des -tь zu -tъ ein, wobei vielleicht auch das Pron. tъ einen Einfluß ausüben konnte.’’ . La motivation, si jamais il y en a eu, du morphe -tъ par le pronom tъ doit donc être restée bien lointaine. Rien ne nous autorise à concevoir les désinences vieilles-macédoniennes à -tъ comme résultant d’une soudure du verbe avec le pronom tъ enclitique, tandis que dans le vieux-prussien nous avons de bonnes raisons de penser ainsi. Nous ne pouvons que regretter profondément que la pénurie des matériaux vieux-prussiens ne nous permette guère d’étudier en détail la distribution de l’élément désinentiel -ts en corrélation avec le sujet du verbe. . Un balte primitif devrait, selon nous, être beaucoup plus proche d’un vieux-prussien à catégories morphologiques peu distinctes que d’un lituanien à catégories morphologiques bien distinctes, riches, pratiquement sans homonymie et à haute motivation sémantique. s

. De prétendus reflets d’archaïsmes i.–e. que l’on croit trouver dans les langues baltes et slaves reposent parfois plus sur une volonté de les y voir que sur des preuves probantes. Nous avons rappelé, à titre d’exemple (8.2 note 6), les difficultés que pose l’allomorphe marginal slave veˇd-eˇ dès que l’on veut rapprocher sa désinence de celle de la forme grecque, primaire et unique, @É*". Dans un article particulier (à paraître en 2002), l’auteur traite des soi-disant thèmes à redoublement baltes et slaves (cf. lit. deda, duoda; vieux-macédonien dastъ) pour lesquels la volonté de les voir parallèles aux thèmes grecs J\20:4, *\*T:4 l’emporte traditionnellement sur les analyses morphologiques internes. Cependant, lesdits thèmes trouvent au sein des systèmes de leurs langues respectives des descriptions plus adéquates, même dans l’optique diachronique, en guise de thèmes élargis et non redoublés.

Sur la paradigmatisation du verbe indo-européen 

Références Abréviations HfM

Det Kongelige Danske Videnskabernes Selskab, Historisk-filologiske Meddelelser. København: Munksgaard TCLP Travaux du Cercle linguistique de Prague SPFFBU Sborník prací filosofické fakulty Brneˇnské (Masarykovy) university Benveniste, E. 1950. ‘‘Actif et moyen dans le verbe’’. Journal de psychologie, janvier-février. Reimpressum in Benveniste (1966: 168–75). Benveniste, E. 1966. Problèmes de linguistique générale I. Paris: Gallimard. Benveniste, E. 1984. Origines de la formation des noms en indo-européen. Paris: Adrien Maisonneuve. Chantraine, P. 1926. Histoire du parfait grec. Paris: Honoré Champion. Dostál, A. 1954. Studie o vidovém systému v starosloveˇnštineˇ [Études du système aspectuel en vieux-slave]. Praha. Erhart, A. 1961. ‘‘Indoevropská préterita s dlouhými vokály a¯, e¯ ’’ [Les prétérits indoeuropéens aux voyelles longues a¯, e¯]. SPFFBU A9: 17–33. Erhart, A. 1983. ‘‘Das lateinische System der verbalen Modi und seine indoeuropäischen Grundlagen’’. SPFFBU A31: 25–30. Erhart, A. 1987. ‘‘Die Nichterscheinung von Numerus in der 3. Person des baltischen Verbs und ihre indoeuropäische Grundlagen’’. Baltistica XXIII(2): 126–30. Erhart, A. 1988. ‘‘Vom protobaltischen zum urslavischen Verbalsystem’’. SPFFBU A36: 39–49. Erhart, A. 1989. Das indoeuropäische Verbalsystem. Université de Brno. Fortunatov, F. F. 1908. [Ф.Ф. Фортунатов] ‘‘Стapocлaвянcкoe mъ в 3-м лицe глaгoлoв’’. Извecmuя Omд. pyccкoгo языкa u cлoвecнocmu AH, т. XIII: 1–44. Havránek, B. 1928 & 1937. Genera verbi ve slovanských jazycích [Les genera uerbi dans les langues slaves] Vol. 1 (1927), 2 (1937). Praha: Královská cˇeská spolecˇnost nauk. Havránek, B. 1939. ‘‘Aspect et temps du verbe en vieux-slave’’. Mélanges de linguistique offerts à Charles Bally, 223–30. Genève. Holt, J. 1943. ‘‘Études d’aspect’’. Acta Jutlandica XV/2. Universitetsforlaget i Aarhus. Hoskovec, T. 1994. Exempla Lituanica sententiarum significationis describendae quaestiones aliquot explicantia. Thèse de doctorat (CSc), Université de Prague. Hoskovec, T. 1995. Exemplum Latino-Graecum exemplis Lituanicis annectendum de uerbi diatheseos sententiarum significatione describendae aut necessitate aut redundantia tractans. Praha (manuscrit). Hoskovec, T. 1997. Verbum Lituanum. Tomus I coniugationes et classes uerborum conectens. (manuscript). Hoskovec, T. 2002. ‘‘Ist lit. duodu ein Reduplikationsstamm?’’ (à paraître). Jasanoff, J. H. 1987. ‘‘The tenses of Latin perfect system’’. In Festschrift for Henry Hœnigswald: On the occasion of his seventieth birthday, ediderunt G.Caedona & N. H.Zide, 177–83. Tübingen: Gunter Narr. Kazlauskas, J. 1968. Leituviu˛ kalbos istorine˙ gramatika [Grammaire historique de la langue lituanienne]. Vilnius: Mintis.

 Tomáš Hoskovec

Komárek, M. 1969. Historická mluvnice cˇeská. Cˇást I. Hláskosloví. [Grammaire historique du tchèque. Tome I. Phonologie et morphonologie]. Praha: Státní pedagogické nakladatelství. Korˇínek, J. M. 1948. ‘‘K pu˚vodu staroslovanského tvaru veˇdeˇ ‘scio’ ’’ [Au sujet de l’origine de la forme vieux-slave veˇdeˇ ‘scio’]. In Slovanské studie. Sborník prelátu prof. Josefu Vajsovi, edidit J.Kurz, 100–4. Praha: Vyšehrad. Kurzová, H. 1993. From Indo-European to Latin. The evolution of a morphosyntactic type. Amsterdam/Philadelphia: John Benjamins. Kuznecov, P. S. 1956. [Л. C. Kyзнeцoв] ‘‘K иcтopии фopм 3-гo лицa нacтoящeгo вpeмeни в pyccкoм языкe’’. Slavia 25: 175–83. Kølln, H. 1969. Oppositions of voice in Greek, Slavic, and Baltic. HfM 43/4. Meillet, A. & J. Vendryes. 1924. Traité de grammaire comparée des langues classiques. Paris: Champion Meillet, A. & J. Vendryes. 1934. Le slave commun (Seconde édition revue et augmentée avec le concours d’André Vaillant). Paris: Champion. Neˇmec, I. 1958. Genese slovanského systému vidového [Genèse du système aspectuel slave]. Praha: Rozpravy cˇeskoslovenské akademie veˇd, rocˇ. 68, sešit 7. Panevová, J. 1980. Formy a funkce ve stavbeˇ cˇeské veˇty [Formes et fonctions dans la structure de la phrase tchèque. Praha: Academia. Studie a práce lingvistické 13. Partee, B. H. 1999. ‘‘Nominal and temporal semantic structures’’. TCLP (nouvelle série) vol. 3: 91–108. Perel’muter, I. A. 1977. [И. A. Пepeльмyтep] Oбщeuндoeвponeu˘cкuu˘ u гpeчcкuu˘ глaгoл (вuдoвpeмeнныe u зaлoгoвыe кameгopuu). Лeнингpaд: Hayкa. a a a Ruipérez, M.S. 1952. ‘‘Desinencias medias primas indoeuropeas sg.1 -(m)ai, sg.2 -soi, sg.3 a -(t)oi, pl.3 -ntoi’’. Eremita 20: 8–31. Šabršula, J. 1962/65. ‘‘Contribution aux problèmes de méthode de la recherche dans le ème domaine de l’aspect — Langues romanes’’. Actes du X Congrès international de linguistique et philologie romanes 1962, 157–74. Paris: Klincksieck 1965. Šabršula, J. 1999. ‘‘Aspect, contexte, distribution’’. TCLP (nouvelle série) vol. 3: 125–33. Schmid, W. P. 1963. Studien zum baltischen und indogermanischen Verbum. Wiesbaden: Otto Harrassowitz. Serbat, G. 1976. ‘‘Les temps du verbe en latin’’. Revue d’Études Latines 54: 308–52. Sgall, P. & alii. 1984. Contributions to Functional Syntax, Semantics and Language Comprehension. Praha: Academia. Sgall, P., E. Hajicˇová et J. Panevová. 1986. The Meaning of the Sentence in Its Semantic and Pragmatic Aspects. Dordrecht: D.Reidel/Praha: Academia. Stang, Chr. S. 1942. ‘‘Das slavische und baltische Verbum’’. Det Norske Videnskaps-Akademi i Oslo, Hist.-Filos. Klasse bd II, N0 1: 39–280. Stang, Chr. S. 1962. ‘‘Die athematischen Verba im Baltischen’’. Scando-Slavica 8: 161–70. Tesnière, Lucien. 1959. Éléments de syntaxe structurale. Paris: Klincksieck. Toporov, V. 1961. [B. Toпopoв] ‘‘K вoпpocy oб эвoлюции cлaвянcкoгo и бaлтийcкoгo глaгoлa’’. Bonpocы cлaвянcкoгo языкoзнaнuя 5: 35–70.

Sur la paradigmatisation du verbe indo-européen

Trost, P. 1958. ‘‘Infinitiv v litevštineˇ’’ [L’infinitif lituanien]. In Studie ze slovanské jazykoveˇdy. Sborník k 70. narozeninám akademika Františka Trávnícˇka, ediderunt Václav Macek et alii, 271–4. Praha: Státní pedagogické nakladatelství. Trost, P. 1981. ‘‘Kategorie cˇísla v baltském slovese’’ [La catégorie du nombre dans le verbe balte]. In Praha – Vilnius. Sborník k 400. výrocˇí založení university ve Vilniusu. Reimpressum in Trost (1995: 321–3). Trost, P. 1995. Studie o jazycích a literaturˇe [Études de langues et de littérature], edidit Jaromír Povejšil. Praha: Torst. Trubeckoj, N. S. 1954 [Trubetzkoy] Altkirchenslavische Grammatik. Schrift-, Laut- und Formensystem, edidit Rudolf Jagoditsch. Wien: Österreichische Akademie der Wissenschaften. Philologisch-historische Klasse 228/4. Vaillant, A. 1950. Grammaire comparée des langues slaves. Tome I, Phonétique. Paris – Lyon: IAC. Vaillant, A. 1966. Grammaire comparée des langues slaves. Tome III, Le verbe. Paris: Klincksieck. Vecˇerka, R. 1961. Syntax aktivních participií v starosloveˇnštineˇ [Syntaxe des participes actifs en vieux-slave]. Praha: Státní pedagogické nakladatelství. Wondrak, W. 1908. Vergleichende slavische Grammatik. Band II. Formenlehre und Syntax. Göttingen: Vandenhoeck und Ruprecht. Zubatý, J. 1901. ‘‘Srovnávací skladba jazyku˚ indoevropských’’ [Syntaxe comparée des langues indo-européennes]. Veˇstník Cˇeské akademie pro veˇdy, slovesnost a umeˇní rocˇ. 10: 507–32. Reimpressum in Zubatý (1954: II, 7–43). Zubatý, J. 1945, 1949, 1954. Studie a cˇlánky [Études et articles] I (1,2), II. Praha: Cˇeská akademie veˇd a umeˇní/Cˇeskoslovenská akademie veˇd. Zubatý, J. 1980. Cˇeské sloveso [Le verbe tchèque], edidit Jaroslav Porák. Praha: Academia



Part III

Topic–Focus articulation

The Russian genitive of negation in existential sentences The role of Theme–Rheme structure reconsidered*

Vladimir Borschev and Barbara H. Partee Moscow and Amherst, Mass.

Introduction

186

.

Babby on ‘‘Declarative’’ and ‘‘Existential’’ sentences

190

.. The scope of negation. Theme and Rheme. Problems .. Sentences with the verb byt’ ‘be’ and referential subject

190 195

2.

Paducˇeva on genitive verbs

196

.

Existential sentences and the ‘‘genitive rule’’

200

4.

Existential sentences: LOC as Perspectival Center

205

.. THING and LOC .. Perspective ... The notion of Perspective and its role in Existential Sentences ... Perspective vs. Theme–Rheme Structure ... Examples ... Perspective and Observer ... Issues about the sources of presuppositions

206 207 207 208 213 220 222

.

On the scope of negation 224 .. Nominative under scope of negation in BE(THING, LOC) sentences 225 .. Genitive of negation apparently not under scope of sentential negation 229

.

Syntactic issues .. Unaccusativity vs. Non-agentivity .. Existential sentences as Impersonal sentences .. LOCative as subject? ‘‘THING’’ as subject? .. Genitive of negation as a Negative Polarity Item .. The nature of the NP in the Genitive of Negation construction

232 232 233 235 235 236

.

Concluding remarks

237

Appendix. Data from Serbo-Croatian

238

 Vladimir Borschev and Barbara H. Partee

Introduction We introduce our topic by citing Yu. D. Apresjan (1985): In some negated sentences of Russian, as is well known, two main case forms are possible — nominative case and genitive case: Pis’mo ne prišlo — Pis’ma ne prišlo.1 The syntactic, semantic, and communicative particulars of the second of these constructions is one of the classic themes of general and Russian grammar, and has given rise to a huge literature. (Apresjan 1985: 292)

Another characteristic of intransitive sentences whose subject is marked with the Genitive of Negation (henceforth GenNeg) is the non-agreement of the ‘‘impersonal predicate’’ with the subject. Russian grammarians often classify such sentences as impersonal, although, as Peškovskij notes, ‘‘These sentences are impersonal only when negated. If one removes the negation, they become personal . . .’’.2 (Peškovskij 1938: 334.) Babby (1980) introduced the terminology ‘‘negated declarative sentences’’ (NDS), for the sentences with nominative subjects, and ‘‘negated existential sentences’’ (NES), for those with genitive ‘‘subjects’’ (there are controversies about subjecthood3), as in (1a) and (2a) (his (81a–b), from Ickovicˇ 1974). We give the corresponding affirmative sentences (ADS and AES) in (1b) and (2b). (1) a. NDS Otvet iz polka ne prišel. answer-nom.m.sg from regiment neg arrived-m.sg ‘The answer from the regiment has not arrived.’ b. ADS Otvet iz polka prišel. answer-nom.m.sg from regiment arrived-m.sg ‘The answer from the regiment has arrived.’ (2) a. NES

b. AES

Otveta iz polka ne prišlo. answer-gen.m.sg from regiment neg arrived-n.sg ‘There was no answer from the regiment.’ Prišel otvet iz polka. arrived-m.sg answer-nom.m.sg from regiment ‘There was an answer from the regiment.’

The affirmative ADS and AES sentences differ (obligatorily) in the order of subject and verb, while in the negative sentences, where the difference between NDS and NES is marked by case, the word order can vary; we return to this important point later. Here are some more standard examples.4 (3) a. NDS: Stok talyx vod ne nabljudalsja. runoff-nom.m.sg melted water neg was.observed-m.sg ‘No runoff of thawed snow was observed.’5

The Russian genitive of negation 

b. NES: Stoka talyx vod ne nabljudalos’. runoff-gen.m.sg melted water neg was.observed-n.sg ‘No runoff of thawed snow was observed.’ (= There was no runoff.) (4) a. NDS: Moroz ne cˇuvstvovalsja. Frost-nom.m.sg neg be.felt-m.sg ‘The frost was not felt.’ (E.g. we were dressed warmly). b. NES: Moroza ne cˇuvstvovalos’. frost-gen.m.sg neg be.felt-n.sg ‘No frost was felt (there was no frost).’ (5) a. NDS: *(#) Somnenija ne byli.6 doubts-nom.n.pl neg were-n.pl b. NES: Somnenij ne bylo. doubts-gen.n.pl neg were-n.sg ‘There were no doubts.’ (6) a. NDS: Lena ne pela. Lena-nom.f.sg neg sang-f.sg ‘Lena didn’t sing.’ b. NES: *(#) Leny ne pelo. Lena-gen.f.sg neg sang-n.sg Among the many puzzles about GenNeg, we will be most concerned with the characterization of the semantics of the construction. One question in this area is why, under sentential negation, some sentences allow only genitive subject, some only nominative subject, and some both. When both are possible, we want to know what the difference is. It is to be hoped that if we understand the semantics of the construction and the lexical semantics of the various verbs, the answers to those questions will follow. Here is another quotation from Apresjan (1985): The key to understanding the construction we are considering can be found in the position taken long ago in the works of S. O. Karcevskij and R. Jakobson, namely that in negative sentences with the genitive case, ‘‘the subject itself is negated’’ (Jakobson), and in negative sentences with nominative case, ‘‘what is negated is not the subject, but its activity’’ (Karcevskij), or, speaking in the words of R. Jakobson, ‘‘just the action’’. (Apresjan 1985: 295) In a certain sense, it is that same claim which is sharpened and developed in the work of Babby (1980). Babby holds that the scope of sentential negation is different in ‘‘existential sentences’’ (ESs) and ‘‘declarative sentences’’ (DSs), and that that difference is in turn determined by the differing communicative structure of these sentences, the structure of Theme and Rheme. According to Babby, in an ES, the entire sentence falls within the Rheme, and hence when the sentence is

 Vladimir Borschev and Barbara H. Partee

negated, the subject as well as the verb falls within the scope of negation. In a DS, on the other hand, the subject is the Theme and the verb (and its complements, if any) is within the Rheme; therefore, under sentential negation, the subject is outside the scope of negation. Many other accounts have been offered from many points of view and in many different theoretical frameworks. In recent years the stream of literature on this theme has not diminished, and evidently not all ‘‘keys to understanding the construction we are considering’’ have been found yet. Thus for example E. V. Paducˇeva stresses the interaction of verbs of perception with this construction and discusses the role of ‘‘Observer’’ (Paducˇeva 1992, 1997). Many works are dedicated to the syntactic aspects of the given construction, and there has been much discussion of the hypothesized centrality of ‘‘Unaccusativity’’ as a pivotal and explanatory property of the verbs whose ‘‘subjects’’ (being underlying Objects) appear in the genitive under negation.7 Other works focus in addition or instead on hypotheses concerning the position(s) of negation with respect to the verbal complex and its relation to the position of arguments in existential and declarative sentences (Bailyn 1997; Brown 1999; Abels 2000). While we agree with Babby (2000) that the assignment of case to the subject is almost certainly mediated by the syntax (since case assignment is unlikely to depend directly on semantics/pragmatics), in this article we remain uncommitted and agnostic about the syntax. We return to some of the syntactic issues in Section 6 below. Our main goal is to discuss some proposals that have been made for the description of the semantics of this construction and the issue of ‘‘different ways of looking at the same situation’’. We will concentrate particularly the approaches of Babby and Paducˇeva, plus our own approaches in Borschev and Partee (1998a, 1998b), considering some of the problems and apparent contradictions among the various structural schemes that have been proposed, and possible directions of resolution. Special attention will be focussed on Babby’s central use of Theme– Rheme structure, its attractions and its problems. In Borschev and Partee (1998a), we followed Babby in his use of Theme– Rheme structure, making certain modifications in his proposals for the sake of the semantics. But it has long been observed in the literature (for example, in Arutjunova 1976, 1997) that ESs may be used with a wide variety of communicative structures, and many believe that the subject can well be the Theme even when it shows up with GenNeg. The acknowledgement of this possibility8 seems to create a major problem for the account of Babby (1980), and at least requires that the arguments be rethought. Our present position is that Babby did indeed carve out an aspect of structure, some kind of ‘‘marking’’ or distinguishing of some constituent(s) of the internal part-whole structure, which determines the choice of the genitive vs. nominative construction. But we no longer agree that the

The Russian genitive of negation 

relevant structure is to be identified with Theme-Rheme (or Topic–Focus) structure in the classical Praguian sense; we suggest that it is a notion of ‘‘taking a perspective’’, a notion we suggested (in slightly different terminology) in Borschev and Partee (1998b) and will explore further here. In sentences (3a)–(6a), the ‘‘marked’’ element is the first constituent, the ‘‘existing object’’ in the terminology of Arutjunova, or the THING, as we have called it for brevity. In sentences (3b)–(6b), on the other hand, it is the LOC(ation) (‘‘domain of existence’’, in Arutjunova’s terms) that is marked, and the situation is looked at in terms of this LOCation. We call this structure of ‘‘marking’’ perspective structure, and the ‘‘marked’’ constituent the Perspectival Center; we introduce these notions more fully in Section 3. It seems that our notion of perspective structure relates in certain important ways to the various types of ‘‘Observer’’ that Paducˇeva has written about; we discuss this relationship in Section 4.2.4. This paper is organized as follows: in Section 1 we describe Babby’s (1980) analysis, the core of which consists of two theses: (i) that the most crucial conditioning factor for GenNeg is that the (indefinite) subject and (weak) verb both fall under the scope of sentential negation, and (ii) that the scope of sentential negation is determined by Theme–Rheme structure. As a result, the existential sentences in which GenNeg occurs are a species of ‘‘Rheme-only’’ sentences. We introduce some problems that have led us to doubt the correctness of the proposed link between Theme–Rheme structure and GenNeg (which has been questioned by others as well). We also introduce some possible problems for the nearly universally accepted principle that GenNeg always occurs within the scope of sentential negation. Also in Section 1, we discuss some problems raised by Babby (1980) concerning sentences with the verb byt’ ‘be’ with definite subjects marked with GenNeg. In Section 2, we discuss Paducˇeva’s approach to GenNeg and the classifications she and others have made of ‘‘genitive verbs’’, particularly her idea that the semantics of the verb, augmented by contextual factors, is crucial in the conditioning of GenNeg. In Section 3, we review several characterizations of ‘‘existential sentences’’ and ‘‘genitive sentences’’, and outline the kernel of the evolving analysis developed in Borschev and Partee (1998a,b) and in the present work. In our earliest work, we followed Babby (1980) in the use of the Theme–Rheme distinction in the analysis of GenNeg, but with greater attention to a potentially implicit but semantically obligatory LOCation role, which was posited as the Theme in existential sentences. Now we are partially convinced that the needed distinction is not identical to the Theme–Rheme distinction, and we introduce a ‘‘perspective structure’’. The new notion of ‘‘Perspectival Center’’ is argued to be needed in place of Babby’s use of Theme. Section 4 expands on a number of issues related to our proposal that a LOCation is always the Perspectival Center in existential sentences. We discuss the

 Vladimir Borschev and Barbara H. Partee

roles we call THING and LOCation in both existential and locative sentences, and the additional perspectival structure which we posit to distinguish between those sentence types. We present arguments for favoring our use of perspectival structure over Babby’s use of Theme–Rheme structure in the analysis of GenNeg. We also present and discuss a structured set of examples, varying the parameters of word order, definiteness of subject, affirmative and negative, and genitive and nominative, adding some comments on intonation and the occasional difficulty of distinguishing sentential negation from constituent negation on the verb and from verum-focus negation. We close Section 4 with some brief discussion of the relation of our perspective structure to Paducˇeva’s use of Apresjan’s Observer, and of some issues concerning the sources of various presuppositions associated with existential and declarative sentences. Section 5 returns to the problem of the scope of negation, and presents, without a satisfying solution, two separate kinds of problems that face the favored hypothesis concerning the direct correlation of scope of negation with the licensing of GenNeg. Section 6 gives a brief sketch of a number of interesting syntactic proposals that have been made for treating GenNeg. We remain agnostic about the syntax but appreciative of its undoubted importance in mediating between the semantics and pragmatics of the construction and the morphological choice of cases. The paper ends with some remarks emphasizing the challenge of finding an integrated view of both the elusive core and the interactions of pragmatics, semantics, lexicon, syntax, intonation and morphology in reaching an eventual understanding of the fascinating subtleties of this important construction.

.

Babby on ‘‘Declarative’’ and ‘‘Existential’’ sentences

. The scope of negation. Theme and Rheme. Problems We noted above that Babby (1980) distinguished ‘‘declarative’’ from ‘‘existential’’ sentences.9 For Babby’s distinction between NDS and NES, the negative sentences, occurrence of nominative vs. genitive is criterial. For the corresponding affirmative sentences, ADS and AES, there is no absolute surface criterion; the usual basis for distinguishing between ADS and AES is to ask which case would be used on the subject if the sentence were negated. It is on this criterion that one can determine that there is normally an obligatory difference in word order in the affirmative sentences, as noted in the introduction. But since word order can be varied with accompanying changes in intonation in certain contexts, word order is not itself criterial in the affirmative case. We note that for almost every NES, one can construct a corresponding ‘‘declarative’’ sentence, although, as shown by example (5a) above, the result will

The Russian genitive of negation

not necessarily be acceptable. The converse is by far less valid; it is very easy to find NDSs with no NES counterpart. Babby’s first main proposal about the distinction is shown in his chart (7) (Babby 1980: 72) below: DSs and ESs differ in their ‘‘scope of assertion/negation’’. (7)

AFFIRMATIVE [Scope of A VP NP] ⇒NEG DECLARATIVE NP [Scope of A VP] ⇒NEG

EXISTENTIAL

NEGATED [ne VP NPgen] NPnom [ne VP]

Thus the declarative sentence (3a) presupposes that there was some runoff of thawed snow and asserts that it was it was not observed, i.e. negates only that it was observed. The corresponding ES (3b) is used to negate the very existence of any runoff of thawed snow.10 The ES also negates ‘‘was observed’’, i.e. it negates the whole sentence; but in this case nabljudalsja ‘was.observed’ functions as a ‘‘weak verb’’ (often described as ‘‘semantically empty’’). The notion of ‘‘weak’’ or ‘‘empty’’ verbs was at the center of the work reported in Borschev and Partee (1998a); it is briefly reviewed in Sections 2 and 3. According to chart (7), a declarative sentence consists of two parts. The subject NP, which remains outside the scope of assertion/negation, and the predicate VP, which is in the scope of assertion/negation. An ES, in contrast, has just one ‘‘part’’ in this sense; the entire sentence is inside the scope of assertion/negation, and such a sentence typically asserts/denies the existence of what the subject NP refers to. Babby cites Kuroda’s (1972) discussion of Brentano and Marty’s distinction between categorical and thetic judgments, and notes that Kuroda lists existential sentences and impersonal sentences as examples of thetic judgments. Categorical judgments, corresponding to Babby’s ‘‘declarative sentences’’, have two parts, one part involved with recognizing the subject, the other part predicating something about that subject; thetic judgments are ‘‘simple’’, have just one part. Babby suggests that the Russian nominative/genitive alternation provides even stronger linguistic evidence for the linguistic reality of the categorical/thetic distinction than the Japanese wa/ga alternation discussed by Kuroda. Babby’s second main proposal is that the scope of assertion/negation can be equated with the Rheme of the sentence according to the division of the sentence into Theme and Rheme (or Topic and Focus). He notes that various authors define the concepts of Theme and Rheme differently, but the majority agree that Rheme is that which is asserted in an affirmative assertive sentence and negated in sentences that have normal sentential negation.11 He states the following principle: ‘‘When an assertive sentence is negated, its scope of negation is determined by its underlying rheme; the theme accordingly falls outside the scope of negation.’’ (Babby 1980: 103) He then expands on this: ‘‘In other words, what I am proposing



 Vladimir Borschev and Barbara H. Partee

here is that the rheme of a negated assertive sentence and the scope of negation in an assertive sentence are different terms for the same thing’’ (Babby 1980: 103). On Babby’s view, an ES, AES or NES, is a ‘‘rheme-only’’ sentence.12 In Prague School literature and Russian linguistics these are known as ‘‘unpartitioned’’ utterances (nerascˇlenennye vyskazyvanija in Russian). Babby’s final formulation of his rule of genitive marking in NESs (his (160)) is given in (8) below. (8)

neg [Rheme V NP] ⇒ [ne V NPGEN] Conditions: a. NP is indefinite b. V is semantically empty

Condition a. requires that the subject of an ES be indefinite, not refer to a definite particular; Borschev and Partee (1998a) argued against this condition, and some of the conditions surrounding the use of definite NP’s in an ES were further discussed in Partee (2000); see Section 6.5. Condition b., that the verb be ‘‘semantically empty’’, was discussed in some detail in Borschev and Partee (1998a), where we offered an account in which the verb retains its literal meaning but the construction effectively reduces its assertional component to little more than ‘‘existence’’ (in a given LOCation.) In our earlier work, we agreed with Babby about the role of communicative structure in the interpretation of existential and declarative sentences. Our principal amendment was our endorsement of the principle that existence is always relative to a ‘‘LOCation’’, which may be implicit.13 We accept Jackendoff ’s (1972, 1990) metaphorical-structural extensions of ‘‘being in a location’’ to include ‘‘being in some state’’, ‘‘occurring in some spatiotemporal region’’, ‘‘being in someone’s possession’’, extending also to ‘‘being in the speaker’s (or an observer’s) perceptual field’’ (Paducˇeva 1992, 1997). Then whereas Babby analyzed ESs as ‘‘Rheme-only’’, with a possible optional Thematic Location, we argued that the LOCation, either given or contextually presupposed, is a semantically obligatory part of the construction and is the Theme. The assertion (Rheme) is that the/a ‘‘THING’’ described by the subject NP exists in that LOCation. The idea that all ‘‘Rheme-only’’ sentences have an implicit topic, understood as something like a situation or a spatiotemporal LOCation, occurs in various forms in the literature. Erteschik-Shir (1997), who introduces the notion of ‘‘stage topic’’, corresponding to an explicit or implicit spatiotemporal argument (à la Kratzer 1995) for such sentences, notes that her distinction between sentences with stage topics and sentences with individual topics is equivalent to Kuroda’s distinction between thetic and categorical judgments, and to the distinction made by Guéron (1980) between presentation sentences and predicational sentences. She also notes that the idea that scene-setting expressions that specify temporal or

The Russian genitive of negation 

spatial background for the sentence also function as topics may be found in Gundel (1974) and Reinhart (1981). Sgall et al. (1986: 202) claim that initial locatives are always topics. Others who claim that an initial locative in an otherwise unpartitioned sentence always functions as a topic are Dahl (1969: 38), Arutjunova (1976: 210–11), Gundel (1974: 34), all cited by Babby (1980: 96), who agrees; he notes that others have claimed that it is part of a complex Rheme (Krylova and Xavronina (1976: 26, [note 2]); Crockett (1976: 241)). So there is very wide agreement that an initial Locative in an existential sentence is generally a topic, and considerable support for the idea that when such a locative topic is not overt in an existential sentence, something like it is implicitly understood. Now, in the light of the observations of our colleagues mentioned in the Introduction and after reviewing Arutjunova (1976, 1997) and other works, we have doubts about the full correlation of the distribution of NESs and NDSs with the postulated difference in Theme–Rheme structure. Thus in examples (9–12) below, it appears to us that our colleagues are correct in claiming that the words sobaki ‘dog-gen.f.sg’, myšej ‘mouse-gen.f.pl’, kefira ‘kefir-gen.m.sg’, and otveta ‘answergen.m.sg’ are the Themes (or part of the Theme) of these sentences. Both their most natural intonation pattern and their (most likely) interpretation in the given contexts support this point of view, which argues against the generalization in (8). (9) Sobaki u menja net. (Arutjunova 1976) dog-gen.f.sg at I-gen not.is I don’t have a dog. [Context: talking about dogs, perhaps about whether I have one.] (10) [Myši v dome est’?] — Net, myšej v dome [mouse-nom.f.pl in house is?] No, mouse-gen.f.pl in house net. (Arutjunova 1997) not.is [Are there mice in the house?] — No, there are no mice in the house. (11) [Ja iskal kefir.] Kefira v magazine ne bylo. [I looked-for kefir-acc.m.sg Kefir-gen.m.sg in store neg was-n.sg ‘[I was looking for kefir.] There wasn’t any kefir in the store.’ (12) [Ja napisal emu i ždal otveta.] Otveta ne [I wrote him and waited.for answer-gen.m.sg] answer-gen.m.sg neg prišlo. came-n.sg [I wrote to him and waited for an answer.] No answer came. Our conception of the communicative structure of sentences is based to a considerable extent on works in the tradition of the Prague School such as

 Vladimir Borschev and Barbara H. Partee

Hajicˇová (1973, 1974, 1984); Sgall, Hajicˇová and Panevová (1986) and works of Russian linguists. For us the Theme is, roughly speaking, what is being talked about in a sentence, which is presupposed to be familiar to the hearer, referring back to something which was either spoken about earlier or else simply well known. The Rheme is new information which the speaker wishes to communicate. We note that Babby, alongside the Theme–Rheme opposition, considers an additional opposition, between old (given) and new information. Judging by his discussion of similar examples, he would consider sobaki ‘dog-gen.f.sg’, myšej ‘mouse-gen.f.pl’, kefira ‘kefir-gen.m.sg’, and otveta ‘answer-gen.m.sg’ in (9)– (12) above to be old information but part of the Rheme. Although we consider this issue far from settled,14 we are now inclined to treat them as (part of) the Theme of the corresponding sentences. Even the relation of GenNeg to the scope of negation is not as clear as it has generally been believed to be. Consider examples (13)–(14). (13) [My nadejalis’, cˇto na seminare budut studenty.] No ni odin [we hoped, that at seminar will.be students] but NI one-nom.m.sg student tam ne byl. student-nom.m.sg there neg was-m.sg [‘We hoped that (some of the) students would be at the seminar. But not a single one of the students was there.’ (14) [My nadejalis’, cˇto na seminare budut studenty.] No ni odnogo [we hoped, that at seminar will.be students] but NI one-gen.m.sg studenta tam ne bylo. student-gen.m.sg there neg was-n.sg [‘We hoped that there would be students at the seminar.] But there was not a single student [or: not a single one of the students] there.’ The sentences (13) and (14) have the same truth conditions if looked at in simple extensional models; in both cases, the expression translated as ‘‘not one student’’ is necessarily under the scope of the sentential negation in its clause.15 Yet only in (14) does that expression occur in the genitive; (13), with the nominative, is equally well-formed, which seems unexpected not only under Babby’s account but under virtually all accounts of GenNeg.16 There is one important difference in the interpretation of (13) and (14). In the sentence with the nominative subject, we must be speaking of a contextually definite group of students. The background sentence, which has the same surface form in both (13) and (14), contains the bare noun studenty ‘students’, which is indeterminate with respect to definiteness or specificity. Given its sentence-final position, studenty cannot easily be definite ‘the students’, but it can have a partitive interpretation ‘some of the students’, and must have the partitive

The Russian genitive of negation 

interpretation for (13) to be felicitous. The nominative ni odin student is also understood partitively: ‘not a single one of the students’. In (14), on the other hand, there is no such restriction. A partitive interpretation is possible but a completely non-specific interpretation is normal. We return to this problem and other issues concerning the scope of negation in Section 5. . Sentences with the verb byt’ ‘be’ and referential subject Babby (1980: 124) (like Arutjunova 1976: 225) does not include sentences like (15) among the class of NESs, even though in all syntactic and morphological respects, including the manifestation of GenNeg, (15) looks like an ordinary NES. (15) Ivana ne bylo na lekcii. Ivan-gen.m.sg neg was-n.sg at lecture ‘Ivan wasn’t at the lecture.’ The problem is the semantics: sentences like (15) whose subject is a proper name or other clearly referential definite NP do not deny the existence of the referent of the subject. And Babby has argued that the primary function of an existential sentence is the assertion (in an AES) or denial (in an NES) of existence of the/a referent of the subject NP. So NESs should normally not permit proper names or other definite NPs as subjects, as Babby indicated in his indefiniteness condition a. in his rule (8) above. But sentences with byt’ (and a few other typically existential verbs) do commonly allow GenNeg with proper names and other definite NPs, although other verbs rarely do. So Babby claims that sentence (15) cannot be an existential sentence because of its definite subject and therefore must be a ‘‘locative sentence’’, a type of NDS, with ‘‘be at the lecture’’ as the negated Rheme. This, however, goes contrary to the generalization about the distribution of GenNeg in Babby’s basic scheme in (7), as well as to the diagnostics developed by Chvany (1975) for distinguishing existential and locative sentences. In Borschev and Partee (1998a), we argued that this was a weakness in Babby’s treatment. The distribution of GenNeg is otherwise accounted for in Babby (1980) by a single rule, and it would seem strange for byt’ to be an exception when byt’ is in a sense a ‘‘basic’’ verb of existence (‘‘being’’) and roughly speaking all NESs can be approximately paraphrased by NESs with byt’, as Babby himself notes.17 (16) a.

Otveta ne prišlo = Otveta ne answer-gen.m.sg neg arrived-n.sg = answer-gen.m.sg neg bylo.

 Vladimir Borschev and Barbara H. Partee

was-n.sg ‘No answer came.’ = ‘There was no answer.’ b. Moroza ne cˇuvstvovalos’ = Moroza ne bylo. frost-gen.m.sg neg be.felt-n.sg = Frost-gen.m.sg neg was-n.sg ‘No frost was felt.’ = ‘There was no frost.’ c. Posudy na stole ne stojalo . = dishes-gen.f.sg on table neg stood-n.sg = Posudy na stole ne bylo. dishes-gen.f.sg on table neg were-n.sg ‘No dishes stood on the table.’ = ‘There were no dishes on the table.’ d. Snega na poljax uže ne ležalo. = snow-gen.m.sg on fields already neg lay-n.sg = Snega na poljax uže ne bylo. snow-gen.m.sg on fields already neg was-n.sg ‘Snow no longer lay on the fields’ = ‘There was no longer snow on the fields.’ In sentence (15), the subject is definite and denotes a concrete entity presupposed to exist, and indeed such sentences differ in some way from the ‘‘classic’’ existential (bytijnye) sentences. But in its form, according to the principles that govern the choice between genitive and nominative under negation, it does not differ at all from the various paraphrases with byt’ ‘be’ in (16). Babby suggests the following explanation: when a locative sentence contains the verb byt’, it is so close syntactically to the form of an existential sentence that Russian grammar ‘‘simply treats them the same with respect to genitive marking’’ (Babby 1980: 124). We argued in Borschev and Partee (1998a), however, that with more careful attention to the interpretation of ‘‘existence’’ in ‘‘existential sentences’’, in particular by claiming that ‘‘existence is always relative to a LOCation’’ (see principle (21) in Section 3), sentence (15) can indeed be interpreted as an ES (an NES). We showed that Babby’s analysis could then work uniformly for GenNeg in sentences with byt’ just as in ESs with lexical verbs. While we now have a different view of the distinction which Babby identified as the Theme–Rheme distinction, we continue to believe that sentence (15) should be included among the class of ESs to be accounted for by a uniform treatment of GenNeg. We return to our analysis in Section 3.

.

Paducˇeva on genitive verbs

The Russian genitive of negation 

Babby’s rule for GenNeg in (8) above specifies that the verb in an NES should be ‘‘semantically empty.’’ Paducˇeva (1997) takes a different approach, and tries to describe in more positive terms the class of verbs that can occur with the genitive construction.18 The class of verbs with which one finds nominative/genitive alternation in negative sentences is not a closed class; but on the other hand, it would be wrong to simply say that any verb can occur with either a nominative or a genitive subject in a negative sentence. The possibility of using a normally nominative-only verb with a genitive subject might be compared with the possibility of using a normally count-noun-only in an environment that demands a mass noun (presumably shifting the noun from a count-noun sense to a related mass-noun sense). Babby (1980) reports that the number of verbs that regularly allow genitive marking on the subject is on the order of 50–100; but he emphasizes the openness of the class and the impossibility of predicting the possibility of the use of genitive solely from the semantics of the verb. Paducˇeva (1997) includes a list of over 300 ‘‘genitive verbs’’; she reports about 60% as attested in texts, the others checked with native speakers’ intuitions (including some about which there is reported disagreement). Paducˇeva also notes that because of the possibility of coercion, no list can be complete. Paducˇeva (1997) refers to the full range of verbs that can occur in such ‘genitive sentences’ as ‘genitive verbs’, and we will sometimes do the same, with the understanding that these are not all lexically specified as such, and not a closed class; sometimes what one encounters is a rather unusual use of a certain verb, aided by a particular context. Various subclasses of typically ‘genitive’ verbs19 are often distinguished in addition to the ‘‘core’’ existential verbs discussed by Arutjunova. One group are verbs which add some aspectual modification to the core ‘‘existential’’ notion, especially inchoatives: voznikat’ ‘arise, come into existence’, pojavljat’sja ‘appear’, ostat’sja ‘remain’. Another group are modal: trebovat’sja ‘be required (be required to exist)’, nužno ‘necessary’. A commonly cited group are (passive forms20 of) perception verbs: nabljudat’sja ‘be observed’, slyšat’sja ‘be heard’. With eventdenoting subjects, a common class is verbs of ‘happening’: proizojti ‘occur, take place’, slucˇit’sja ‘happen’. Paducˇeva (1997) aims to give a unified semantics for the genitive construction, and she sees the semantics of the main verb as central. She considers that: in a negative sentence with the genitive construction, one of two [semantic] components is present: I. ‘X does not exist in the World/Location’; II. ‘X is not (there is no X) in the perceptual space of the Subject of consciousness’ (Paducˇeva 1997: 103)

She sets the task of seeking a semantic invariant or close semantic relatedness

 Vladimir Borschev and Barbara H. Partee

among the verbs which can occur in the genitive construction, which she calls genitive verbs. She argues that: [their] definitions21 (or lexical semantic analyses) include an existential or a perceptual component, and moreover with a communicative status such that that component of the verb’s meaning is negated when the sentence is negated: that component may be an assertion or an implication, but it cannot be a presupposition. The affirmative analogs of components I and II have the form: I¹. ‘X exists in the World/Location’; II¹. ‘X is located in the perceptual space of the Subject of consciousness.’

In order to schematize this approach, some elaboration is necessary: what is a definition or lexical representation (what type of object is it, from a more or less formal point of view?), what does it mean that components I and II ‘‘are present in the sentence’’? Evidently, for Paducˇeva , as for us (cf. Borschev and Partee 1998a, 1998b), the semantic content of a sentence is equivalent (to a first approximation) to a collection of propositions which are true in every situation in which the given sentence is true. A definition, or lexical semantic analysis, is also a collection of propositions,22 which Paducˇeva further subdivides into presuppositions, assertions, and implications.23 These propositions are what constitute the ‘‘components’’ of a definition. In some sense Paducˇeva and Babby describe the genitive construction from different sides: Babby is primarily concerned with the semantics of the construction itself, and Paducˇeva primarily with the semantics of ‘‘genitive verbs’’. Her thesis may be formulated as follows: If the speaker wishes to express ‘‘components’’ I or II, then he or she uses in the negative sentence a genitive verb and is (therefore?) obliged to use the genitive construction. However, the relevant ‘‘semantic component’’ of the verb does not actually have to be lexical; it may arise from the context or be an optional addition by the speaker him/herself. There are several further respects in which Paducˇeva ‘s approach contrasts with Babby’s and with our variant of Babby’s approach, including her proposals about a separate subclass of perceptual verbs and her distinctions among various senses of the verb byt’ ‘be’. The division of genitive verbs into existential and perceptual is one of the main theses of Paducˇeva (1992, 1997). The semantics of perceptual verbs, and the splitting off of a ‘‘perceptual’’ meaning, merit particular attention. Thus Paducˇeva (1992) states that alongside the existential meaning of the verb byt’, there exist in addition two distinct locative meanings: byt’1, with her perceptual component of meaning II¹, as in example (17a), and byt’2 , which lacks such a component, as in example (17b).24 (17) a.

Geologicˇesko

partii

ne bylo1

na baze.

The Russian genitive of negation 

geological-gen.f.sg party-gen.f.sg neg was1-n.sg at base ‘The geological party25 was not at the base.’ (Paducˇeva 1992: 55 (3¹b)) b. Geologicˇeskaja partija ne byla2 na baze. geological-nom.f.sg party-nom.f.sg neg was2-f.sg at base ‘The geological party was not at the base.’ Paducˇeva ‘s perceptual component II’ is connected with the notion of an observer 26 (nabljudatel’) (a perceiving subject, a subject of consciousness). In particular, for example, consider the lexical semantic analysis of the verb byt’1 (an optionally genitive verb although it is a ‘‘locative verb’’ rather than an ‘‘existential verb’’). Its basic argument structure is as in (18), but in addition there may optionally occur the semantic component ‘‘Subject of consciousness imagines himself located at place Y at moment t0’’ (that is, together with the individual X), which then licenses GenNeg as in (17a). (18) X byl1 v meste Y [v momente t0]. X was1 in place Y at moment t0 The meaning of byt’2, with which only a nominative subject is possible, is also connected to an ‘‘observer’’, but in this case the observer’s frame of reference is the utterance situation (in the simplest case), yielding what Paducˇeva calls a ‘‘retrospective observer’’, in contrast to the ‘‘synchronous observer’’ with byt’1. We find the notions of these (often implicit) observers interesting and substantive,27 and the description of the semantics of perception verbs is clearly an important part of the study of genitive sentences. In our own work we have nevertheless concentrated on what is common to perception verbs and other genitive verbs, treating ‘‘being perceived’’ as a species of ‘‘existing’’ in the (metaphorical) ‘‘location’’ of the perceptual space of the observer. One particularly interesting possibility that is raised by some parts of Paducˇeva ‘s discussion is the possibility that some (perhaps many) verbs have noticeably different senses when used with genitive and when used with nominative in negative sentences. For verbs such as nabljudat’sja ‘be observed’ and cˇuvstvovat’sja ‘be felt’, for instance (cf. examples (3) and (4)), as well as for the verb byt’ ‘be’, the verb is felt to be ‘‘more stative’’ when used with genitive and ‘‘more agentive’’ when used with nominative.28 We would hope that such differences in the interpretations of the verbs could be described in terms of regular semantic shifts influenced by the semantics of the constructions in which they occur, rather than as simple stipulated polysemy. We will return to the issue of the possible interactions between lexical sense-

 Vladimir Borschev and Barbara H. Partee

shifting and perspective-shifting in Section 4.2.4, where we examine the relation between Paducˇeva’s and others’ investigations of the role of the Observer and our notion of perspective structure.

.

Existential sentences and the ‘‘genitive rule’’

We have mentioned that various authors use the terms ‘‘existential sentences’’ (bytijnye predloženija) and ‘‘genitive sentences’’. The expression ‘‘genitive sentences’’ in this context means, by definition, negative sentences with an NP subject in the genitive which would be in the nominative in a corresponding affirmative sentence, or sometimes the affirmative sentences of which those are the negation. The expression ‘‘existential sentence’’ is used by some authors, including Babby29 (1980) and Borschev and Partee (1998a,b) as fully equivalent to ‘‘genitive sentence’’, representing a hypothesis about the core semantics lying behind the prototypical cases of the construction and extended in some systematic way to the more marginal or novel cases. Arutjunova (1997:57) characterizes bytijnye predloženija ‘existential sentences’ as ‘‘a syntactic type of sentence asserting the existence in the world or some fragment of it of objects of some class or other.’’ As bytijnye glagoly ‘existential verbs’ she cites the prototypical existential verb byt’ ‘be’ plus ‘‘its equivalents and analogs’’ sušcˇestvovat’ ‘exist’, imet’sja ‘be available’, byvat’ ‘be (frequentative)’, vodit’sja ‘be (found)’, vstrecˇat’sja ‘be encountered’, popadat’sja ‘[‘passive’ of ‘impersonal’ of ] come across’. (19) V lesu popadajutsja belye in woods come.across.pres.pl.refl white-nom.m.pl griby. mushroom-nom.m.pl In the woods one comes across white mushrooms. (Arutjunova 1997:57) Arutjunova cites the GenNeg phenomenon as a property that existential sentences all have, but she does not suggest the converse and thus does not seem to equate the class of existential sentences with the class of sentences in which GenNeg occurs. She speaks of prototypical existential verbs and of verbs which are semantically close to them or ‘‘analogs’’ of them. She may have in mind a distinction between a narrower and more literal notion of existential sentences and a broader notion that covers the entire range of genitive sentences. For Paducˇeva (1997), it is clear that ‘‘existential verbs’’ are just a proper subclass of ‘‘genitive verbs’’, which she divides into those which contain her ‘‘existential’’ semantic component I, and those which contain her ‘‘perceptual’’

The Russian genitive of negation 

semantic component II. But since we do not believe that the presence of a ‘‘genitive verb’’ is sufficient to license a genitive sentence, we prefer to follow the lead of Babby and Arutjunova in seeing the literally ‘‘existential’’ sentences as a model case which provides the basis for a construction whose use is extended to sentences which may be only in some metaphorical sense asserting or denying the ‘‘existence’’ of the referent of the subject. All of the mentioned authors, including the present writers, allow extension of the notion of ‘‘existential sentence’’ to those in which there are certain modal or aspectual operators added to the core verbal notion. So our principal question is what the conditions on the ‘‘genitive rule’’ are.30 In principle, of course, we should not have to stipulate a ‘‘genitive rule’’ covering the entire GenNeg construction; the assignment of genitive case should be a consequence of the interplay of a number of principles, and whatever is ‘‘special’’ about the Russian GenNeg construction should be identified and ‘‘localized’’ to an appropriate part of Russian grammar. One example of ‘‘localization’’ of this sort can be found in the hypothesis (Pesetsky 1982; Pereltsvaig 1997) that there is a phonologically empty quantificational head in Russian which is a negative polarity item with various other constraints on its occurrence, which itself occurs in the nominative or accusative, and which, like many other Russian quantifiers, governs the genitive case in its complement. Under such a hypothesis, the problem becomes one of explaining the distribution of that null quantifier, so as not to predict that genitive NPs can occur in every kind of sentence in which a nominative or accusative NP might occur under the scope of negation. We mention this and other syntactic hypotheses in Section 6. But in the meantime, we concentrate on the semantic factors, including communicative structure, that seem to be crucial in GenNeg, remaining agnostic about how these are reflected in the syntax. We will consider the full range of ‘‘genitive verbs’’, including those that can behave as genitive verbs only with the help of strong contextual factors. But we will take as the central paradigm case the ‘‘relation of existing’’ in a situation of ‘existing’, with the verb byt’. Among the central notions needed for understanding existential sentences, Arutjunova distinguishes three components in a ‘‘classical’’ existential sentence: a ‘‘Localizer’’ (‘‘Region of existence’’), a name of an ‘‘Existing object’’, and an ‘‘Existential Verb’’: (20) V ètom kraju (Localizer) est’ (Existential Verb) in that region is/are lesa (name of ‘‘Existing Object’’). forests-nom.m.pl ‘There are forests in that region.’ We have used different terms for the same notions. Our terms are probably not

 Vladimir Borschev and Barbara H. Partee

optimal, but they are shorter (and closer to Paducˇeva’s): the ‘‘Localizer’’ is the name of our LOCation, and the ‘‘Existing Object’’ is our THING.31 As for the verb, we usually schematize and simply call it BE. The core principles behind our analysis are as follows.32 (21) ‘‘EXISTENCE IS RELATIVE’’ PRINCIPLE: Existence (in the sense relevant to AESs and NESs) is always relative to a LOC(ation). We discuss the principles that determine which LOCation is relevant in a given case in Borschev and Partee (1998a). The distinctions among locations associated with Perspectival Center, Perspectival non-Center, and background existence make it possible to make sense of GenNeg examples which deny the existence of the THING in a certain LOCation, possibly a perceiver’s perceptual field, while presupposing existence of that THING in ‘‘the actual world’’ or some other LOCation invoked in the interpretation of the sentence in the given context. The distinctions discussed in Borschev and Partee (1998a) were based on Theme–Rheme structure; we would now systematically modify them, replacing ‘Theme’ by ‘Perspectival Center’. (22) The Common Structure of ‘‘Existence/location situations’’ and their descriptions: BE (THING, LOC) We share the view of many writers that in many cases, including the core cases with the verb byt’, existential sentences and ‘‘declarative’’ sentences (including ‘‘locative’’ sentences), have the same verb and the same participants but a different organization. Abstracting away from the difference, both describe situations involving some THING ‘being’ in some LOCation. An ‘‘existence/location situation’’ may be structured either from the perspective of the THING or from the perspective of the LOCation. Let us use the term Perspectival Center for the participant chosen as the point of departure for structuring the situation. (Our Perspectival Center will play the role that ‘‘Theme’’ played for Babby (1980).) We have no special term for the complementary term, the part of the situation that is not the Perspectival Center; we will sometimes just call it a predicate. We expand on the notion of Perspective Structure in Section 4. (23) PERSPECTIVE STRUCTURE: In the following, we underline the Perspectival Center. BE (THING, LOC): structure of the interpretation of a Locative (‘‘Declarative’’) sentence. BE (THING, LOC): structure of the interpretation of an Existential sentence.

The Russian genitive of negation 

There are many ways to imagine different syntactic structures corresponding to these different structures of interpretation: unergative vs. unaccusative, personal vs. impersonal, NP as external argument vs. implicit event argument as external argument. The differences might be located only in the semantics, but that would be a problem for compositionality if we posited neither a lexical difference in the verbs nor a structural difference in the sentences. So while we remain agnostic about the syntax, we believe that it is likely that what we are discussing is the semantics that corresponds to some syntactic distinction. Perspective structure has much in common with Theme–Rheme structure. Just as we said that the Theme is what the speaker is talking about and the Rheme contains the new information the speaker wishes to convey about the Theme (what the speaker predicates of the Theme), we can also say that Perspective Structure involves picking out one participant, here the THING or the LOC, and describing it. We say more about the differences between the two notions in Section 4.2.2. (24) PERSPECTIVAL CENTER PRESUPPOSITION: Any Perspectival Center must be normally be presupposed to exist. Principle (24) allows us to derive the same presuppositions that were derived in Borschev and Partee (1998a) from the correlation of greater presuppositionality with the Theme of the sentence (Hajicˇová 1973, 1974, 1984; Peregrin 1995; Sgall et al. 1986). In particular, from this principle it will follow that the nominative subjects in NDSs are normally presupposed to exist, whereas in NESs, only the LOCation is normally presupposed to exist, and the perspectival structure does not provide any existence presupposition for the THING. Of course there may be presuppositions derived from other sources. But we will argue that ESs have a ‘‘marked’’ structure in comparison to the ‘‘unmarked’’ structure of Locative (Declarative) sentences, and that choosing a marked structure that avoids a presupposition of existence invites an implicature of likely non-existence. Another way of stating the core of our analysis is to say that when the LOC is chosen as the Perspectival Center, then we look at the situation in terms of the LOC and ‘‘what’s in it’’. In the case of an NES, the main assertion is that in that LOCation x, ‘‘NEG (BE (THING, in x)). In an NDS, on the other hand, our Perspectival Center is the THING, and of that THING, we state something about its location, namely that it is not in the given LOC. (25) NES PRINCIPLE: An NES denies the existence of the thing(s) described by the subject NP in the Perspectival center LOCation. This statement may be too strong in the cases where the verb of the sentence is

 Vladimir Borschev and Barbara H. Partee

not byt’ ‘be’. In any case, until we are able to derive this or some related principle compositionally, we are not in a position to be sure exactly how strong it should be. In the meantime, this is our analog of Paducˇeva ‘s statement that a genitive sentence must contain Semantic Component I. In Borschev and Partee (1998a), we related principle (25) to the following principle, where ‘‘V’’ represents any lexical verb.33 (26) PRESUPPOSED EQUIVALENCE: An NES presupposes that the following equivalence holds locally in the given context of utterance: V(THING, LOC) ⇔ BE(THING, LOC) In the general case, we assume that verbs have their normal literal meaning, which in most cases is not simply ‘‘exist’’ or ‘‘be’’. Then in order for the presupposition to be satisfied in a given context of use, some further assumptions must be available in the context which together with the compositional semantics of the sentence will entail the equivalence. If the GenNeg construction is used, the presupposition must be satisfied for the use of the sentence to be felicitous, and the hearer uses whatever contextual information is available to support an accommodation of the presupposition. (Examples involving the interaction of additional ‘‘axioms’’ deriving from lexical semantics, encyclopedic knowledge, and local contextual information are given in Borschev and Partee 1998a.) But for it to be possible to satisfy the Presupposed Equivalence, the semantics of the verb must not be directly incompatible with it. If the meaning of the main verb of a sentence is inconsistent with the presupposed equivalence, then the use of GenNeg is impossible; this is standardly the case with Agentive verbs, for instance. But it may be possible in some cases for the verb to be coerced to a weaker meaning compatible with the presupposed equivalence in the given context, in which case GenNeg may be possible.34 And what about cases in which the use of GenNeg is obligatory and nominative is impossible, as in examples (5a–b)? It is clearly not right to say that if the Presupposed Equivalence follows directly from the lexical semantics of the verb, as it evidently does for the verb byt’ ‘be’ itself (at least in its existential sense, if it has several), then GenNeg is obligatory and the Nominative alternative is impossible. We have already seen many pairs of NES/NDS examples with the verb byt’ ‘be’ alternating between GenNeg and the Nominative construction (and see the systematic variations in examples (34)–(39) in Section 4.2.3.) What makes GenNeg obligatory and an NDS impossible in example (5) is not only that the verb is byt’ ‘be’, but a combination of factors. To interpret (5a), the (impossible) NDS, it is necessary to take somnenija ‘doubts’ as the Perspectival Center of the sentence, and to try to interpret the predicate ne byli ‘were not’ as predicating some property of those doubts. A good example of an NDS with somnenija ‘doubts’ as its subject is given by Paducˇeva (1992: 53):

The Russian genitive of negation 

(27) Somnenija ne iscˇezli. doubts-nom.n.pl neg disappeared-pl ‘The doubts did not disappear.’ (And in fact GenNeg is impossible with the verb iscˇezat’ ‘disappear’, since the lexical semantics of that verb is unsalvageably incompatible with the Presupposed Equivalence.) But what about the predicate in (5a)? While it is normal to leave a LOCative complement of byt’ implicit when it is the Theme, it is virtually impossible for the verb byt’ by itself to be used as an assertive main predicate.35 (In this it differs from sušcˇestvovat’ ‘exist’, which is regularly used intransitively to assert existence, with ‘‘in the world’’ as its default ‘‘LOCation’’.) And even with sušcˇestvovat’ ‘exist’, an asserted denial of existence using an NDS is only possible when the subject has some independent status in some possible world other than the one in which its existence is being denied. For example, the subject may be presupposed to exist in some mythical, fictional, theological, subjective or other world and its existence is denied in the actual (or other contextually relevant) world, or vice versa. See examples (28a–b) from Paducˇeva (1997: 111). (With byt’ ‘be’, it seems to be impossible to consider two alternative worlds or LOCations if the Perspectival Center is the THING and there is no overt locative complement expressed.) (28) a.

Dlja nego ne sušcˇestvoval obšcˇezavodskoj rasporjadok for him neg existed-m.sg factory-wide schedule-nom.m.sg dnja. (Paducˇeva (1997: 111). From Panova, ‘‘Kružilixa’’) day-gen.m.sg ‘For him, the factory schedule did not exist.’ b. Ètot real’nyj mir sejcˇas ne sušcˇestvoval this real-nom.m.sg world-nom.m.sg now neg existed-m.sg dlja nee. (From Gladkov, ‘‘Povest’ o detstve’’) for her ‘Now this real world did not exist for her.’

Thus there are many factors conspiring against (5a); it is almost impossible to use bare intransitive byt’ ‘be’ to mean ‘exist’, and even if that were possible, the subject somnenija ‘doubts’ is almost impossible to interpret as an individual having something like an independent existence in some world other than the one in which the nonexistence is asserted. One factor or the other might be overcome in a strong enough context, but not both. We illustrate the workings of these various principles below, with examples. Some of the foundational questions about where such principles come from remain open for further research.

. Existential sentences: LOC as Perspectival Center

 Vladimir Borschev and Barbara H. Partee

There seems clearly to be a distinction, discussed by many authors in many frameworks, involving a contrast in two kinds of sentences each having the parts we have called ‘‘BE (THING, LOC)’’. One kind of sentence is ‘‘ordinary’’, and has the ‘‘THING’’ as ordinary subject. This kind of sentence doesn’t really have a name except when put in contrast with the other kind; this is Babby’s ‘‘Declarative Sentence’’, more often called ‘‘Locational’’, and often subsumed within the larger class of ‘‘Predicational’’ sentences. This also seems to be an instance of the Brentano/Marty ‘‘categorical judgment’’. The ESs do not have that ordinary structure, but exactly what structure they do have is controversial. In some sense they seem to be turning the predication around: saying of the LOCation that there is THING in it, or that it has THING in it. If the LOC is implicit, this is a ‘‘thetic judgment’’. And insofar as various authors (see Erteschik-Shir (1997) for discussion) have argued that thetic judgments should always be understood as predicating something of a situation or of a spatiotemporal location, then perhaps statements with an explicit LOC also express thetic judgments. But in what way and at what ‘‘level’’ or ‘‘levels’’ of structure is the predication ‘‘turned around’’? Babby (1980) proposed that the difference is a difference at the level of Theme–Rheme (or Topic–Focus) structure. A number of linguists including Babby (2000) have proposed differences in syntactic structure. We propose a difference in Perspectival Structure. But that is no more than a change in terminology until we can say more about what Perspectival Structure is and by what principles it relates to other better-recognized aspects of pragmatic, semantic, and syntactic structure. We make some suggestive remarks below, but many questions remain open. The main thing we claim, and this too needs deeper exploration, is that the structural divisions Babby drew were more or less correct, but the distinction he identified cannot be identical to the widely recognized distinction between Theme and Rheme. So the weakest form of our present view is simply that Babby’s structures need new labels, and we are suggesting that the new labels should relate to our intuitions about looking at a situation from alternative perspectives. A more substantive claim is embedded within this view, however: namely, that there is an important level of structure which does not so far have any standard representation in any theory that is familiar to us. . THING and LOC Above we schematized the structure common to existential and locative situations as involving a THING and a LOCation related by a two-place predicate (BE in the prototypical case, other V’s in the more extended cases.) In place of our LOCation, one might propose the term SCENE or STAGE36 (Russian scena ‘stage’), in the sense of the stage on which the action takes place, for ESs, and similarly for

The Russian genitive of negation 

what are often called ‘‘presentational’’ sentences. But we will continue to use LOC. This scheme is analogous to the proposal of Jackendoff (1972, 1990) to metaphorically broaden the notion of ‘‘being in a location’’, mentioned in Section 1.1. One could say that THING and LOC are roles of the verb byt’, but it is undoubtedly better to consider them roles of the participants of the situation (or state) of existing or of being located. Thus, in examples (11), (15), and the examples on the right-hand side in (16), THING is (what is denoted by) kefir ‘kefir’, Ivan ‘Ivan’, otvet ‘answer’, etc., and LOC is (what is denoted by) v magazine ‘in the store’, na lekcii ‘at the lecture’, and an understood ‘‘here’’ in the case of (16a). The LOC may be given explicitly, as in (11), (15), and (16c,d), or it may be implicitly understood, as in (16a,b). It is important that existence is always understood with respect to some LOCation. An implicit LOCation must be given by the context. This is usually ‘‘here’’ or ‘‘there’’, ‘‘now’’ or ‘‘then’’: at the place and time where someone is awaiting an answer which ‘‘didn’t arrive’’, for (16a), or ‘‘feeling (or not feeling) the frost’’, for (16b). The THING and the NP denoting it may belong to various sorts: concrete objects (including humans), massy stuff (kefir, money, etc.), events, pluralities, abstract entities (justice, problems); one area in need of further investigation is the difference in interpretation when the NP is an individual-level indefinite and when it is a kind-denoting term (Carlson 1977/80). Assertions of non-existence are presumably semantically somewhat different in the two kinds of cases. The problem of definite NPs is discussed further in Partee (2000). Note that the denial of existence of a THING in a LOCation does not contradict the existence of the THING outside the given LOCation, for instance ‘‘in the world’’. Thus, for instance, the use of GenNeg in sentence (15) to deny the existence of Ivan at the lecture is compatible with the presupposition of existence ‘‘in the world’’ carried by the proper name Ivan ‘Ivan’. When our Perspectival Center is the lecture, the relevant LOC is the location of the lecture, and in that LOC there is no Ivan. But the use of the proper name is itself associated with a conventional, larger LOCation,37 the world in which the people we know exist (and continue to exist when we don’t see them). Thus our rejection of the requirement that the THING be expressed by an indefinite NP is intimately connected to our adoption of the thesis that existence is always relative to a LOCation. The much greater frequency of indefinite over definite NPs in ESs may be linked to the consideration that when the THING is a definite particular, the speaker is more likely to make a description from the perspective of that THING. But that is not an absolute preference; sentence (15) is natural because we may sometimes be more interested in noting the absence of Ivan from the lecture than in talking about where Ivan was.

 Vladimir Borschev and Barbara H. Partee

. Perspective .. The notion of Perspective and its role in Existential Sentences The structure of a situation of existing cannot be reduced to its decomposition into the participants THING and LOC, since a situation of ‘‘being located’’ (corresponding to a ‘‘declarative’’ or ‘‘locative’’ sentence) has the same participants. We have proposed that the distinction between a situation of existing and a situation of being located involves a choice of perspective, of a point of view, from which the speaker, or sometimes the subject of a higher clause in the sentence, looks at that situation. The speaker, of course, is the one who chooses the form of expression; but if the relevant clause is an embedded one, the speaker may be representing the point of view of a higher subject of a propositional attitude. And even in the case of a simple sentence, if it occurs as part of a narrative, then the point of view of someone other than the ‘‘author’’ may be represented. The situation may be described with the THING as Perspectival Center, or with the LOCation as Perspectival Center. When the THING is chosen as Perspectival Center, its existence is presupposed, and the sentence speaks of its LOCation and potentially about other properties or states or actions in the situation. When we choose the LOCation as Perspectival Center, the sentence speaks about what THINGs there are or are not in that situation and potentially about what is happening in the situation. The choice of Perspectival Center, as so described, has much in common with the choice of Theme (Topic) on the one hand (see Section 4.2.2), and with the choice of grammatical Subject on the other (see Section 6): all three notions involve structuring something (a situation, a proposition, or a sentence) so that one part is picked out and the rest is in effect predicated of it. We believe that the perspectival structuring is first and foremost a (cognitive) structuring of the situation that will be talked about. How that structuring will be reflected in the linguistic structure of the corresponding expression may vary from language to language. Our main hypothesis with respect to the case at hand is that the role of perspective in ESs is to indicate a choice among different kinds of situations within the class of situations that share the basic participant-structure V(THING, LOC): when the LOC is marked as Perspectival Center, the result is an existential sentence,38 as stated in the Perspective Structure principle (23) in Section 3. As noted, we consider it likely that this result is mediated by a difference in choice of syntactic structure, but are agnostic as to what the syntactic difference is. But we do not believe that perspective structure directly affects or is affected by Theme– Rheme structure; we turn to this topic in the next section. .. Perspective vs. Theme–Rheme Structure

The Russian genitive of negation 

While the choice of perspective structure and the choice of Theme–Rheme structure may often coincide (in the sense that the Perspectival Center is a natural choice for Theme), we believe that they are different notions and in principle independent of one another. Perspectival structure is basically a structuring at the model-theoretic level, a structuring of the situation the sentence describes. If the job of semantics is to associate linguistic expressions with the non-linguistic ‘realia’ they refer to or describe, then we take perspectival structure as a property of those ‘realia’, in the same sort of domain as the properties that distinguish telic from atelic eventualities, or agents from experiencers, or the denotata of mass nouns from the denotata of count nouns. All of these properties reflect cognitive structuring of the domains that we use language to talk about, and are not simply ‘‘given’’ by the nature of the external world. Correspondingly, all of them are properties with respect to which we find differences from language to language. Two verbs in different languages, which seem to be otherwise perfect translations of one another, may nevertheless differ as to whether one of their arguments counts as a ‘‘Theme’’ (in the ‘‘theta-role’’ sense) or an ‘‘Experiencer’’. Hence the verbs may govern accusative in one language and dative in the other, even when the general principles for the choice of Accusative and Dative are the same in the two languages. (E.g. the verb meaning ‘call’, as to call someone on the phone, takes dative in some of the Indo-European languages, such as Russian zvonit’ and Spanish llamar, and accusative in others, such as English call and German rufen, anrufen.) This presumably represents a different schematization of the situation-type, a different decision as to whether a given participant in the situation-type counts as an ‘‘Experiencer’’ or not. Theme–Rheme structure is first and foremost a matter of information structure in discourse, although as Sgall et al. (1986) have argued, it belongs in semantics and not only in pragmatics. The information structure chosen depends not only on the situation to be described but also on what aspects of that situation have been mentioned most recently, which aspects are presumed familiar to the hearer, etc. The kind of example which has most strongly convinced us that distinction we need for explaining the distribution of GenNeg is not standard Theme–Rheme structure is our ‘‘kefir example’’ (11), repeated below as (29). (29) [Ja iskal kefir.] Kefira v magazine ne [I looked-for kefir-acc.m.sg] Kefir-gen.m.sg in store neg bylo. was-n.sg ‘[I was looking for kefir.] There wasn’t any kefir in the store.’ In (29), kefira ‘kefir’, in the genitive, is nevertheless part of the Theme. The evidence

 Vladimir Borschev and Barbara H. Partee

is twofold. In the first place, the rules governing the interplay of word order and intonation in Russian have been very well studied and repeatedly argued to be intimately bound up with Theme–Rheme structure (Kovtunova 1976; Yokoyama 1986; Russkaja Grammatika 1980; similarly for Czech and in general in Sgall et al. 1986.). According to those principles, kefira should be the Theme or part of the Theme in (29). Secondly, according to virtually all theories of communicative structure, the Rheme of one sentence is a favored candidate to become the Theme of the following sentence: one sentence introduces something as important, and the following sentence picks up that topic and says something about it. And in example (29), the background sentence clearly introduces kefir in the Rheme, so it is natural that kefira should be the Theme (or part of the Theme) of the following sentence. In this example the Rheme is probably only the final ne bylo ‘wasn’t’, and v magazine ‘in the store’ is also part of the Theme. According to the most general rule of the place of negation in Russian (Paducˇeva 1974: 154–5; see also Sgall et al. 1986), negation is associated with the Rheme. A minimally contrasting example, in which kefira is indeed Rheme and the sentence fits Babby’s pattern as he predicts, is (30) below. (30) [Ja zašel v magazin.] V magazine ne bylo kefira. [I went into store.] in store neg was-n.sg kefir-gen.m.sg ‘[I went into the store.] In the store there wasn’t any kefir.’ A different kind of minimal pair is given below in (31) and (32). In this pair of sentences, the sentence-initial Theme is the same (on/ego ‘he-nom/he-gen ), anaphorically referring to the Rheme Petja ‘Petja’ of the preceding sentence. In (31), the THING Petja is chosen as the Perspectival Center: we consider Petja, and where he was, and we give the partial information that he was not at the lecture. In (32) the LOCation is the Perspectival Center; this suggests that in our search for Petja, we went to the lecture expecting to find him, but Petja was not among those at the lecture.39 (31) [Ja iskal Petju.] On ne byl na lekcii. [I looked.for Petja.] he-nom.m.sg neg was-m.sg at lecture. [I looked for Petja.] He wasn’t at the lecture. (32) [Ja iskal Petju.] Ego ne bylo na lekcii. [I looked.for Petja.] he-gen.m.sg neg was-n.sg at lecture. [I looked for Petja.] He wasn’t at the lecture.40 We note again that our proposed perspective structure corresponds quite exactly with the division Babby (1980) proposed between Theme and Rheme: our Perspectival Center corresponds to Babby’s Theme. The difference could be called purely terminological and in a sense it is; the substantive issue is that Babby

The Russian genitive of negation

believed that the structure he had identified could be equated with the independently motivated Theme–Rheme structure, and we believe that it cannot be. The issues concerning the scope of negation mentioned in the Introduction and discussed in Section 5 also raise substantive problems for what seemed to be additional independent motivation for using Theme–Rheme structure as the determinative structure. But we do not consider this issue closed, and we do not consider our arguments against the use of Theme–Rheme structure totally conclusive. There are several possible kinds of arguments that may be given in defense of Babby’s original proposal, or which suggest modifications that could preserve the main structure of Babby’s analysis. Babby’s own suggestion (Babby 1980: 114–20) is that in NESs, the word order no longer expresses topic–focus structure, since that is now disambiguated by the genitive marking, but rather a difference in old (given) vs. new information. Babby cites Kovtunova (1976) and Gundel (1974) in support. He cites Gundel as characterizing ‘‘given’’ as ‘‘what you were talking about or what I was talking about before’’, while ‘‘topic’’ is ‘‘what I am talking about now’’. And he claims that in NESs like our kefir example (11/29), the genitive NP is pre-verbal if it is ‘‘old’’ or ‘‘given’’ in that sense (contextually bound), even though it is Rheme in the given sentence. It is well-known that while Theme is often ‘‘old’’, and Rheme often ‘‘new’’, these correlations are by no means absolute, and Babby thinks that is what is going on in the NESs with initial genitive NP. To support this suggestion, one would need an account of word order and intonation in these terms.41 A similar idea can be found in Paducˇeva (1995: 119–20), also citing Kovtunova (1976). Paducˇeva discusses the ‘‘dislocation of part of a complex rheme’’, in which the ‘‘rheme proper’’ is left at the end of the sentence with the main accent, and the remainder of the Rheme is dislocated leftward, usually but not always to sentence-initial position; the dislocated part receives a secondary stress with falling intonation. Paducˇeva (personal communication) holds that the kefir example (11) can have either of two communicative structures: in a context such as we present in (11), kefira would indeed be Theme and unaccented, but the same sentence could occur with no prior mention of kefir, only of the store, and in that case kefira would be a dislocated part of the Rheme, with a secondary falling stress. But note that if Paducˇeva is correct about that, it would not support Babby’s approach, because on Paducˇeva’s approach, if kefira is a dislocated part of the Rheme it remains indefinite, and it is only when kefira is part of the Theme that it would be interpreted as definite. Babby was trying to account for a potentially definite interpretation still being part of the Rheme. A second argument begins with the assertion that the LOCation (explicit or implicit) in examples like (11 / 29) is still Theme. The basis for this assertion is that it cannot be contrasted. The Appendix contains similar data from Serbo-



 Vladimir Borschev and Barbara H. Partee

Croatian, but in which genitive is used in both affirmative and negative ESs, and where differences in contrastability suggest that nominative subjects are Themes and genitive subjects are Rhemes, no matter what the word order.) If the LOC is still Theme, then either the THING must be part of the Rheme, or else the Rheme is only the (negation plus the) copula. The latter case is a possibility, but its semantics is mysterious: what exactly is being negated in that case? Does the negation still have sentence scope? Or is this verum-focus, focus only on the Affirmative/Negative polarity?42 But this issue should be deferred until we have solved the other puzzles of scope of negation discussed in the next section. The not fully conclusive outcome of this argument is that if just one or the other of LOC and THING is Rheme in the kefir sentences, it would seem that it is still the THING kefir that is Rheme; but matters may well be more complex. A different suggestion comes from Erteschik-Shir, and this is really an argument for greater subtlety of Theme–Rheme (Topic–Focus) structure, with internal hierarchical structure and other enrichments not dealt with in our simple adaptations of Babby’s simple one-level Theme–Rheme structures. Erteschik-Shir (1997) has a very relevant notion of stage-topic, and has an articulated system that allows for nested topic–focus structures. She suggests (p.c.) analyzing our kefir example as follows in her terms, with an initial implicit stage-topic, and kefira as a subordinate topic:43 (33) sTOPt[ KefiraTOP-sub [v magazine ne bylo]FOC-sub]FOC On this proposal, both Babby and the critics could be correct, although the crucial aspects of the structures would not be isomorphic to Babby’s. On Erteschik-Shir’s approach, existential (thetic) sentences are characterized by having a Stage-topic, and normal ‘declarative’ or ‘predicational’ sentences have an NP, usually the subject, as topic. On the other hand, when there is an implicit Stage-topic, which is not pronounced, the subordinate topic–focus structure of the explicit part of the sentence presumably takes over in determining the word order and intonation of the pronounced sentence. Other relevant work includes the proposals of Junghanns and Zybatow (1997) concerning the information structure of Russian; they follow Jacobs (1992) in working with two binary distinctions, Topic–Comment and Focus-Background, and under their proposal there are instances of thetic (Comment-only) sentences with the unaccusative subject moved into preverbal position although it is not topic. They also have examples of all-focus sentences which contain an added sentence-initial topic. A further complication that arises in trying to evaluate the Theme–Rheme and other structures of the affirmative and negative sentences we have been considering is touched on in Babby’s (1980: 119–20) discussion of Givón’s observations about the different pragmatic situations in which affirmative and

The Russian genitive of negation 

negative sentences are likely to be used. Givón (1975) notes that negative speech acts are typically uttered in contexts that are presuppositionally richer than the contexts typical for the corresponding affirmatives. Paducˇeva (p.c.) has raised our awareness of the related fact that not every affirmative sentence has a corresponding negative sentence that can be safely assumed to have the same Theme–Rheme structure. It is all too easy when testing the word order and intonation pattern of a negative sentence to pronounce it as a denial of an implicit preceding affirmative sentence (hence inviting a verum-focus pattern), and not always very easy to conjure up a natural context which does not involve such a denial. In fact, we can see that not every affirmative sentence has a corresponding negative sentence at all, and some of the clearest examples of this are presentational sentences like Here comes the bus, which are close cousins of affirmative ESs. We conclude this subsection by reiterating that we are inclined to believe that there are sentences like the kefir example in which the THING manifests GenNeg and is the Theme of the sentence. We are therefore inclined to prefer the introduction of our Perspective Structure over Babby’s original use of classic Theme– Rheme structure for distinguishing Declarative from Existential sentences, but we consider the issues far from settled. .. Examples In this section we list a set of examples with systematic manipulations of three factors, not in all possible variations, but in several of the more significant patterns. We group the examples according to their values on the three parameters: Declarative/Existential,44 Affirmative/Negative, and ‘‘thVLoc’’ Word Order (THING–V–LOC)/‘‘locVTh’’ Word Order 45 (LOC–V–THING). The subjects vary: one is a proper name, one a quasi-proper-name (otec ‘father’), one a bare singular count noun (doktor ‘a doctor/the doctor’), one a bare plural count noun (studenty ‘students/some students/the students’), and one a mass noun (vodka ‘vodka’). There is also some variation in the semantic nature of the LOCative expressions, particularly with respect to whether they denote simple spatial locations or more event-like entities with a strong intrinsic temporal dimension. At one extreme is na konferencii ‘at the conference’, which can only be understood as including a temporal as well as spatial dimension. There are two which can but need not be understood as implying something like an event (na more ‘at the sea’ (easily eventive) and v gorode ‘in town’ (less easily eventive). And one, v magazine naprotiv ‘in the store across the way’, is almost never interpreted as having any intrinsic temporal dimension. We first list the sentences and then discuss them.46 (34) ADS (decl, Aff, ThVLoc)

 Vladimir Borschev and Barbara H. Partee

a.

Petrov byl na konferencii. petrov-nom.m.sg was-m.sg at conference ‘Petrov was at the conference.’ b. Otec byl na more. father-nom.m.sg was-m.sg at sea ‘Father was at the sea (at a definite time / at some time in his life).’ c.

Doktor byl v gorode. doctor-nom.m.sg was-m.sg in town ‘The doctor was in town.’ d. Studenty byli na konferencii. students-nom.m.pl were-m.pl at conference ‘The students were at the conference.’ e. Vodka byla (prodavalas’) v magazine naprotiv.47 vodka-nom.f.sg was-f.sg (was.sold-f.sg) in store opposite. ‘Vodka was (was sold) in the store across the way.’ (35) NDS (decl, Neg, ThVLoc) a. Petrov ne byl na konferencii. Petrov-nom.m.sg neg was-m.sg at conference ‘Petrov was not at the conference.’ b. Otec ne byl na more. father-nom.m.sg neg was-m.sg at sea ‘Father was not at the sea (at a definite time/at any time in his life).’ c. Doktor ne byl v gorode. doctor-nom.m.sg neg was-m.sg in town ‘The doctor was not in town.’ (=‘The doctor did not go to town.’) d. Studenty ne byli na konferencii. students-nom.m.pl neg were-m.pl at conference ‘The students were not at the conference.’ e. Vodka ne prodavalas’ v magazine naprotiv. vodka-nom.f.sg neg was.sold-f.sg in store opposite. ‘Vodka was not sold in the store across the way.’ (36) NDS (decl, Neg, LocVTh) a. Na konferencii ne byl Petrov. at conference nEG was-m.sg Petrov-nom.m.sg ‘Petrov was not at the conference.’ b. Na more ne byl otec. at sea neg was-m.sg father-nom.m.sg

The Russian genitive of negation 

‘Father was not at the sea (at a definite time/at any time in his life).’ c. V gorode ne byl doktor. in town neg was doctor-nom.m.sg ‘The doctor was not in town.’ d. Na konferencii ne byli studenty. at conference neg was-m.pl students-nom.m.pl ‘The students were not at the conference.’ e. V magazine naprotiv ne prodavalas’ vodka. in store opposite neg was.sold-f.sg vodka-nom.f.sg ‘In the store across the way, vodka was not sold.’ (37) AES (ex, Aff, LocVTh) a. Na konferencii byl Petrov. at conference was-m.sg Petrov-nom.m.sg ‘Petrov was (among those) at the conference.’ b. Na more byl otec. at sea was-m.sg father-nom.m.sg ‘Father was at the sea (on a certain occasion at the sea/at some time in his life48).’ c. V gorode byl doktor. in town was-m.sg doctor-nom.m.sg ‘There was a doctor in town.’ d. Na konferencii byli studenty. at conference was-m.pl students-nom.m.pl ‘There were students at the conference.’ e. V magazine naprotiv byla (prodavalas’) vodka. in store opposite was-f.sg (was.sold-f.sg) vodka-nom.f.sg ‘In the store across the way there was vodka (sold).’ (38) NES (ex, Neg, LocVTh) a. Na konferencii ne bylo Petrova. at conference neg was-n.sg Petrov-gen.m.sg ‘Petrov was not (among those) at the conference.’ b. Na more ne bylo otca. at sea neg was-n.sg father-gen.m.sg ‘Father was not (among those) at the sea (on a certain occasion).’ c. V gorode ne bylo doktora. in town neg was doctor-gen.m.sg ‘There was no doctor in town.49’ d. Na konferencii ne bylo studentov. at conference neg was-n.sg students-gen.m.pl ‘There weren’t any students at the conference.’

 Vladimir Borschev and Barbara H. Partee

e.

V magazine naprotiv ne bylo vodki. in store opposite neg was-n.sg vodka-gen.f.sg ‘In the store across the way there wasn’t any vodka.’

(39) NES (ex, Neg, ThVLoc) a. Petrova ne bylo na konferencii. Petrov-gen.m.sg neg was-n.sg at conference ‘Petrov was not (among those) at the conference.’ b. Otca ne bylo na more. father-gen.m.sg neg was-n.sg at sea ‘Father was not (present) at the sea (on a certain occasion).’ c. Doktora ne bylo v gorode. doctor-gen.m.sg neg was in town ‘The doctor was not in town.’50 d. Studentov ne bylo na konferencii. students-gen.m.pl neg was-n.sg at conference ‘The students were not at the conference.’ e. Vodki ne bylo v magazine naprotiv. Vodka-gen.f.sg neg was-n.sg in store opposite. ‘There wasn’t any vodka in the store across the way.’ First a note about intonation patterns, since intonation and word order have to be taken into account together; the same word order with different intonation patterns may represent different syntactic structures and different information structures (see especially Yokoyama 1986). In general, we have intended that our sentences be read with neutral intonation insofar as possible, with no contrastive stress. Under neutral intonation in Russian (IK-1 in Russkaja Grammatika 1980: 96–122), there is a level tone followed by a final fall, with no strong emphasis at the point of the fall. Nevertheless, we are not sure that we have succeeded in limiting ourselves to examples which lend themselves naturally to neutral intonation. There may be examples in this paper (possibly including some which are part of the standard stock of examples on this subject) which require some special intonation, in which case any arguments in which they play a role may require closer scrutiny. We have noted some cases in earlier sections in which we are unsure that neutral intonation is a natural choice, and we will note below some uncertainty about some of the examples above. In the declarative sentences (34)–(36), with the THING as Perspectival Center, the situation is looked at in terms of the properties and location of the THING. For that reason, it is almost impossible for the THING to be indefinite; it is almost impossible to look at a situation from the point of view of an indefi-

The Russian genitive of negation 

nite THING. (More must eventually be said about kinds, individuals, generic expressions, specific vs. non-specific indefinites, etc. But to a first approximation it is safe to say that the thing chosen as Perspectival Center must be presupposed to exist and in general must be identifiable.) In the a. and b. examples, we have a proper name and the quasi-proper name ‘‘father’’. In (c) and (d) we have the bare nouns doktor ‘doctor’ and studenty ‘students’, which in Russian are unspecified for definiteness. In the declarative sentences (34)–(36) they can only be understood as definite, even in (36), where the NP is sentence-final and hence Rheme, a position and communicative status that normally favors an indefinite interpretation. (Of course, if the affirmative pattern (34) were to be inverted, the final NP would be understood as indefinite. But that is precisely pattern (37), which is unambiguously an existential sentence.) In the (e) examples, the subject is vodka ‘vodka’, which can be understood as the name of a kind, and hence implicitly definite. In some of the (e) examples, the verbs byt’ ‘be’ and prodavat’sja ‘be sold’ seem to be interchangeable, in some not; we are not entirely sure of the reasons. In the ESs (37)–(39), the LOC is the Perspectival Center. In (37) and (38), it is also the Theme. But in (39), although on our view LOC is still the Perspectival Center, it is not clear that it is Theme. If these sentences can be pronounced with neutral intonation (IK-1), then they appear to be cases in which the THING is marked with GenNeg but is Theme, and the LOC is part of the Rheme. But it is not clear that such a pronunciation is possible; the preferred pronunciation of those sentences seems to have heavy stress on ne ‘NEG’, suggesting that this may be an inverted structure of some kind, and the LOC, though final, may still be Theme. What the status of the ‘‘preposed’’ GenNeg THING is is not clear. According to the intuitions of the first author, the ‘‘marked’’ word orders in both the NDS (36) and the NES (39) are not completely natural without a marked intonation. Because we are interested in the problem of the alternative word orders in the NESs, (38) vs. (39), we also included both word orders for the NDSs for comparison, (35) and (36). But it must be noted that the locVTh word order in the NDS (36), in which the Perspectival Center is not the THEME, is not entirely natural; it seems to require a particular context such as the asking of a negative question like Who was not at the conference? Who was not at (or has never been to) the sea? In that case this is not a normal negated sentence, since it answers a question which contains a negated predicate. A more natural LOC-initial order for an NDS than the locVTh order of (36a) would be the locThV order in (40). We suspect that with this word order (where the stress most naturally falls on ne ‘NEG’), the Rheme may be only the negated verb, or only the negation itself. In that case this would be a case of verum-focus, as conjectured above for our kefir example (see footnote 43). In neither case would (36) be an ordinary NDS according to Babby’s scheme. (Babby himself did not discuss alternative word orders in NDSs; (36) and (40) are our additions to the discussion.)

 Vladimir Borschev and Barbara H. Partee

(40) Na konferencii Petrov ne byl. at conference Petrov-nom.m.sg neg was-m.sg ‘Petrov was not at the conference.’ And as for the marked thVLoc order in the NES (39), we already noted above that it may not be possible to pronounce that sentence with neutral intonation; if (39) would normally be pronounced with contrastive stress on ne ‘neg’, then it too may be a case of verum-focus. The first author considers the thLocV word order in (41) below a more natural word order for ESs in which the THING, marked with GenNeg, is initial. (41) (ex, Neg, ThLocV) a. Petrova na konferencii ne bylo. Petrov-gen.m.sg at conference neg was-n.sg ‘Petrov was not (among those) at the conference.’ b. Otca na more ne bylo. father-gen.m.sg at sea neg was-n.sg ‘Father was not (present) at the sea (on a certain occasion).’ c. Doktora v gorode ne bylo. doctor-gen.m.sg in town neg was ‘The doctor was not in town./There was no doctor in town.’ d. Studentov na konferencii ne bylo. students-nom.m.pl at conference neg was-n.sg ‘The students were not at the conference./There were no students at the conference.’ e. Vodki v magazine naprotiv ne bylo. vodka-nom.f.sg in store opposite neg was-n.sg ‘There wasn’t any vodka in the store across the way.’ It is interesting to note that this is the word order that occurs in our kefir example (11, 29), in the examples (9, 10) which Arutjunova uses to argue that the THING marked by GenNeg can be Theme, and also in our problematic scope-of-negation example (14). The same thLocV word order occurs in all of the byt’-paraphrases of the lexical-verb examples in (16). So the complications and uncertainties concerning the kefir example discussed in Section 4.2.2. carry over to this whole set of examples. In fact, a review of the examples in Babby(1980) reveals that there are no NESs with the marked thVLoc order of (39), although there are many with initial THING marked with GenNeg. But virtually all of them are either thLocV like (41) or else have no overt LOC constituent at all, as in (3b, 4b, 12), and in all of the weak-verb examples of (16). Examples with the thVLoc word order of (39) have, however, received some attention in the literature, and one of the interesting things noted about them is

The Russian genitive of negation 

that the bare NPs in sentence-initial position in these sentences must be understood as definite, leading some authors to doubt that these are ESs. Brown (1999: 85) discusses the contrasts among the following three sentences,51 of which (42b) is of the thVLoc order. (42) a.

V gorode net doktora. in town neg.is doctor-gen.m.sg ‘There is no doctor in town.’ b. Doktora net v gorode. doctor-gen.m.sg neg.is in town ‘The doctor is not in town.’ c. Doktor ne byl v gorode. doctor-nom.m.sg neg was-m.sg in town ‘The doctor was not in town.’ = ‘The doctor did not go to town.’

These sentences pose an interesting challenge to the notion of existential sentence; note that only (42a) (like (39c–e)) translates into an existential there-sentence in English, and yet both (42a) and (42b) show GenNeg. Since the existence of the doctor is presupposed, both Chvany and Brown consider (42b) to be a locative sentence rather than an existential sentence. Chvany designs her rules so that GenNeg applies to both existential and locative sentences with ‘‘existential’’ (lexical) byt’, and accounts for the nominative in (42c) (an NDS, =(35c)) by appealing to an agentive variant of byt’ that has a deep structure subject NP (as opposed to her basically Unaccusative analysis of existential/locative byt’. Brown has a direct account of the opposition of (42a) vs. (42c), with the higher position of the subject in (42c) at the relevant point in the derivation corresponding both to the assignment of nominative case and to the subject’s being outside the scope of negation. She believes that doktora in (42b) receives its ‘‘presuppositional interpretation’’ because it is outside the scope of negation,52 and that that is accomplished by its overt movement to some leftmost position, possibly Topic Phrase (Brown 1999: 90–1 and footnote 33.) For Babby, (42a) and (42c) would be normal NES and NDS respectively; (42b) might be an example in which the genitive subject, although part of the Rheme, occurs initially because it is ‘‘old information.’’ But since doktora, as a simple bare nominal, is unspecified for definiteness, it is unclear how to account for its obligatory definiteness in this example. Babby gives several examples in which a genitive-marked subject is initial because it is ‘‘old’’, but in which it is nevertheless still indefinite. Let us first consider the short discourse in (43) [/136/]: in the first sentence, the word sobaku (ACC) ‘dog’ is mentioned for the first time. Later on, the word sobak (GEN) ‘dogs’ is mentioned again, this time as the ‘‘subject’’ in an NES; the word

 Vladimir Borschev and Barbara H. Partee

order is NPgen- verb since sobak is old information: (43) [/136/] . . . devocˇka – Katja – zavela – sobaku . . . Do – ètoj – pory – ni-ukogo – iz – žilcov – sobak – ne – vodilos’. . . . girl – Katja – got – dog (acc) . . . up – to – then – no-one – fromtenants – dogs (Gen pl.) – neg – had (n.sg.) ‘. . . little Katja got a dog . . . Up to then none of the tenants had dogs.’ (Izvestija 1978). (Babby 1980: 117)

In fact it seems that in all NESs with the thVLoc pattern of (39), the subject is obligatorily interpreted as definite, even though it is marked with GenNeg. In contrast, in those with the thLocV pattern of (41), the initial GenNeg-marked subject may be interpreted as either definite or indefinite, and preferably indefinite. We do not pretend to understand all of the factors and their interactions. In fact it seems that no one has a fully satisfactory explanation at this point of the sentences that appear to be ‘‘declarative’’ or ‘‘existential’’ but which do not have the canonical word order of examples (34)–(35) and (37)–(38). The apparently greater freedom of word order in the negative sentences remains interestingly problematic. Also remaining only partially understood is the fact that an underspecified indefinite/definite bare noun phrase must be understood as definite in all declarative sentences and in one word order in NESs, that of (39). We resist the claim that those are not ESs, as we resisted it in the case of sentences with referential subjects and the verb byt’ ‘be’ discussed in Section 1.2. We return to some of these problems (without full resolution) in Section 5, in connection with problems concerning the scope of negation. .. Perspective and Observer Our use of the notion of perspective is in some ways close to the use of the notion of Observer in the work of Paducˇeva (1992, 1997). When the Perspectival Center is the LOCation, there is normally a moment or interval of time associated with it as well;53 some kinds of LOCation expressions (like at the lecture) are inherently as much temporal as spatial. Apart from atypical domains like mathematics, one can say that the marking of a LOC as Perspectival Center, like the notion of a stage on which a play takes place, always includes a temporal dimension as well,54 the time at which something exists or is taking place at that LOC. Paducˇeva’s ‘‘synchronous Observer’’ can be conceived as being located at the LOC at that time; or perhaps as ‘‘observing that LOC’’ at that time. (We believe that our Perspectival Center is what Paducˇeva’s Observer observes, not where the Observer stands; see below.) On the other hand, when the THING is the Perspectival Center, it is conceived as an independently existing thing, with no intrinsically associated temporal moment or interval. For this reason, in the case of ‘‘declarative’’ sentences where the THING is the Perspectival Center, and the situation is described in terms of

The Russian genitive of negation 

the THING, one may follow Paducˇeva in speaking of a ‘‘retrospective Observer’’, particularly in the case where ‘‘reference time’’ and ‘‘reference location’’ are associated either with the speaker’s utterance time and place or with the thoughts or perceptions of some textually identified ‘‘subject of consciousness’’. There are some differences between our notion of perspective and the notion of the Observer, however. (And there may well be more than one relevant notion of each sort in connection with various levels of sentence and discourse structure.) In perception verbs, the LOC is not the place where the Observer stands, but the ‘‘scene/stage’’ the observer sees, the observer’s ‘perceptual field’. The LOC can be named by egocentric (or ‘‘protagonist-centric’’) expressions like blizko ‘nearby’, nedaleko ‘not far away’, naprotiv ‘opposite, across the way’ (cf. examples (34d — 39d), vperedi ‘ahead’, nalevo ‘on the left’, etc. Such expressions do not denote the position of the observer but rather presuppose an observer from whose point of view the descriptions are to be interpreted. Our ‘‘Perspectival Center’’ is therefore not Bühler’s origo, the ‘‘I, here, now’’ center. If we take the metaphor of the movie camera, Bühler’s origo is the eye of the cameraman, and our Perspectival Center is what the camera is tracking. When the Perspectival Center is the THING, the cameraman will follow the THING as it moves and keep it in the picture; when the Perspectival Center is the LOCation, individuals may come into the scene and leave it again, while the cameraman doesn’t move the camera, or moves it to ‘‘survey the scene’’. A still photograph, like a written sentence out of context, may be unspecified as to perspective: a picture showing a person crossing a street may be a picture of the person or a picture of the street. Erteschik-Shir’s notion of a ‘‘stage-topic’’, mentioned earlier, is also very compatible with our suggestion of LOC as Perspectival Center in an existential sentence, where the Perspectival Center is what the ‘‘implicit Observer’’ or the ‘‘imagined camera’’ is visually tracking. The idea of a ‘‘stage’’ on which events are unfolding implies observers or an audience who are looking at the stage from a given location. The audience corresponds to Paducˇeva’s Observer, and the stage to our LOCation. As a related matter, T. E. Yanko (manuscript 1998), citing discussions with Yu. Bronnikov, emphasizes the fact that sentences with GenNeg subject use the impersonal form of the verb. The impersonal form of the verb denotes an event which takes place with no visible effort of the subject, as a result of hidden forces or by itself, accidentally, uncontrolled . . . . Let us compare [(44)], which means that Maša, being of sound mind and clear memory and possessing a will, decided not to visit us. While [(45)] may be understood as indicating that it was a matter of fate that Maša didn’t visit us . . . (Yanko 1998: 1)

(44) Maša

priezžala v Moskvu, no u nas ne byla.

 Vladimir Borschev and Barbara H. Partee

Maša-nom.f.sg came to Moscow, but at us neg was-f.sg ‘Maša came to Moscow, but she hasn’t been to us (didn’t come to us.)’ (45) Maši u nas ne bylo. Maša-gen.f.sg at us neg was-n.sg ‘Maša wasn’t (didn’t happen to be) at our place.’ The non-agentivity of the subject of an NES is a matter to which we will return briefly in Section 6.1., in connection with the Unaccusative hypothesis. This nonagentivity has been noted by a number of authors. Chvany (1975: 158) compares the ‘‘polite’’ (46a) (her (4.47a)) with the ‘‘less polite’’ (46b) (her (4.48a)). (46) a.

Menja ne budet na vašem koncerte. I-gen neg will.be-3.sg at your concert ‘I will not be at your concert.’ (Non-agentive) b. Ja ne budu na vašem koncerte. I-nom neg will.be-1.sg at your concert ‘I will not be at your concert.’ (Agentive)

Recall from Section 2 that Paducˇeva posited a systematic polysemy in the verbs used in the various constructions with byt’ and other verbs. Roughly, her proposal was that non-agentive verbs are used in NESs and agentive verbs in NDSs. We can put her proposal together with our perspective in the following way. For us, the difference in perspective structure is the basic difference between ESs and DSs. But one could well imagine that a given verb contains lexical information about its preferred perspective structure. (See in particular Paducˇeva (in press), for discussion of the obligatory THING-centered perspective of the verb naxodit’sja ‘to be located’ as compared with the preferred LOCationcentered perspective of prisutsvovat’ ‘to be present’.) With many verbs, we are able to shift our perspective from one with THING as Perspectival Center to one with LOCation as Perspectival Center; it would be natural to suppose that such a shift in perspective is very often accompanied by a shift, perhaps subtle and slight, in verb meaning. As has been noted by many authors, when we are paying attention to some human individual and talking about him or her (whether this is reflected in choice of that individual as Topic or as subject or as Perspectival Center), we are inclined to view that individual as an Agent of the events he/she participates in. When we are just watching a scene unfold, with LOCation as the Perspectival Center, we are more inclined to look at processes as just ‘‘happening’’, by themselves or as if by themselves, or at least without any attention to what their causes might be. Such shifts could in principle be signaled in a number of ways, including alternations in Theme–Rheme structure, alternations among related verbs, and

The Russian genitive of negation 

choices of different syntactic structures. Such a view of this family of phenomena is compatible with different degrees of ‘‘grammaticization’’ of the means of expressing these different perspectives, within and across languages, and with differences in how the expressive power is shared among lexicon, syntax, and information structure. .. Issues about the sources of presuppositions In all of the literature and discussion of the Russian GenNeg construction, there is widespread agreement about the differences in presuppositions in various different examples, but much less agreement about the degree of strength of the various presuppositions and about their sources, i.e. how their presence is to be explained. In Borschev and Partee (1998a), some of the crucial presuppositions were consequences of the Theme–Rheme structure, by principles articulated in Hajicˇová (1973, 1974, 1984), and having challenged the correctness of Babby’s assumptions, we are not sure how to derive them from other principles. The challenge for a theory like Babby’s based on Theme–Rheme structure amounts to the following. It does not appear to be possible to simultaneously maintain all of the following, at least not in a simplistic form: (47) 1. Theme–Rheme articulation is indicated (in Russian) by certain patterns of intonation and word order. 2. Theme–Rheme articulation is a determinant of the scope of negation. 3. Being in the scope of negation is a necessary condition for GenNeg in ESs. 4. The central difference between NESs and NDSs is their Theme– Rheme structure, from which their differences in scope of negation and in presuppositions follows. We have argued in favor of maintaining (1), since it seems to represent the most solidly established criterion for that distinction, and to replace (4) by an appeal to a somewhat similar but not identical distinction which we call perspectival structure. But since perspectival structure is a new concept, there is no preexisting theory to connect it with scope of negation or presuppositions. In Section 5 below we will raise and discuss some challenges to principle (3); if principle (3) above fails, then (2) becomes moot. But we would still like to try to say something about presuppositions, since they play a role in distinguishing existential from declarative sentences. As a working hypothesis, we will assume that any element chosen as a Perspectival Center must be presupposed to exist (in the relevant LOCation, which may or may not be the actual world). Elements that are not part of the Perspectival Center may or may not be presupposed to exist, but any presupposition of exis-

 Vladimir Borschev and Barbara H. Partee

tence for them would have to come from other sources and not from the perspectival structure. In an ES, since the main assertion is an assertion of existence in some LOCation, they cannot be presupposed to exist in that LOCation, or else the sentence would be anomalously redundant. We thus allow for definite subjects in AESs and NESs, as in the a. and b. examples in the example set (34)–(39). Those who classify sentences with definite subjects as automatically disqualified from being counted as ESs (we have cited examples in which Babby, Brown, Chvany, and Paducˇeva have done that) may be failing to accept or to exploit sufficiently Principle (21), that existence is always relative to a LOCation. The Russian GenNeg construction seems to use this principle to permit a much broader range of ESs (as manifested by GenNeg) than English permits with there-sentences. Other issues related to the sources of presuppositions are discussed in Borschev and Partee (1998a), including the presupposition of existence of the LOCation in ESs and the potential absence of such a presupposition in DSs. But the issues relating to definiteness in ESs are among the areas that call for further study; see e.g. Ward and Birner (1995) and subsequent discussion.

.

On the scope of negation

As noted in the Introduction, we have recently discovered that even the generally agreed-on statement that GenNeg is a reliable indicator of occurring in the scope of sentential negation is not quite correct. We have apparent counterexamples in both directions. In the paradigm cases of NES vs. NDS, such as (3a–b), (4a–b), (35c) vs. (38c), (35d) vs. (38d), there is unquestionably a difference in the scope of negation. But there are problem cases. Since negation is very often a much more complex matter than it seems at first sight, the apparent counterexamples to the claim that GenNeg occurs only under the scope of sentential negation that we will discuss may not be absolute. But before turning to our two classes of problem cases in Sections 5.1. and 5.2., let us just note one interesting problem that arises in trying to determine the scope of negation. Suppose we have a negated sentence containing a proper noun, as in (48). (48) John is not in the house. It might seem intuitively obvious that John is outside the scope of negation in that sentence. But in fact, as far as ordinary truth-conditions go, there is no way to decide. One can represent the semantic interpretation of the sentence either as in (49a) or as in (49b); the results will be equivalent. If what were moving in and out

The Russian genitive of negation 

of the syntactic scope of the negation operator were a weak quantifier like a boy, the truth conditions would indeed be different, but if it’s a proper name or a presuppositional or anaphoric definite description,55 its structural scope will not be reflected in any semantic difference in the result. ¬ [ In (John, 4z(house(z)))] [informally: NOT (John in the house)] b. 8x [¬ In(x, 4z(house(z)))] (John) [informally: John has the property: NOT in the house ]

(49) a.

We mention this basic fact as an example of ways in which the scope of negation is not open to immediate intuition; in examples like (48), theoretical arguments are likely to be crucial in deciding whether the subject should be included under the scope of the sentential negation. These observations are directly relevant to some of the disputed examples discussed above, such as the definite-subject readings of the examples (39c–d). If the THING-first word order arises because the subject is definite and Theme, then it would be as scope-insensitive as a proper name. Its presuppositionality would be independent of whether it is in the scope of negation or not, just as the presuppositionality of a proper name normally is. The sentences could then be argued to be NESs by the same arguments we have used for the proper-name examples. (But there is more to explain in the case of bare nouns; a proper name is interpreted as definite unless there is a very strong context to coerce it otherwise; the factors that make a bare noun in Russian preferentially interpreted as indefinite or definite (or non-presuppositional vs. presuppositional) require much more discussion.) The important thing to note is that a definite interpretation of the subject is not by itself proof that the subject is outside the scope of negation, though that is a natural first hypothesis. We turn now to our problematic examples. . Nominative under scope of negation in BE(THING, LOC) sentences Babby (1980) argued, and we as well as most other researchers had been in agreement on this matter, that GenNeg in intransitive sentences occurs under the scope of negation, and nominatives in such sentences are outside the scope of negation. But it turns out56 that both of the following are perfectly good sentences: (50) Nikto tam ne byl. NI.who-nom.m.sg there neg was-m.sg ‘No one was there.’ (51) a.

Nikogo

tam ne bylo.

 Vladimir Borschev and Barbara H. Partee

or

NI.who-gen.m.sg there neg was-n.sg ‘No one was there.’ b. Tam ne bylo nikogo. there neg was-n.sg NI.who-gen.m.sg ‘No one was there’

It seems clear that (50) is a DS and (51a, b) are ESs, in Babby’s terms; they have all the same differences in presuppositions, natural contexts, etc. as the well-known pair (35b)–(39b) discussed by Apresjan (1980) in connection with differences in ‘reference time’, and by Borschev and Partee (1998a, b), repeated below as (52a–b): (52) a.

Otec ne byl na more. father-nom.m.sg neg was-m.sg at sea ‘Father was not at the sea (at a definite time/at any time in his life).’ b. Otca ne bylo na more. father-gen.m.sg neg was-n.sg at sea ‘Father was not (present) at the sea (on a certain occasion).’

But it seems that nikto in (50) must be in the scope of the sentential negation ne just as much as nikogo in (51). Is there some way out of this that preserves the otherwise apparently valid generalization about genitive vs. nominative correlating with scope of negation? Babyonyshev, Fein, Ganger, Pesetsky, and Wexler (ms. 1998) discuss the acquisition of GenNeg by Russian children ages 3–6, and show that they master GenNeg with the direct object of transitive verbs earlier than they master it with intransitive verbs. That paper includes a number of constructed minimal pair examples with NPs of the form nikakie (nom acc.pl) / nikakix (gen.pl), in various positions where GenNeg is licensed. The authors’ judgment is that in such positions (under the scope of sufficiently local sentence negation) both forms are possible though the genitive form is likely to be preferred when the NPI nikakie/nikakix is included. But the very possibility of the nom acc form raises the same question as the occurrence of nikto raised above.57 The examples in which Babyonyshev et al. suggest that nom acc and gen are in free variation under scope of negation are all object cases, so we haven’t reproduced them above and we do not put much weight on them. But we have also found apparently analogous subject cases. (53) a.

Odin student tam ne byl. one-nom.m.sg student-nom.m.sg there neg was-m.sg ‘One student was not there.’ (One student has never been there/one student wasn’t there at an understood occasion.) b. Odnogo studenta tam ne bylo.

The Russian genitive of negation 

one-gen.m.sg student-gen.m.sg there neg was-n.sg ‘One student wasn’t there.’ (on a given occasion) (54) a.

Ni odin student tam ne byl. NI one-nom.m.sg student-nom.m.sg there neg was-m.sg Not one student was there. (No student was there.) b. Ni odnogo studenta tam ne bylo. NI one-gen.m.sg student-gen.m.sg there neg was-n.sg ‘There was not one student there.’ (‘There was no student there.’)

In terms of the logical scopes of the negation operator and the quantifier ‘‘one’’, it seems that the (truth-conditional part of the) meanings of both (53a, b) could be represented approximately as:58 (55) One student x [ neg [ x was there] ] And the (truth-conditional part of the) meanings of both (54a,b) could be represented as: (56) neg [one student x [ x was there ]] So the standard notion of scope of negation seems not to distinguish between the interpretations of the nominative and genitive NPs in these examples. This is different from the behavior of comparable English quantifiers in existential there-sentences compared to their behavior in plain categorical sentences. (57) a. One student wasn’t there. (One student was absent) b. There wasn’t one student there. (No student was there.) One might have expected the nominative sentence (53a) to be like (57a) and the genitive (53b) to be interpreted like (57b). And it is a surprise that (54a), in the nominative, is perfectly fine. Part of the explanation (thanks to Elena Paducˇeva (p.c. and 1992) is that Russian odin ‘one’ is more ‘specific’ than English one. This odin1, since it is specific, may be as scope-insensitive as a definite. Compare the ‘‘wide-scope-only’’ behavior, i.e. scope insensitivity, of English a certain: (58) I couldn’t find a certain book I wanted. And ni odin ‘not one, not a single’ is not (semantically) just the negation of odin, it’s something much stronger, like ‘not a single’. So this odin2 may be a Negative Polarity Item. There are thus two different things going on in those examples, which the examples illustrate nicely. One is degree of referentiality (odin1 and (ni-) odin2), +/- specific respectively. The other is the nom gen choice, which is a

 Vladimir Borschev and Barbara H. Partee

central concern of this paper. For Paducˇeva (1992), the choice involves two different verbs byt’, with the more agentive one taking nominative, and the more stative one taking genitive. But in Paducˇeva (1997), and in our work, the choice would rather be seen as depending on whether the sentence as a whole is interpreted as an existential sentence or not, not (or not only, for Paducˇeva) as an ambiguity in the verb. It turns out that Babby (1980) has an example similar to (54a,b) (59) a.

Tam ne roslo ni odno derevco. there neg grew-n.sg. NI one-nom.m.sg. tree-nom.m.sg. ‘Not a single tree was growing there.’ b. Tam ne roslo ni odnogo derevca. there neg grew-n.sg. NI one-gen.m.sg tree-gen.m.sg ‘There wasn’t a single tree growing there.’ (Babby 1980: 66)

Babby asserts that even when such a pair of sentences seem to mean essentially the same thing, there is always a subtle functional difference. But is it consistent with Babby’s statements to have a nominative in (59) at all? Babby makes slightly conflicting statements on this point. In his central Chapter 3 on the subject, ‘‘Genitive Marking and the Scope of Negation’’, the chapter which includes his chart (7), he says: The key to understanding the ‘‘genitive of negation’’ in general, and genitive subject NP marking in particular, can be stated as follows: /78/ In Russian, an NP in a negated sentence can be marked with the genitive case if and only if it is in the scope of negation. . . . It is only in NESs that the subject NP falls in the scope of negation; thus it is only in NESs that the subject NP is marked genitive. (Babby 1980: 69)

Footnote 10 to Chapter 3: . . . I would like to emphasize once again that /78/ is a necessary but not sufficient condition for genitive marking to operate, i.e., all NPs marked genitive in Russian negated sentences must be in the scope of negation, but it is not the case that all NP’s in the scope of negation are marked genitive. There are a number of other conditions that must be satisfied if an NP in the scope of negation is to be marked genitive. (Babby 1980: 76)

We believe that the ‘‘if and only if ’’ in Condition /78/ was meant to be ‘‘only if ’’; then Babby would be consistently claiming that we will sometimes find nominatives in the scope of negation, but will never find genitives that are not in the scope of negation. So from this much, it is not inconsistent with Babby’s principal hypothesis that we may find examples like (50), (54a), and (59a), with nominatives inside the scope of negation.

The Russian genitive of negation 

But for Babby, a second important hypothesis is that the scope of negation is determined by the Theme–Rheme structure, and that NESs and NDSs differ in their Theme–Rheme structure. And it is central to that distinction that the subject of an NDS is outside the scope of assertion and accordingly outside the scope of sentential negation. So we should not find NDSs like (50), (54a), and (59a): the only time we should find nominative NPs inside the scope of negation is when we have an NES which fails to meet some of the further conditions for genitive marking. But Babby makes it clear that he considers the pair in (59) a good example of an NDS/NES pair. Perhaps the problem thus posed by an apparent NDS with its subject in the scope of negation was part of the reason for his careful hedge later in Footnote 10: It should also be pointed out that I have introduced two separate hypotheses in this chapter: 1) a sentence’s scope of negation is determined by the scope of assertion; and 2) an NP can be marked with the genitive only if it is in the scope of negation. Since (1) and (2) are separate hypotheses, one can be shown to be wrong without necessarily involving the other. Only hypothesis 2 is crucial for our analysis of NESs. (Babby 1980: 76–7.)

It seems clear that one of the two hypotheses, presumably the first, must be wrong if Babby’s own examples (59a–b) are to be taken as a pair of an NDS and an NES respectively. At first glance it would seem that the second, more central, hypothesis is challenged by example (53b), where we appear to have a GenNeg that is outside the scope of negation, violating the principal hypothesis. But this violation may be only apparent: the NP odnogo studenta ‘one student-gen.m.sg' is a specific indefinite, and hence following Kratzer (1998) and others, it may be analyzed by a choice function and may display many of the same properties as proper names and definite NPs. In other words, a specific indefinite, like a definite, may ‘‘sit inside the scope’’ of negation, and nevertheless semantically appear to escape that scope by virtue of its referential properties. (No ‘‘movement’’ is required. Movement or some comparable mechanism is needed for obtaining ‘‘wide scope’’ readings only for those NP’s for which wide and narrow scope relative to negation are semantically distinct.) A real challenge to Babby’s principal hypothesis comes in the next subsection. . Genitive of negation apparently not under scope of sentential negation The examples above concerned nominatives under the scope of negation in positions where one might have expected GenNeg to be obligatory, and an apparent but probably not real case of GenNeg not under the scope of negation. We have also found real examples of the opposite sort, with occurrences of GenNeg which

 Vladimir Borschev and Barbara H. Partee

more strongly appear not to be under the scope of sentential negation. We introduce this section with a real-life example involving a direct object GenNeg not in scope of negation. The two authors were out collecting mushrooms on August 4, 2000. We had started early in the morning, and in the first places we looked, it appeared that we had succeeded in being the first ones there, maybe for some days. Later, towards noon, we were in places where others had evidently been before us. But we still found some. The second author [BHP] was thinking about how to say in Russian, ‘‘[Other people have been here, but] there could be some they didn’t find.’’ She made some incorrect attempts, after which the first author [VB] corrected her with (60a) In later discussion VB said that (60b–d) are also possible, though (60b) seems somehow not quite as natural. (60) a.

Oni mogli cˇego-nibud’ ne najti. they could something-gen (non-specific) neg find-inf ‘There could be something (non-specific) they didn’t find.’ b. ?Oni mogli cˇto-nibud’ ne najti. they could something-acc (non-specific) neg find-inf ‘There could be something (non-specific) they didn’t find.’ c. Oni mogli cˇto-to ne najti. they could something-acc (specific) neg find-inf ‘There could be something (specific) they didn’t find.’ d. Oni mogli cˇego-to ne najti. They could something-gen (specific) neg find-inf ‘There could be something (specific) they didn’t find.’

What’s interesting about (60a) in particular is that cˇto-nibud’ is often characterized as a weak kind of polarity item that is licensed by various modals, interrogatives, and higher-clause negation, but cannot occur with clausemate negation (as its licenser). So (61), for instance, is impossible, in contrast to (62), (63), which are both ok. (61) *Oni cˇego-nibud’ ne našli. they something-gen (non-specific) neg found. (62) Oni nicˇego ne našli. they nothing neg found. ‘They didn’t find anything.’ (63) Oni cˇto-to/cˇego-to ne našli. they something-nom/gen (specific) not found. ‘There was something they didn’t find.’ So (60a) seems to show that in the case of object GenNeg, at least, the object does

The Russian genitive of negation 

NOT have to be in the scope of negation, it just has to be the syntactic direct object in a ‘negative VP’ or ‘negative sentence’. We have not studied that construction though, so we cannot comment on the factors that may be involved.59 Here is a constructed example involving the subject of an existential construction. We find the same result. (64) Možet byt’, cˇego-nibud’ u nego net. may be, something-gen (non-specific) at him neg.is ‘Maybe there is something he doesn’t have.’ An appropriate context for (64) would be shopping for a present for someone who has just about everything he needs, and hoping you can think of something that he doesn’t have that you could get for him. Možet byt’ ‘maybe’ in (64) is a verbal construction that forms a higher clause; that is what licenses the -nibud’ form, which cannot occur within the scope of its clausemate negation. The same kind of construction can also be found with a higher modal verb taking the lower clause in infinitive form, as in (65); and conversely, examples like (60a–d) but with higher mozhet byt’ and a finite clause complement are also fine, as in (66). Cˇego-nibud’ moglo u nego ne byt’. something-gen (non-specific) could at him neg be-inf ‘There could be something he doesn’t have.’ b. U nego moglo cˇego-nibud’ ne byt’. at him could something-gen (non-specific) neg be-inf ‘There could be something he doesn’t have.’ c. Moglo u nego cˇego-nibud’ ne byt’. could at him something-gen (non-specific) neg be-inf ‘There could be something he doesn’t have.’ ‘Maybe there is something (non-specific) they didn’t find.

(65) a.

(66) Mozhet byt’, oni cˇego-nibud’ ne našli. may be, they something-gen (non-specific) neg found. ‘Maybe there is something (non-specific) they didn’t find.’ So putting together the examples in Sections 5.1 and 5.2, it seems that occurring in the semantic scope of negation is not necessary for genitive and not impossible for nominative. Babby (1980) acknowledged the latter, although it is not without problems for his theory, as noted above. The data in Section 5.1. rely on the behavior of ni-phrases: if nikto and ni odin student are NPIs, and must be in the semantic scope of negation, then nominative can be in the semantic scope of negation.

 Vladimir Borschev and Barbara H. Partee

The data in Section 5.2 rely on the behavior of -nibud’–phrases. It has been argued that they cannot occur in the semantic scope of clausemate negation, but we can find them in the genitive form where that genitive is apparently indeed (syntactically) licensed by clausemate negation. (And see Brown (1999: 94–8) for some related observations.) This suggests that there should be some relevant distinction between being syntactically ‘in the scope of negation’ [e.g. c-commanded by it in surface structure with no intervenors of some relevant sort] and semantically in the scope of negation [the usual semantic notion]. Brown (1999) is one proposal of that sort, and we will mention it in Section 6 below. Although we remain syntactically agnostic, at least the second author strongly suspects that there are very likely some syntactic as well as semantic differences that distinguish NESs from NDSs. The fact that some Slavic languages have completely syntacticized the rules for GenNeg also lends some plausibility to the hypothesis that there is syntactic mediation of the phenomenon in Russian as well.

. Syntactic issues Many authors who have worked on the problem of Russian NES have proposed syntactic distinctions of one kind or another between NES and NDS. And Babby (2000), while preserving the Theme/Rheme ideas of Babby (1980), also proposes that the assignment of GenNeg in NESs should be mediated by the syntax, and makes a specific syntactic proposal. The NP within Rheme of his analysis of ESs is syntactically characterized as an NP dominated by V-bar in his recent analysis; his NP Theme is syntactically one that has gotten to a higher position in the tree, either by starting there or by movement. A number of authors point to Diesing’s (1992) tree-splitting hypothesis as a relevant parallel. See also the connections suggested between Theme–Rheme structure and analogs of Diesing’s tree-splitting hypothesis in Hajicˇová, Partee and Sgall (1998). Syntactic analyses of Russian ESs, motivated largely by the problem of the GenNeg, are often put in direct parallel with analyses of Italian existential sentences, with references to work by Belletti and critiques of it by Williams and others; and Perlmutter has worked on both and sees them as very parallel. We do not consider ourselves experts on the syntax of these sentences, but compositional semantics can’t be done without considering syntax. So let us briefly review some of the main ideas that have been proposed about the syntax of existential sentences. . Unaccusativity vs. non-agentivity

The Russian genitive of negation 

The GenNeg construction is found in Russian not only on the ‘‘subjects’’ of intransitive verbs in ESs, but on the objects of transitive verbs in almost all kinds of sentences. And many have claimed that the verbs that can be used in ESs are all Unaccusative verbs (Chvany 1975; Pesetsky 1982; Bailyn 1997; Brown 1999), intransitive verbs whose surface subject is an underlying object. The idea that GenNeg applies only to underlying objects has great appeal in unifying the GenNeg construction. Babby (2000) argues against it. His strongest arguments are against equating the post-verbal ‘‘subject’’ position in ESs with the direct object position in transitive sentences. He also suggests, less conclusively, that some ESs involve Unergative verbs; but there it might be possible to appeal to type-shifting of verbs from one sort to the other, something Pesetsky explicitly acknowledged as a consequence of the Unaccusative hypothesis for Russian. Unaccusativity gives at most a necessary condition for GenNeg, not a sufficient one. In particular, no one suggests that all and only sentences containing Unaccusatives are ESs. Questions not automatically answered by the Unaccusative hypothesis: 1. Does the underlying object in an NES become surface subject? In an AES? 2. Are minimal AES/ADS pairs both Unaccusative, with different derived surface structures (if so, what?), or are AES/ADS pairs Unaccusative/Unergative? 3. A related question: What governs ‘‘optional’’ GenNeg assignment in NDS/ NES pairs? Babyonyshev (1996) relates the alternation of genitive with nominative and accusative in Russian to the ‘weak-strong’ ‘‘Diesing effects’’ effects found when NPs in Germanic languages are left in their base-generated positions or topicalized. Bailyn (1997) is another who supposes that both ESs and DSs involve Unaccusative verbs, with DSs involving a raising of the THING NP to a position higher than scope of negation. Brown (1999) builds on Bailyn’s view, with her negation position higher in the tree than Bailyn’s in order to account for negated subjects in all kinds of sentences. Brown analyzes GenNeg as a phenomenon that affects NPs that start ‘‘low enough’’ in the tree. They may move higher in the course of a derivation, and many of their optional properties are explained as reflecting a choice of interpreting the ‘‘head’’ or the ‘‘tail’’ of the chain created by their movement. Semantically, it is widely agreed that when the same verb may be used in both ESs and DSs, it receives a more agentive reading in the DS and a more stative reading in the ES. Those who are skeptical about the Unaccusative hypothesis generally believe that much of what it is trying to explain should be explained in terms more directly relating to Agentivity.

 Vladimir Borschev and Barbara H. Partee

. Existential sentences as Impersonal sentences There is considerable debate in the literature as to whether NESs (and even more controversially, AESs) are Impersonal sentences. Perlmutter and Moore (1999) propose that both AESs and NESs involve Unaccusative sentences in which the underlying object does not become the surface subject. We sketch Perlmutter’s proposal in extremely brief form. Perlmutter was concerned only with syntax; the extension of his ideas to relate to semantics and pragmatics is our suggestion. 1. In ‘‘DS’’, there are two possibilities. a. verb Unergative, initial subject is the surface subject, in both affirmative and negative sentences. b. verb Unaccusative, underlying object has become surface subject and has remained a surface subject. 2. ‘‘ES’’ are impersonal sentences, whose surface subject is a null dummy expletive pronoun, classified as neuter singular and nominative. The previous subject is demoted to become ‘‘brother-in-law’’ of the dummy expletive. In affirmative sentences, the ‘‘brother-in-law’’ NP is nominative, in negative sentences it is genitive. 3. Like many other authors, Perlmutter classifies the verbs that participate in the ESs as ‘‘Unaccusatives’’, having an underlying object and no underlying subject. Thus the genitive NP in an NES starts out as an object, then becomes a subject, then becomes ‘‘brother-in-law’’ of the dummy expletive subject. (This is similar to the structure proposed for English existential sentences with There.) Perlmutter’s rule for GenNeg in Russian is that an underlying direct object that is not the surface subject may be genitive under the scope of negation. (This unifies the ESs with the transitive sentences in which the object gets the GenNeg marking.) The distinction between ES and DS, with those verbs that allow both, would be whether the NP in question becomes the surface subject (DS) or the dummy expletive becomes the surface subject. We would propose that it may be the choice of ‘‘LOCation’’ as the ‘‘Perspectival Center’’ of the sentence that is the functional trigger for choosing the impersonal construction, and that this syntactic difference could therefore be the way the distinction is reflected in the grammar. See the similar semantic correlate of Locative inversion in Chichewˆa discussed by Bresnan and Kanerva (1989). A three-way contrast is possible in the case of verbs which can be either Unergative or Unaccusative: a DS (either affirmative or negative) with Unergative verb and THING subject, a DS with Unaccusative verb and THING subject, or an ES with Unaccusative verb and the THING as brother-in-law to a null expletive subject. The verb plavat’ ‘swim, float’ is one which in Russian can be understood either as Unergative (and agentive) or Unaccusative (and non-agentive). We give (67a–c) from Perlmutter and Moore (1999), adding (67d) as an unergative

The Russian genitive of negation 

example; if the lexical items of (67d) were used in the NES structure of (67b–c), the (unnatural) interpretation might tend to be one of bodies floating in the pond, or some other loss of agentivity. (67) a.

Kuvšinki ne plavali v prudu. NDS (Unaccusative) water-lilies-nom neg floated-pl in pond ‘Water lilies were not floating in the pond.’ b. V prudu ne plavalo kuvšinok. NES (Unaccusative) in pond neg floated-n.sg. water-lilies-gen ‘There weren’t any water lilies floating in the pond.’ c. Kuvšinok ne plavalo v prudu. water-lilies-gen neg floated-n.sg in pond ‘There weren’t any water lilies floating in the pond.’ Scrambled NES (Unaccusative) d. Malcˇiki ne plavali v prudu. boys-nom neg swam/(floated) in pond ‘Boys/the boys were not swimming (floating) in the pond.’ NDS (Unergative (or unaccusative))

In both (67b) and (67c), kuvšinok ‘water-lilies’ is in the predicate predicated of the Perspectival Center; in (67b), the unmarked order, it is also in the Rheme. But in the scrambled (67c), appropriate if we have just been talking about water lilies, it is Theme: here it names a type (which exists in the ‘universe of discourse’, the Reference LOCation) of which no tokens exist in the stated LOCation. . LOCative as subject? ‘‘THING’’ as subject? Few authors go so far as to suggest that the LOCative can become the subject in an ES (as it can in Chichewˆa), but Kondrashova (1996) does make such a suggestion for ESs with the copula verb. Babby (2000) is in a sense at the opposite extreme: he argues that the THING element is still the subject even in an NES or AES, but that unlike in an NDS or ADS, the subject NP is not in ‘‘canonical’’ subject position. The roles of subject are ‘split’ in ESs, Babby suggests, with the Locative typically fronting to satisfy the Extended Projection Principle (the requirement that every sentence have something that functions as a ‘‘subject’’ in a suitable structural sense.). The THING argument remains post-verbal, but not in a direct object position; it is sister to a V-bar, where it can be in a predication relation. But Babby (2000) and Perlmutter and Moore (1999) are not really so far apart; they propose very similar impersonal constructions for ESs and may differ primarily on criteria for calling something a subject. And probably only Perlmutter has impersonal AESs. What these proposals and other related ones have in common is proposing a

 Vladimir Borschev and Barbara H. Partee

more syntactic version of the idea that what is a canonical subject in a DS somehow stays inside the VP in an ES, or at any rate stays ‘‘lower’’, or at least starts out ‘‘lower’’. It is as if a Perspectival Center LOCation ‘‘wants’’ to be subject of an ES, but since it is structurally unable to become subject, some non-canonical structure must be found. Proposals for the structure of a ‘‘NegP’’ (NEG Phrase, a functional projection) exploit the ‘‘lowness’’ of the subject of an ES to account for the fact that the subject of an ES is under the scope of sentential (or VP) negation, and the subject of a DS is not. . Genitive of negation as a Negative Polarity Item A number of authors, including Pesetsky (1982) and Pereltsvaig (1997), have taken the fact that the GenNeg construction is almost invariably (but see Section 5) found under the relatively local scope of negation as grounds for seeking to assimilate it to negative polarity phenomena. Pesetsky suggests that the GenNeg construction reflects the presence of a null NPI quantifier (analogous to English any) which in turn governs the genitive in its complement, as many quantifiers do. Pereltsvaig extends Pesetsky’s analysis to explain the interactions between GenNeg and aspect. These analyses face some difficulties when the NP in the GenNeg construction is a pronoun or a definite NP headed by a demonstrative or an explicit quantifier. This is a very large topic which we can barely mention here. . The nature of the NP in the Genitive of Negation construction We noted earlier that Babby (1980) made a strong claim that one of the conditions on the NES construction was that the NP must be indefinite; but that he himself has numerous counterexamples. But indefinites are clearly the unmarked case. Steube and Späth (1999) include some discussion of Borschev and Partee (1998a) and make it clear that the notions of ‘‘non-referentiality’’ and ‘‘indefiniteness’’ often mentioned in connection with the GenNeg cannot be simply dismissed just because of the existence of examples with proper nouns or demonstrative ètot ‘this’. These authors also note that the full range of factors affecting Russian word order is not yet fully understood. When a proper name or a definite is used in an ES, then usually one of several factors is present. (i) The LOCation may be understood as stage-level: (15) says there was no Ivan at lecture, i.e. no stage of Ivan in the given spatiotemporal location. (ii) The NP is understood as naming a kind or a type, and the sentence asserts there is no instantiation of that kind or type at the given LOCation. (iii) The NP is definite but ‘‘intensional’’: it names something that has been intensionally

The Russian genitive of negation 

established in the discourse, of which the sentence says there is no extension in the given LOCation, as in example (i) in Note 56, from (Babby 1980:118.) Similarly, the disputed ‘‘topics’’ in examples like the kefir example do not have all the prototypical properties of topics. It appears that all the disputed examples involve intensionality. Note, for instance, that while the noun kefira ‘kefir’ is ‘‘old’’, ‘‘already under discussion’’, in the second sentence in (11), it is not ‘‘referential’’. Babby (1980) gives interesting examples of definite NPs, even pronouns, in the GenNeg, all of them sentence-initial and all intensional. His example in note 56 shows a non-referential pronoun in the GenNeg. These issues are discussed in Partee (2000), but many fundamental questions remain open.

.

Concluding remarks

The problem of Genitive of Negation in Russian existential sentences is endlessly interesting and difficult because it relates so strongly to so many different kinds of principles from morphology to pragmatics, and because the structures involved are of necessity highly theory dependent and hence subject to change with changes in theoretical frameworks. There are many more questions about it than we have raised here. As noted by Bailyn (1997) and Abels (2000), until recently there seemed to be an either-or attitude about functionalist vs. syntactic explanations, but now it is generally acknowledged that parts of both may be necessary to an adequate account. While we are still far from a full account, a picture may be emerging. The Perspectival Center status of the LOCation and the corresponding not-ordinarysubject status of the THING are both marked choices. A language which simply let one make the LOCation the subject would align subject and Perspectival Center, and syntactic predicate with what is predicated of it; that would represent a full ‘syntacticization’ of the distinction. On the other hand, one could imagine a language in which there was no difference except word order, and the existential sentence was realized just by making the Perspectival Center the Theme and the rest of the sentence the Rheme, indicated by word order. Russian seems to do something in between; the best way to characterize it syntactically is still not clear, but as functional and formal approaches are brought together the pieces are beginning to fall into place.

Appendix Data from Serbo-Croatian60 (A-1) AF/GEN [Na stolu] je bilo (raznih) knjiga i cˇasopisa. on table was-n.sg (various-gen.pl) books-gen and magazines-gen ‘On the table were (various) books and magazines.’ Note: unlike Russian, Serbo-Croat also has genitive subjects in some affirmative sentences; namely, in the same ones that have genitive under negation. (A-2) NEG/GEN [Na stolu] nije bilo (*raznih) knjiga i cˇasopisa. On table neg-was-n.sg (various-gen.pl) books-gen and magazines-gen ‘On the table there weren’t any books and magazines.’ Note: the impossibility of the added adjective holds also for Russian, and for English sentences with ‘‘any’’ under negation. (A-3) AF/GEN : *Knjiga i cˇasopisa je bilo na stolu. books-gen and magazines-gen was-n.sg on table Note: in this word order genitive is impossible in an affirmative sentence. (A-4) AF/NOM [Knjige i cˇasopisi] su bili na stolu (a ne na podu). books-nom and magazines-nom were-pl on table (and not on floor) ‘(The) books and magazines were on the table (and not on the floor.)’ (A-5) NEG/GEN Knjiga i cˇasopisa nije bilo [na stolu]. Books-gen and magazines-gen neg-was-n.sg on table ‘There weren’t any books and magazines on the table.’ Note: Unlike in A-4, this word order is OK with genitive in a negative sentence. (A-6) NEG/NOM [Knjige i cˇasopisi] nisu bili na stolu, vec na podu. Books-nom and magazines-nom neg-were-pl on table (but on floor) ‘The books and magazines were not on the table (but on the floor.)’ (A-7) NEG/GEN *Knjiga i cˇasopisa nije bilo na stolu, vec na podu. Books-gen and magazines-gen neg-was-n.sg on table but on floor

The Russian genitive of negation 

The last sentence is impossible. In (A-6), [knjige-nom i cˇasopisi-nom] is Theme; but in (A-7), [knjiga-gen i cˇasopisa-gen] can’t be Theme but it is Rheme — because you can’t contrast two locations in (A-7) as you can in (A-6). So here in (A-7) you have still the Theme [na stolu], which, naturally, can’t be contrasted. It follows that in (A-5), [Knjiga-gen i c’’asopisa-gen] is a Rheme,1 even though it is sentence-initial. And even when you have this sentence in the following text, similar to the kefir example, the same holds, even if the subject is pronominalized. (A-8) (Trazila sam svoje knjige i cˇasopise.) ‘I was searching for my books and magazines.’ Ali mojih knjiga i cˇasopisa (or: Ali njih) nije bilo but my-gen.pl books-gen and magazines-gen (or: but they-gen) neg-was-n.sg na stolu. on table. ‘But my books and magazines (or: But they) were not on the table. As in Babby’s analysis of similar Russian cases, ‘‘my books and magazines’’, or ‘‘they’’, while already introduced and hence not ‘‘new’’, is still a Rheme.

Notes * We are grateful for valuable discussions and comments to Tijana Ašic´, Leonard Babby, John Bailyn, Michael Brody, Wayles Browne, Maria Nella Carminati, Catherine Chvany, Bernard Comrie, Östen Dahl, Nomi Erteschik-Shir, Hana Filip, Kai von Fintel, Jan Firbas, Eva Hajicˇová, Uwe Junghanns, Hans Kamp, Knud Lambrecht, Fred Landman, Elena Paducˇeva, Jaroslav Peregrin, Asya Pereltsvaig, David Perlmutter, Maria Polinsky, Ellen Prince, Ekaterina Rakhilina, Petr Sgall, Andreas Spaeth, Anita Steube, Leonard Talmy, Vladimir Turovsky, Tanya Yanko, and Annie Zaenen; our gratitude does not imply their agreement with our conclusions, for which we take full responsibility. We are also grateful to audiences and conference co-participants at FASL 6, the ‘‘Bridges and Interfaces’’ conference in Prague 1998, the spring 1998 Mathesius Institute, the Max Planck Institute for Evolutionary Anthropology, the University of Leipzig, the Leipzig conference in honor of Anita Steube, IATL 2000 in Tel Aviv, the Third Workshop on Text Structure at the University of Texas, Austin, and the University of Connecticut Cognitive Science Program. This work was supported in part by a grant to the first author from the Russian Foundation for Basic Research, Project No. 96–06–80315a, in part by a grant from the National Academy of Sciences in 1997 to the second author under the program for Collaboration in Basic Science and Engineering to host the first author for collaboration on the project ‘‘Towards an Integration of Formal and Lexical Semantics: Meaning Postulates and Fine-grained Sortal Structures’’, and in part by the National Science Foundation under Grant No. BCS-9905748 to the second author for the collaborative project ‘‘Integration of Lexical & Compositional Semantics: Genitives in English and Russian’’. .

Pis’mo ne prišlo – Pis’ma ne prišlo. letter-nom.n.sg neg came-n.sg – Letter-gen.n.sg neg came-n.sg ‘The letter didn’t come.’ ‘No letter came.’

 Vladimir Borschev and Barbara H. Partee

In glossing our examples, we use the following abbreviations: acc=accusative; f=feminine; gen=genitive; m=masculine; n=neuter; nom=nominative; pl=plural; sg singular; 1=first person; 2=second person; 3=third person. We use boldface to highlight the relevant occurrences of nom and gen on nouns and n.sg on non-agreeing verbs. We do not gloss irrelevant morphology. . Perlmutter and Moore (1999) consider even the affirmative counterparts of these sentences, where the ‘‘subject’’ is necessarily nominative, to be impersonal constructions; most linguists do not. See Section 6.2. . In Section 6 we discuss two debated issues concerning these ‘‘subjects’’, the Unaccusative Hypothesis according to which they are all underlying objects (Section 6.1), and the question of whether NESs are impersonal (Section 6.2). In the meantime, we will simply use the term ‘‘subject’’ as the simplest label for the NPs we are discussing. . Examples (3) and (4) are from Ickovicˇ (1974) (cited in Babby 1980), and example (5) from Babby (1980). We note that after many years of work on GenNeg, a large collection of examples has been accumulated and has in a sense become common property. These examples, which highlight various aspects of the construction, are analyzed in many works. We largely take our examples from this collection (sometimes with variations), drawing principally on examples cited by Ickovicˇ (1974), Babby (1980), Apresjan (1980, 1985), and Paducˇeva (1992, 1997). . This is Babby’s translation (Babby 1980: 59); in most cases he translates nominative subjects in NDSs with the definite article, and we believe a definite article would be appropriate in this case as well. . The status of examples like (5a) as ‘‘ungrammatical (*)’’or ‘‘infelicitous or semantically anomalous without a special context (#)’’is a theory-dependent matter that does not have a straightforward answer. Examples (5a) and (6b) are very hard to improve by manipulating the context; other ‘‘bad’’ examples are much easier to improve by that means. . Pesetsky (1982) argued in favor the hypothesis that Unaccusativity is a crucial part of the explanation of GenNeg, as do Perlmutter and Moore (1999) in the framework of Relational Grammar. Babby (2000) argues against the hypothesis. Most Western Slavists accept the hypothesis, so that behavior with respect to GenNeg is frequently invoked as a principal criterion for classifying Slavic verbs as unaccusative or not. . This fact was brought to our attention by our colleagues Tanya Yanko, Elena Paducˇeva, and Sandro Kodzasov, and was confirmed by Jan Firbas and our Prague colleagues Eva Hajicˇová and Petr Sgall. Babby considered and rejected this ‘‘fact’’ in his 1980 book; we discuss his response below. . The most common term in Russian corresponding to Babby’s ‘‘existential sentences’’ is bytijnye predloženija; see Arutjunova (1976, 1997), Arutjunova and Širjaev (1983). The contrasting term ‘‘declarative’’ is not widespread in the English-language literature, and is perhaps not the best choice. The Russian alternative preferred by Arutjunova and Paducˇeva is ‘‘sobytijnye’’ predloženija, ‘eventive sentences’. . Cf. the discussion of these examples in Apresjan (1985): ‘‘In contemporary terminology in cases of the type of Stoka talyx vod ne nabljudalos’ [=(3b)], Moroza ne cˇuvstvovalos’

The Russian genitive of negation 

[=(4b)], there is no presupposition and one asserts the nonexistence or complete absence of the object spoken about: there was no runoff of melted snow at all, there was no frost. In contrast, the sentences Stok talyx vod ne nabljudalsja [=(3a)], Moroz ne cˇuvstvovalsja [=(4a)] most often carry the presupposition of existence of the object spoken about and assert only that it did not carry out the given action or was not in the given state: some runoff of melted snow in principle took place, but no observations were made of it; there was frost, but it could not be felt (because, for instance, everyone was warmly dressed).’’ (Apresjan 1985: 295–6.) . Narrow, or contrastive, ‘‘constituent negation’’ is a separate story; we leave it aside, since it does not trigger GenNeg effects. (See Babby 1980: 104–8.) However, there is room for debate about the status of the negation in a number of the examples here and in the literature, particularly examples with final ne ‘neg’ + Verb, especially those with final ne ‘neg’ + byt’ ‘be’. Examples of the last sort in this paper include (5, 9, 10, 11 (=29) , 13, 14, the right-hand sides of 16a–d, 40, 41a–e, 44 , 45, 46). This class of examples is discussed in Sections 4.2.2 and 4.2.3. . Babby notes that an existential sentence may have an additional Locative constituent, which if present will be outside the Rheme. One of our modifications of Babby’s work is to give the Locative a more important role; see below. . The claim that existential be-sentences always have an obligatory Locative argument is also made by Yokoyama (1986), Kondrashova (1996), Comorovski (1995). Chvany (1975) emphasizes that no such argument is syntactically obligatory, while Kondrashova (1996) goes so far as to make the location the subject at her level of NP-structure. . In Section 4.2.2. we mention a proposal by Erteschik-Shir according to which the sentences (9–12) are indeed Rheme-only as claimed by Babby, and kefira and the like have the status of a subordinate Topic (Theme) inside the main Focus (Rheme). This ‘‘subordinate Theme’’ could correspond to Babby’s ‘‘old information’’. . The ni ‘not’ of ni odin, ni odnogo ‘not one’ always (except in certain constructions that are irrelevant here) requires the co-presence of sentential negation ne in its clause. Russian is a ‘‘Negative Concord’’ language, and expressions like ni odin ‘not one’ and nikto ‘no one’ are generally analyzed as negative polarity items of a special kind (see Brown 1999). We therefore gloss ni simply as ‘NI’ rather than translating it as ‘not’. . We return to examples of this kind, including some noted by Babby (1980), in Section 5. . ‘‘The lexical verb in NESs can always be replaced by byt’ or one of its functional equivalents without affecting the sentence’s primary assertion.’’ (Babby 1980: 17) In Babby’s defense, it should be mentioned that byt’, as an exceedingly common verb, could be expected to tolerate irregularities. Wayles Browne has pointed out to us that in Serbo-Croatian, there is no object GenNeg, and subject GenNeg happens only with the (irregular) ‘‘there is’’ verb; and that in Polish and Slovenian, object GenNeg is completely obligatory (no semantic conditions), whereas subject GenNeg happens only with the ‘‘there is’’ verb. Thus not only do subject and object GenNeg not pattern together in all Slavic languages, but some ‘be’ verbs have special properties with respect to GenNeg in several Slavic languages.

 Vladimir Borschev and Barbara H. Partee

. We sometimes use the shortened expression ‘genitive construction’ for ‘GenNeg in alternation with nominative’, since none of the other genitive constructions of Russian are discussed in this paper. Similarly, in this paper, whenever we say GenNeg, we mean GenNeg in alternation with nominative unless we explicitly state otherwise; we do not discuss the relation of this construction to GenNeg in alternation with accusative except in Section 6.1 in our brief discussion of the Unaccusative hypothesis. . Not all of the main predicates in existential sentences are tensed verbs; some are participles, and some are impersonal predicates like nužno ‘necessary’. In the past or future they would be accompanied by a tensed form of the copula byt’ ‘be’, which is phonologically null in the present tense in both affirmative and negative sentences. This copula byt’ is distinguished by Chvany (1975) from the existential verb byt’, which she treats as a real lexical verb occurring in both locative and existential sentences, which differ for her in whether the NP argument ends up as surface subject or not. This lexical verb byt’ has the special present tense negative form net ‘(there) is not’ in existential sentences. In present tense affirmative sentences, it is either null or has the form est’ (an irregular and invariant form which occurs with all persons and numbers); the distribution of est’ is a big topic which we will not touch. In the past and future, the forms of copula byt’ and lexical byt’ are identical. . Most of the verbs which end with the suffix -sja are ‘‘reflexive passives’’, sometimes ‘‘middles’’, but we will simply refer to them here as passives. . The Russian tolkovanie is hard to translate; it is somewhere between ‘explication’ and ‘definition’, and in practice is an extensive decompositional semantic analysis of a given lexical item. . This is of course a simplification, ignoring the question of types (among other things). We have in mind the Russian tradition of representing the meaning of a verb as a conjunction of propositions, each containing free variables standing for the arguments of the verb. Something resembling a formal semantic (but highly ‘decompositional’) lexical entry could then be obtained by lambda abstraction over the given variables. . The term ‘‘implications’’ in this sense is not familiar in formal semantics. To illustrate the distinctions, Paducˇeva (1997: 104) gives a sample lexical entry for the verb postroit’ ‘build.perfective’, as follows (we translate into English): ‘‘Y built X from Z’’ = (i) before t, X did not exist (presupposition); (ii) at t Y acted with a goal: in a definite manner: affected Z (assertion); (iii) this had as a result: now X exists (implication). The notion of implication here may be related to the notion of ‘‘resultant state’’ of an accomplishment verb, but we will not follow up on these points here. . Example (17a) is example (3¹b) in Paducˇeva (1992: 55); example (17b) is our addition based on her statements there. . The subject NP in (17a), although genitive, is obligatorily interpreted as definite. It is presumably for this reason that Paducˇeva classifies this sentence as locative rather than existential and byt’1 as a locative rather than existential verb. We discuss the definiteness issue in connection with examples (39c–d) in Section 4.2.3.

The Russian genitive of negation 

. The linguistic notion of an observer is looked at in detail in the works of Ju. D. Apresjan, in particular in Apresjan (1986). . The point of view of an implicit observer also plays a role in Fillmore’s analysis of English come and go (Fillmore 1971), and in a number of insightful works by Carlota Smith and by Paducˇeva about the manipulation by authors of fiction of the points of observation from which the reader is ‘shown’ the unfolding scene (see, for example, Smith and Whitaker 1985; Paducˇeva 1994, 1996.) . In Section 6.1 we briefly discuss the Unaccusative hypothesis, according to which GenNeg with intransitive verbs affects only Unaccusative verbs, whose subject is an underlying direct object, and which are necessarily non-agentive. The use of nominative in a negative intransitive sentence, on that approach, signals that the subject is an underlying subject, i.e. that the verb is ‘‘Unergative’’; and such verbs are typically interpreted as Agentive. On that view, there are indeed different verbs used with nominative and genitive, and the fact that for many verbs there appears to be a choice would have to be recast in terms of lexical ambiguity, perhaps with some regular sense-shifting rules. . As noted earlier, Babby (like Arutjunova) makes an exception in the case of genitive sentences with the verb byt’ and a definite subject, which he classifies as ‘‘locative’’ rather than existential. . We use this term in a broad sense; we do not rule out base generation. The conditions may be of various sorts, including conditions on possible interpretations. . While for us, both THING and LOC are in the first instance elements of models, we often use the same terms to identify the corresponding expressions. On the one hand, this may be slightly careless terminology, but on the other hand, it corresponds to the use of a linguistic term such as Agent, which is regularly used to identify both a role of a participant in an event and the expression in a sentence that denotes the participant that plays that role. Our Russian terms for these two participants are VEŠCˇ’ ‘thing’ and MESTO ‘place’. . Principles (21, 22, 25, 26) are from Borschev and Partee (1998a) with little change. In that work, Principle (23) was stated in terms of Theme–Rheme structure, so in that work Principle (24) was not needed. . The statement as given is evidently too strong. It is meant as a first approximation, to be amended to take account of the modal and aspectual modifications of basic existential meanings discussed in Section 2. . We see the openness of the class of ‘‘genitive verbs’’ as resulting from the possibility of such coercion. Even an agentive verb like pet’ ‘sing’, used in examples (6a–b) as a case where GenNeg is impossible, can sometimes be coerced into a ‘genitive verb’ with the help of a very strong context. We plan to discuss this issue in future work. . Descartes’ statement, which is rendered in English as I think; therefore I am is rendered in Russian as Ja myslju; sledovatel’no ja sušcˇestvuju ‘I think; therefore I exist’. There may be occasional stylistically marked intransitive uses of byt’ ‘be’ with no complement to assert existence, but if it is possible at all (and it may not be) then only with proper names or other definite subjects. . Erteschik-Shir (1997) speaks of ‘‘stage-topics’’, and we like the concept of a stage on

 Vladimir Borschev and Barbara H. Partee

which the events described in a sentence take place. There is a problem with the already existing usage of the English word stage in a different sense, in stage-level vs. individual-level predicates (Carlson 1977; Kratzer 1995). Russian scena unambiguously conveys the sense of a stage on which a performance takes place, but its English cognate scene unfortunately has a much more static meaning, ‘what one sees before one, or what is depicted in a picture.’ So in Russian we could use SCENA alternatively with MESTO, but in English we will continue to use LOC. . Cf. Barwise and Perry’s (1981) notion of a ‘‘Resource Location’’. And cf. examples (28a–b) above, which illustrated the similar possibility of NDSs with denials of existence in one world (a LOCation) of entities presupposed to exist in another world. . Some conditions probably need to be placed on the choice of verb as well; there may be dynamic sentences that may also have the structure V(THING, LOC) with LOC as Perspectival Center that should not be classed as existential. However, we cannot simply require that the Presupposed Equivalence (26) hold, since that is a requirement only on NESs and not on AESs; and it is a requirement on the interpretation of the sentence in context, not on the verb. At this point the most we can say is that an affirmative sentence with semantic structure V(THING, LOC) and the LOC as Perspectival Center is existential if its ‘‘corresponding negative sentence’’ meets (26) in the given context. . Examples of this kind are also given by Paducˇeva to illustrate her distinction between ‘‘retrospective Observer’’ (in (31)) and ‘‘synchronous Observer’’ (in (32)); as we note below in Section 4.2.4, her synchronous Observer corresponds to the cases in which for us the LOCation is the Perspectival Center. . We have given these two sentences the same translations, because the difference felt between them by a native speaker of Russian, described in the paragraph preceding the examples, does not easily translate into English (see Chvany 1975: 157–8). Existential sentences with indefinite subjects usually translate into there-sentences in English, but that option is not normally possible with definite subjects. . The Serbo-Croatian data in the Appendix offers some potential support but also some problems for Babby’s claim. The data indicate that the LOC in a sentence with a genitive subject is still Theme, even if the subject is sentence-initial, indirectly supporting Babby’s claim that the subject is Rheme. But the word order restrictions in Serbo-Croatian undermine one of Babby’s arguments about Russian NESs with initial subject: see Note 62. . This suggestion was made to us for this example by Petr Sgall (personal communication). For a discussion of verum-focus in general, see Höhle (1992). . It is irrelevant to the main argument whether v magazine ‘in the store’ is part of the subordinate focus, part of the subordinate topic, or a modifier of the stage-topic (another suggestion made to us by Nomi Erteschik-Shir). What’s crucial is that the main topic is the Stage-topic, and kefir is part of the main focus (Rheme). . This parameter is not entirely independent of the other two. In the affirmative sentences, it is determined by thVLoc vs. LocVTh word order (assuming neutral intonation); hence there are only six sets, not eight. In the negative sentences, Declarative sentences are those with Nominative case and Existential sentences those with Genitive. In

The Russian genitive of negation 

addition to the parameter decl/Ex, we indicate the sentence types using Babby’s labels ‘ADS’, ‘NDS’, ‘AES’, ‘NES’. . Since our simple sentences have three relevant constituents, there are in principle six different word orders to consider, without counting the position of negation in the negative sentences. We principally consider the two ‘‘main’’ word orders, locVTh and thVLoc. Other word orders do occur, and we will mention two of them in the discussion. But a fuller study of word order, which would only be fruitful if accompanied by a serious study of intonation, is far beyond the scope of this work. See Yokoyama (1986) for a serious study of communicative structure, word order, and intonation in Russian. . The ‘‘doctor’’ examples here are an extension of a set of examples discussed in Brown (1999: 85), who in turn draws on discussion of the same examples in Chvany (1975). The examples discussed by Brown and Chvany are in (42) below, with further discussion of our own. The ‘father at the sea’ examples start from the classic pair (35b, 39b) discussed by Ickovicˇ (1974), Apresjan (1980), Paducˇeva (1992), Borschev and Partee (1998a,b), and others. . As a declarative sentence with neutral word order, this sentence answers a question such as ‘‘Where was the vodka?’’ or ‘‘Where was vodka (generic) available?’’ The sentence can also be read as a topicalized ES; in that case the main stress would be on the verb byla. In the present tense, both versions would be possible with prodaetsja ‘is sold’; but if the verb were just byt’ ‘be’ in the present tense, they would be disambiguated by est’ ‘is’, with main stress, (existential) vs. null (declarative). . Although this sentence seems most naturally interpreted with the sea as the Perspectival Center considered at some particular time or occasion, it’s possible to understand it as an answer to the question ‘‘Who has ever been to the sea?’’; this still has the sea as Perspectival Center, we believe, but without a specific associated time. But see Note 55. . The simple existential translation given represents the most natural reading without any special context. But it is also possible to interpret the sentence analogously to the a. and b. sentences, about a definite doctor, who we perhaps hoped or expected to find at some town-event (here ‘‘in town’’ is understood in an event-like way, like ‘‘at the sea’’ in b., or ‘‘at the conference’’ in a..) It might even be possible to get a similar reading for (d), but it’s harder (i.e. it requires a richer context) to interpret the bare plural studenty ‘students’ as a definite group than to interpret the bare singular doktor ‘doctor’ as a definite individual. . The obligatory definiteness of the subject in this pattern, which we have still labeled NES, is discussed later in this subsection. . Sentence (42a)=(78) from Brown (1999: 85)=(2.2a) from Chvany (1975: 45). Sentence (42b) is Brown’s (79a),=(2.5b) from Chvany (1975: 47). Sentence (42c) is Brown’s (79b), which is a variant (substituting doktor for Ivan) of (4.44a) from Chvany (1975: 157). . But we should ask: Is doktora in (42b) outside the scope of negation? Is it interpreted as definite because it’s outside the scope of negation, or is it scope-irrelevant because it’s definite? If there is to be any hope of preserving the correlation of scope of negation with the occurrence of GenNeg , one would have to be able to argue for the latter option. We say a bit more about this issue in Section 5.

 Vladimir Borschev and Barbara H. Partee

. See the translations of the ‘father at the sea’ examples, (34b)–(39b). . The claim that a LOCation which is the Perspectival Center is always associated with a time is at odds with what we said in Note 49 about the possible translations of (37b), an AES. It is unexpected, on our view, that (37b) allows a reading in which it asserts that Father was at the sea at some time in his life, a temporal specification that is associated with Father and not with the sea. It would fit our claims better if the sentence on that reading turned out to be an instance of an inverted Declarative sentence, with Father as the Perspectival Center. It is possible that the affirmative sentences in (34) and (37) are ambiguous, with the dominant interpretation as given, plus a marked interpretation, probably with a marked intonation, on which the sentences in (34) are inverted versions of those in (37) and vice versa. . These qualifiers are meant to exclude a genuinely attributive or quantificational definite noun phrase from this claim. Such possibilities are relevant to the discussion of possible definite descriptions as GenNeg subjects of NES, as in Babby’s example (i), discussed in Partee (2000). (i) Navodcˇik . . . ždal komandy. No ee ne posledovalo. %gunner . . . awaited command but it-gen-f-sg neg followed-n-sg %‘The gunner waited for the command (to fire). But it didn’t come.’ . Thanks to the students in Partee’s seminar at RGGU in Moscow in spring 2000, especially Masha Frid, Lena Model, Yura Lander, and Zoya Efimova. Data from a subsequent ‘‘Google’’ search of Russian with Latin transliteration: For (50) in three variant word orders, 2 citations; for (51) in three variant word orders, 63 citations. In each of the nominative citations, there is a definite group presupposed in the context, over which the nikto can be understood as quantifying. . The first author and some of our Russian colleagues have some doubts about quite a number of the examples that have nom acc nikakie, though they do not categorically exclude all of them. . Since tam ‘there’ is a definite LOCation, the ‘scope’ of tam, which could structurally vary in (53a) vs. (53b), doesn’t really have any logical import. . There is a brief discussion of some possibly related examples in Brown (1999: 94–8), who notes that various examples of ‘‘attenuated’’ negation, i.e. negation accompanied by various other operators, do allow -nibud’ words. It seems that perhaps when some other ‘‘operator’’ expression is licensing the -nibud’ word, the presence of negation does not block it. So it may be simply that clausemate negation cannot license -nibud’, but can cooccur with it if it is not the closest operator. This calls for further study. . For these data and the accompanying remarks we are indebted to Tijana Ašic´, formerly of the University of Belgrade. Sentences are labelled AF or neg depending on whether they are affirmative or negative, and NOM or GEN according to the case of the subject. The THEME (according to her judgment) is in square brackets. . It is interesting and puzzling that this word order is ungrammatical in the affirmative sentence only; this data would seem to argue against Babby’s conjecture that the greater freedom of word order in Russian NESs and NDSs as compared to their affirmative

The Russian genitive of negation 

counterparts is explained by the morphological disambiguation of Theme/Rheme in the negative case. Here we have morphological disambiguation in the affirmative sentences as well, but still less freedom of word order. . Note [BHP and VB]: We now see that this indirect argument that the THING is still Rheme rests on some unexamined assumptions that bear closer examination. In the Russian equivalent of the NES (A-5), with thVLoc word order, it is impossible to contrast either the THING or the LOCation. So if contrast is a test for Rheme, then neither is behaving like a proper Rheme in these sentences. Further investigation is needed.

References Abels, K. 2000. The genitive of negation in Russian. Ms., University of Connecticut. Apresjan, J. D. 1974. Leksicˇeskaja Semantika [Lexical Semantics]. Moscow: Nauka. Apresjan, J. D. 1980. Tipy informacii dlja poverxnostno-semanticˇeskogo komponenta modeli ‘‘Smysl ⇔ Tekst’’ [Types of Information for the Surface-Semantic Component of the Model ‘‘Meaning ⇔ Text’’]. Vienna: Wiener Slavistischer Almanach, Sonderband 1, and Moscow: Škola ‘‘Jazyki Russkoj Kultury.’’ Apresjan, J. D. 1985. ‘‘Syntaksicˇeskie priznaki leksem.’’ [Syntactic features of lexemes]. Russian Linguistics 9: 289–317. Apresjan, J. D. 1986. ‘‘Deiksis v grammatike i leksike i naivnaja model’ mira.’’ [Deixis in grammar and lexicon and the naïve model of the world]. Semiotika i Informatika 28:5–33. Arutjunova, N. D. 1976. Predloženie i ego smysl [The Sentence and its Meaning]. Moscow: Nauka. Arutjunova, N. D. and Širjaev, E. N. 1983. Russkoe predloženie. Bytijnyj tip (struktura i znacˇenie). [The Russian Sentence. The Existential Type (Structure and Meaning)]. Moscow: Russkij jazyk. Arutjunova, N. D. 1997. Bytijnye predloženija [Existential Sentences]. Enciklopedija ‘‘Russkij jazyk.’’ Moscow: Bol’š aja Rossijskaja Enciklopedija, 57–9. Babby, L. 1980. Existential Sentences and Negation in Russian. Ann Arbor, Michigan: Karoma Publishers. Babby, L. ms. 2000. The genitive of negation and unaccusativity. Ms., Princeton University. Babyonyshev, M.A. 1996. Structural Connections in Syntax and Processing: Studies in Russian and Japanese. Ph.D. dissertation, MIT. Distributed by MIT Working Papers in Linguistics, Cambridge, MA. Babyonyshev, M., Fein, R., Ganger, J., Pesetsky, D., and Wexler, K. 1998. The maturation of grammatical principles: evidence from Russian unaccusatives. Ms., MIT, Cambridge, MA, Bailyn, J. F. 1997. ‘‘Genitive of negation is obligatory.’’ In Annual Workshop on Formal Approaches to Slavic Linguistics: The Cornell Meeting 1995. (= FASL 4), W. Browne, E. Dornsich, N. Kondrashova, and D. Zec. (eds), 84–114. Ann Arbor, Michigan: Michigan Slavic Publications. Barwise, J. and Perry, J. 1981. ‘‘Situations and attitudes.’’ The Journal of Philosophy 78:668–91.

 Vladimir Borschev and Barbara H. Partee

Borschev, V. and Partee, B. H. 1998a. ‘‘Formal and Lexical Semantics and the Genitive in Negated Existential Sentences in Russian.’’ In Formal Approaches to Slavic Linguistics: The Connecticut Meeting 1997,Ž. Boskovic´, S. Franks and W. Snyder (eds), 75–96. Ann Arbor: Michigan Slavic Publications. Borschev, V. and Partee, B. H. 1998b. ‘‘Bytijnye predloženija i otricanie v russkom jazyke: semantika i kommunikativnaja struktura.’’ [Existential sentences and negation in Russian: semantics and communicative structure]. In Dialogue ‘98: Computational Linguistics and its Applications, A. S. Narin’yani (ed), 173–82. Kazan: Xéter. Bresnan, J. K., Jonni, M. (1989). ‘‘Locative Inversion in Chichewˆa: A Case Study of Factorization in Grammar.’’ Linguistic Inquiry 20: 1–50. Brown, S. 1999. The Syntax of Negation in Russian: A Minimalist Approach. Stanford: CSLI Publications. Calabrese, A. 1992. ‘‘Some remarks on focus and logical structures in Italian.’’ In Harvard Working Papers in Linguistics, vol. I, S. Kuno and H. Thráinsson (eds), 91–127. Cambridge: Harvard University Department of Linguistics. Carlson, G. N. 1977. Reference to Kinds in English. Ph.D. dissertation, University of Massachusetts, Amherst. Published 1980 by Garland Press, New York. Chvany, C. V. 1975. On the Syntax of BE-Sentences in Russian. Cambridge, MA: Slavica. Comorovski, I. 1995. ‘‘On quantifier strength and partitive noun phrases.’’ In Quantification in Natural Languages, E. Bach, E. Jelinek, A. Kratzer and B. H. Partee (eds), 145–77. Dordrecht: Kluwer. Crockett, D. 1976. Agreement in Contemporary Standard Russian. Cambridge, MA: Slavica. Dahl, Ö. 1969. Topic and Comment: A Study in Russian and General Transformational Grammar. (Acta Universitas Gothoburgensis, Slavica Gothoburgensia, 4) Stockholm: Almkvist and Wiksell. Diesing, M. 1992. Indefinites. Cambridge, MA: MIT Press. Erteschik-Shir, N. 1997. The Dynamics of Focus Structure. Cambridge: Cambridge University Press. (= Cambridge Studies in Linguistics 84) Fillmore, C. 1971. Lectures on Deixis. Bloomington: Indiana University Linguistics Club. Givón, T. 1975. ‘‘Negation in language: pragmatics, function, ontology.’’ In Working Papers in Linguistic Universals 18, 59–116. Stanford: Language Universals Project, Department of Linguistics, Stanford University. Guéron, J. 1980. ‘‘On the syntax and semantics of PP extraposition.’’ Linguistic Inquiry 11, 637–78. Gundel, J. 1974. The Role of Topic and Comment in Linguistic Theory. Ph.D. dissertation, University of Texas at Austin. Reproduced by Indiana University Linguistics Club, Bloomington, 1977. Hajicˇová, E. 1973. ‘‘Negation and Topic vs. Comment.’’ Philologia Pragensia 16:81–93. Hajicˇová, E. 1974. ‘‘Meaning, presupposition and allegation.’’ Philologia Pragensia 17:18–25. Hajicˇová, E. 1984. ‘‘On presupposition and allegation.’’ In Contributions to Functional Syntax, Semantics and Language Comprehension, P. Sgall (ed), 99–122. Amsterdam: John Benjamins. Hajicˇová, E., Partee, B. H., and Sgall, P. 1998. Topic–Focus articulation, tripartite structures, and semantic content. Dordrecht: Kluwer.

The Russian genitive of negation 

Höhle, T. 1992. ‘‘Über Verum-Fokus im Deutschen.’’ Informationsstruktur und Grammatik. J. Jacobs (ed), 112–41. Opladen: Westdeutscher Verlag. Ickovicˇ, V. A. 1974. ‘‘Ocˇerki sintaksicˇeskoj normy.’’ [Remarks on the syntactic norm]. In Sintaksis i norma. G. A. Zolotova (ed), 43–106. Moscow: Nauka. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge: MIT Press. Jackendoff, R. 1990. Semantic Structures. Cambridge: MIT Press. Jacobs, J. (ed) 1992. Informationstruktur und Grammatik. (=Linguistische Beiträge, Sonderheft 4.) Junghanns, U. and Zybatow, G. 1997. ‘‘Syntax and information structure of Russian clauses.’’ In FASL 7, The Cornell Meeting, W. Browne, E. Dornisch, N. Kondrashova, and D. Zec (eds), 289–319. Ann Arbor: Michigan Slavic Publications. Kondrashova, N. 1996. The Syntax of Existential Quantification. Ph.D. dissertation, University of Wisconsin, Madison. Kovtunova, I. I. 1976. Sovremennyj russkij jazyk: porjadok slov i aktual’noe cˇlenenie predloženija [Modern Russian: word order and the communicative structure of a sentence]. Moscow: Russkij jazyk. Kratzer, A. 1995. ‘‘Stage-level and individual-level predicates.’’ In The Generic Book, G. N. Carlson and F. J. Pelletier (eds), 125–75. Chicago: University of Chicago Press. Kratzer, A. 1998. ‘‘Scope or pseudo-scope? Are there wide-scope indefinites?’’ In Events in Grammar, S. Rothstein (ed), 163–96. Dordrecht: Kluwer. Krylova, O. A. and Xavronina, S. A. 1976. Word Order in Russian Sentences. Moscow: Russkij jazyk. Kuroda, S.-Y. 1972. ‘‘The categorical and the thetic judgment (evidence from Japanese).’’ Foundations of Language 2, 153–85. Paducˇeva, E. V. 1974. O Semantike Sintaksisa: Materialy k Transformacionnoj Grammatike Russkogo Jazyka [On the Semantics of Syntax: Materials toward the Transformational Grammar of Russian.] Moscow: Nauka. Paducˇeva, E. V. 1992. O ‘‘semanticˇiskeskom podxode k sintaksisu i genitivnom sub’’ekte glagola BYT’.’’ [On the semantic approach to syntax and the genitive subject of the verb BYT’ ‘BE’]. Russian Linguistics 16:53–63. Paducˇeva, E. V. 1994. ‘‘From Puškin to Nabokov: narrator in the position of observer.’’ Russian Literature 36, 353–70. Paducˇeva, E.V. 1996. Semanticˇeskie Issledovanija: Semantika vremeni i vida v russkom jazyke; Semantika Narrativa. [Semantic Studies: The Semantics of Tense and Aspect in Russian; The Semantics of Narrative.] Moscow: Škola ‘‘Jazyki Russkoj Kultury.’’ Paducˇeva, E. V. 1997. ‘‘Roditel’nyj sub’’ekta v otricatel’nom predloženii: sintaksis ili semantika?’’ [Genitive of Subject in a negated sentence: syntax or semantics?]. Voprosy Jazykoznanija 2: 101–16. Paducˇeva, E. V. 2000. ‘‘Definiteness effect: the case of Russian.’’ In Reference and anaphoric relations, K. von Heusinger (ed), 133–46. Dordrecht: Kluwer. Paducˇeva, E. V. In press. ‘‘Ob otraženii kommunikativnyx aspektov semantiki slova v ego tolkovanii.’’ [On the reflection of the communicative aspects of the semantics of a word and its definition]. To appear in Studien zur Syntax und Semantik der slavischen Sprachen: Festschrift for Andrzej Bogusławski, V. Xrakovskij (ed).

 Vladimir Borschev and Barbara H. Partee

Partee, B. H. 2000. ‘‘Topics under negation: But the answer never came.’’ In Johannes Dölling and Thomas Pechmann (eds), Prosodie — Struktur — Interpretation. Linguistische Arbeitsberichte 74, 43–57. Leipzig: University of Leipzig. Peregrin, J. 1995. ‘‘Topic and focus in a formal framework.’’ In Discourse and Meaning: Papers in Honor of Eva Hajicˇová, B. H. Partee and P. Sgall (eds), 235–54. Amsterdam: John Benjamins. Pereltsvaig, A. 1997. ‘‘The genitive of negation and aspect in Russian.’’ Ms., The Hebrew University of Jerusalem/McGill University, Montreal. Perlmutter, D. and Moore, J. 1999. Syntactic universals and language-particular morphology: Russian impersonals. Talk given at the Max-Planck Institute for Evolutionary Anthropology, Leipzig, April 1999. Pesetsky, D. 1982. Paths and Categories. Unpublished Ph.D. dissertation. Cambridge: MIT. Peškovskij, A. M. 1938. Russkij sintaksis v naucˇnom osvešcˇ enii [Russian syntax in a scientific light]. 6nd edition. Moscow: Gosucˇpedgiz. Reinhart, T. 1981. ‘‘Pragmatics and linguistics: an analysis of sentence topics.’’ Philosophica 27, 53–94. Russkaja grammatika (1980 edition) ed. by N. Ju. Švedova. Akademija nauk SSSR, Institut russkogo jazyka. Moscow: Nauka. Sgall, P., Hajicˇová, E. and Panevová, J. 1986. The Meaning of the Sentence in Its Semantic and Pragmatic Aspects. Dordrecht: Reidel and Prague: Academia. Smith, C. and Whitaker, J. 1985. ‘‘Some significant omission: Ellipsis in Flaubert’s ‘Un Coeur Simple.’’’ Language and Style 14, 251–92. ‘‘Determination in German and Russian.’’ Travaux de Cercle Linguistique de Prague (novelle série), v. 3. Prague. Ward, G. and Birner, B. 1995. ‘‘Definiteness and the English existential.’’ Language 71: 722–42. Yanko, T. E. 1998. [No title; notes on Borschev and Partee (1998b)] Ms. Moscow. Yokoyama, O. T. 1986. Discourse and Word Order. Amsterdam: John Benjamins.

Synonymy vs. differentiation of variant syntactic realizations of FSP functions

Libuše Dušková Prague

0. Introduction The present article deals with the second participant in verbal action as it appears on the syntactic level and the level of functional sentence perspective (FSP).1 The interaction between the two levels will be considered where the second participant in verbal action has the FSP function of theme.2 This involves contextual boundness3 of the second participant insofar as context-independent second participants are as a rule rhematic.4 The constructions under consideration include the passive, in which a thematic second participant appears as the subject, active clauses with the second participant as thematic object occupying the regular postverbal position, and active clauses with the second participant constructed as fronted object. Fronted objects moreover raise the question of their relation to the passive, and the cleft sentence in which the second participant appearing as thematic object in the underlying noncleft form is focused. The aim of the present paper is to find out to what extent these structures are interchangeable, and if differentiated, to ascertain the features to which the differentiation is due.

.

Analysis

The following discussion is based on an analysis of three samples of contemporary fiction (see Sources), each providing 500 clauses taken from continuous text, i.e. a total of 1,500 clauses. To eliminate textual diversity, attention was paid only to the authors’ monologue, i.e. direct speech was excluded on the ground of the differences between monologue and dialogue on the one hand, and the different role of intonation in speech and writing on the other. In written language the role of intonation as an indicator of FSP is of minor importance insofar as in this

 Libuše Dušková

medium intonation largely constitutes a concomitant feature of an FSP structure primarily indicated by other means, viz contextual boundness, syntactico-semantic structure and linear arrangement (Firbas 1992: 10–11, 115, Sgall et al. 1986: 3.10– 3.13). In addition, since the representation of some of the structures under discussion is not sufficiently illustrative in my corpus, I also include examples collected variously from other fiction. . The passive A major syntactic realization of the second participant with the FSP function of theme is found in the passive. .. Second participant with the function of theme The passive is usually described as a device serving to achieve basic distribution of communicative dynamism (CD).5 This capacity of the passive partly results from the fact that the subject, the second participant, has the FSP function of theme. Equally important is the postverbal section of a passive clause, which usually contains the rheme. In studies of the FSP role of the passive the function of the rheme is mostly described in connection with the by-agent (cf. Dušková 1971), the first participant in verbal action, as in ex (1). (1) British art has constantly been changed and enriched by contact with European art. However, the function of the rheme in a passive clause is not confined to the by-agent. The rheme can be realized by any context-independent postverbal element, except scene-setting adverbials. In (2) the postverbal adverbial is an adjunct of purpose, in (3) an adjunct of manner, and in (4) the object of a ditransitive passive verb. (2) The sacrifice has been made for nothing. (3) The family were now seen to advantage. (4) We may yet be spared undue publicity. Scene-setting adverbials,6 whether contextually bound or not, are thematic, as in active clauses, cf. the temporal setting in (5) and the contextually bound prepositional phrase in (6). (5) He got stuck in the underground last Monday. (6) This was accepted between them. (Adams 18) In the absence of postverbal elements, the rheme is usually the verb, as in (7). (7) All our good intentions were forgotten.

Variant syntactic realizations of FSP functions 

Statistical counts (Dušková 1971; Hedviková 1996) show that passives with by-agents or quasi-agents7 almost invariably display the basic distribution of CD, and this is precisely why the passive is described as a special device of FSP (cf. Mathesius 1975: 101): it serves to rearrange the information structure of the corresponding active clause containing a context-independent subject and a contextually bound object so as to achieve basic distribution of CD. .. Passives with thematic by-agents However, as shown by ex (8), passives with a thematic by-agent are not entirely ruled out. (8) I am so terribly upset by all this. Here the by-agent is contextually bound and the rheme is constituted by the verb with its intensification. Two points should be noted in this case: first, it is an example taken from speech, which ensures recourse to intonation. And secondly, upset is one of the attitudinal verbs which are found more frequently in the passive than in the active, the passive being moreover stative, which endows it with adjectival features. Another counterexample to the FSP function of the passive is provided by passives with rhematic subjects. Admittedly, they are rare (3 examples out of 127, i.e. 2.4% in my material). A characteristic type of a passive clause with a rhematic subject is illustrated by (9). (9) Near one of the posts, a hammer and a few nails had been left behind. (Adams 19) The rhematic function of the subject is due not only to its context-independence, but also to the semantics of the verb, which implies existence/appearance on the scene, and to contextual boundness of the adverbial element. That is, the passive in this case implements the presentation scale.8 .. The passive of ditransitive verbs The FSP function of the passive is also demonstrated by the passive of ditransitive verbs, which normally serves to achieve basic distribution of CD: whichever object in the active is less dynamic becomes the subject of the passive, the more dynamic object preserving its object function, as in (4) and (10). (10) The ring was left to the eldest daughter. In (4) the theme is the indirect object of the active, whereas in (10) the theme is constituted by the direct object of the active. However, as shown by ex (11), this is not always the case.

 Libuše Dušková

(11) Strength was given this opposing theory by the meteorological report stating the velocity of the wind. Here the only contextually bound element is the indirect object, which could be constructed as the subject. The context independent subject (direct object of the active) would then stand in postverbal position before the rheme proper,9 implemented by the by-agent. Cf. (11¹). (11¹) This opposing theory was given strength by the meteorological report stating the velocity of the wind. In (11) the initial position of a context-independent element gives this element more prominence than the position consistent with the basic distribution of CD, in a similar way as in the case of fronted object (see 1.3). . Active clauses with thematic objects Since one of the FSP functions of the passive is to achieve initial placement of a thematic element, the object of the corresponding active, it is legitimate to inquire into active clauses with thematic objects, and consider them with respect to passivization. Statistical data show these two realization structures to be about equally frequent: in my corpus of 1,500 clauses the passive was represented by 127 and the active with thematic objects by 136 instances, i.e. 8.5% and 9%, respectively. Expectedly, not all active clauses containing a transitive verb and an object lend themselves to passivization. Apart from structural and semantic restrictions, a constraint arises from the FSP structure itself. Significantly, an active clause with a thematic object passivizes primarily where the subject is the rheme. In most active clauses with a thematic object, however, the subject is also contextually bound, often operating as the theme proper.10 This is reflected in its form of realization. In my corpus half of the subjects of finite clauses with thematic objects are personal pronouns (45 out of 89). Passivization of such clauses would result in sentences with a thematic by-agent, which, as we have seen, are very rare, precisely because what the passive primarily serves to achieve is final or postverbal placement of the rheme, the agent in the syntactic function of adjunct. Where the subject is thematic, the motivation for constructing the first participant as a postverbal clause element is thus lacking. Accordingly, passivization of clauses containing both thematic subject and thematic object is as a rule applicable only if the agent can be suppressed, which is usually ruled out by the semantic relevance of the agent. Hence passivization of sentences like (12) and (13), which require expression of the agent, is conceivable only under special contextual conditions. (12) he greeted them politely11 (Adams 23)

Variant syntactic realizations of FSP functions 

(12¹) they were greeted politely by him (13) they followed us all the same (Raverat 129) (13¹) we were followed by them all the same Passivization is even less likely where the subject is pronominal and the object, though contextually bound, is a noun phrase, the rheme being constituted by a final adverbial, as in (14). (14) She was evaluating the interview in the light of subsequent events. (James 352) (14¹) The interview was being evaluated by her in the light of subsequent events. On the other hand, passivization is more feasible where the subject, though contextually bound, is a noun phrase and the object a personal pronoun. It is again the verb that constitutes the rheme, but the subject is diathematic,12 i.e. it is more dynamic than the object, as in (15). (15) Her initial testiness had surprised him (James 350) (15¹) He was surprised by her initial testiness. While (15¹) is presumably more acceptable than either (13¹) or (14¹), owing to the respective forms of realization of the first and the second participant, ex (16) demonstrates an even more important aspect, which presumably also asserts itself in the preceding examples of this group (12)–(15). (16) [‘‘Are you all right, Blackie? I mean, do you want someone to go home with you?’’]13 The thought appalled Blackie. (James 195) The subject resumes the most activated semantic content, expressed in the immediately preceding clause, whereas the object refers to an element further to the left in the text. Both noun phrases are again contextually bound, and their form of realization is equally weighty. Their respective positions in the sentence appear to be due to the position of the sentence in the text, specifically to the linear arrangement of the referents of the two noun phrases in the preceding stretch of text. It may be concluded that active clauses with thematic objects do not in fact constitute an alternative realization of passive clauses with expressed by-agent insofar as they lack the second relevant feature of transitive active clauses liable to passivize, viz a rhematic subject. Even in the case of the most likely candidates for interchangeability illustrated by exx (15) and (16), the choice between the two structures is subject to textual factors.

 Libuše Dušková

. Fronted objects Although the regular position of thematic objects appears to be the postverbal, there are also thematic objects in initial position, as a result of fronting. These should again be considered with respect to passivization, and obviously also with respect to postverbal thematic objects. Fronted objects are not only thematic, but also rhematic. These two structures should be distinguished because they are motivated by different factors. Consider exx (17) and (18). (17) a.

[Pen would have liked to have been present at the rendezvous,] but this offer Piers declined. b. [A.: ‘‘I am the most selfish creature alive. I never do anything to please anyone but myself.’’] B.:‘‘That I know to be untrue.’’ c. [A.: ‘‘There are things in my life you don’t know anything about.’’ B.: ‘‘Good God, I should hope there were! I’ve only known you for a month.’’] A.: ‘‘And some of them you wouldn’t like.’’

(18) a. Wonderful memory your grandmother has, Miss Smith. b. Damn silly names these horses have! c. A lot of trust I get around this house, don’t I? Whereas the examples under (17) illustrate fronted objects that are thematic, the preceding context showing them to be contextually bound, the examples in (18) contain context-independent objects constituting the rheme. Both structures are variants of postverbal objects, as illustrated by (17¹) and (18¹). (17¹) a. . . . but Piers declined this offer. b. . . . I know that to be untrue. c. . . . And you wouldn’t like some of them. (18¹) a. Your grandmother has a wonderful memory, Miss Smith. b. These horses have damn silly names! c. I get a lot of trust around this house, don’t I? Neither in (17¹) nor in (18¹) does the change in the position of the object affect the FSP function of the object. The factor determining the FSP function of the object regardless of its position is its contextual boundness in (17): accordingly, it is thematic; and the context-independence in (18): it is hence the rheme. Fronted objects obviously represent a marked variant which shares one feature in both instances, viz initial placement makes both thematic and rhematic objects more prominent. Otherwise, the markedness of the two types is of a different kind.

Variant syntactic realizations of FSP functions 

Fronting of a thematic object not only makes the object more prominent, but also allows the rheme to occupy the final position. It thus contributes to basic distribution of CD. At the same time it also serves as a connective link with what precedes — this is amply demostrated in all three examples under (17). Fronted thematic objects thus have not only intrasentential, but also suprasentential function, as a means of cohesion. A fronted rhematic object, on the other hand, has only intrasentential function. Initial placement makes it not only more emphatic than it would be in its regular postverbal position, but also indicates emotive colouring, expression of an attitude. The emotive aspect of a fronted rhematic object manifests itself in concomitant features such as intensifying imprecations (damn), and irony (in (18c) a lot is used in the opposite sense). Unlike fronting of a thematic object, fronting of a rhematic object results in a deviation from the basic distribution of CD, viz the rheme precedes the theme. This again testifies to the emotive nature of fronted rhematic objects. Evidently, it is only the fronted thematic object that can be considered as a variant of the structures discussed so far, i.e. the passive, and thematic object in postverbal position. As regards the relation between the passive and an active clause with a fronted thematic object, part of what has been said about thematic objects in postverbal position applies even here. The subject is again thematic, but in contrast to clauses with postverbal thematic objects in which it can constitute either the theme proper or the diatheme, with a fronted thematic object it can only serve as the theme proper since the diatheme is implemented by to the fronted object. Apart from the constraints imposed on passivization by the necessity to express the agent, whose FSP function is different in active clauses with thematic objects and in clauses underlying the passives with a by-agent, there is also a difference in the FSP function of the fronted thematic object and that of the subject of the passive. As has been explained, fronted objects are diathematic, whereas the subject in the passive largely constitutes the theme proper. Accordingly, active clauses with fronted thematic object do not provide a variant of the FSP structure of the passive. Nor do sentences with fronted thematic objects provide a variant of active clauses with postverbal thematic objects. As has been pointed out, fronted thematic objects have a distinctly cohesive function. In contrast to other contextually bound elements, which are also inherently cohesive, the cohesive function of fronted thematic objects consists in ‘‘providing direct linkage with what has preceded’’ (Quirk et al.: 18.20). . The cleft sentence Fronted object should also be considered in respect of its relation to the cleft

 Libuše Dušková

sentence since both fronting and the cleft sentence serve to give prominence to a clause element. As appears from previous studies of the cleft constructions (Dušková 1993; Dvorˇáková 1988), it-clefts focus both thematic and rhematic elements of the underlying noncleft clause. In view of the point under discussion, only it-clefts focusing thematic elements will be taken into account. Focusing an element that is thematic in the underlying non-cleft sentence appears to be the primary function of it-clefts. Specifically, it-clefts most frequently highlight the subject (almost 50%, see Dvorˇáková 1988: 35–46) and scene-setting adverbials (30%), i.e. elements which appear initially and are mostly context-dependent. Focused objects are the least frequent of the three syntactic functions: they account for 21%. Though thematic in the underlying noncleft sentence, in the it-cleft all these elements operate as the rheme, since the subordinate clause of the it-cleft typically presents its content as known, and hence presupposed. Considering these facts, as well as the fact that fronted objects constitute the FSP function of diatheme, not that of the rheme, commutability between a fronted thematic object and an object focused by means of an it-cleft appears to be unlikely. This is confirmed by the cleft forms of example (17): (17¹a–c). (17¹) a.

Pen would have liked to be present at the rendevous, * but it was this offer that Piers declined. b. A: I am the most selfish creature alive. I never do anything to please anyone but myself. B: *It is that which I know to be untrue. c. A: There are things in my life you don’t know anything about. B: Good God, I should hope there were! I’ve only known you for a month! A: *And it is some of them that you wouldn’t like.

As indicated by the asterisks, the cleft forms are textual misfits. The reasons for the misfit are to be sought in the differences in the FSP structure between the forms with fronted object and the it-clefts. In (17) the fronted object constitutes the diatheme, whereas the rheme is implemented by the verb in a. and c., and the object complement in b. On the other hand, in the cleft forms (17¹a–c) he diathematic element of the underlying structure becomes the rheme, and the rheme of the underlying structure is backgrounded in the subordinate clause, which presents it as known. Moreover, the cleft constructions involve certain presuppositions which are absent in the particular contexts, and the absence of the presuppositions is reflected in the forms with fronted objects. Thus the cleft form (17¹a) presupposes that Piers declined some offer, and the only point in question is which offer. On the other hand the only presupposition of (17a) is that an offer had been made. Similarly in (17¹b–c).

Variant syntactic realizations of FSP functions 

Nevertheless, as shown by example (19), the function to give prominence to an element, which fronting and clefting have in common, may occasionally manifest itself in an appropriate context. It is to be noted, however, that the cleft construction in the last clause of (19) does not represent the more common type with given information in the subordinate clause, reflected in its weak stress, but the informative-presupposition cleft with a normally stressed subordinate clause (Prince 1978). (19) when I sent that letter, all I was interested in was a leave of absence. That’s still all I’m interested in. I haven’t given any thought to the matter of a contract, and I don’t think I’m prepared to think about it right now. A leave of absence I asked for, and it’s a leave of absence that I want. Significantly, all elements in the last sentence are contextually bound: the passage is about a leave of absence, and the context preceding the last sentence implies that the speaker has asked for it and wants it. The highlighted elements in both clauses are the verbs. The clause with the fronted object performs its usual function: it serves to give prominence to a contextually bound element, and at the same time leaves the final position to the rheme. The it-cleft achieves information focus by the intonation centre on the last word of the subordinate clause. We may even consider the possibility of interchanging the two constructions. (19¹) a. A leave of absence I asked for, and a leave of absence I want. b. ?It’s a leave of absence that I asked for, and it’s a leave of absence that I want. c. ?It’s a leave of absence that I asked for, and a leave of absence I want. Even though (19) is presumably preferable to (19¹a), the latter impairs neither the FSP structure nor the textual fit. Objections are conceivable on stylistic grounds: repetition of the same structure tends to create a monotonous effect. (19¹b, c), as the question marks indicate, appear to be less acceptable. (19¹b) tends to be interpreted as an it-cleft with a presupposed subordinate clause; hence the only element presented as newsworthy is the focused leave of absence, the other relevant component of both the semantic and the information structure being lost. In (19¹c) the more emphatic form precedes the less emphatic one. As explained above, whereas the fronted object constitutes the diatheme, the focused element, even in the type with new information in the subordinate clause, implements the rheme, at least within its communicative subfield. Hence a focused element is given more prominence than a fronted one. On the other hand the arrangement in (19), displaying the fronted object in the first clause and the cleft in the second, achieves an information climax, which is absent in (19a) and (19b), while (19c) in fact results in an anticlimax.

 Libuše Dušková

.

Conclusions

To conclude, the foregoing discussion appears to have shown that the FSP structure of the different syntactic realizations of the second participant in verbal action with the FSP function of theme is more or less differentiated. Consequently, interchangeability is largely ruled out. Clarification of instances in which it is conceivable calls for further study.

Notes . Functional sentence perspective is defined as the distribution of degrees of communicative dynamism (CD) over the elements of a sentence. In written language it is determined by the interplay of three factors, linear modification, semantic structure, and contextual boundness (Firbas 1992: 10–11). . In the Prague school treatment of FSP, the theme and the rheme are defined as elements carrying, respectively, a low and a high degree of CD (Firbas 1992: 72–3). . Contextual boundness means retrievability of information from the (immediately relevant) preceding context (Firbas 1992: 11). . Cf. Firbas’s discussion of context-independent objects (1992: 42 ff.). . Communicative dynamism (CD) is defined as ‘‘the relative extent to which a linguistic element contributes towards the further development of the communication’’ (Firbas 1992:8); the degree of CD corresponds to ‘‘the variation in communicative value’’ in Quirk et al. (1985:18.3); the degree of CD is also referred to as the information load, or newsworthiness, of a linguistic element in a sentence. By basic distribution of CD is meant a linear arrangement which displays a gradual rise in CD (Firbas 1992:10), i.e. the order Theme–Transition–Rheme. Basic distribution of CD is comparable to the principle of end focus as presented in Quirk et al. (1985:18.3), and with the concept of systemic ordering in Sgall et al. (1986:3.13). . Cf. Firbas’s setting (1992: 49 ff.). . Agents taking lexically determined prepositions other than by (be worried about, interested in, etc.), cf. (Svartvik 1966: 102–4). . For the concept of the presentation scale, see Firbas (1992: 66 ff.). This scale typically contains an active intransitive verb of appearance or existence on the scene with the first participant constituting the rheme. The different syntactic forms of the presentation scale and their differentiation are treated in Dušková (1998). . The most dynamic element in the rhematic section (cf. Firbas 1991: 71–2). . The least dynamic element in the thematic section (cf. Firbas 1992:80–1, and Svoboda 1983). . The examples preserve the punctuation as found in the source, i. e. where the excerpt is a medial clause of a sentence it has no final punctuation mark and the first word is not capitalized.

Variant syntactic realizations of FSP functions 

. The most dynamic element in the thematic section (cf. Svoboda 1981 and 1983). . Where the context is needed, it is added in square brackets.

References Dušková, L. 1971. ‘‘On some functional and stylistic aspects of the passive in present-day English’’. Philologica Pragensia 14: 117–43. Dušková, L. 1993. ‘‘On some syntactic and FSP aspects of the cleft construction in English’’. Acta Universitatis Carolinae — Philologica 1, Prague Studies in English 20: 71–87. Dušková, L. 1998. ‘‘Syntactic forms of the presentation scale and their differentiation’’. Linguistica Pragensia 8: 36–43. Dvorˇáková, B. 1988. Cˇeské proteˇjšky anglické vytýkací konstrukce (Czech counterparts of the English cleft sentence). Diploma dissertation. Prague: Charles University. Firbas, J. 1992. Functional sentence perspective in written and spoken communication. Cambridge University Press. Hedviková, O. 1996. Developmental tendencies in the passive use: comparing the present state of the passive with the situation at the beginning of this century. Diploma dissertation. Prague: Charles University. Mathesius, V. 1975. A functional analysis of present day English on a general linguistic basis. Prague: Academia. Prince, E. F. 1978. ‘‘A comparison of wh-clefts and it-clefts in discourse. Language 54: 883–906. Quirk, R., S. Greenbaum, G. Leech and J. Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. Sgall, P., E. Hajicˇová and J. Panevová. 1986. The meaning of the sentence in its semantic and pragmatic aspects. Dordrecht: Reidel. Svartvik, J. 1966. On voice in the English verb. The Hague: Mouton. Svoboda, A. 1981. Diatheme. Brno: Masaryk University. Svoboda, A. 1983. ‘‘Thematic elements’’. Brno Studies in English 15: 49–85.

Sources Adams R. 1974. Watership Down. Penguin Books. James P. D. 1994. Original Sin. London: Faber and Faber. Raverat G. 1952. Period Piece. London: Faber and Faber.

Topic–Focus articulation as generalized quantification

Jaroslav Peregrin Prague1

Recent results of Partee, Rooth, Krifka and other formal semanticians confirm that topic–focus articulation (TFA) of sentence is relevant for its semantics. The essential import of TFA, which is more apparent in case of a language with relatively free word order such as Czech than in case of English, has been traditionally intensively studied by Czech linguists. In this paper we would like to indicate the possibility of the account for TFA in terms of the theory of generalized quantifiers, drawing on the results of both these groups of theoreticians. The basic intuition which we accept as our point of departure is the intuition of topic as the ‘‘semantic subject’’ and focus as the ‘‘semantic predicate’’; we point out that the role of topic is to specify the entity the sentence is ‘‘about’’ (thereby triggering a presupposition), while that of the focus is to reveal a characterization of this entity, and usually a characterization that is in some sense exhaustive. Then we show that it may be plausible to consider topic and focus as arguments to an implicit generalized quantifier, which may get overridden by an explicit focalizer.

0. A preliminary remark: symbols in semantics There are two principal ways to employ formal or symbolic means to semantically analyze a sentence (more generally an expression or an utterance). The first of the ways consists in using symbols to picture some aspects of the sentence to emphasize or to point out those aspects of its syntactic structure which we consider relevant for its semantics. We may, for example, depict the loves statement John loves Mary as JOHN —→ MARY to emphasize that John is the agent of the action expressed by loves and Mary its objective; or we may indicate the same by writing [John]Ag loves [Mary]Ob. It is the graphical shape of such a representation that is important: it is just the graphics that does the job of explication. We are free to employ symbols of any kind, the only point is to emphasise the relevant, and to suppress the irrelevant, features of the sentence analyzed.

 Jaroslav Peregrin

The other way to use formal means is to do logical analysis, and its nature is essentially different. Its primary aim is not to depict features of the sentence’s structure, but rather to point out the proposition which is to explicate the meaning of the sentence (and to point out this proposition means — in effect — to point out the truth conditions of the sentence, or the contribution the sentence brings to truth conditions of suprasentential wholes, i.e. the ‘‘context-change potential’’ of the sentence). We again use symbolic means: we may, for example use the formula LOVES(JOHN, MARY) of the first-order predicate calculus to schematize John loves Mary. However, now it is not the graphical shape of the symbols which is relevant — we know that the formula we have used is equivalent to many other formulas and it would have been the same if we have used any of them — the proposition pointed out would be the same. The basic difference between the two approaches is that the latter works with closed systems of symbols with strictly defined equivalence — in their case, it is not the formula itself that does the explication, but rather the proposition pointed out by it. And it is also only this approach that facilitates semantic analysis in the strict sense (cf. Lewis 1972).

.

Topic and Focus within the Prague Tradition

Contemporary semantic theories are inclined to view focus as something extraordinary, as something which a sentence may, or may not, have, and which becomes semantically relevant only when it gets combined with a focus-sensitive operator. The Praguian approach, on the other hand, considers TFA as the universal basis of semantic structuring of sentence. Let us briefly recapitulate its basic tenets — for a more detailed exposition see Sgall, Hajicˇová and Panevová (1986) and also Sgall (1995) and Hajicˇová (1995). The syntactic characterization of sentence underlying this approach works with the concept of dependency. The default order of items depending on a verb (i.e. the order of thematic roles and adverbials) is considered to be fixed for a given language; it is called the systemic ordering. However, this order together with the order of other items not depending directly on the main verb is modified in a concrete utterance, so that the resulting ‘‘deep’’ order is that of the communicative dynamism (CD). The CD order of contextually bound items dependent on the same head is determined by the speaker’s discourse strategy rather than by grammar; on the other hand, the CD of the unbound items dependent on the same head is in accordance with the systemic ordering. An item is less dynamic than its head iff the dependent item is bound. The least dynamic element of the sentence constitutes the topic proper. All the contextually bound items depending on the main verb together with all that

Topic–Focus articulation as generalized quantification 

depends on them and together with the main verb if this is contextually bound, constitute the topic of the sentence; the rest of the sentence constitutes the focus. Hence the topic/focus classification is exhaustive: every element of the sentence belongs either to the topic or to the focus.

.

Topic and Focus as ‘‘Subject’’ & ‘‘Predicate’’

The basic idea is that the topic–focus articulation has to do with the dynamism of discourse. A linguistic utterance says — in the prototypical case — something new about something old. There is a subject which the sentence is about, and there is a predicate which is ascribed to the subject. The utterer first points out an element of the common ground shared by the participants of the discourse and then reveals something which he proposed to be added to the common ground. If his utterance is successful, then the new item is added and it can become the starting point of a following utterance. In the prototypical case, this semantic notion of subject and predicate coincides with the syntactic notion — the grammatical subject refers to what the sentence is about, and the grammatical predicate expresses what is said about it. However, in general almost any syntactic part of sentence can act as the semantic subject, leaving the rest of the sentence for the role of semantic predicate. Thus, while under normal condition the sentence (1) expresses the ascription of the property of walking to the individual John, its variant (2) (with the stress on John) seems rather to express something like the ascription of the property ‘‘being instantiated by John’’ to the ‘‘entity’’ walking. John walks JOHN walks

(1) (2)

An instructive way to see topic and focus is to see them just as ‘‘semantic subject’’ and ‘‘semantic predicate’’. Topic is that which the sentence is about, by which it gets anchored to the common ground, and focus brings about the very information the sentence was assembled to convey. What does this mean for logical analysis? The straightforward way to analyze (1) is (1¹); and it seems that to account for the fact that in (2) the roles of John and Walks get exchanged it is enough to employ the mechanism of lambdaabstraction and to use the well-known Montagovian trick of type-raising. (2) thus seems to be analyzable as (2¹). Walk(John) 8f.f(John)(Walks)

(1¹) (2¹)

However, this idea is misguided, because it disregards the very sense of logical

 Jaroslav Peregrin

analysis. The point is that (1¹) and (2¹) are equivalent formulas ((2¹) is straightforwardly lambda-reducible to (1¹)), and hence (2¹) fosters precisely the same analysis as (1¹). To analyze something as a ‘‘semantic subject’’ it is not enough to provide a formula in which it corresponds to the subject — because we can always provide another, equivalent formula in which it would correspond to the predicate. This means that if we want to speak about ‘‘semantic subject’’ and ‘‘semantic predicate’’, we must understand these terms as expressing two different ways of contributing to truth conditions. We cannot simply say that the contribution of the predicate differs from that of the subject in that the truth conditions of the whole sentence result from applying the former to the latter; because we saw it is easy to provide an alternative analysis. The real semantic difference can be seen as stemming from the fact that the subject purports to point out something known. If it fails to do so, the sentence is not false, but rather infelicitous, because it makes the listener wonder what the speaker is talking about. On the other hand, if the subject succeeds in pointing out a known anchor and the predicate fails to be true, the sentence is simply false. In other words, the subject is the matter of presupposing, whereas the predicate is the matter of stating. What kind of logic do we need to account for this? First, we need a logic that is partial: its formulas can possess not only the values T and F, but can also have no truth value at all (i.e. they can be not only true or false, but also ‘‘infelicitous’’). Then we can define what can be called the ‘‘propositional content’’ of an expression (a propositional content of a statement is the statement itself, that of a term is the proposition claiming the existence of the corresponding individual and that of a predicate is the proposition claiming the non-emptiness of the corresponding class); and we can define ‘‘presuppositional’’ predication as predication posing the presupposition associated with the subject. Let us do it — for the sake of simplicity — on the level of extension (everything can be straightforwardly relativized to possible worlds); thus, ¦¦X¦¦ will denote the extension of X (i.e a truth value if X is a sentence). Definition 1. (‘‘propositional content’’ of X) ¦X¦=¦¦X¦¦ if X is a sentence =¦¦∃y.y=X¦¦ if X is a term =¦¦∃y.X(y)¦¦ if X is an unary predicate =T otherwise Definition 2. (‘‘presuppositional’’ predication) ¦¦P{S}¦¦=T iff (¦S¦=T & ¦¦P(S)¦¦ =T) =F iff (¦S¦=T & ¦¦P(S)¦¦ =F) =0 iff ¦S¦=F

Topic–Focus articulation as generalized quantification 

Only now we can give the analyses of (1) and (2) in spirit of the above considerations — viz (1¹¹) and (2¹¹). Walk{John} 8f.f(John){Walk}

(1¹¹) (2¹¹)

These two formulas now express different propositions: although there are no circumstances under which one of them would be true and the other false, there are circumstances under which one is false and the other lacks a truth value altogether (if there is no John, then (2¹¹) yields F while (1¹¹) yields 0; if nobody walks then it is the other way round). If we take the sentence John loves Mary, there are formally nine straightforward ways to draw the boundary between topic and focus; they lead to the analyzes (3a) through (3h) (where x, y are variables for individuals, p for properties of individuals, r for binary relations among individuals, f for properties of propositions and g for properties of properties of propositions).2 John loves Mary 8y.love(y,Mary){John} 8x.love(John,x){Mary} 8p.p(Mary){8 y.love(John,y)} 8p.p(John){8 x.love(x,Mary)} (8f.f){love(John,Mary)} 8g.g(love(John,Mary)){8 f.f} 8r.r(John,Mary){love} 8x,y.love(x,y){John,Mary}

(3) (3a) (3b) (3c) (3d) (3e) (3f) (3g) (3h)

Out of these, (3e), in which the whole sentence is contained in the topic, appears to be ruled out — a sentence seems to have to bring in something new. All the other ways of TFA are acceptable. Moreover, there are other possibilities to draw the boundary, non-aligned to the boundary of words. Thus, to respond to the statement John loved Mary by stating John LOVES Mary means to make only the present tense grammateme of the verb into the focus.3 What is important is that in case of sentences with a generic noun phrase the subject-predicate structuring that is yielded by the TFA determines the scope and thus can mean differences not only regarding felicity conditions, but also regarding truth conditions proper, like in case of (4). Every man loves one woman 8P.P(8x.∃!y.(woman(y)&love(x,y))){8 p. x.(man(x)→p(x))} 8P.P(8y. x.(man(x)→love(x,y))){8 p.∃!y.(woman(y)&p(y))} 8p.∃!y.(woman(y)&p(y)){8 y. x.(man(x)→love(x,y))}

(4) (4a) (4b) (4c)

 Jaroslav Peregrin

8p. x.(man(x)→p(x)){8 x.∃!y.(woman(y)&love(x,y))}

(4d)

Besides this, there is another semantically relevant feature of TFA. The focus not only predicates something of the topic; it appears to tend to exhaust all what there is to be predicated about the topic. Uttering (2), we not only state that walking is instantiated by John, but also that John is the only (or at least ‘‘the principal’’) walker. (2) would be inappropriate in the situation where John would be only one of a plenty of those who walk around. To be able to account for the phenomenon of exhaustiveness, we have to introduce the concept of alternative.4 We assume that for every entity there is a class of alternatives (of the same type), which count as its alternatives. We can introduce a functor ALT such that if e is an entity of the type t, then ALT(e) is a set of entities of type t (containing e). The trivial version of ALT would be such that ALT(e) would be the set of all the entities of the type t. In fact, it is clear that the nature of ALT is pragmatic — what counts as an alternative to a given item and what not is given by the context; and from this point of view it seems to be reasonable to see ALT as something which gets changed by ongoing utterances. This means that a logic which incorporates an adequate version of ALT would have to be dynamic — changing ALT in the way dynamic logic of Groenendijk and Stokhof (1991) changes valuation of discourse markers. Anyway, once we have ALT, we can define the ‘‘exhaustive’’ predication as predication such that the predicate has no alternative which would also hold of the subject, i.e. as follows (where ¦¦X¦¦I means the extension of X under the interpretation I and I[y→Y] denotes the interpretation which is just like I save for the single exception that I(Y)=y. Definition 3. (‘‘exhaustive’’ predication) ¦¦P!(S)¦¦ I =T iff ¦¦P(S)¦¦ I =T & p.[p∈ALT(¦¦P¦¦I)&¦¦ P(S)¦¦ I[p→P]=T→¦¦P¦¦I =p] Combining definitions 2 and 3 we arrive at the concept of ‘‘exhaustive presuppositional’’ predication. Definition 4. (‘‘exhaustive presuppositional’’ predication) ¦¦P!{S}¦¦I =T iff ¦S¦I=T & ¦¦P(S)¦¦ I=T & p.[(p∈ALT(¦¦P¦¦I)&¦¦ P(S)¦¦ I[p→P]=T)→¦¦ P¦¦I=p] = F iff ¦S¦I=T & (¦¦P(S)¦¦ I=F v ∃p.[p∈ALT(¦¦P¦¦I)&¦¦ P(S)¦¦ I[p→P]=T&¦¦P¦¦I ≠ p])) = 0 iff ¦S¦I=F Now we can analyze (3) as (3a¹) through (3h¹). 8y.love(y,Mary)!{John} 8x.love(John,x)!{Mary} 8p.p(Mary)!{8 y.love(John,y)}

(3a¹) (3b¹) (3c¹)

Topic–Focus articulation as generalized quantification 

8p.p(John)!{8 x.love(x,Mary)} (8f.f)!{love(John,Mary)} 8g.g(love(John,Mary))!{8 f.f} 8r.r(John,Mary)!{love} 8x,y.love(x,y)!{John,Mary}

(3d¹) (3e¹) (3f¹) (3g¹) (3h¹)

(3a¹) through (3h¹) differ not only in their felicity conditions, but they may also differ in truth conditions (this depends on how we interpret the operator ALT). If both John and Tom love Mary and if Tom counts as an alternative to John, then (3d¹) is false, whereas (3c¹) may well be true (in the case when Mary is the only person loved by John). However, it seems not to be adequate to take TFA to always amount to this exhaustive version of predication. The extent to which the focus of a given utterance really purports to exhaust all what there is to be said about the topic (relative to context) seems to vary (the heavy stress on focus usually strongly facilitating exhaustiveness), and thus the TFA should be considered to be somewhere in between simple presuppositional and exhaustive presuppositional predication.

.

Introducing the generalized quantifier PRED

The linguistic significance of the theory of generalized quantifiers, as elaborated by van Benthem and others, bears on the idea that the truth value of a sentence can be seen as the matter of a pair of sets being in certain relation.5 This theory seems to be in accordance with certain intuitions regarding topic and focus, namely with the intuition that ‘‘topic specifies a class and focus lists the elements of the class’’. This is in fact an idea underlying the first attempts of semantic capturing of the Prague notion of topic and focus within the framework of intensional logic (Materna and Sgall, 1980; Materna, Hajicˇová and Sgall, 1987). This approach is also closely allied to that based on the concept of tripartite structure as proposed by Partee (1995). There is a straightforward way from the analysis by means of strong presuppositional predication given above to the analysis by means of a generalized quantifier. Let us introduce the (type-polymorphic) generalized quantifier PRED in the following way. Let us assume that we have a sentence analyzed as P!{S}; we shall show how to turn this analysis into the new one. We shall distinguish two cases. If P is P=8p.p(T) for some T (i.e. if P is a ‘‘type-raised T’’), like in case (3c¹), then we turn P!{S} into PRED(S,8 x.x=T); thus we turn (3c¹) into (3c¹¹). PRED(8y.love(John,y),8 x.x=Mary)

(3c¹¹)

In the other case (when P is not a 8p.p(T), e.g. (3a¹)), we turn P!{S} into PRED(8p.p(S),P); thus we turn (3a¹) into (3a¹¹).

 Jaroslav Peregrin

PRED(8p.p(John),8 y.love(y,Mary))

(3a¹¹)

However, in this new way we can account also for cases which are not so easily analyzable in the old one. Thus, we can straightforwardly analyze (5) as (5¹) and (6) as (6¹). John loves MARY, MINIE and HILLARY PRED(8x.loves(John,x), 8 x.(x=Mary∨x=Minie∨x=Hillary) John loves those who BILL loves PRED(8x.loves(John,x), 8 x.loves(Bill,x))

(5) (5¹) (6) (6¹)

What is the character of the generalized quantifier PRED introduced in this way? It can be characterized by three points. (1) PRED(A,B) implies that all Bs are As, hence B⊆A; which accounts for the intuition that ‘‘focus lists only members of the topic-class’’. (2) PRED(A,B) presupposes nonempty A (if A is empty, then PRED(A,B) is ‘undefined’); which is the formal counterpart of the intuition that ‘‘topic assumes knowledge shared by communicants’’. Then there is the most controversial point: (3) PRED(A,B) holds only if B exhausts, or ‘almost exhausts’, A (‘‘focus lists all, or the most important, members of the topic-class’’). If we took (3) as stating that PRED(A,B) implies that B⊇A, then PRED would turn out to be very close to ONLY. But it would be probably necessary to distinguish at least two versions of PRED: the basic version defined only by means of (1)+(2), and the exhaustive version, PREDexhaust, defined by (1)+(2)+(3). The idea is that PRED is the default, implicit generalized quantifier which can be overridden by an overt focalizer. Thus, (7) is analyzed as (7¹), whereas (8) as (8¹).6 John loves only Mary ONLY(8x.loves(John,x),8 x.x=Mary) John loves only those who BILL loves ONLY(8x.loves(John,x),8 x.loves(Bill,x))

(7) (7¹) (8) (8¹)

However, this does not mean that a word like only would automatically assume the place of PRED — in cases with the so called ‘‘free focus’’ the quantifier expressed by the focalizer stays embedded within the topic — viz (9¹). JOHN loves only Mary PRED(8x.ONLY(8x.loves(y,x),8 x.x=Mary),8 x.x=John)

(9) (9¹)

The problem of focalization can thus be seen as the problem of overriding PRED. In some cases, like (7) and (8), it is overriden, in other cases it is not. Partee (1995) points out, that ‘‘adverbial’’ quantifiers are usually focalizers, whereas ‘‘determiner quantifiers’’ are usually not. Negation can be considered as a principal focalizer — as the primary aim of negation seems to be to deny that ‘‘the focus holds of the topic’’ (see Hajicˇová, 1995), it can be seen as changing PRED into

Topic–Focus articulation as generalized quantification 

another, complementary quantifier. And possibly also the relationship between PRED and PREDexhaust can be understood in this way: we might see PREDexhaust as a non-default quantifier which may in some cases (especially in cases with the heavy stress on focus) override the default PRED. However, let us once more stress that if what we are after is semantics proper, i.e. truth conditions, then we must be careful about claims to the effect of presence of quantifiers or other similar items within that which the sentence expresses, means or denotes. Quantifiers are constituents of formulas, and logical formulas are, as we have stressed, only ways to point out propositions. The same propositions can be pointed out in various ways, some of them employing a quantifier, others not. Let us consider the sentence (10). Every man walks

(10)

In fact, it is sentences of this kind, which oil the wheels of linguist’s embracing the theory of generalized quantifiers. We have seen that the intuition behind generalized quantifiers in linguistics stems from the fact that the truth of the sentence of the shape Det N V can be computed by comparing the extensions of N and V in the way prescribed by Det. Thus, (10) is true if and only if the extension of man is included in the extension of walk. The TFA-based approach proposed in this paper now might seem to detach the tools of the theory of generalized quantifiers from this powerful intuition. However, this is not exactly true, because the analysis we proposed need not contradict the traditional one. Let us, for the sake of illustration, assume the standard TFA of (10), with every man being in the topic and walks in the focus. If we disregard presupposition and the requirement of exhaustiveness, then our analysis becomes equivalent to the standard one. To see this, let us consider the analysis (10¹) of (10) yielded by our approach. It is easy to see that (10¹) is — under the given conditions — equivalent to (11), which can be straightforwardly reduced to (12). And if we define the generalized quantifier EVERY in the usual way (i.e. EVERY(A,B) iff ¦¦A¦¦⊆¦¦ B¦¦), then (12) is further equivalent to (13). PRED(8p. x.man(x)→p(x),8 p.p=walk) (p=walk)→( x.man(x)→p(x)) x.man(x)→walk(x) EVERY(man,walk)

(10¹) (11) (12) (13)

Thus it does not always make sense to ask whether the semantic analysis of a sentence contains this or that quantifier. What does make sense is to ask whether if we decide to employ a quantifier, its operands coincide with topic and focus (i.e. whether the quantifier can be seen as overriding PRED).

 Jaroslav Peregrin

. Conclusion The aim of this paper was to survey the ‘‘logic’’ of topic and focus and to indicate how this ‘‘logic’’ can be done justice to by means of the currently popular formal approaches to semantics. We have claimed that it is the theory of generalized quantifiers, which could provide for a simple and adequate framework to capture TFA. Contemporary semantic theories are inclined to view TFA as something that is relevant only for the scope of focus-sensitive operators. The Praguian approach, on the other hand, considers TFA as the universal basis of semantic structuring of sentence. The approach proposed here can be — I believe — seen as compatible with the approach of Rooth and Krifka — the cases of focus-sensitive operators are considered only special cases of the general pattern. Moreover, the approach can be seen as a step towards an elaboration of Partee’s insight regarding the tripartite structuring of sentence.7

Notes . The paper originally appeared in P. Bosch and R. van der Sandt (eds) 1995. Proceedings of ‘‘Focus and natural language processing’’, 49–57. Heidelberg: IBM Deutschland. . If S is a sentence with the focus F, then the general way to analyze it would be as 8x.S[x→F]{F} (where S[x→F] stands for S in which x is substituted for F; x being the variable of the appropriate type). . This shows that we need semantic representation fine-grained enough to allow for capturing all possible TFAs; everything that can by itself constitute the focus must obtain its own counterpart within semantic representation. We see that the tense grammateme constitutes such an item. This might seem an obstacle, but is it not rather a test of what are the ultimate constituents of semantic representation? . Employed for the first time — at least to my knowledge — by Rooth. . And it can thus be seen as a sort of vindication of the Aristotelian, syllogistic approach to logic. . We can — in a sense — say hat also negation behaves as a focalizer, because in the typical case it can be seen as replacing PRED by the complementary generalized quantifier NPRED. This is to say that in the typical case it is the TFA that determines the scope of negation — the scope being the focus. See Hajicˇová (1995) and also Peregrin (1996).

References Groenendijk, J. and Stokhof, M. 1991. ‘‘Dynamic Predicate Logic’’. Linguistics and Philosophy 14: 39–101.

Topic–Focus articulation as generalized quantification 

Hajicˇová, E 1984. ‘‘On Presupposition and Allegation’’. In Contributions to Functional Syntax, Semantics and Language Comprehension, P. Sgall (ed.), 99–122. Amsterdam: Benjamins and Praha: Academia. Hajicˇová, E. 1995. ‘‘Topic, focus and negation’’. In Proceedings of Focus and natural language processing, P. Bosch and R. van der Sandt (eds.), Heidelberg: IBM Deutschland. Jakobs, J. 1991. ‘‘Focus Ambiguities’’. Journal of Semantics 8: 1–6. Krifka, M. 1991. ‘‘A Compositional Semantics for Multiple Focus Constructions’’. In Proceedings of SALT I, Cornell Working Papers 11. Krifka, M. (ms). ‘‘Focus, Quantification and Dynamic Interpretation’’. Manuscript. Lewis, D. 1972. ‘‘General Semantics’’. In Semantics of Natural Language, D. Davidson and G. Harman (eds.). Dordrecht: Reidel. Materna, P. and Sgall, P. 1980. ‘‘Functional Sentence perspective, the Question Test, and Intensional Semantics’’. SMIL 1–2: 141–60. Materna, P., Hajicˇová, E. and Sgall, P. 1987. ‘‘Redundant Answers and Topic–Focus Articulation’’. Linguistics and Philosophy 10: 101–13. Partee, B. 1991. ‘‘Topic, Focus and Quantification’’. In Proceedings of SALT I. Cornell Working Papers 10. Partee, B. 1995. ‘‘Focus, quantification and semantics-pragmatics issues’’. In Proceedings of Focus and natural language processing, P. Bosch and R. van der Sandt (eds.). Heidelberg: IBM Deutschland. Peregrin, J. and Sgall, P. 1986. ’’ An Attempt at a Framework for Semantic Interpretation of Natural Language’’. Theoretical Linguisitcs 13:37–73. Peregrin, J. 1996. ‘‘Topic and Focus in a Formal Framework.’’ In Discourse and Meaning, B. Partee and P. Sgall (eds.), 235–54. Amsterdam: Benjamins. Rooth, M. 1992. ‘‘A Theory of Focus Interpretation’’. Natural Language Semantics 1: 75–116. Sgall, P., Hajicˇová, E. and Panevová, J. 1986. The Meaning of the Sentence in Its Semantic and Pragmatic Aspects. J. L. Mey (ed.). Dordrecht: Reidel and Prague: Academia. Sgall, P. 1995. ‘‘Focus and focalizers’’. In Proceedings of Focus and natural language processing, P. Bosch and R. van der Sandt (eds). Heidelberg: IBM Deutschland.

Information structure and the partition of sentence meaning*

Klaus von Heusinger Constance

.

Introduction

The article argues for a different perspective on information structure. Traditionally, information structure is modeled by a partition of the ordinary meaning in two parts. In contrast to this, I argue for a two-level semantics in which one level corresponds to the meaning expressed in the sentence and the other level to the background meaning. Theories of information structure traditionally assume a partition of the sentence meaning in (at least) two parts: an information unit, often called ‘‘focus’’, corresponding to the focused phrase, and a complementary unit which corresponds to the remainder of the sentence meaning, sometimes called ‘‘background’’. The partition of the sentence meaning also reflects the idea that the focused phrase is more informative or provides new information, a fact which also seems to be reflected in the phonological (or syntactic) prominence of the focused phrase. In these theories, the ordinary meaning of the whole sentence must be composed of the two informational units in a subject–predicate manner. The additional background or presuppositional information is derived from the background unit and focus-sensitive operators are translated into quantifiers that range over alternatives (or p-sets) that are generated by the focused phrase. I discuss three problems of this view: (i) The ordinary meaning of a sentence is only indirectly derived from the two informational units. (ii) The meaning of the focused expression receives a too important position with respect to the meanings of the other expressions. (iii) There is no satisfactory rule for the generation of alternatives to definite NPs, indicating that the idea of alternatives must be revised. I propose a different information structure, which consists of two parallel informational units: a foreground (or ordinary) meaning, and a background meaning (or a presupposition skeleton). The foreground represents the meanings of all expressions in the sentence, while the background represents all expressions but the focused ones, which are replaced by designated variables. The focused

 Klaus von Heusinger

expressions indicate the ‘‘difference’’ between the two informational units, rather than an independent unit. The high informativeness of focus follows from this architecture and the approach accounts for the discussed problems: (i) The ordinary meaning of a sentence is directly derived from the foreground. (ii) The focused expression merely indicates the difference between both informational units. (iii) Focus-sensitive operators are translated into quantifiers that range over functions from the background into the foreground. The remainder of the paper is structured as follows: In Section 2, I discuss the early models of information structure. I will concentrate on the tacit assumptions in those models. In the course of the presentation of these theories, I try to make several implicit assumptions explicit. In particular, most approaches rely more or less on the following two poorly defined or scientifically obsolete notions: (i) the ‘‘Aristotelian’’ subject–predicate structure, (ii) the simplistic dichotomy of figureground from psychological gestalt theory. Section 3 presents classical models of information structure, which assume the communicative basis of this structure. The Prague School transformed the early models into linguistic theories, which then were adapted by the American structuralistic schools. Section 4 discusses semantic theories of information structure by the example of Alternative Semantics, one of the most advanced theory of focus and background structure. However, it will be shown that even this theory has problems with simple facts because of the assumption of the partition of the sentence into two parts. Section 5 gives some general perspective on the alternative view that the information structure consists in two levels of meaning representation. The section ends with a program for further research.

.

The background of information structure

Information structure, a name that was introduced by Halliday (1963), has received different names as listed in table (1): (1) Terminology for informational dichotomy Notion Used by Psychological subjectvon der Gabelentz 1869, Paul 1880 psychological predicate Theme–Rheme Ammann 1928: Thema-Rhema, Mathesius 1929, Prague School (Daneš, Firbas), Halliday 1967 Topic–comment von der Gabelentz 1869, Reinhart 1982 Topic–focus modern Prague School: Sgall, Hajicˇová and Benešová 1973

Information structure and the partition of sentence meaning1 

Presupposition-focus Background focus Old/given–new Open proposition–focus Notional subject– notional predicate

Chomsky 1971, Jackendoff 1972 Chafe 1976 for contrastive focus, Jacobs 1982 Halliday 1967, Chafe 1976 Prince 1981 É. Kiss 1995

The primary function of this structure is understood as inducing a partition of the sentence into two informational units, which are independent of the grammatical, i.e. syntactic-semantic, subject–predicate organization. Nevertheless, the units are defined with respect to the grammatical sentence: They can be reunited to the whole in an extra-grammatical subject–predicate manner, as illustrated in (2): (2) Information structure S Syntactic representation NP

[Mary

VP V

NP

prefers]

[CORDUROY]

Background

Focus

Surface structure

Informational representation

The theoretical basis for this additional subject–predicate structure varies according to the background theory of the researcher. E.g., von der Gabelentz (1869) introduces the pair psychological subject — psychological predicate according to his view that psychology is the ultimate base for language structure. The Prague School used the terms theme — rheme and later topic — comment, which are both borrowed from traditional rhetoric and philology. Chomsky (1971) uses focus — presupposition, indicating the semantic nature of the distinction. É. Kiss (1995: 7) uses the distinction notional subject — notional predicate, trying to abstract from the grammatical connotations. This very idea of the dichotomy was transmitted unnoticed from the psychological tradition of linguistics to the anti-psychological one, which has based linguistic functions on informational concepts. All these different approaches have in common that the dichotomy of the sentence is explained in terms of a subject–predicate structure. However, it were the problems of the grammatical subject–predicate structure that motivated the additional information structure.

 Klaus von Heusinger

. The logic of the subject–predicate structure The contrast between subject and predicate is not only a linguistic distinction, it also plays an important role in traditional philosophical disciplines such as epistemology, logic, and metaphysics, i.e. ontology, and in more recent fields such as psychology or information theory. The nature of the contrast is still very controversial and confusion only arises in the attempt to define the concepts at one level, more often confusion is caused by relating the distinction of one level with the distinction on another level. In what follows, I present some of the main concepts of the subject–predicate structure in linguistics, epistemology, logic and metaphysics.1 The linguistic use of predicate and subject as constituents in a sentence goes back to Plato and Aristotle. The two categories indicate two different functions in the sentence and they are often associated with the noun and the verb in a simple sentence. Typological research has shown that subject and predicate are not universal categories; so some linguists have proposed to replace this contrast by the contrast between topic and comment. However, topic and comment cannot be regarded as purely grammatical, i.e. syntactic, categories since they have more to do with the knowledge or understanding of the participants of what is being discussed, i.e. the aboutness aspect, than with the constituent structure. Problems in the definition of a sentence in terms of subject–predicate structure caused the abolishment of these concepts in syntactic theory proper.2 However, the basic notion survived in the generative rule of expanding a sentence to an NP and a VP (S → NP VP). The present situation in linguistics can be summarized by saying that subject and predicate are useful descriptive terms, but they do not have a clear grammatical definition. In epistemology, the contrast between subject and predicate is defined in terms of which part of the sentence serves to identify what is being discussed, and which part serves to describe or to characterize the thing so identified. Often the grammatical and epistemological subject and predicate are identical, in particular in simple sentences. However, there are several instances in which they differ, e.g. passive constructions, sentences with dummy subjects etc. More complex sentences, in which a relation between two objects is described, cause other problems for the epistemological view of the sentence structure. In the sentence Bill was hit by John the context has to decide which is the epistemological subject.3 In other words, the epistemological view is as context dependent in its determination as the use of topic in linguistic theory. In logic, there are two main conceptions of sentence structure: (i) the traditional subject–predicate view, which goes back to Aristotle’s pioneering work, and (ii) Frege’s view of the functor-argument structure of the sentence. The Aristotelian or traditional view assumes that subject and predicate are both general terms

Information structure and the partition of sentence meaning1 

(i.e. concepts or predicates in the modern view) connected by the copula, schematically as S is P. The sentence asserts a quantificational relation between the subject and the predicate, which is expressed in the copula some . . . is, or all . . . are. (3) a. Some Athenian is big. b. All Athenians are Greeks.

(A is B for some A) (A is G for all A)

This view treats the subject and the predicate symmetrically and allows the inversion of the expression for certain inference rules (Some big person is Athenian, Some Greeks are Athenians). However, this view was criticized since (i) the symmetry does not hold with negation, (ii) there is no simple way to describe sentences with singular terms, and (iii) multiple quantification, as in Every Athenian hates some king, is not possible. Frege’s new view is based on the concept of the atomic sentence, which consists of a functor and an argument: Fa.. Singular (or atomic) sentences like (4a) are represented by the application of the predicate to the argument, while general sentences are represented by complex sentences with quantifiers as in (4b): (4) a. Socrates walks. Walk(socrates) b. Every Athenian hates some king. x [A(x) → ∃y [K(y) & H(x,y)]] In this way, Frege solved (i) the problem of negation, (ii) the representation of singular terms, and (iii) multiple quantification. The copula does not receive any semantic content but is understood as a surface form of flexion. In order to define the relations expressed in a sentence, he introduces new formal expressions. Frege (1879, vii) explicitly argues for the replacement of the terms subject and predicate by argument and function, respectively:4 Insbesondere glaube ich, daß die Ersetzung der Begriffe Subject und Prädicat durch Argument und Function sich auf die Dauer bewähren wird. Man erkennt leicht, wie die Auffassung eines Inhalts als Function eines Argumentes begriffsbildend wirkt. Es möchte ferner der Nachweis des Zusammenhanges zwischen den Bedeutungen der Wörter: wenn, und, nicht, oder, es giebt, einige, alle u.s.w. Beachtung verdienen.

The metaphysical (i.e. ontological) distinction between subject and predicate is closely correlated to the substance-attribute distinction, which has caused major disputes in philosophy. In particular, it is controversial whether subjects and predicates are related to ontologically different kinds of objects, or whether they can be assumed to be of a similar nature. For example, Plato gives ontological prominence to predicates, since they never change, while subjects easily change. Aristotle, however, gave ontological standing to both subjects and predicates. Frege accepted the subject–predicate dualism as a basis for his metaphysics, which he built in parallel to his logic. As already mentioned, he replaces the

 Klaus von Heusinger

traditional notion of subject and predicate by argument and function. The argument is defined as complete or saturated object, while the predicate is an unsaturated item, or what Frege called a function, representing a concept. Together they form a sentence, another kind of saturated item. (5) Frege’s view of the logical and ontological structure of a sentence (i) traditional description subject — predicate (ii) Frege’s logical form argument — function corresponds to singular term — general term (iii) ontological type saturated — unsaturated (iv) extension individual — class of individuals It was only Frege’s definition of the sentence structure as function-argument structure that gave the fundamental semantic definition of a sentence. Since this definition is not dependent on syntax, pragmatics or psychology, semantics was able to establish its own research domain. To sum up, subject and predicate are useful descriptive terms for describing sentence structure. However, there is no clear definition of these concepts, but rather a bundle of historically related claims from as different fields as syntax, epistemology, logic, and metaphysics. In particular, the Aristotelian concept of subject–predicate in grammar has created more problems than it has solved. These two points, the vagueness of the notion of the subject–predicate structure, and its highly problematic use in linguistic theory, should be borne in mind when this concept is transferred to or used in other domains such as psychology or information theory. In other words, it does not help the understanding to apply the notion of subject–predicate at any level, but it rather adds more confusion. . Hermann Paul and psychologism During the course of the last century, psychology became the epistemological basis for linguistics. Hermann Paul’s ‘‘Prinzipien der Sprachgeschichte’’ is not only one of the most influential books on general linguistics, but also one of the nicest examples of the influence of psychological theory on general linguistics and linguistic foundations. Since the first edition of the Prinzipien from 1880, there were repeatedly new editions, which however did not change the overall view of Paul’s system. Paul (1880, 121) defines the sentence as the linguistic expression that connects several (psychological) concepts or groups of concepts in the mind (‘‘Seele’’) of the speaker. Furthermore, the sentence is the tool to generate these concepts and their connection in the mind of the hearer: Der Satz ist der sprachliche Ausdruck, das Symbol dafür, dass sich die Verbindung

Information structure and the partition of sentence meaning1 

mehrerer Vorstellungen oder Vorstellungsgruppen in der Seele des Sprechenden vollzogen hat, und das Mittel dazu, die nämliche Verbindung der nämlichen Vorstellungen in der Seele des Hörenden zu erzeugen.

In his extensive discussion of sentence structure, Paul (1880, ch. 15–16) introduces most of the relevant aspects of the relation between the psychological subject–predicate structure and the grammatical subject–predicate structure, of which I only will mention the following three (for a more detailled discussion see von Heusinger 1999, sect. 3.1.2,) (6) Main aspects of Paul’s conception of sentence structure (i) the psychological fundaments of linguistics (ii) the partition of the sentence into two parts (iii) the question test as a criterion for thematicity .. The psychological fundaments of linguistics The terms psychological subject and psychological predicate were introduced by von der Gabelentz (1869), who compared the sequence of thoughts or psychological concepts with the sequence of linguistic expressions in a sentence (A similar distinction was already suggested by Weil 1844). Von der Gabelentz then distinguished two levels: the grammatical level and the psychological level of composition. Von der Gabelentz defines the psychological subject as ‘‘that about which the hearer should think’’, and the psychological predicate as ‘‘that what he should think about it’’. It is interesting to note that the distinction between the two parts is defined with respect to the mental or cognitive state of the hearer. Paul (1880, 263) adopted the terminological distinction easily, since he conceives every grammatical category on the basis of a psychological one: ‘‘Jede grammatische Kategorie erzeugt sich auf Grundlage einer psychologischen.’’ .. The partition of the sentence into two parts One of the main assumptions of Paul is that the sentence is divided into (at least) two parts. Paul (1880, 124) says that each sentence consists of at least two elements. These elements are not equal but differ in their function. They are termed subject and predicate. These grammatical categories are based on a psychological relation. We have to distinguish between psychological and grammatical subject and predicate because they do not always coincide. But the grammatical relation is always formed on the grounds of the psychological relation. Jeder Satz besteht demnach aus mindestens zwei Elementen. Diese Elemente verhalten sich zu einander nicht gleich, sondern sind ihrer Funktion nach differenziert. Man bezeichnet sie als Subjekt und Prädikat. Diese grammatischen Kategorien beruhen auf einem psychologischen Verhältnis. Zwar müssen wir

 Klaus von Heusinger

unterscheiden zwischen psychologischem und grammatischem Subjekt, respektive Prädikat, da beides nicht immer zusammenfällt, wie wir noch im Einzelnen sehen werden. Aber darum ist doch das grammatische Verhältnis nur auf Grundlage des psychologischen auferbaut.

.. The question test as criterion for thematicity The clearest test for the psychological predicate is the constituent question. Paul illustrates this with the simple assertion (7), which can be the reaction to the different questions (7a)–(7d). Depending on the question, the psychological predicate of the sentence differs, while the grammatical structure remains the same. In (7a) the psychological predicate is the locative Berlin, in (7b) it is the time adverbial morgen, and in (7d) it is the grammatical subject Karl: (7)

Karl fährt morgen nach Berlin. ‘Karl goes to Berlin tomorrow.’ a. Wohin fährt Karl morgen? ‘Where does Karl go tomorrow?’ Karl fährt morgen nach berlin. b. Wann fährt Karl nach Berlin? ‘When does Karl go to Berlin?’ Karl fährt morgen nach Berlin. c. Wie reist Karl nach Berlin? ‘How does Karl travel to Berlin?’ Karl fährt morgen nach Berlin. d. Wer fährt morgen nach Berlin? ‘Who goes to Berlin tomorrow?’ Karl fährt morgen nach Berlin.

The question test is still one of the most fundamental tests for information structure. However, it is not clear whether the question test illuminates the information structure in its sentential or its discourse aspect. For Paul, who does not distinguish these two aspects, there is no difference between the aboutness and the discourse anchoring of the psychological predicate. In more recent approaches, the question test is generally associated with the focus, i.e. with the new information given in a discourse or with the discourse anchoring. To sum up, Paul discusses the notion of psychological structure vs. grammatical structure very informally. However, it was shown that he already recognizes most of the relevant aspects of information structure. In the course of this argument it will be illustrated that even most of the more recent theories do not overcome the informality of Paul’s presentation. For if they were more explicit, they would have to acknowledge that the basic concepts like subject–predicate are not linguistically well defined.

Information structure and the partition of sentence meaning1 

. The communicative shift At the beginning of this century, the interest in the communicative (or social) function of language had increased, which was most obviously manifested in the ‘‘Cours de linguistique générale’’ of Ferdinand de Saussure (1916). The meaning of a sentence is evaluated with respect to its contribution to the communication between the participants. For example, Ammann (1928: 2) focuses on two points: first that a sentence is primarily a message (‘‘Mitteilung’’). Due to its nature, a message consists of two parts, which closely correspond to the sentence organization into subject and predicate.5 Ammann then argues that the informational structure of the message is the basis for dividing the grammatical structure into subject and predicate, rather than assigning primacy to the dichotomy of psychological subject and predicate. Thus, he notes that it is not important that psychological structure does not always coincide with the grammatical structure, but what counts is that it is the information unit that causes the dichotomy of subject and predicate. In order to distinguish between the grammatical structure of the sentence, the psychological structure of concepts or ideas, and the informational structure of the message, Ammann introduces a new pair of terms: theme and rheme (‘‘Thema’’ and ‘‘Rhema’’). Rheme is borrowed from the Greek grammatical tradition, where it refers to the verb, in contrast to onoma which refers to the name or subject. Ammann (1928:3) concludes that these terms do not carry any connotations since the classical concepts are only used in the Latin translations. Ammann and others have recognized that the sentence is organized not only by syntactic structure, but also by other principles. However, they were unable to locate this organizational level in linguistic theory proper. Instead they referred to other fields like psychology and information theory. One reason may have been that semantics as a proper subfield of linguistics was not yet established. It was only the work of Frege, Russell, Carnap, Montague and their followers that established semantics as one of the core disciplines of linguistics with independent levels of representations, abstract objects, and rules operating on them. . Gestalt theory: The concept of figure and ground In the course of the nineteenth century, the syntactic definition of the sentence became more elaborate. At the same time it became obvious that the syntactic description does not cover all aspects of sentence meaning. Differences in the presentation of the sentence content were attributed to an underlying psychological structure (as the contemporary epistemological basis for language function). The meanings of words were assumed to be psychological concepts or mental ideas, and sentence meaning was understood in terms of operations on those ideas. In particular, the

 Klaus von Heusinger

dichotomy of the sentence was understood as expressing a dichotomy of mental ideas. If conjoined they would yield the content of the sentence as their product. One important movement in psychology was the so-called Gestalt theory. According to this school, perception functions as a whole gestalt and not by constructing something out of small units. The gestalt perception includes two different parts: figure and ground. The figure is recognized only against the ground, which is the principle for many optical illusions as in (8) and (9), where one and the same stimulus (the line) is perceived differently depending on the ground. In (10), one and the same stimulus is structured differently into figure and ground, giving rise to two ways of conceiving it. Here, we can either see a chalice or two faces, but not both at the same time. (8)

(9)

(10)

Another feature of Gestalt theory is the direct relation between stimulus and perception. There is no intermediate level of representation. Even though Gestalt theory was mainly developed on vision, it was also used for the explanation of other perceptual channels, such as speech. Thus the idea of the dichotomy of the sentence organization not only found additional support in this psychological movement, it also inherited the terms figure and ground from it. The figure represents the prominent or highlighted part, while the ground represents the given or less informative material of the sentence. The idea of a direct correspondence between stimulus and function was instantiated by the direct correspondence between intonational highlighting and communicative highlighting. Later, communication and information theory replaced psychology as the epistemological background for linguistic research. Sentences are analyzed with respect to their communicative functions. However, the dichotomy of figure and ground was adopted and the idea of highlighting and of the direct relation between intonational prominence of an expression and its information content were inherited from the earlier psychological treatments. This one-to-one relation is assumed without any more abstract level or representation. Hirst and Di Cristo (1998: 28 ff.) summarize the situation and draw the connection to the Prague School as follows:6 The basic idea behind all work in this area is that communication takes place against a background of shared knowledge so that the way a listener interprets an utterance will be partly dependent on the (situational) context in which the utterance occurs. This idea follows a more general principle which had been proposed in the beginning of the century by the Czech psychologist Wertheimer [1886–1943], one

Information structure and the partition of sentence meaning1 

of the founders of the Gestalt School of psychology, according to which the perception of a stimulus, particularly in the case of vision, generally consists of attributing a structure in which one part of the stimulus, called the figure, seems to stand out against the rest of the stimulus, called the ground ( . . .). Under different conditions, the same stimulus can be structured differently into figure and ground, giving rise to a number of familiar optical illusions. The first person to apply a similar idea to language was another Czech, the linguist Mathesius [1882–1945], founder of the Prague Linguistic Circle, whose theory of Functional Sentence Perspective was taken up and developed by other linguists of the Prague School such as Daneš and Firbas.’’ ( . . .)

.

Information structure in structuralism

. The Prague School At the end of the last century the psychological foundation of sentence structure led to an additional level of descriptions for the organization of the sentence using contrasts like psychological subject and psychological predicate (von der Gabelentz 1869). Paul (1880) developed this contrast and Ammann (1928) shifted the point of interest from psychology to communication and introduced the terms theme and rheme. At the same time, Mathesius (1929) used these terms for describing word order phenomena in Czech and other Slavic languages. Unlike the approaches discussed so far, the Prague School integrated the distinction between theme and rheme into the grammatical system. The most characteristic feature of the Prague structuralists, in contrast to other structuralist schools, was the functional approach. Language is understood as a tool for communication and the information structure is important for both the system of language and for the process of communication.7 Mathesius reformulates the contrast between the grammatical subject– predicate and the organization of the message of a sentence in his functional sentence perspective. Weil (1844) had already noted that the sequence of words and the sequence of thoughts do not always correspond. Thus, Mathesius working on word order in Slavic languages proposes the thematic structure of a sentence as a linguistic level of analysis, which is independent of the subject–predicate relation. The functional sentence perspective was further developed by a series of researchers. Firbas (1964), argues that information structure is not a dichotomy but rather a whole scale, or hierarchy, or what he calls communicative dynamism. Daneš (e.g. 1970) extends the thematic relation of the sentence to one of a text, and the newer Prague School (Sgall, Hajicˇová and Benešová 1973 and Sgall, Hajicˇová and Panevová 1986) uses the contrast of topic and focus and gives an account of how to integrate this structure into a grammatical model. Peregrin (1995) attempts a formalization of the topic–focus articulation in terms of

 Klaus von Heusinger

structured proposition and dynamic logic. Hajicˇová, Partee and Sgall (1998) give a comparison between the Prague School and the semantic tradition since Montague. It was Halliday (1967) who introduced the Praguian distinction of theme and rheme into American structuralist linguistics. For lack of space, I cannot present an adequate discussion of the Praguian theory of functional sentence perspective or the topic–focus-articulation. Excellent overviews and comparisons with contemporary grammatical theory are presented in Sgall, Hajicˇová and Benešová (1973), Sgall, Hajicˇová and Panevová (1986) and Hajicˇová, Partee and Sgall (1998). Here, only two aspects should be touched upon: first the two sides of information structure, the sentence internal aspect and the textual aspect; and second the extension of the Theme–Rheme structure to texts. Daneš (1970: 134) describes the two faces of information structure or what he calls ‘‘utterance organization’’ or ‘‘utterance perspective’’: (1) Taking for granted that in the act of communication an utterance appears to be, in essence, an enunciation (statement) about something (question should be treated separately), we shall call the parts theme (something that one is talking about, topic), and rheme (what one says about it, comment). (2) Following the other line, linking up utterance with the context and/or situation, we recognizes that, as a rule, one part contains old, already known or given elements, functioning thus as a ‘starting point’ of the utterance, while the other conveys a new piece of information (being thus the ‘core’ of the utterance). But, as in most cases, the two aspects coincide, we shall, in our following discussion, disregard the said distinction. Daneš (1970: 137f) notes that the two aspects are not independent, since one can extend the idea of the topic of a sentence to the discourse. He introduces three ways sentential topics can be concatenated in a discourse: (i) simple linear progression as in (11), a progression with a continuous theme, and (iii) the exposition of a split theme. (11) simple linear progression of a theme T1 → R1 ↓ T2 (= R1) → R2 ↓ T3 (= R2) → R3 ... . Halliday and the American structuralists While the Praguian approach locates the information structure in the syntaxsemantics interface, Halliday postulates an independent level for information

Information structure and the partition of sentence meaning1 

structure. He is in fact the first who uses the term information structure and establishes an independent concept of it. His main preoccupation was to account for the structure of intonation in English. Since phrasing does not always correspond to syntactic constituent structure, Halliday (1967:200) postulates a different structural level as the correlate to phrasing (his ‘‘tonality’’): Any text in spoken English is organized into what may be called ‘information units’. ( . . .) this is not determined ( . . .) by constituent structure. Rather could it be said that the distribution of information specifies a distinct structure on a different plan. ( . . .) Information structure is realized phonologically by ‘tonality’, the distribution of the text into tone groups.

Halliday assumes two structural aspects of information structure: the informational partition of the utterance, and the internal organization of each informational unit. He calls the former aspect the thematic structure (Theme–Rheme) and the latter aspect is treated under the title givenness. The thematic structure organizes the linear ordering of the informational units, which corresponds to the Praguian view of Theme–Rheme (or topic-comment, or topic–focus) and is organized according to the principle of aboutness. The theme refers to that informational unit that comprises the object the utterance is about, while the rheme refers to what is said about it. Halliday assumes that the theme always precedes the rheme. Thus Theme–Rheme are closely connected with word order, theme being used as a name for the first noun group in the sentence, and theme for the following: ‘‘The theme is what is being talked about, the point of departure for the clause as a message; and the speaker has within certain limits the option of selecting any element in the clause as thematic.’’ (Halliday 1967: 212). This is the aspect that is determined by the ‘‘ontological structure’’ of the message as described by Ammann (see above). The second aspect refers to the internal structure of an informational unit, where elements are marked with respect to their discourse anchoring: ‘‘At the same time the information unit is the point of origin for further options regarding the status of its components: for the selection of point of information focus which indicates what new information is being contributed’’ Halliday (1967: 202). Halliday calls the center of informativeness of an information unit information focus. The information focus contains new material that is not already available in the discourse. The remainder of the intonational unit consists of given material, i.e. material that is available in the discourse or in the shared knowledge of the discourse participants. Halliday (1967: 202) illustrates the interaction of the two systems of organization with the following example (using bold type to indicate information focus; // to indicate phrasing). Sentence (12a) contrasts with (12b) only in the placement of the information focus in the second phrase. The phrasing, and thus the thematic structure, is the same. On the other hand, (12a)

 Klaus von Heusinger

contrasts with (12c) in phrasing, but not in the placement of the information focus. However, since the information focus is defined with respect to the information unit, the effect of the information focus is different. (12) a. //Mary//always goes to town on Sundays// b. //Mary//always goes to town on Sundays// c. //Mary always goes to //town on Sundays// Phrasing and its correlate thematic structure is independent of information focus: ‘‘But the interpretation of information focus depends on where it is located relative to the information unit, so that it is the distribution that partially determines the focus and not the other way round’’ (Halliday 1967: 202). The main progress initiated by the work of Halliday is the assumption of an independent level of information structure. This structure is closely related to the discourse and assigns the features given or new to the expressions in a sentence. However, what was still not solved is the definition of given vs. new. Halliday (1967: 211) himself defines ‘‘given’’ information as being treated by the speaker as ‘‘recoverable either anaphorically or situationally’’. New information, on the other hand, is characterized by at least three formulations: (i) ‘‘new’’ information is said to be focal ‘‘not in the sense that it cannot have been previously mentioned, although it is often the case that it has not been, but in the sense that the speaker presents it as not being recoverable from the preceding discourse’’ or (ii) new information is ‘‘contrary to some predicted or stated alternative’’, or (iii) new is what is ‘‘replacing the WH-element in a presupposed question’’, as illustrated in the examples (13)–(15), respectively: (13) A. Why don’t you have some French toast? B. I’ve forgotten how to make French toast. (14) (John’s mother voted for bill.) No, she voted for john. (15) Who did John’s mother vote for? She voted for john. These three characterizations of new suggest three different kinds of focus, as in (16): (16) Three kinds of focus (i) informative in assertion/addition (ii) contrastive in assertion/contradiction (iii) focus-question in question/answer pairs Since it is not clear whether there is one coherent definition of new or of focus, it seems more plausible to find a definition of given. Halliday himself relates givenness to ‘‘anaphorically recoverable’’.

Information structure and the partition of sentence meaning1 

There have been other directions in which the concept of givenness was characterized. For example, Chafe (1976) and others reinterpreted the givenness of Halliday as a cognitive or psychological category which is applied to the objects in the mental states of the speakers. Another direction is represented by Chomsky (1971) and Jackendoff (1972), who replaced the concept of given-new by presupposition-focus. The notion of presupposition was already a basic concept in semantics even though not fully understood. . Information packaging In his seminal paper givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View, Chafe (1976) discusses aspects — or ‘‘statuses’’ as he calls them — of nouns. Chafe is interested in the way discourse is structured. He assumes that discourse is organized according to the beliefs of the speaker about the beliefs of the hearer, rather than according to the semantic content of linguistic expressions. Chafe (1976:28) illustrates this with the famous metaphor of ‘‘information packaging’’: I have been using the term packaging to refer to the kind of phenomena at issue here, with the idea that they have to do primarily with how the message is sent and only secondarily with the message itself, just as the packaging of toothpaste can affect sales in partial independence of the quality of the toothpaste inside. Our starting point, then, is that the packaging phenomena relevant to nouns include the following: a. the noun may be either given or new ; b. it may be a focus of contrast; (c) it may be definite or indefinite.; (d) it may be the subject of its sentence; (e) it may be the topic of its sentence; and (f) it may represent the individual whose point of view the speaker is taking, or with whom the speaker empathizes.

Chafe extends Halliday’s givenness into psychological models of the consciousness of speaker and hearer: Givenness. What is it? The key to this distinction is the notion of consciousness (. . .). Given (or old) information is that knowledge which the speaker assumes to be in the consciousness of the addressee at the time of the utterance. So-called new information is what the speaker assumes he is introducing into the addressee’s consciousness by what he says. (Chafe 1976: 30)

This definition of information structure is very similar to the definition of psychological structuring of von der Gabelentz (see above). Chafe (1976:30) himself notes that although this use of given and new is often misleading and a different use such as ‘‘activated’’ would be more appropriate, he will continue to use the old pair: Terms like ‘‘already activated’’ and ‘‘newly activated’’ would convey this distinction more accurately, but are awkward; we will probably have to live with the terms ‘‘given’’ (or ‘‘old’’) and ‘‘new.’’

 Klaus von Heusinger

Prince (1981; 1986: 208) picks up Chafe’s packaging idea and states that speakers tailor sentences in various ways to (their assumptions about) their interlocutors or, in more detail: Information in a discourse does not correspond simply to an unstructured set of proposition; rather, speakers seem to form their utterances so as to structure the information they are attempting to convey, usually or perhaps always in accordance with their beliefs about the hearer: what s/he is thought to know, what s/he is expected to be thinking about.

The idea of information packaging was further developed by Vallduví (1990), who assumes an information structure that consists of three parts merging the most prominent aspects of information structure into one: focus-background and topic-comment. The question (17) introduces John as a topic and focuses on the object that John drinks. The predicate drink is both part of the comment as well as part of background in (18). Vallduví (1990: 55) proposes the structure (19) corresponding to his information structure in (20). He proposes a main partition into focus and ground (corresponding to the notion of focus-background), and a second partition of the ground (background) into link and tail (topic and the rest of the background), yielding his three informational units: focus, link and tail. (17) What does John drink? (18)

Topic-comment and focus-background structure Topic Comment John drinks beer Background Focus

(19) Information packaging (Vallduví) Link Tail Focus John drinks beer (20) Information structure as focus-link-tail Sentence Focus

Ground Link

Tail

Like Chafe and Prince, Vallduví describes the situation very informally using, for example, Heim’s metaphor of file cards, but he does not attempt to give a semantic analysis of the phenomena presented. Vallduví (1990: 2) claims that information packaging does not effect truth conditions: ‘‘While their propositional

Information structure and the partition of sentence meaning1 

contents are the same, they do not provide the same information ( . . .).’’However, he does not give a clear account of what he means by information. . Focus and presupposition Chomsky (1971: 199ff) discusses information structure in the context of his distinction between deep structure and surface structure. It is the deep structure that determines the meaning of a sentence. However, if it can be shown that intonational contrasts, which only effect the surface, exhibit systematic meaning contrasts, then the model is threatened. Even though Chomsky does not give a final answer, he discusses several examples and some approaches to describe them. What should be of interest here is his treatment of the contrast given-new, or in his terms: presupposition-focus. Focus is defined ‘‘as the phrase containing the intonation center’’ (Chomsky 1971: 200). In a first approach Chomsky explains the focus-presupposition dichotomy in the following way. (22) is a reasonable answer to (21a)–(21d), all of which have the presupposition that John writes poetry somewhere. The presupposition is informally described as that part of the sentence that is conveyed independently of the speech act or the negation made in the sentence. The presupposition corresponds to the sentence minus the focus element. Chomsky proposes the deep structure (23) for (22), in which the presupposition is the embedded sentence and the focus is part of the matrix predicate is in the garden. (21) a. Does John write poetry in his study? b. Is it in his study that John writes poetry? c. John doesn’t write poetry in his study. d. It isn’t in his study that John writes poetry (22) No, John writes poetry in the garden. (23) [the place where John writes poetry] [is in the garden] presupposition: John writes poetry at a place focus: garden Chomsky (1971: 205) informally describes the construction of such a deep structure: ‘‘The focus is the phrase containing the intonation center, the presupposition, and the expression derived by replacing the focus by a variable.’’ Then in a quasi formalization, he proposes to associate each sentence with a class of pairs (F, P) where F is a focus and P a presupposition indicating the range of possible interpretations. Since the function of focus is to ‘‘determines the relation of the utterance to responses, to utterances to which it is a possible response, and to other sentences in the discourse’’, the range of possible interpretation can do so.

 Klaus von Heusinger

For example, (22) is a proper answer to (21a) since they share the same presupposition that John write poetry at some place. Even though Chomsky is not very explicit in his description of focus — he is more interested in the relation between surface structure and deep structure — he assumes like Halliday that focus is intonationally marked. In contrast to Halliday Chomsky (1971: 205) makes the claim that the interpretation of focus is to be integrated into semantics proper: The notions ‘focus’, ‘presupposition’, and ‘shared presupposition’ ( . . .) must be determinable from the semantic interpretation of sentences, if we are to be able to explain how discourse is constructed and, in general, how language is used.

. The semantics of information structure The theories of information structure discussed above assumed an independent level of organization of the sentence. They explicitly state that this is not a semantic level since a great number of effects attributed to information structuring does not effect the sentence meaning in terms of propositional content (e.g., Chafe 1976; Vallduví 1990). Consequently, approaches to information structure employ other fields like psychology for an explanation of informational contrasts. However, the informational organization does not only interact with other linguistic levels, but also plays an essential role in several linguistic aspects of meaning. For example, it is essential for the construction and the coherence of a discourse, for the choice of anaphoric elements, for the interpretation of the adequate speech act and so on. More importantly, information structure is necessary for the interpretation of sentences with focus-sensitive particles like only, also, too or adverbs of quantification like always, sometimes etc. Here, we find truthconditional effects of different informational organizations of the text, which is the primary sign that the latter is of semantic import. Therefore, semantic theories of information structure are built on the analysis of focus-sensitive particles. Their contribution to sentence meaning depends on the informational structuring of the sentence, which is considered to consist of a focus part and a complement to this focus. This complement is known under different terms such as presupposition, frame, open proposition, shared knowledge or background. I will use background as a neutral term, defined as complement of the focus unit. The focus-background effects are informally illustrated by the following three examples. In the simple assertion (24), the focus on Fred is said to induce a presupposition that Sam talked to someone and that there are reasonable alternatives that would have been good choices for this someone, as well. (24) Sam talked to fredF.

Information structure and the partition of sentence meaning1 

(24¹) It is part of the background or of the shared knowledge that Sam talked to someone. There are alternatives to Sam under discussion. In (25), the focus particle only is translated into an operator that quantifies over alternatives to Fred and asserts that if such an alternative fits the background, then it is identical with the focus Fred. This corresponds to the paraphrase that nobody but Fred is such that Sam talked to him. (25) Sam only talked to fredF. a. Nobody but Fred is such that Sam talked to him or her. b. for every element x that is a reasonable alternative to Sam, if x fits the background ‘‘Sam talked to x’’ then x is identical with Fred The adverb of quantification in (26) asserts that for most occasions in which Sam invites someone to the movies, it is Fred who is invited. Here, the adverb of quantification is translated into an operator which takes two arguments and compares the two: the set of times when Sam invites someone to the movies and the set of time when he invites Fred. The operator states that most time points of the first set are also in the second set. Here, the operator does not quantify over the focus, but the focus indicates how to form the proper domains for the quantifier. (26) Sam usually invites fredF to the movies. a. For most times in which Sam invites someone to the movies, he invites Fred to the movies b. Most {t ¦ Sam invites someone to the movies at t} {t ¦ Sam invites Fred to the movies at t} The following general picture can be drawn: The surface structure receives a syntactic representation in terms of constituents and an informational representation, which is a partition into background and focus. The focus induces or evokes alternatives in one way or other. The value of the focus, i.e., the semantic denotation of the focused expression, is part of the set of alternatives, which are sometimes called p-set. The background corresponds to a semantic structure that I will call here the p-skeleton, following Jackendoff ’s (1972) presuppositional skeleton, other expressions are open proposition, frame etc. The p-skeleton is formed by substituting the focused expressions by appropriate variables. The p-skeleton, the set of alternatives, and the value of the focus combine to form the assertion and the presupposition of the sentence. The presupposition is formed by existential closure, i.e., by existential quantification over the free variables that had been substituted for he focused expression. The presupposition in (24) is that Sam talked to someone. Applying the value of p-skeleton to the focus value yields the assertion that Sam talked to Fred, as illustrated in diagram (27).

 Klaus von Heusinger

(27) Information structure and focus-semantics S Syntactic representation NP

VP

[Sam

V

NPF

talked]

[to Fred]F

Background

p-skeleton

∃ Presupposition: Existential closure of p-skeleton

Focus

Alternative/p-set

Surface structure

Informational representation

Semantic representation

+ Assertion: applied to focus value

Meaning

The ‘‘focus-meaning’’ of (24) is the felicity condition that Sam could have talked to other individuals in the alternative set induced by the focus on Fred. Thus, the focus in (24) influences the felicity conditions, rather than the truth conditions. However, a sentence with focus particles exhibit different truth conditions depending on the location of the focus. Focus particles can be translated into quantifiers that operate on the background (p-skeleton), the alternatives and the focus meaning. E.g., the assertion of (25) is that nobody but Fred is such that Sam talked to him. Thus the focus, i.e. the meaning of the focused constituent, receives the function of the most important argument for the quantifier. Semantic theories of focus differ in whether they directly access focus and background, or only one, or none. Direct association with focus or movement (or structured proposition) theories (Jacobs 1983; von Stechow 1991; Krifka 1992) assume that the focus is moved to a position adjoined to the focus operator at the level of Logical Form. The focus leaves a trace in its original position which is interpreted as a variable. These theories translate traditional informational partition of the sentence into the semantic representation. The in situ approach of Alternative Semantics posits an additional level of interpretation, the alternative interpretation, which is computed in parallel with the ordinary semantics.

Information structure and the partition of sentence meaning1 

. Alternative Semantics Alternative Semantics distinguishes between two dimensions of meaning, the ordinary meaning ¦¦ ¦¦O and the alternative meaning ¦¦ ¦¦A. The alternatives are formed by the function ALT applied to the ordinary meaning of the focused expression. The alternative value of an expression is a set containing elements of the same type as its ordinary meaning. In this sense, the alternative meaning of a basic expression is derived from the correspondent ordinary value. The alternatives are projected in parallel to the composition of the ordinary meaning. Since there are two semantic dimensions, we have to define the interpretation rules for both dimensions. The ordinary interpretation (28a) does not see the focus feature F and, therefore, interprets a focused expression like the unfocused one. The alternative interpretation of a focused expression (28b) creates the set of alternatives. The alternative semantics of an unfocused expression (28c) is the singleton containing the ordinary semantic value, which maintains the same type for the alternative values of all expressions — focused or unfocused. For purpose of illustration, the general schema (28) is instantiated for proper names in (29) and for intransitive verbs in (30): (28) a. ¦¦"¦¦O=¦¦"F¦¦O b. ¦¦"F¦¦A=ALT(¦¦"¦¦O)=Dtype(¦¦"¦¦O) c. ¦¦"¦¦A={¦¦"¦¦O} (29) a. ¦¦c¦¦O=¦¦cF¦¦O=c¹ ∈ De b. ¦¦cF¦¦A=ALT(c¹)=De c. ¦¦c¦¦A={c¹} (30) a. ¦¦V¦¦O=V¹ ∈ D〈e,t〉 b. ¦¦VF¦¦A=ALT(V¹)=D〈e,t〉 c. ¦¦V¦¦A={V¹} The interpretation of composition rules must be formulated in both ordinary and alternative semantics. For the time being, there is only one composition rule: the application of a predicate to its arguments. The ordinary semantic function of this composition is functional application as in (31). The alternative function of functional application (32) is more complex since it must warrant that the alternatives that are generated by a focused expression can be projected. It is a set formed by all possible expressions X(Y) that are derived from the application of an element X of the first alternative set to an element Y of the second alternative set. (31) ¦¦" $¦¦O=¦¦"¦¦O(¦¦$¦¦O) (32) ¦¦" $¦¦A={X(Y)¦ X∈¦¦"¦¦A, Y∈¦¦$¦¦A} For instance, the application of a predicate to its focused argument is the

 Klaus von Heusinger

functional application (33a) of its meaning to the meaning of the argument. The alternative semantics (33b) denotes a set that includes all expressions that result from the application of the predicate to one of the alternatives to the argument. The alternative set (33c) generated by the VP talk to samF includes the interpretations of all VPs of the form talk to y, where y is an alternative value to Sam. This is the set of individuals that have the property of talking to someone. (33) a. ¦¦V(cF)¦¦O =V’ (c¹) b. ¦¦V(cF)¦¦A ={X(y)¦ X ⊆ ¦¦V¦¦A, y ∈ ¦¦cF¦¦A} = {X(y)¦ X ⊆ {V¹} y ∈ ALT(c¹)} c. ¦¦talk to samF¦¦A={X(y)¦ X ⊆ ¦¦talk¦¦A, y ∈ ¦¦samF¦¦A} ={talk¹ (y)¦ y ∈ ALT(s)} ={d¦ ∃y ¦ talk¹ (y)(d)} The definition of the meaning (34) for the (adverbial) focus-sensitive operator only operates on both aspects of the meaning of an expression ". When applied to a VP, the ordinary meaning ¦¦VP¦¦? expresses the presupposition, whereas the alternative meaning ¦¦VP¦¦! determines the domain of quantification for the operator. There is no property in the set of alternatives that holds of x other than the property that is identical with the ordinary meaning. Here, the operator does not need two disjoint parts of the meaning of the expression as in the LF-movement account. It rather works with both dimensions of the meaning. Thus, the focused expression is not directly involved in the semantics of the operator. It merely generates alternatives, which then are projected to the alternative meaning of the whole phrase. (34) ¦¦only VP¦¦O=8x [¦¦VP¦¦O(x) &

P ∈¦¦VP¦¦! P(x) → P=¦¦VP¦¦O]]

We can now analyze sentence (35). In (35a), the focused expression sueF generates a set of alternatives, whereas the alternative interpretations of Mary, John and introduce form singletons containing the ordinary meaning. The ordinary semantics of the application of the predicate introduce to its arguments Sue and John yields the property introd¹ (s)(j), as in (35c). The alternative value of this application is the set of properties consisting of introducing someone (i.e., an alternative value to Sue) to John. The semantics of only asserts in (35d) that there is only the one property, which consists of introducing Sue to John (and there is no other property of introducing someone else to John). This combines in (35e) with the subject and yields the correct semantic representation for the sentence, namely that Mary introduces Sue to John. Furthermore, for all predicates that are formed by the description introduce someone to John if they hold of Mary, then they are identical with the property of introducing Sue to John.

Information structure and the partition of sentence meaning1 

(35) Mary VP[only VP[introduced SueF to John]] a. ¦¦SueF¦¦O=s ¦¦SueF¦¦A=ALT(s)=De b. ¦¦Mary¦¦O=m ¦¦Mary¦¦A={m} ¦¦John¦¦O=j ¦¦John¦¦A={j} ¦¦introduce¦¦O=introd¹ ¦¦introduce¦¦A={introd¹} c. ¦¦introduced SueF to John¦¦O=introd¹ (s)(j) ¦¦introduced SueF to John¦¦A={introd¹ (x)(j)¦ x ∈ ALT(s)} e.g. {introd¹ (s)(j), introd¹ (s)(j), introd¹ (j)(j), . . .} d. ¦¦only introduced SueF to John¦¦O=8x [introd’ (s)(j)(x) & P ∈{introd¹ (y)(j)¦ y ∈ ALT(s)} P(x) → P=introd¹ (s)(j)]] e. ¦¦Mary only introduced SueF to John¦¦O=introd¹ (s)(j)(m) & P ∈{introd¹ (y)(j)¦ y ∈ ALT(s)} P(m) → P=introd¹ (s)(j)]] . A problem for focus semantics Focus semantics, like LF-movement theories or Alternative Semantics, are exemplified with proper names. However, proper names are quite exceptional because they do not contain linguistic material that could interact with grammatical structure. The only linguistic information they contribute is that they are of type e. For instance, the focused proper name Sue in (36) are associated with alternatives that are of the same type, namely type e. In LF-movement theories, these alternatives form the domain of quantification for the operator only. Thus, (36) is true if nobody but Sue is such that Sam talked to her or him. In Alternative Semantics, the alternatives are projected to the VP level and the operator only quantifies over VP-alternatives, yielding — in principle — the same truth conditions as in the LF-movement approach. Here, we investigate association with focus in definite NPs in order to understand more about the interaction between the semantics of focus and the semantic contribution of complex NPs as in (37)–(42): (36) Sam only talked to [sueF]NP. (37) Sam only talked to [john’s spouse]NP. (38) Sam only talked to [the first american astronaut in space]NP. (39) Sam only talked to [john’s mother]NP. (40) Sam only talked to [the woman who introduced bobF to John]NP. (41) Sam only talked to [john’s sister] NP. (42) Sam only talked to [the dutchF professor]NP. (36) can be paraphrased as (36¹) indicating that in a formal analysis the focus-

 Klaus von Heusinger

sensitive operator quantifies over alternatives to the focused expression Sue. We could also use this strategy for (37) and (38): In (37), we quantify over persons who have a spouse and, in (38), we quantify over nationalities. (36¹) Nobody but Sue is such that Sam talked to him/her. (37¹) Nobody but John is such that Sam talked to his/her spouse. (38¹) There is no nationality but being American such that Sam talked to the astronaut of that nationality. While this strategy seems feasible for these two examples, the correspondent paraphrase (39¹) for (39) predicts that John is the only child of his mother. In a situation in which Sam talked to Ann and to nobody else, and Ann is the mother of John and of Bob, the quantification over alternatives to the possessor, as in (39¹), predicts contrary to fact that sentence (39) is false. For the operator only does not quantify over alternatives to John, but over alternatives to a larger constituent, e.g. john’s mother as illustrated by the paraphrase in (39¹¹). Similarly, the paraphrase (40¹) predicts contrary to fact that sentence (40) is false if Sam had talked to Diana and to nobody else, and Diana had introduced Bob to John and Bill to John. Again, the operator only quantifies over the set of women who introduce someone to John, rather than over alternatives to Bob. (39¹) Nobody but John is such that Sam talked to his/her mother. (39¹¹) No mother but John’s mother is such that Sam talked to her. (40¹) There is nobody but Bob such that Sam talked to the woman who introduced him/her to John. (40¹¹) There is no women who introduces someone to John but the woman who introduces Bob to John such that Sam talks to her. A second issue concerns the uniqueness condition expressed by the definite NP, which claims in (41) that John has only one sister and in (42) that there is only one Dutch professor at the contextual given situation. However, this uniqueness condition cannot be maintained in the set of alternatives, which is illustrated by (41¹) and (42¹): In a situation where Sam talked to John’s sister and to one of the two sisters of Bob and to no one else, the paraphrase (41¹) predicts contrary to fact that (41) is true, since Bob sisters are not in the alternative sets. The expression Bob’s sister is not defined because it violates the uniqueness condition for definites. Similarly, in a situation where Sam talked to the Dutch professor and to one of three German professors the analysis in (42¹) predicts that the sentence is true because it does not contain any German professors. It is obvious that the domain

Information structure and the partition of sentence meaning1 

of quantification in (41) is formed by sisters, rather than by unique sisters of a person, and the operator in (42) quantifies over professors, and not over unique professors with respect to their nationality. The paraphrases in (41¹¹) and (42¹¹) illustrate that the alternatives to a definite NP do not carry the uniqueness condition of the ordinary meaning of the definite NP. (41¹) There is no one but John such that Sam talked to his/her sister. (41¹¹) There is no sister but John’s sister such that Sam talked to her. (42¹) There is no nationality but being Dutch such that Sam talked to the professor of that nationality. (42¹¹) There is no professor but the Dutch professor such that Sam talked to him. This informal discussion of the computation of alternatives (illustrated in terms of Alternative Semantics, but it holds for the moving approach as well, see Krifka 1996) illustrates the problems of all approaches to focus that assume a focusbackground structure, which provides the arguments for the focus particles (for the detailed argument compare von Heusinger 1997, 1999: ch. 5).

.

A Foreground–Background Semantics

What is called for is a discourse semantics that provides the adequate semantic objects that can be used as arguments of focus operators. In contrast to the dichotomy of the sentence in focus and background I propose two representational levels, the foreground and the background. The latter is what Jackendoff and others call the p-skeleton. The foreground representation contains all material supplied by the sentence. The background contains the foreground minus the focused expression which is substituted by designated variables. The interpretation of the foreground yields the ordinary meaning, while the existential closure over the designated variables of the background yields the alternatives. This is informally illustrated by the interaction of an adverb of quantification with a focused constituent, as in (43). The operator ranges over sets that are constructed from the foreground and the background representation as in (43b). (43) Sam usually takes fredF to the movies. a. For most times in which Sam takes someone to the movies, he invites Fred to the movies. b. Most {t ¦ Sam takes someone to the movies at t} {t ¦ Sam takes Fred to the movies at t}

 Klaus von Heusinger

(44) Schematic mapping relations for adverbs of quantification Sam usually takes FRED to the movies Sam takes X to the movies Sam takes Fred to the movies usually MOST ({s|∃X T(s, X, to-m}, {s|T(s, f, to-m})

However, I do not define focus operators as ranging over denotations of these two representations, but as ranging over the representations themselves. They express relations between the different representations of a sentence. Since the background contains the same material as the foreground, save the focused expressions which are replaced by designated variables, one can describe the relation between background and foreground in terms of underspecification: The background is underspecified with respect to the foreground. The adverb usually can be defined as ranging over functions that map the underspecified background onto the specified foreground, as in (45). (45) Foreground–Background representation for focus Sam usually takes FRED to the movies Sam takes X to the movies Background

MOST(f)

Sam takes Fred to the movies Foreground

Alternatively, one can also represent the two representations as two DRSs that differ with respect to the representation of the focused element. In the background it is represented as the variable X, while in the foreground representation its content is fully represented. The adverb of quantification is translated into an operator over (possible) functions that map the background onto the foreground. (46) a. Sam usually takes fredF to the movies. b. x, y x, y Most h(X) Sam(x) X(y) x takes y to the movies

Sam(x) Fred(y) x takes y to the movies

The theory of information structure sketched in this section radically differs from

Information structure and the partition of sentence meaning1 

the approaches discussed earlier, which present the information structure as a dichotomy of the sentence content into focus and background. As it was argued, the psychological gestalt-notion of figure-ground has haunted linguistic for some time. And if one likes to use this often misused metaphorical concept, then I suggest it should be used as in the same way Caffi (1997: 437) applied it to illustrate pragmatic presuppositions: In order to clarify the concept of presupposition, some authors have compared speech with a Gestalt picture, in which it is possible to distinguish a ground and a figure. Presuppositions are the ground; what is actually said is the figure. As in a Gestalt picture, ground and figure are simultaneous in speech; unlike the two possible representations in the Gestalt picture, speech ground and figure have a different status, for instance with respect to the possibilities of refutation. What is said, i.e., the figure, is open to objection; what is assumed, i.e., the ground, is ‘shielded from challenge’ (Givón 1982: 101). What restricts the analogy is the fact that discourse is a dynamic process; the picture is not. When communicating, one is constantly asked to choose what to put in the foreground and what in the background. Discourses and texts are therefore multilevel constructions. Presuppositions represent at least a part of the unsaid.

Like presuppositions, the background expresses another aspect of the sentence, but not as a distinct unit of the sentence, but as a distinct representation of the sentence. To summarize the main assumption of the Foreground–Background semantics for information structure: 1. The information structure of a sentence consists of the foreground and the background. Both terms refer to representations at a discourse representational level. 2. The foreground corresponds to the representation of the whole sentence (modulo certain particles), while the background corresponds to the same representation save for the focused expressions, which are substituted by designated variables. 3. This information structure is described in the framework of DRT, i.e. in an explicit linguistic theory of sentence and discourse semantics. 4. The information structure into foreground and background provides important discourse objects (DRSs, construction trees, etc.) that are essential for the definition of discourse relations between sentences in a discourse. 5. The classical concept of givenness is understood as a structural condition of DRSs at the representational level. More precisely, givenness can be defined in terms of isomorphy between parts of DRSs. In other words, an expression is given if it is already present in the discourse representation. 6. In this view, the classical informational concept focus does not constitute a basic unit. The focused expression merely indicates the difference between the

 Klaus von Heusinger

background (i.e. the given material) and the foreground (i.e. the whole sentence). The focused expression is represented as a designated variable in the representation. 7. Focus-sensitive particles and adverbs of quantification are translated into operators that range over functions from the (underspecified) background onto the fully specified foreground. In this way, association with focus is reconstructed as a discourse operation similar to contrast or information focus. These main assumptions of the Foreground–Background semantics have led to a different view of information structure. Information structure is understood as part of discourse semantics, and therefore, as part of linguistic description. Information structure certainly effects sentence processing, psychological models and computational questions of language, but in the way described here, it is a linguistic level with linguistic objects. The particular view defended here might have raised more questions than it has solved. In particular, it is still an open issue over what kind of domains focus operators range: alternatives, p-sets, presuppositions, DRSs or functions between DRSs. Certainly more research is necessary to elaborate this question.8 Furthermore, the theory presented has to be tested against a wide range of data. Another point concerns the different levels of the ‘‘unsaid’’ such as presupposition, background, implicatures etc. There are subtle differences between these notions and it seems that a sentence contributes much more to a discourse than its foreground and its background.

Notes * This article is the revised version of a talk that I gave at the conference Bridges and Interfaces: Form, Meaning and Function in Prague. I received very valuable comments from the audience. In particular, I like to thank Eva Hajicˇová and Petr Sgall, who gave me very helpful comments for revising the paper. All remaining shortcomings are mine. This research was supported by a Grant from the Dr. Meyer-Struckmann-Stiftung and a Heisenberg-Grant from the German Science Foudation. . The presentation of the subject–predicate structure in these four fields are motivated by the lexicon article of Garver (1967). . Ries (1894) gives a very critical evaluation of traditional and logical definitions of the sentence. Ries (1931) collected more than 140 definitions for a sentence and showed that none of them can be used as a basic tool for defining syntax. . Depending on the intonation of the spoken sentence Bill was hit by John certain preferences can be indicated. . Van Heijenoort gives the following English translation of Frege ([1879] 1967: 7): ‘‘In particular, I believe that the replacement of the concepts subject and predicate by argument

Information structure and the partition of sentence meaning1 

and function, respectively, will stand the test of time. It is easy to see how regarding a content as a function of an argument leads to the formation of concepts. Furthermore, the demonstration of the connection between the meanings of the words if, and, not, or, there is, some, all, and so forth, deserve attention.’’ . Ammann (1928: 2): ‘‘Und in der Tat läßt sich aus dem Begriff der Mitteilung heraus auch eine Deutung jener Zweiheit gewinnen, die wohl am engsten mit dem Begriff des Satzes zusammenhängt: der Zweiheit von Subjekt und Prädikat — wie dies zuerst wohl G. von der Gabelentz getan hat.’’ . The concept or the metaphor of figure-ground is very pervasive in several disciplines such as linguistics and philosophy, to name only the most relevant for this study: Givón (1982: 128) uses it for describing his understanding of language: ‘‘The central fact that epistemology must eventually contend with, I believe, is the rise of temporary, illusive but nevertheless ‘real’ islands of relative firmament and order out of the inherently chaotic universe of experience. Ultimately, I believe, figure-ground pragmatics — the idea that the picture is stable and ‘real’ only as long as the frame remains fixed — must play a central role in such an enterprise.’’ Reinhart (1984: 779) explains the structure of narration with this concept: ‘‘The distinction between foreground and background in narrative has for its counterpart the figure-ground distinction proposed in gestalt theory.’’ Dretske (1972: 411) uses this metaphor in his description of contrastive focus: ‘‘What distinguishes contrastive statements is that they embody a dominant contrast, a contrastive focus, a featured exclusion of certain possibilities. Something similar to a figure-ground distinction is at work in these statements.’’ And finally Caffi (1997: 437) employs this notion to illuminate the structure of presupposition: ‘‘In order to clarify the concept of presupposition, some authors have compared speech with a Gestalt picture, in which it is possible to distinguish a ground and a figure. Presuppositions are the ground; what is actually said is the figure.’’ . There is, however, a certain connection between a psychological and a purely communicative and linear ordering. The psychological subject is often identified with the ‘‘justheard’’, while the psychological predicate is identified with the ‘‘expected’’ (von der Gabelentz 1901: 369): ‘‘Offenbar ist es dies, dass ich erst dasjenige nenne, was mein Denken anregt, worüber ich nachdenke, mein psychologisches Subject, und dann das, was ich darüber denke, mein psychologisches Prädicat, und dann wo nöthig wieder Beides zum Gegenstande weiteren Denkens und Redens mache.’’ . Von Heusinger (to appear) discusses the relation between the information structure developed here and the discourse structure in terms of Discourse Representation Theory.

Bibliography Ammann, H. [1928] 1962. Die menschliche Rede. Sprachphilosophische Untersuchungen. 2. Teil: Der Satz. Darmstadt: Wissenschaftliche Buchgesellschaft. [Reprint of the edition Lahr im Schwarzwald: Moritz Schauenburg] Caffi, C. 1997. ‘‘Presupposition, Pragmatic‘‘. In Concise Encyclopedia of Philosophy of Language, P. Lamarque (ed), 437–44, Oxford: Pergamon.

 Klaus von Heusinger

Chafe, W. 1976. ‘‘Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View’’. In Subject and Topic, C. Li (ed), 25–55, New York: Academic Press. Chomsky, N. 1971. ‘‘Deep Structure, Surface Structure, and Semantic Representation’’. In Semantics, D. D. Steinberg and L. A. Jakobovitz (eds), 193–217, Cambridge: CUP. Daneš, F. 1970. ‘‘One Instance of Prague School Methodology: Functional Analysis of Utterance and Text’’. In Method and Theory in Linguistics, P. Garvin (ed), 132–46, Paris: Den Haag. Dretske, F. 1972. ‘‘Contrastive Statements’’. The Philosophical Review, 411–374. Firbas, J. 1964. ‘‘On Defining the Theme in Functional Sentence Analysis’’. Travaux Linguistique de Prague 1, 267–80. Frege, G. [1879] 1977. ‘‘Begriffsschrift. Eine der arithmetischen nachgebildete Formelsprache des reinen Denkens’’. In Begriffsschrift und andere Aufsätze Mit E. Husserls u. H. Scholz’ Anmerkungen, I. Angelelli (ed), 2nd ed. Darmstadt: Wissenschaftliche Buchgesellschaft. — English translation in: J. van Heijenoort (ed). From Frege to Gödel. Cambridge/Mass.: Harvard University Press. Gabelentz, G. von der 1869. ‘‘Ideen zu einer vergleichenden Syntax. Wort und Satzstellung’’. Zeitschrift für Völkerpsychologie und Sprachwissenschaft 6, 376–84. Gabelentz, G. von der 1901 Die Sprachwissenschaft ihre Aufgabe, Methoden u. bisherigen Ergebnisse. 3rd ed.; reprint of the 2nd ed. Leipzig 1901. Tübingen: Narr. Garver, N. 1967. ‘‘Subject and Predicate’’. In The Encyclopedia of Philosophy Vol. 7, P. Edwards (ed), 33–6. Givón, T. 1982. ‘‘Logic versus Pragmatics, with Human Language as Referee: Toward an GEmpirically Viable Epistemology’’. Journal of Pragmatics 6, 81–133. Hajicˇová, E., Partee, B. and Sgall, P. 1998. Topic–Focus Articulation, Tripartite Structures, and Semantic Content. Dordrecht: Kluwer. Halliday, M. 1963. Intonation and Grammar in British English. Den Haag: Mouton. — Reprints of: ‘‘The Tone of English’’. In Archivum Linguisticum 15.1 (1963): 1–28. And: ‘‘Intonation in English Grammar’’. In Transaction of the Philological Society 1963: 143–69. Halliday, M. 1967. ‘‘Notes on Transitivity and Theme in English. Part 1 and 2’’. Journal of Linguistics 3, 37–81; 199–244. Heusinger, K. von 1997. ‘‘Focus in Complex Noun Phrases’’. In Proceedings of the Eleventh Amsterdam Colloquium, P. Dekker and M. Stokhof and Y. Venema (eds), 49–54, ILLC/Department of Philosophy, University of Amsterdam. Heusinger, K. von 1999. Intonation and Information Structure. The Representation of Focus in Phonology and Semantics. Habilitationsschrift. University of Konstanz. Heusinger, K. von (to appear). ‘‘Information structure and discourse semantics’’. In Prague Bulletin of Mathematical Linguistics. Hirst, D. and Di Cristo, A. 1998. Intonation Systems. A Survey of Twenty Languages. Cambridge: Cambridge University Press. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge/MA: MIT Press. Jacobs, J. 1983. Fokus und Skalen. Zur Syntax und Semantik von Gradpartikeln im Deutschen. Tübingen: Niemeyer.

Information structure and the partition of sentence meaning1 

É. Kiss, K. 1995. ‘‘Introduction’’. In Discourse Configurational Languages, K. É. Kiss (ed), 3–27, New York; Oxford: Oxford University Press. Krifka, M. 1992. ‘‘A Framework for Focus-Sensitive Quantification’’. In Proceedings from the Second Conference on Semantics and Linguistic Theory (SALT 2), C. Barker and D. Dowty (eds.), 215–36, Ohio State University. (Working Papers in Linguistics No. 40) Krifka, M. 1996. Frameworks of the Representation of Focus. ESSLI 96 Conference on Formal Grammar. Mathesius, V. 1929. ‘‘Zur Satzperspektive im modernen Englisch’’. Archiv für das Studium der modernen Sprachen und Literaturen 155, 200–10. Paul, H. 1880. Prinzipien der Sprachgeschichte. 5th ed. Halle: Niemeyer. Peregrin, J. 1995. ‘‘Topic and Focus in a Formal Framework’’. In Discourse and Meaning. Papers in Honor of Eva Hajicˇová, B. H. Partee and P. Sgall (eds), 235–54, Amsterdam: Benjamins. Prince, E. 1981. ‘‘On the Inferencing of Indefinite-this NPs’’. In A. Joshi and B. Webber and I. Sag (eds), Elements of Discourse Understanding, 231–50, Cambridge: CUP. Prince, E. 1986. On the Syntactic Marking of Presupposed Open Propositions. Parasession Papers. CLS 22, 208–22. Reinhart, T. 1982. ‘‘Pragmatics and Linguistics. An Analysis of Sentence Topics’’. Philosophica 27, 53–94. — Preprint in: IU Linguistics Club. Reinhart, T. 1984. ‘‘Principles of Gestalt Perception in the Temporal Organization of Narrative Texts’’. Linguistics 22(6), 779–809. Ries, J. [1894] 1967. Was ist Syntax? Ein kritischer Versuch. Darmstadt: Wissenschaftliche Buchgesellschaft. — Reprint of 2nd ed. Prague (1927); 1st ed. Marburg: Elwert. Rooth, M. 1985. Association with Focus. PhD Dissertation at University of Massachusetts, Amherst. Graduate Linguistics Student Association (GLSA): University of Massachusetts, Amherst. Saussure, F. de [1916] 1968. Cours de linguistique générale. Edited by v. C. Bally and A. Sechehaye. Paris: Payot. Sgall, P., Hajicˇová, E. and Benešová, E. 1973. Topic, Focus and Generative Semantics. Kronberg/Taunus: Scriptor. Sgall, P., Hajicˇová, E. and Panevová, J. 1986. Meaning of the Sentence in its Semantic and Pragmatic Aspects. Edited by Jacob Mey. Dordrecht: Reidel. Stechow, A. von 1991. ‘‘Current Issues in the Theory of Focus’’. In Semantik. Ein internationales Handbuch der zeitgenössischen Forschung, A. von Stechow and D. Wunderlich (eds), 804–25, Berlin: de Gruyter. Vallduví, E. 1990. The Informational Component. PhD Dissertation. University of Pennsylvania. Weil, H. [1844] 1978. The Order of Words in the Ancient Languages Compared with that of the Modern Languages. New edition with an Introduction by Aldo Scaglione. Amsterdam: Benjamins. — Translation of: De l’ordre des mots dans les langues anciennes comparées aux langues modernes. Paris 1844.

Section IV

General views

Freedom of language Its nature, its sources, and its consequences

Petr Sgall Prague

.

Conventions and norms in natural language

. Attitudes towards norms Natural languages exhibit several layers of norms or conventions, which correspond to D. Lewis’ (1969) characterization of ‘convention’. However, languages are not rigorously restricted by these norms, most of which lack an explicit formulation (the main exception in this respect concerns point 2. below). The layers of norms differ both in the conditions or factors of their acquisition and in the respective sanctions. Among these layers there are: 1. the norm of the language (codified or not), acquired on the basis of innate properties (which perhaps are not as complex as assumed in the Minimalist theory; cf. Sgall 1998; 2000) and of the environment, in part deliberately; the speakers’ adherence to this norm is determined first of all by their desire to be well understood without difficulties (without much time and effort to be devoted to decoding and understanding); cases have been found in which the speaker disobeys a rule for the sake of becoming well understood (J. McCawley, p.c.); 2. the codified norm of the Standard, consisting in a relatively highly explicit formulation of the basic norm 1., usually modified in a larger or smaller number of points (mostly concerning details of morphemics and similar grammatical issues); the acquisition of the codified norm is based on recognized authorities, including school education, with further prescriptive activities or without them; it should be reflected in linguistics that some speakers adapt more easily and more thoroughly than others to the requirements of school and of other instances taking care of the Standard (risking, in the extreme, even to become ridiculous in informal situations), while others do not adapt that easily, sticking to the colloquial usage they had adopted in their childhood; as Joos (1961: sect.

 Petr Sgall

3) states, it is ‘‘the highest compliment possible among mature people’’ to use intimate style, thus leaving the usual ‘‘guardedness’’; furthermore, it deserves more attention that conditions for these attitudes differ from one language community to the other (cf. Sgall et al. 1992); the sanctions concern means specific to school (marks or other kinds of evaluation), to editorial activities, criticism, and so on; 3. regularities and conventions in the domain of discourse, especially the conventional and conversational implicatures, postulates and maxims of Grice (1975) and others, and the conditions of different kinds of speech acts in the sense of Austin (1962) and Searle (1969;1983); the usage of these regularities is acquired on the basis of rational attitudes, by imitation, in part also by education; the respective sanctions belong to the domains of easy understanding and of the appropriateness of behavior; 4. in the realm of style, attitudes towards 1.–3. get differentiated more subtly, which is based on imitation and deliberate learning; here again, the appropriateness of behavior, one’s identification with a social group and the effectiveness of messages addressing members of the same or of another group are relevant. . No strict limitations An individual’s adherence to any of these degrees of norms and conventions is not strictly sanctioned. Languages are sharply restricted neither by their grammatical (lexical, phonological) rules, by a codification of their norm, nor by implicatures or style. Not even the world we live in and speak of (or our knowledge of the world) is immediately relevant: not only in fairy tales or when e.g exaggerating, the speakers are free to depart from the plausibility of the literal content, as far as they believe to be understood. Further such irregularities concern on the one hand utterances that can occur when discussing language itself, on the other hand when using conditional, negation, and so on. Thus, questions such as whether wind can open a door, or whether someone might have the right for the sky to be blue, and so on, and so forth, do not actually concern language: this very statement documents that sentences with these and similar combinations of lexical units are not only grammatically well formed, but also display their meaning (in the linguistic sense, which can more precisely be understood as the underlying sentence structure), even though some of such sentences carry only propositions not assigning the value ‘true’ to any possible world (except in metaphorical uses). Thus (as has been mentioned in our discussions with S. Kuno, Ch. Fillmore, R. Jackendoff and others), e.g., Martin provided for Mary to live under a blue sky for whole years cannot express a true statement, but with another modality the possibility of an

Freedom of language

assignment of the value ‘true’ may be provided, cf. — with different degrees — such modalities as he wanted to . . ., he promised . . ., he could not . . . It is true that in several countries patriotic political tendencies have led to the rise of language laws concerning protection of the given language against foreign influence and/or limitations concerning the use of its non-standard forms. However, mostly such restrictions are neither far-reaching in their range, nor accompanied by serious sanctions, and in some cases formulations of the laws may appear ridiculous (e.g. when the community councils in a country with large rural areas and strong minorities are formally required to use the standard form of the national language in their sessions). No wonder that such laws are not taken too much seriously by relevant layers of the language communities. It follows that F. de Saussure’s comparison of language with the game of chess is not precise, since chess rules differ from language conventions in being compulsory (I owe this remark to J. Peregrin, p.c.). . Need to be understood The question then arises what are the main factors that determine the degree of adherence to the norms and conventions of language. Not having the possibility to go into issues such as individual sociolinguistic attitudes or communicative situations, we want to concentrate on one of the factors. Speakers are free to depart from the norm (or from the plausibility of the literal content), as far as they believe to be understood easily enough (cf. Sgall et al. 1986: 32 f.); the speakers themselves decide which kinds and numbers of figurative expressions, of hyperbolic formulations, of deletions, and so on, are appropriate or tolerable in the given circumstances (with this or that addressee, context, content, etc.). Among the factors that are immediately relevant in this respect, especially for deletion, there is repetition (including quotation), and also the close connection between utterances representing a direct continuation of preceding context, including coordinated clauses or answers to questions. In utterances displaying the character of such a direct continuation, deletions of many kinds are possible (e.g. Jim thinks bananas or Flowers to Mary are fully acceptable after What has Jane bought yesterday? or in He gave a bracelet to Nicole; a book to John; flowers to Mary, respectively). The possibilities of systematic and other deletions are manifold and they are difficult to specify as being restricted by grammatical rules, cf. Chomsky (1975: 119). However, certain language specific restrictions do exist even in the case of such a contextual deletion, cf. the shortened answers in E. Have you met Harry? — I have. versus Cz. Potkals Harryho — Potkal [met], where the repetition of either the auxiliary in Czech, or the lexical verb in English (without repeating the object) should be described as grammatically excluded. Let us add illustrations of these possibilities from different layers of Czech:



 Petr Sgall

a.

Czech is a pro-drop language, i.e. (especially in its relatively formal styles) the subject pronoun has a zero form (in any of the three persons); b. as in many languages, also in Czech the whole topic of an utterance in a dialogue may often be left out, cf. (Kde jsi ho našel? ‘Where have you found him?’) — Doma (‘At home’). (Procˇ to tam dáváš? ‘Why do you put it there?’) — Aby to nenašel Jirka (‘For George not to find it’ (Potkal jsi Milenu? ‘Have you met Milena’) — Potkal (lit. ‘Met’, i.e. ‘I have’). c. in a non-standard colloquial style there is the option to pronounce individual word forms with most different shortenings; these are limited neither to unaccented parts of the words, nor to their final or to their beginning parts, as can be documented by the fact that e.g. the word padesátník (‘fifty-heller coin’) can be pronounced either as pade or as sátník, leaving out the other half of the word; d. moreover, there are the well known cases of minimal deletions, which often represent not only individual events in the process of communication, but penetrate into the language system — cf. the loss of individual phonemes and/or even of marked distinctive features; thus, palatalization gets lost e.g. in the development of the Czech infinitives (dát, rather than dát’ from dáti, etc., for almost all verbs), or the colloquial pronunciation of the name Boticˇ (a brook in Prague, pronounced with [t’] in the standard, but with [t] in most everyday conversations) or of porád ‘all the time’ vs. the standard porˇád. The possibilities of deletion (as well as of ‘neglected speech’ and of other layers of redundancy reduction) are highly manifold and the speakers’ choice consists of many options, the repertoire (and hierarchy) of which cannot be easily described sytematically. The speakers find their ways by choosing among the different layers of economy and of redundancy. Also the choice of more or less precise expressions belongs here (with underspecification — ambiguity and indistinctness — being possibly resolved by the addressees on the basis of context). . Tradition and innovations A human being does not have a free choice in deciding which language to adopt. As Stich (1991) points out, the child is pushed to learn the language of its environment and gets acquainted, step by step, with its basic structure and its idiosyncrasies (cf. also McCawley 1992 on the acquisition of complex syntactic structures). During this process the child has to abandon the systematic, analogical image of the language system it has internalized and to acquire all (or almost all) the historically determined exceptions and idiosyncrasies in the morphemics, syntax, lexicon, phraseology and other layers of the language. Being so equipped, the speaker may act against the tradition of the language and the group uniformity, making use of the principles of freedom, individual independence and

Freedom of language

creativity. Not only in poetry, in humor, etc., but also in everyday speech and then in other domains of communication the individual gets aware of his/her individual responsibility both in the choice of means of expression, style, etc., and in creating new ones or new combinations of them. Speakers often deliberately avoid on the one hand to use superfluous effort in expressing themselves and, on the other, to speak in a way that would not be easily understood and accepted as appropriate (both in what concerns norms of behavior and suitability for the given situation). However, they feel free to use most different kinds of daring innovations and of expressive strengthening (that may give rise to new synonyms or intensify the use of existing ones), to make their speech interesting, inventive, witty, thus maximizing the effectivity of their messages and the attractivity of their images, i.e. their potential influence. As the poet writes, ‘‘in our language we look for the most elementary freedom — to be able to express our most intimate thoughts. And this is the basis of any freedom’’ (Seifert 1984). These words certainly concern also the way of expression of our thoughts (and emotions, attitudes, and so on).

.

The sources of the freedom of language

The intrinsic freedom of language, i.e. the freedom of the speaker’s choice in all the aspects mentioned (and in others) follows from the interactive nature of language. Several aspects of this relationship may be briefly mentioned: a. speakers are human individuals with their free will and with different backgrounds and psychologies; as Bach (1996) puts it, ‘‘humans are not just speaking animals, they are also language-creating animals’’; the language norms constitute just a starting point for communication, as Arutjunova (1994) states (in a somewhat different connection); thus, speakers decide more or less consciously and consistently whether to adhere or not to this or that norm in a given point; together with the unlimited variability of communicative situations and of (situational and verbal) contexts this leads to an indefinite number of possible combinations of sentences occurring in a discourse, which cannot be captured by any collection of ‘‘text patterns’’ or ‘‘textemes’’ (although various kinds of hierarchies of these combinations certainly can be specified as soon as the factors exhibiting maximal impact are identified); speakers are free also in choosing the degree of their sincerity and forthrightness, as well as in the degree of precision in the semantics of their utterances (cf. esp. Dönninghaus (1999; in press); b. already the quite elementary possibility of repeating own or others’ formulations, of quoting, recalling, etc., leads to complications in the structure of language; one of the aspects of this effect is the fact that in repetitions and in



 Petr Sgall

formulations recalling a part of the preceding co-text the topic–focus articulation of sentences may change, and that any expression may appear in the topic of a sentence; even a focus sensitive particle with its ‘focus’ and scope may then be included in the topic (in the contextually bound part) of an utterance, cf. e.g. (If everybody expected the weather to deteriorate,) then even the most audacious participants might start thinking of finishing the expedition, where the ‘focus’ of even is (expressed by) the most audacious participants (which recalls everybody), whereas the focus of the sentence is might start thinking of finishing the expedition (cf. Hajicˇová, Partee and Sgall 1998); c. repetitions may give rise to further aspects of complicating the language structure: they evoke the possibility of deletions and of other shortenings mentioned above (as we have seen, there are deletions of most different kinds and layers, from dropping a phoneme down to the ellipsis of the whole topic of a sentence); repetitions are further connected with different kinds of modifications of repeated parts of formulations, since there naturally is a tendency not to repeat literally; here belongs e.g. the already mentioned shortening of answers (not to repeat too much from the question), as well as the expressive strengthening, which supports the use of synonyms; d. anaphoric and other quotations of (sectors of) co-text allow even for selfreference, certain kinds of which (instantiated by the classical Liar’s sentence) represent sentences not expressing any proposition (truth conditions); this is connected with the absence of a clear boundary line between object language and metalanguage (cf. also issues of logophoric contexts, as discussed by A. Kratzer and others).

.

The consequences of the freedom of language

. Variability of expression Among the consequences of the freedom of language there is, first of all, the well known high degree of variability of linguistic forms within language systems, which was discussed already by Mathesius (1911; cf. also the attention he paid in his later writings to what he called the speakers’ communicative needs), as one of the corner stones at the beginning of synchronic linguistics. The speaker chooses among the varying items, creates new collocations and words; what starts as an individual usage can find broader response, the individual’s choice can lead to a change in the system. The variability of the expression forms and of their functions leads to the possibility of a ‘teleological’ (goal oriented) development of the language system.

Freedom of language

The lexicon steadily gets enriched, with the individual speakers’ role coming to the fore, as stressed especially by Ch. Fillmore’s (1982; 1985) ‘semantics of understanding’ (see also Waszakowa 1997) and by cognitive linguistics (which, as Kiefer 1995: 99 points out, opens new horizons, although it should not be understood as substituting older trends in linguistic research). The lexicon keeps growing in always new idiomatic collocations and complex predicates, technical terms and other neologisms, which come from different sources, some of which are: 1. the historically emerging need to have a denomination for a new concept; this concerns technical terms, but also other words, be they borrowed from other languages (often in a narrower sense than they have in the source language, cf. Cz. kolaborant, which does not refer to a collaborator in general, but to a quisling) or created using productive means of word formation; 2. an emotionally marked synonym is first used individually for the sake of marking one’s speech as specific, then it gets stabilized in the language system as bearing a stylistic value of expressiveness, and eventually it may substitute the original, stylistically neutral word, cf. examples such as French tête ‘head’ from Lat. testa, which already in Colloquial (‘‘Vulgar’’) Latin substituted the older caput, or Cz. strašný, ohromný in the meaning of ‘big’. It is important to see that also grammatical means steadily get enriched both in what concerns new forms of expression and their functions: new complex prepositions and conjunctions emerge (i.e. get grammaticalized), etc. It is important to see also the emergence of new complex predicates; in contemporary Czech this process appears to be quite intensive, cf. such recently spreading verb complexes as e.g. in To jsme se nemeˇli v úmyslu ucˇit, lit. ‘This we had-not-theintention-to learn’ (built in analogy to To jsme se nechteˇli ucˇit ‘This we did not want to learn’), Dlužili jsme neˇjaké daneˇ židovské obci ‘We owed some taxes to the Jewish Community’ (Ota Pavel), Jen nám radili studovat chemii ‘They only recommended us to study chemistry’ (cf. Hajicˇová et al. 1998 on ‘Taglicht’s sentences’). Certain properties of these and other types of collocations, connected with the word order positions of their parts, document their lexicalized character; their parts are connected with each other more closely than just by regular syntactic relations, which means that they can best be understood as idiomatic phraseological units. . Language development Another layer of individual decisions determines, in many languages (i.e., in certain stages of their slow development based on choice in variation), the repeated reduction of grammatical morphemes, which is connected with the



 Petr Sgall

emergence of new function words, cf. the typological ‘Spirallauf ’ of G. von der Gabelentz. This typological spiral does not cover the whole domain of typological change and cannot be understood as the backbone of language development as a whole (for which the just mentioned enrichment in most different layers of language constitutes a more important line). In any case, the typological spiral plays an important role, and it can be seen that this aspect of development is closely connected both with the speakers’ tendency for economy and the necessity for them to express themselves with a certain degree of redundancy to be easily understood. The teleological aspect of language development, and thus also of language structure, was highly stressed by R. Jakobson, N. Trubetzkoy and other members of the classical Prague School; cf. esp. Trost (1989), and also Leška (1987;1994), who discusses sources of the looseness of the system of langue in this connection; as Sériot (1999) points out, this view was rooted in the influence of older Russian traditions. Although this aspect of language and its development is not strong enough to enable the linguists to predict individual changes or to specify the respective regulating mechanisms, it clearly is present, and, as we have just mentioned, it is relevant both for the synchronic study of language and for a description of its development. . Universality and its limitations The free variability of language means and their functions is a precondition for its universality. Without a certain degree of indistinctness of language meanings (i.e. of the units of the layer of functions of expressions in the language system) it would not be possible to capture with limited means the unlimited range of the world we perceive and speak of, cf. Marty (1908), Karcevski (1929), Mathesius (1942) or Putnam’s (1975) ‘stereotypes’ and the Gricean (1975; 1982) notions of ‘licence’ and ‘ideal limits’. The fuzzy or indistinct demarcation of linguistic meanings is not only a precondition of the universality of natural language, but also one of its consequences: the complexity of language makes it impossible for any exact rules to be precisely applied in communication. The universality of natural language is, in fact, restricted, cf. especially E. Keenan’s (1975) findings on relative clauses: both in Czech and in English a sentence with a relative pronoun cannot exactly render what colloquial Czech expresses by the particle cooccurring with an anaphoric pronoun e.g. in Ten pán, co jsem jeho a jeho dceru vcˇera potkal, lit. ‘The gentleman what him and his daughter I yesterday met’ (Czech shares this construction with certain southGerman dialects and e.g. with Hebrew). The source of another layer of semantic limitations is constituted by the systemic ordering of complementations, which always is present in the ‘underlying word order’ in the focus of a sentence (cf.

Freedom of language 

Sgall et al. 1986); e.g. the Czech sentence Pavel vždycky pu˚sobil svými výroky neˇjaké obtíže is translated precisely neither by Paul always caused some DIFFICULTIES with his statements (where the with-group belongs to the topic, although in Cz. it clearly is a part of focus), nor by Paul always caused some difficulties with his statements (with normal intonation, i.e. with the intonation center in the final position, where the difference in the degrees of communicative dynamism — and thus also in the preferred scopes — of the two parts of the focus gets inversed). Due to the difference in the systemic orderings of the two languages an actually true translation is impossible in such cases. Still another layer of limitations of the universality of natural language can be found in the fact that speakers may find lacunes in the repertoires of lexical units when they want to characterize certain objects or situations. Thus, e.g., we do recognize individual faces and distinguish them from each other, but we do not always find language means to describe the individual differences in detail (I owe this observation to B. H. Partee, p.c.). As analytic philosophy tells us (see e.g. Peregrin 1992), we are limited by the structure of our language in perceiving and understanding the world. To quote a poet again, we may recall Frynta (1993): ‘‘A nadto je nám rˇecˇ . . . metodou orientace ve veškerém jsoucnu’’ (‘And, moreover, language is for us a method of orientation in the whole universe’). Thus the limits of the freedom of language seem not to be narrower than those of the freedom of human mind. . Complexness of language Even with such limitations, the freedom of language together with its relative universality leads to its extreme complexness. It is not easy to imagine a system that would be fundamentally more complex than natural language. Let us just recall that this does not exempt the linguist from his obligation to describe language as being (seemingly paradoxically) maximally complex and, at the same time, organized along a core simple enough for a child to be acquired without explicit learning. We are committed to look for as simple a patterning of this core as possible, to avoid the assumption of an unnecessarily complex collection of ‘‘innate ideas’’; cf. Section 4 below for a discussion of this issue. With its freedom, language gets into situations which were not present in the primary conditions of its existence and for which its structure cannot easily be adapted, i.e. into situations causing collisions on most different levels. Many of such collisions come into being with new requirements yielded by the development of civilization, be it in its technical aspects or in requirements of different layers of communication in the society. This concerns individual lexical units, cf. the problems connected with the English pronouns he and she, or with you (with which the Plural function often

 Petr Sgall

has to be made clear by the use of you all, you guys, etc.). In German, similar difficulties accompany the Plural forms of the ‘polite’ pronoun Sie vs. the Plural Ihr; cf. also nouns such as German LehrerInnen or Czech ucˇitelé/ky, as compared to E. teachers: the E. nouns of this group do not distinguish gender, so that after an occurrence of e.g. his neighbor it may be difficult to decide which gender of the coreferring pronoun should be used (and whether we face an ambiguous noun).1 Grammar rules often show collisions in cases of morphemic exceptions, which with frequent words are broadly recognized and used (e.g. the suppletion of Cz. cˇloveˇk — lidé ‘(hu)man — people’ or Loc.Sing. ve dne vs. o dni/dnu ‘day (with different prepositions), but may recede if they concern relatively rarely occurring words, so that many speakers are not aware of their existence. In Czech, similar collisions concern the declension of proper names belonging to restricted inflectional classes; some of the family names originally were identical to common nouns and now differ from them in their oblique cases, such as Hrabeˇho vs. hrabeˇte (Gen./Acc. of ‘earl’, Švecem vs. ševce (Gen./Acc. of ‘shoemaker’); other cases are such as e.g. the Voc. of the local name Hradec, which according to the respective inflectional paradigm would be *Hradcˇe; however, due to the unproductivity of this paradigm (at least with inanimate nouns), the actually used (although of course not frequent) form is Hradci, analogical to the more productive declension class of stroj ‘machine’. Another group with similar collisions concerns the feminine family names of foreign origin, which do not always display the regular Cz. suffix -ová: with complex names such as ThurnTaxisová the suffix usually is used only at the latter part of the name (although according to the codified norm it should occur also with its first part), with some names ending in a vowel, Czech speakers are used to forms such as Dijkstra (although e.g. Parteeová does occur, perhaps due to the fact that it is used first of all by linguists).2 It is well known that along with these and similar extreme cases, there are many other kinds of collisions between different regularities in language structure. The influence of other languages may cause collisions such as those concerning the declension forms of different Czech loan nouns, or pronunciation and spelling in English, and so on. Other collisions concern conversion and classification of word classes, as well as their syntactic properties and conditions of their occurrence; cf. e.g. Cz. constructions such as já rád sladké ‘I like sweet (things to eat)’, with a word that is morphemically characterized as pronoun or adjective, but occupies the syntactic position of a finite verb (even with the corresponding analytic preterite já jsem rád). The hierarchy of the levels of language gets complicated by collisions of many different kinds. Let us just recall a few examples that are more or less typical for many languages:

Freedom of language 

a. There are cases in which an opposition that basically pertains to morphemics is used more or less systematically as also relevant for meaning. Although function words basically occupy the same position as function morphemes (their analytic character belongs to the level of morphemics, rather than to the underlying structure), their use occasionally may obtain semantically relevance. Thus, it deserves further discussion whether (or, under which conditions) in certain kinds of coordination structures a preposition perhaps is repeated only if relevant semantic conditions are met; this may concern the Czech construction pro prˇedsedu a pro rˇeditele (E. for the chairman and for the director), which is used only as referring to two different persons (K. Oliva, p.c.). b. In the so-called ‘second instance’ sentences (or, perhaps, in all sentences with a specific ‘narrow focus’), the intonation center (sentential stress) may fall on a syllable that prototypically does not carry the word stress (on a prefix in English, on an ending in Czech, etc.). Such a marked accentuation indicates that only one of the derivational or grammatical values carried by the given word form is in the focus, all the rest of the sentence belonging to its topic, referring to ‘given’ information, cf. Cz. (Ne že to udeˇlá.) On to udeˇLAL, cf. E. (It is not only that he will make it.) He HAS made it. c. Direct speech often is described as an object clause dependent on the introducing verb. While this approach is adequate for sentences such as Jim told us: You are invited, there are also cases in which the direct speech contains a string of more than one sentence (which can hardly be understood to function as a single dependent part of a complex sentence), and cases in which the introducing verb has another direct object than the direct speech, cf. Czech: ‘‘Meˇli byste tam prˇijít,’’ sdeˇlila nám Marie své prˇání (lit. You ought to come there,’’ told us Mary her wish). d. Idiomatic expressions, as was mentioned above, often behave (in the functional sense) as lexical units, although their morphemic form is that of a collocation of more than one words. However, there are several degrees between this behavior and the nature of a prototypical syntactic combination of words, i.e. degrees of grammaticalization. A similar scale can be seen there where primarily a simple opposition of an analytic grammatical form and a free syntagmatic combination is expected; cf. also the marked use of function words discussed under a. above. Further irregularities, which cannot be discussed here, are connected with quotations and other specific contexts, in which connotations determined by the morphemic forms of words, their etymologies, and so on, are relevant for communication. The underlying representations (or Logical Forms) are often claimed to be disambiguated. However, the boundary-line between ambiguity and indistinctness in the linguistic sense perhaps is not identical with the corresponding opposition

 Petr Sgall

as handled in logic or in formal semantics. Certain phenomena, such as N. Chomsky’s ‘systematic ambiguity’ (book as a text and as a physical object, warm weather vs. warm sweater, etc.), or the differences of group reading and distributive reading, as well as J. Hintikka’s branching quantifiers usually are not understood as distinguished by the structure of natural language (not even in its underlying structure), although they are seen as cases of ambiguity in formal semantics.3 Also the difference between the linguistic patterning of quantifier scopes (on the basis of word order, or of communicative dynamism) and their logical analysis points to the possibility that different degrees of underspecification are needed in the two kinds of description (cf. Hajicˇová et al. 1998: 86 f.).4 It may even be possible to ask whether the compositionality of natural language is complete, i.e. is not accompanied by peripheral disturbances in such domains.

. Core and periphery in linguistics and in mathematics . Transition zones and periphery We have seen that a high degree of openness of the system belongs to the consequences of the freedom of natural language, which is one of its constitutive properties. One of the manifestations of this openness may be seen in the existence of grey zones between various layers in language systems, as well as around these systems themselves, cf. the issue of degrees of grammatical well-formedness of sentences. The grey zones include layers in which lacunes, deviations and collisions of most different kinds occur. The resulting complexity of natural language, some aspects of which have occasionally been called ‘‘tremendous’’, leads to the question whether and how the relatively high degree of accessibility of language can be understood. How is it possible that children acquire the language of their community so easily? As is well known, N. Chomsky answers this question by postulating a rich (though ‘minimal’) set of ‘innate ideas’, underlying the ‘universal grammar’ that constitutes the core of each of the natural languages (as we know them today). The fundamental distinction between the core of language and its periphery (which does not understand the core as relatively large and the periphery as narrow in any aspect) has been well known in linguistics for the whole time of its modern development. The Neogrammarians distinguished between the ‘sound laws’ (Lautgesetze, regular phonetic changes) and the disturbing impact of ‘analogy’ or other factors, the classical Prague School worked with marked and unmarked (or primary and secondary) phenomena, and the ‘second generation’ of the School recalled the opposition of ‘center’ and ‘periphery’ in the title of the second volume of their ‘Travaux’ renewed in the 1960s: Les problèmes du centre et de la périphérie

Freedom of language

du système de la langue (Travaux linguistiques de Prague 2, Prague:Academia 1966). Often such a view contrasts with the standard mathematical approach, with which it is required more or less explicitly that the domain including both the core and the periphery should be observed from a generalizing point of view: first a domain is to be characterized that certainly would cover the whole of the relevant space, although perhaps this domain would be broader than necessary. It is supposed that the complex object can be described only on the basis of such a general framework. . Classical standpoints in mathematics and in linguistics The difference between these two fundamental standpoints may be schematically characterized by Figure 1.

Figure 1.

The standard mathematical approach starts with the specification of the outer rectangle and of all general conditions and factors relevant for the space defined in this way. On the other hand, the typical approach of linguistics as an empirical science starts with an examination of the inner ‘rounded’ space, in which the typical relationships obtain, and then proceeds to the specific study of the individual deviations, exceptions, collisions, and so on. On the one hand, then, there are approaches well founded on methods of mathematics and logic, which examine the underlying structure of sentences as being patterned in a way that can be adequately handled by means of type theory, lambda calculus, intensional (or, perhaps, situational) semantics. On the other hand, for a linguist educated in the classical structuralist research, this view appears to correspond to the outer quadrangle, i.e. to include ingredients which are not directly necessary for a description of the core of language. From this viewpoint these means, proper to formal semantics, are more useful for a semantic or semantico-pragmatic interpretation of underlying structures than immediately for an investigation of the patterning of these structures themselves; these means appear to be appropriate for an analysis of the structure of human intellect (‘mind’), which includes language inferencing (based on phenomena such as factual



 Petr Sgall

and contextual knowledge, which display different degrees of richness and consistency with different speakers and timepoints), rather than for the structure of language. The linguist then would prefer to look first of all for means directly needed to handle the core of the language system, and then to identify, step by step, the individual ‘marked’, deviant properties and phenomena of different languages. Preferably, the core should be handled by means patterned simply enough to represent the core of language as accessible for the child’s language acquisition. Another illustration may be offered by certain phenomena for the handling of which the mathematical complexity theory has been recalled. What I have in mind here is the high degree of ‘‘free’’ word order that is typical for inflectional languages such as Czech, Russian, or Latin. If I understand well, it can be maintained on the basis of complexity theory that the rich possibilities of word order patterns containing interruptions (discontinuities) of syntagms yield the impossibility of formulating a general procedure that would parse every Czech sentence in a polynomially restricted time; cf. Holan et al. (2000). However, from a viewpoint such as that characteristic of the Prague School approach mentioned above (or for another approach based on the opposition of center or prototype and periphery or ‘secondary cases’) it may be claimed that this disadvantage is not really crucial: If we may assume that an extremely large majority of the Czech sentences occurring in a properly compiled corpus (such as the Czech National Corpus) can be parsed in a relatively short time span, then it is not basically relevant that in extremely rare cases such utterances may be found for which this does not hold. (Let us recall that in any case such difficult conditions obtain if one wants not only to find one or more well-formed structures for a sentence, but also to disambiguate the sentence’s occurrence in a given context.) It seems then to be relatively adequate to work with underlying representations as tree-based structures the left-to-right order of whose nodes meets the condition of projectivity (or of the continuity of constituents). The prototypical case then can be seen in a regular correspondence between the underlying structure and the morphemic and phonemic form of the sentence, both in the left-to-right order of items and in the expression of lexical and grammatical underlying items by morphemes. The peripheral cases, represented by the irregular arms between the rounded space and the quadrangle in Fig. 1, have to be described step by step, using specific means with an additional cost.5 . An alternative in mathematics It is then of great importance that in contemporary mathematics there are schools whose fundamental approach appears to come closer to the empirical viewpoint characterized above than the approach of the standard formalisms based on classical set theory. In Czech mathematics, an approach of this type has been

Freedom of language 

formulated by Vopeˇnka (1979;2000), whose ‘alternative set theory’ is based on concepts such as indistinctness; while the core of his ‘semi-sets’ is clearly patterned, their periphery (close to the ‘horizon’) is characterized by uncertain membership and thus resembles the grey zone of the language system, or the various unclear boundaries within this system (with their deviations and collisions). The first attempt at specifying an application of this recent mathematical perspective to problems of linguistic indistinctness was offered by Novák (1993). It may be expected that a further elaboration of such approaches to those phenomena that constitute the effects of the freedom of language brings a more appropriate way of handling the problems connected with the openness of language systems than the way connected with the usual kind of mathematical thought. . Core, periphery and transition The opposition between core (or prototype) and periphery does not exhaust the problems concerned. As Holenstein (1975) notes, R. Jakobson’s (and Greenberg’s) implication laws display an important advantage, if compared to the prototype model: they make it possible to distinguish different degrees on a scale between very prototypical phenomena and those belonging to the outer periphery. In the Praguian typology of languages, based on the approach of V. Skalicˇka, this role belongs to probabilistic implication laws (see Sgall 1995 and the writings quoted there). The way of thinking based on ‘prototype’, ‘center’ vs. ‘periphery’, or on implication laws can be used to formulate explicit results in the shape of classical mathematical systems only with strict limitations; it appears to come closer to ‘semi-set’ based formalisms, especially if the ‘horizon’ is assumed to be relatively low in natural language. Describing language systems does not require to enumerate long scales of more and less well-formed structures, and in many respects it seems appropriate to postulate the horizon to lie not far beyond the number two. Thus, e.g., the position of the word accent is known to be determined as the first, the last or the last but one syllable (or mora) of a word form, or of a phonetic segment; similarly, if a complementation of a verb may occur more than once with a given governor (as, e.g., the temporal adverbial with all or, at least, many verbs), then it is possible to find examples with two or three of its occurrences (as in Yesterday Paul came late to us at night), but it is not easy to find more such occurrences in a single clause, although no grammatical rule determines such a limitation.

.

Conclusions

A linguistic descriptive framework may reflect the high degree of variability and openness of the described language systems in being oriented towards a maximally

 Petr Sgall

economical account of the core of language, acquired by children, and in displaying more expensive means for the description of more or less peripheral phenomena. The conviction of many logicians according to which such a pattern of thought as that of the proposition calculus is innate to humans (or that ‘‘natural numbers were created by God’’, in L. Kronecker’s words, see Weber 1893: 19) may be seen as a good reason for preferring linguistic descriptive frameworks that aim at capturing the core as being patterned in a very simple way. Looking for interfaces and bridges between the different existing frameworks, one might then be interested to see them formulated in a way open for a wider discussion and clearly pointing out what specific descriptive methods or viewpoints are present and what is seen as the core of language. If this core basically is looked for in the (underlying) structure of the sentence, then the adequacy and economy of its description in different frameworks should be examined and compared. A similar comparison is desirable as for the relative complexity of the treatment of relationships to other layers within language and beyond it — in the cognitive domain, in the process of communication and also in phonetics and acoustics. It might be of interest for such discussions to pay attention to frameworks claiming to be able do describe the core of language at an especially ‘low cost’: simple tree-based networks which can be univocally represented by parenthesized strings of symbols with indices taken from a fixed repertoire, relatively low number of nodes in the networks, much of grammatical information included in the lexicon, so that a small set of general principles (rules or declarative constraints) are sufficient to handle the core of language. Dependency based underlying syntax has been found not only to meet these conditions, but also to allow for an adequate account of the topic–focus articulation (information structure) as of a basic aspect of the language system. Such an integrated view of the core of language allows for an intuitively ‘natural’ explanation of such peripheral cases as those in which the opposition of topic and focus is relevant not only for the contextual combinability of sentences, but also for their semantics proper, even for their truth conditions (cf. the discussions on pairs of sentences such as He works on his dissertation in the weekends and In the weekends he works on his dissertation). The ‘aboutness’ relation between focus and topic is relevant for the intension of the sentence in certain cases, and this can be described without extensively increasing the cost of the core description if the mentioned opposition is seen as present in the underlying structures of sentences. An underlying sentence representation meeting these requirements, which might be based on the starting points that have been offered in the Functional Generative Description, could constitute an appropriate input for semantic(opragmatic) interpretation in the sense of formal semantics, handling the intension of the sentence with the use of lambda calculus, type theory and other suitable

Freedom of language 

devices. With a relativization to different contexts, the intension of the sentence (basically, a Carnapian function from possible worlds into truth values) could then be understood as underlying the content of an utterance (i.e. of a sentence occurrence) seen as an operation on the hearer’s state of memory, i.e. as determining a context-update potential. It may be added that the necessity for linguistics to work with notions such as ‘dependent’ or ‘obligatory’ does not contradict the fundamental character of the freedom of natural language. If the theory of Minimalism requires that the presence of individual syntactic items always has to be specifically licensed, it may be objected that such a view contradicts the very requirement of the economy of description, and thus of minimalism. There are no specific reasons to compare the nature of language in a straightforward way with ‘necessity’ in the domain of natural sciences (as done e.g. by Pazuchin 1983, whose views are rather unclear also e.g.in what concerns the relationship of grammatical correctness and truth). The structure of natural language has come into existence and has developed for millennia in a spontaneous (although partly goal-oriented) way under conditions determined by the process of human communication. It has not been constructed along the principles of logical analysis. Thus, there is no wonder that the peripheral parts of language structure exhibit complications of most different kinds and layers. As V. Skalicˇka used to say, every unit of the language system is a syndrome of various properties, not all of which are present with every occurrence of that unit. If it is recognized that the boundary between the core and periphery of language is complex in most different dimensions, this means that indistinctness may be found everywhere in the domain of language and its use. If the description is to reflect this character of language and to avoid Procrustean methods, then we may admit the necessity of a very broad use of underspecification. This can then concern even the opposition of ambiguity and indistinctness, as mentioned at the end of Section 3.5 above. The presence of asymmetries and devitations of most different kind is connected with the fact that a natural language primarily exists as internalized and intentionally used by its speakers, as Schnelle (1996) has convincingly concluded. It is true that speakers acquire their language to a higher or lesser degree, from which it follows that also H.-H. Lieb’s notion of the language as an abstract, ideally existing object has to be taken into account; however, this abstract object can only be specified on the basis of observable behavior of language users. When speaking explicitly about a natural language, these two explicata of the presystemic notion should be distinguished, with the richness of complex phenomena in zones ‘close to the horizon’ being proper to either of them in most different layers of language structure and of its interfaces with the process of communication as well as with the layers of cognition and of physical sounds.

 Petr Sgall

Notes * The work reported on in this paper has been carried out under the projects GACˇR 405/96/K214 and MŠMT LN00A063. — A preliminary version of the present paper was published in The Prague Bulletin of Mathematical Linguistics No. 73–4, 2000. . In the process of communication, such limitations are present as those concerning the use of here on phone; cf. also now (when I write this) I believe . . . vs. now (when you are reading this) you see . . . in a letter. . Many further cases of collisions in Czech morphemics can be found in Šimandl (2000); let us recall just a small sample of cases exhibiting oscillating usage: gender in contexts such as Oscarˇi byli prˇideˇleni vs. Oscary byly prˇedány ‘the Oscar prizes were alloted’ (Anim. vs. Inanim.); number in cases like s peˇstmi i ocˇima sevrˇenýma/mi, lit. ‘with fists and eyes closed’, with the adjective in agreement either with the noun form ocˇima (i.e., in Dual) or with peˇstmi (Plural); oblique case forms e.g. with local names of foreign origin, such as the Gen. Lukky and Dat./Loc. Lucce (with the Czech pronunciation of -c-) from Ital. Lucca. . It must be admitted that some of these phenomena (i.e. the ‘systematic ambiguity’) hardly would be classified as cases of vagueness, since the different interpretations are delimited from each other without ‘‘fuzzy’’ boundaries. However, indistinctness may be understood in a broader sense than ‘vagueness’; cf. also the differences in the specification of reference — the word house does not have as many meanings as there are houses (in extension or in intension), i.e. no linguistic ambiguity is present here, but even so for semantic interpretation separate entities (referents) are needed. . The phenomena of quantification pertain to different parts of language structure, documenting again the richness of the often not quite regular interrelations of different aspects of morphemics, underlying structure (meaning) and semantics (content); cf. Partee’s (1999) analysis of determiner, adverbial and morphemically expressed quantification as related to the mass/count division of nouns and to the process/event division of verbs or verb forms. . The picture in Figure 1 is too schematic and should not suggest that the opposition of prototypical and peripheral phenomena is limited to the core of language and to its interfaces with other domains. On the contrary, such an opposition is to be seen and reflected also in many subdomains within the core, which, of course, is far from being just a flat ‘domain’; it has its shallower and deeper layers that can be found as pertinent to many dimensions distinguishable under an appropriate viewpoint. Thus e.g. Skalicˇka’s (1962) characterization of the central (deepest, in a sense) parts of syntax as being anthropocentric points to a difference beteeen fundamental sentence patterns such as a. Agent–Action (–Patient) or b. Property–its Bearer and secondary (in a sense peripheral) further constructions. Along with a. Jim ran, Jim built a house we also find such marked, secondary examples as Sun is shining, with the subject being patterned as if it were an Agent, and b. along with The town is large there is Amsterdam is larger than Leiden and Leiden is smaller than Amsterdam, in which while one of the two members of a relation has to be chosen as the Bearer, the other one is embedded into the Property. With verbs such as buy/sell, one of

Freedom of language 

the participants has to be chosen as Agent, although an active involvment (at least agreement) of the other also is present. It may be recalled that due to the combinatory properties of these syntactic role, in the Functional Generative Description we work with dependency relations of an Actor/Bearer, a Patient (which also covers Property in these examples), and others, to their head words, and we understand a correlate of Action as the function of a verb; also Copula may then be classified as a verb.

References Arutjunova, N. D. 1994. ‘‘Etika a pravda’’ [Ethics and truth]. Slovo a slovesnost 55: 116–20. Austin, J. L. 1962. How to do things with words. Cambridge, Mass.: Harvard University Press. Bach, E. 1996. The politics of Universal Grammar. The President’s speech at the Annual Meeting of the Linguistic Society of America, San Diego. Chomsky, N. 1975. Reflections on language. New York. Dönninghaus, S. 1999. Sprache und Täuschung. Wiesbaden:Harrassowitz. Dönninghaus, S. (in press). ‘‘Der Bereich des Quantitativen aus der Sicht der kognitiven Linguistik’’. In: Quantität und Graduierung als kognitive und semantische Kategorien., H. Jachnow and B. Norman, (eds), Wiesbaden:Harrassowitz. Fillmore, C. J. 1982. ‘‘Frame semantics’’. In: Linguistics in the morning calm. 111–37. Seul: Hanshin. Fillmore, C. J. 1985. ‘‘Frames and the semantics of understanding’’. Quaderni di semantica VI, 2: 222–55. Frynta, E. 1993. Zastrˇená tvárˇ poezie [The covert face of poetry]. Prague: Nakladatelství Franze Kafky. Grice, P. 1975. ‘‘Logic and conversation’’. In: Syntax and semantics 3, P. Cole and J. L. Morgan (eds), 41–58. New York: Academic Press. Grice, P. 1982. ‘‘Meaning revisited’’. In: Mutual knowledge, Smith N. V. (ed.), 223–43. London and New York. Hajicˇová, E., Partee, B. H. and Sgall, P. 1998. Topic–Focus Articulation, Tripartite Structures, and Semantic Content. Dordrecht: Kluwer. Holan, T., Kubonˇ, V., Oliva, K. and Plátek, M. 2000. ‘‘On complexity of word order’’. Revue T. A. L. 41, No. 1. Holenstein, E. 1975. Roman Jakobsons phänomenologischer Strukturalismus. Frankfurt am Main. Joos, M. 1961. ‘‘The five clocks’’. In International Journal of American Linguistics 28, No. 2/5; published as a book in New York: Harcourt, Brace, Jovanovich 1967. Karcevskij, S. 1929. ‘‘Du dualisme asymétrique du signe Linguistique’’. In: Travaux du Cercle linguistique de Prague 1, 88–93; a Czech translation in: Principy strukturní syntaxe I, S. Machová (ed), 26–30. Praha: SPN (MFF UK), 1974. Keenan, E. 1975. ‘‘Logical expressive power and syntactic variation in natural language’’. In: Formal semantics of natural language, E. Keenan (ed), 406–21. Cambridge University Press.

 Petr Sgall

Kiefer, F. 1995. ‘‘Cognitive linguistics: A new paradigm?’’. In: Linguistics in the morning calm 3. Selected papers from SICOL-1992, 93–111. Seoul:Hanshin. Leška, O. 1987. ‘‘Teleology and Saussure’s dichotomy ‘synchrony’ :: ‘diachrony’. On the teachings of the Prague School’’. In: XIV. Internationaler Linguistenkongreß — Abstracts. Berlin, 194. Leška, O. 1994. ‘‘Hierarchy and overlap of language levels’’. In: Luelsdorff et al. 1994: 133–40). Translated from: ‘‘Ierarxija jarusov stroja jazyka i ix perekryvanie’’. In: Edinicy raznyx urovnej grammaticˇeskogo stroja jazyka i ix vzaimodejstvie, 20–7. Moskva: AN SSSR, 1969. Lewis, D. 1969. Convention. A philosophical study. Cambridge, Mass. M. I. T. Press Marty, A. 1908. Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie 1, Halle/S. Mathesius, V. 1911. ‘‘O potenciálnosti jevu˚ jazykových’’. In: Veˇstník Královské cˇeské spolecˇnosti nauk: Trˇída filosoficko-historicko-jazykoveˇdná 2, 1–24. Translated as On the potentiality of the phenomena of language in Vachek 1964: 1–32). Mathesius, V. 1942. ‘‘Rˇecˇ a sloh’’ [Speech and style]. In: Cˇtení o jazyce a poezii, 11–102, Praha — Separately Otázky a názory, vol. 61, 1966, Praha: Cˇs. spisovatel. McCawley, J. 1992. ‘‘A note on auxiliary verbs and language acquisition’’. Journal of Linguistics 28:445–51. Novák, V. 1993. The alternative mathematical model of linguistic semantics and pragmatics. New York: Plenum Publishing Corporation. Partee, B. H. 1999. ‘‘Nominal and temporal semantic structure: Aspect and quantification’’. In: Prague Linguistic Circle Papers 3: 91–108. Amsterdam/Philadelphia:Benjamins. Pazukhin, R. 1983. Linguistic necessity and linguistic theory. Katowice: Uniwersytet S´la˛ski. Peregrin, J. 1992. Words and worlds. An essay on foundations of logic, semantics and ontology. Prague: Charles University. Putnam, H. 1975. ‘‘The meaning of ‘meaning’’’. In: Language, mind, and knowledge. Minnesota Studies in the Philosophy of Science VII, K. Gunderson (ed), 131–93. Schnelle, H. 1996. ‘‘Beyond new Structuralism’’. In: Theoretical linguistics and grammatical description, R. Sackmann (ed.), 277–86. Amsterdam/Philadelphia: John Benjamins. Searle, J. R. 1969. Speech acts. Cambridge University Press. Searle, J. R. 1983. Intentionality. Cambridge University Press. Seifert, J. 1984. ‘‘Interview with Zagorka Živkovicˇ’’. Dagens nyheter 2.12.1984. Quoted from: Janouch F. 1995: Šel básník chudeˇ do sveˇta, 295–8. Prague: Cˇeský spisovatel. Sériot, P. 1999. Structure et totalité. Paris: P. U. F. Sgall, P. 1995. ‘‘Prague School typology’’. In: Approaches to language typology. Masayoshi Shibatani and Theodora Bynon (ed), 49–84. Oxford: Clarendon Press. Sgall, P. 1998. ‘‘Revisiting the classification of the dependents’’. In: Issues of valency and meaning. Studies in honour of Jarmila Panevová, 15–26. Prague:Karolinum. Sgall, P. 2000. ‘‘English syntax in Functional Generative Description. Topic–Focus articulation. Syntax and semantics’’. In: Rudiments of English linguistics, Štekauer P. (ed), 225–64. Prešov: SLOVACONTACT. Sgall, P., Hajicˇová, E. and Panevová, J. 1986. The meaning of the sentence in its semantic and pragmatic aspects. Praha:Academia — Dordrecht:Reidel.

Freedom of language 

Sgall, P., Hronek, J., Stich, A. and Horecký, J. 1992: Variation in language: Code switching in Czech as a challenge for sociolinguistics. Amsterdam/Philadelphia: Benjamins. Šimandl, J. 2000: ‘‘Morfologická problematika v jazykové poradneˇ’’ [Morphemic issues in the linguistic consultation center]. Naše rˇecˇ 83: 57–76, 113–31, to be continued. Skalicˇka, V. 1962: ‘‘Das Wesen der Morphologie und der Syntax’’. In: Acta Universitatis Carolinae: Slavica Pragensia IV, 123–7. Prague. Stich, A. 1991: ‘‘A co jazyk?’’ [And what about language?] Prˇítomnost 7:26. Trost, P. 1989: ‘‘Interlingualer und intralingualer Lautwandel.’’ Linguistica generalia IV, 7–10. Prague:Charles University Vopeˇnka, P. 1979: Mathematics in the alternative set theory. Lecture. Leipzig. Vopeˇnka, P. 2000: Úhelný kámen evropské vzdeˇlanosti a moci [The corner-stone of European culture and power]. Prague:Práh. Waszakowa, K. 1997: ‘‘O nowych zjawiskach leksykalnych w s´wietle semantyki rozumienia.’’ In: Semantyczna struktura słownictwa i wypowiedzi., R. Grzegorczykowa and Z. Zaron (ed.), 9–24. Warszawa:Ossolineum. Weber, H. 1893: ‘‘Leopold Kronecker’’. In: Jahresberichte der Deutschen Mathematikervereinigung 2:5–31.

The natural order of cognitive events

Philip A. Luelsdorff Regensburg

There is a natural order of cognitive linguistic events, e.g. things and states change and events happen or occur, some before others, some at the same time as others, and some after others, and the more one knows about the things, states, and events in question, the more deterministic the natural orders become. Tense and aspect are the major exponents of temporality in English. These are traditionally defined in terms of three points — the speech time (S), the event time (E), and the reference time (R). In this paper, the English tenses are defined in terms of (S, E) (i.e. without R), and the English aspects are defined in terms of (E1, E2) (also without R). Moreover, each pair of tenses selects an aspect, and each aspect selects a pair of tenses. The subject of sequence of tenses (SOT) in sentential and discoursal contexts is often discussed, but the subject of the natural orders of tense, aspect, tense-aspect, modality, complementation, and mood is largely ignored. In this paper, these natural orders are defined, and an attempt is made to explain the orders specified. The positions taken have a number of implications for the study of tense and aspect in language acquisition, language attrition, and language pathology. With respect to the latter, some symptoms and syndromes of possible ‘‘temporal agnosia’’ are defined and discussed.

0. Introduction Cognitive linguistic events are mental and physical representations which exhibit the textlinguistic properties enumerated in (1) (Dressler 1994). For Dressler, ‘‘The model is functionalist insofar (as) it subordinates textual operations to the two main functions of language, the communicative and the cognitive ones,’’ i.e. function determines form. (1) Standards of Textuality (Dressler 1994) a. Cohesion e. Informativity b. Coherence f. Situationality c. Intentionality g. Intertextuality d. Acceptability h. Interpretability

 Philip A. Luelsdorff

The extended Buhlerian model of language (Bühler 1934; Luelsdorff 1984, 1986, 1987) is a related attempt to capture the essence of textuality. Within the organon model of language, spoken and written events are understood in terms of cognitive linguistic factors, the functions they serve, and the validity claims they make. Thus, whereas natural textlinguistics focuses on validity claims, the extended organon model of language also focuses on factors and functions. An outline of the factors, functions, and validity claims of speech events is given in (2), where the interested reader is referred to Luelsdorff (1987) and the references cited therein for further details. (2) Speech events Factor a. Speaker b. Hearer c. Objects d. Channel e. Code f. Message g. Utterance h. Inner Form i. Attitude j. Intention

Function Expressive Appellative Referential Phatic Metalinguistic Aesthetic Communicative Constitutive Adaptive Selective

Validity Sincerity Normative Adequacy Equality Intelligibility Beauty Intersubjectivity Executibility Conformity Clarity

A contrastive analysis of the similarities and differences between the models of textuality presented in (1) and (2) would be a rewarding exercise in intertextuality. For the time being, however, we only wish to emphasize the obvious but often overlooked fact that all the factors, functions, and validity claims of speech events are saturated in frequencies, rendering the relations between and among them more or less determinate. In order to be able to express the frequentative, determinate character of cognitive linguistic events, Chesnokov and Luelsdorff (1991) propose the adoption of determinacy form (DF), as restated in (3). (3) Determinacy form x,z → y, I=m and C=n, where I=N(xy)/N(x) and C=N(xy)/N(y) According to DF, a subject (x) determines (→) a predicate (y), when the determinacy x → y is bound by (zero or more) binders (z), with frequency (N)-based measures of accuracy (I) and completeness (C). DF is the form of all cognitive and cognitive linguistic events, be they content → content, expression → expression, content → expression, or expression → content. Moreover, DF cannot be learned because it is needed to learn, with the result that DF is a cognitive universal. The subject of this paper is the natural temporal order of cognitive events with special reference to English tense and aspect, modality, reported speech,

The natural order of cognitive events 

complementation and mood. There are vast differences in linguistic opinion about the nature and number of English tenses and aspects. For example, Reichenbach (1947) envisages twelve tenses (six non-extended (non-progressive), six extended (progressive)); Hornstein (1990), following Reichenbach (1947) envisages six nonextended, but omits six extended; Palmer (1989) maintains that there are two tenses (past, present, but no future), one aspect (progressive), and one phase (perfect); and, most recently, Enç (1996) maintains that there is one tense (past), otherwise present -s is an agreement marker, and futurate will is a modal. By way of preview, we propose that English has four tenses (FUT(URE), PRES(ENT), PAST, GEN(ERIC)), in that temporal order, and four aspects (TO FUT(URE), PROG(RESSIVE), PERF(ECT), PASS(IVE)), in that temporal order. The tenses express the relative temporal relations of anteriority (<), simultaneity (=), or posteriority (>) between the time of the speech event (S) and the time of the narrated event (E), i.e. without Reichenbach’s reference point (R). The aspects express the same relative temporal relations between the time of one event (E1) and the time of another (E2). Moreover, the same relative temporal relations are used to describe the temporal relations between and among the tenses, aspects, and tensed aspects. In general, aspect temporality is found to constrain the temporality of tense. Essentially the same approach is used to describe natural temporal orders among the modals, the complementizers, and. the moods. The paper concludes with some remarks on possible temporal disorders directed mainly at clinical linguists and neuropsychologists.

.

The natural order of cognitive events

Cognitive linguistic events are situational types definable in terms of a limited number of types of predicates and roles. For example, ‘‘Arthur (Agent) writes plays (Factitive),’’ ‘‘Plants (Precedent) bear fruit (Processive),’’ and ‘‘Bubbles (Patient) burst.’’ Although opinions differ about the nature and number of roles, it seems clear that predicates determine roles and vice versa, with more or less accuracy (I) and completeness (C). In English, the above-mentioned situation types are tensed and aspected. Tenses function to express the temporal relations between speech time (S) and event time (E),9 whereas aspects express the temporal relations between the time of one event (E1) and the time of another (E2). The temporal relations in question — anteriority, simultaneity, and posteriority — are relative (as opposed to absolute) in that each tense is defined in terms of the temporal relations between S and E, and each aspect is defined in terms of the temporal relations between E1 and E2. The English tenses and aspects are defined in (4), where X < Y means that X is anterior to Y (alternatively, Y is posterior to X), and X=Y means that X is

 Philip A. Luelsdorff

simultaneous with Y (alternatively, Y is simultaneous with X). The formulation X ≤ Y appears to be simpler but brings with it bidirectionality. (4) English tenses and aspects a. Tenses b. Aspects 1. FUT: S < E 1. TO FUT: E1 < E2 2. PRES: S=E 2. PROG: E1=E2 3. PAST: S > E 3. PERF: El > E2 4. GEN: (1–3) 4. PASS: E1 > E2 According to (4), the English tenses are FUT(URE), PRES(ENT), PAST, and GEN(ERIC),10 where GEN (e.g. ‘‘The earth revolves around the sun’’) holds of FUT, PRES, and PAST, and the English aspects are TO FUT(URE), PROG (RESSIVE), PERF(ECT), and PASS(IVE). TO FUTs have the form V1 TO V2 (e.g. have to leave, decide to resign, etc.), PROGs have the form BE V, PERFs have the form HAVE V, and PASSs have the form BE V. Other aspects have been posited for English (e.g. inceptive, repetitive, continuative, terminative, etc.), but these can be accommodated within the framework proposed (e.g. begin to work is a TO FUT, keep working is a PROG, and have finished working is a PERF PROG). PASS is an aspect, not a voice, because ACT(IVE)s (activities, processes) express processes or causes, and PASSs express their products, results, or effects. To the extent that processes precede products, or causes precede effects, ACTs precede PASSs. If Cézanne paints a landscape, the result is a landscape painted by Cézanne. If Luciano disappoints Maria, the result is Maria’s being disappointed by Luciano. Among the aspects defined in (4), it is not obvious why PROG is defined as E1=E2 and why PASS is defined as E1 > E2. In the case of PROG, Givón (1993) writes that ‘‘The progressive perspective on an event is often established through bringing the observer onto the scene in the middle of the event, when it was already going on. This is most commonly accomplished by depicting in an adjacent clause the entry of ‘‘a witness onto the scene.’’ For example, ‘‘When I came in, he was kicking the couch,’’ in which the entry and the kicking (the event in progress) are simultaneous. Again, ‘‘the progressive (‘imperfective’) aspect . . . is typically used to depict events that are simultaneous to a contiguous clause-event.’’ Therefore, Givón provides some justification for defining PROG as E1=E2.11 In the case of PASS, Palmer (1989) discusses active-passive correspondences, agentless passives, the passive and transitivity, restrictions on the passive, passive functions, passive gradience, pseudo-passives, semi-passives, statal passives, getpassives, lexical passives, and adverbial passives. With respect to statal passives. Palmer writes: ‘‘Any verb that has a passive may also have a statal passive,’’ e.g. ‘‘The boy is hurt, The trees are cut down, The exams are finished, The garden is dug,’’ and that the -EN forms that function in this way are essentially perfect in meaning and refer to a resultant present state. Palmer’s observation that passives

The natural order of cognitive events 

are ‘‘essentially perfect in meaning’’ supports the claim that PASS is E1 > E2. On the definitions of the English aspects given, both PERF and PASS are defined as E1 > E2, where PERF E1=have, and PASS E1=be. On this analysis of PERF, E1 follows E2, and the tenses on E1 and E2 are accordingly constrained. The formal near identity of PERF and PASS has a functional explanation. Just as the PASS expresses resultant states (cf. above), the PERF expresses ‘‘the element of result,’’ which, according to Hajicˇová, Panevová and Sgall (1973), is ‘‘. . . a state after the completion of an action rather than a mere result of an action; resultivity, of course, implies that the action concerned precedes the time point to which the resulting state relates, but this temporal relationship merely follows from the extralinguistic relation12 between the resulting state and the action that has led to it.’’ If, as we suggest, PERFs postdate non-PERFs and PASSs postdate non-PASSs, it might be asked if PASSs postdate PERFs and if PERF PASSs postdate both PERFs and PASSs. Compare the chronologies in (5). Here, it seems evident that the order of succession is (a–c–b–d). (5) Statal temporality a. John felled the tree. b. John has felled the tree. c. The tree was felled. d. The tree has been felled. The tenses and aspects are ordered by the relation ‘‘x precedes y’’ (equivalently, ‘‘y follows x’’). In the case of the tenses, FUT precedes PRES, PRES precedes PAST, and PAST precedes GEN. Alternatively, GEN follows PAST, PAST follows PRES, and PRES follows FUT. For example, (1) ‘‘John will be absent’’ (FUT), (2) ‘‘John is absent’’ (PRES). (3) ‘‘John was absent’’ (PAST), and (4) ‘‘John is always absent’’ (GEN). In the case of the aspects, TO FUT precedes PROG, PROG precedes PERF, and PERF precedes PASS. Alternatively, PASS follows PERF, PERF follows PROG, and PROG follows TO FUT. For example, (1) ‘‘Maria will begin to sing’’ (TO FUT), (2) ‘‘Maria will be singing’’ (PROG), (3) ‘‘Maria will have sung’’ (PERF), and (4) (1, 2, or 3) can undergo PASS, i.e. ‘‘The aria will begin to be sung’’ (TO-FUT PASS), ‘‘The aria will be being sung’’ (PROG PASS), and ‘‘The aria will have been sung’’ (PERF PASS). NOT and NOA are given in (6). (6) Natural order of tenses and aspects13 a. NOT: FUT < PRES < PAST < GEN b. NOA: TO FUT < PROG < PERF < PASS The relations ‘‘precede’’ and ‘‘follow’’ are transitive relations. Therefore, since FUT precedes PRES, and PRES precedes PAST, FUT also precedes PAST, and since TO FUT precedes PROG, and PROG precedes PERF, TO FUT also precedes PERF. Conversely, since PAST follows PRES, and PRES follows FUT, PAST also follows

 Philip A. Luelsdorff

FUT, and since PERF follows PROG, and PROG follows TO FUT, PERF also follows TO FUT. The orders of tense and aspect given in (6) do not imply that every linguistic event is cognized, produced, or comprehended in the orders indicated. On the contrary, we can conceptualize and express an event in the PRES or PAST without having conceptualized or expressed that event in the FUT. What the orders do mean is that every event has a FUT, a PRES, and a PAST, in that order, and a TO FUT, PROG (with familiar exceptions), and a PERF, in that order, irrespective of whether all the tenses and aspects of the event are actually cognized and expressed or not. Just as it is not possible for a PRES not to have had a FUT and for a PAST not to have had a PRES, it is not possible for a PERF not to have had a PROG and for a PROG not to have had a TO FUT. According to the definitions of tenses and aspects in (4), tenses are R(S,E), and aspects are R(E1,E2), where S is the time of the speech event, E is the time of the narrated event, and R is one of the relative temporal relations of anteriority, simultaneity, or posteriority. Tense shifts can occur in indirect speech, indirect questions, and conditions. On the present analysis, each event (E) is tensed (T), with the result that each event (E1, E2) of each aspect is tensed. Tensed aspects therefore have the form E1T1 (TO) E2T2, where T1 tenses E1, and T2 tenses E2. Here, the use of the symbol ‘E’ to refer to (1) time of event and (2) event is equivocal, but ought not lead to misunderstanding. Now, each aspect determines the pairs of tenses with which it cooccurs and vice versa, the constraints on aspect-tensing given by the definitions of the tenses and aspects in (4). For example, if El=E2 (PROG), then T1=T2. In other words, PROG aspect, by definition, requires the simultaneity of E1 and E2, and this aspectual simultaneity imposes temporal simultaneity on T1 and T2. In this manner, the definitions of the tenses and aspects are held to constrain the natural order of tense-aspect (NOTA) expressed by the determinacy schema in (7). According to (7), if the temporal relation between E1 and E2 is Ri (anterior, simultaneous, posterior), then the temporal relation between T1 and T2 (the tenses on E1 and E2) is Rj (anterior, simultaneous, posterior) and vice versa. (7) Natural Order of Tense-Aspect (NOTA) Ri(E1,E2) → Rj(T1,T2), where Ri=Rj, and I=m and C=n. NOTA embodies the small set of testable hypotheses given in (8). The first hypothesis concerns anteriority, the second concerns simultaneity, and the third posteriority. (8) Tense–aspect Hypotheses14 a. H1: E1 < E2 → T1 < T2, I=m and C=n. b. H2: E1=E2 → T1=T2, I=m and C=n. c. H3: E1 > E2 → T1 > T2, I=m and C=n.

The natural order of cognitive events 

By way of interesting example, consider the case of ‘‘past participial’’ -EN, which, like PROG -ING, can mean FUT, PRES, or PAST, as in PERF ‘‘will have/have/had eat-en’’ and their PASS correspondences. These examples support hypothesis H3, because in PERFs (and corresponding PASSs), E1 > E2 and, consequently, T1 > T2. The PRES PERF, which is the ‘‘noisiest’’ tensed aspect in English, has several manifestations. First, there is the case of the PRES PERF non-PROG, as in ‘‘John has solved the problem.’’ Here, HAVE is in the PRES, and the MV is in the PAST. Second, there is the case of the PRES PERF PAST PROG, as in ‘‘John has been solving the problem’’ (but is not solving the problem at the moment). Here, HAS is in the PRES and BEEN SOLVING is in the PAST PROG. Third, there is the case of PRES PERF PAST and PRES PROG, as in ‘‘John has been solving the problem’’ (and is still solving the problem at the moment). The PRES PERF PROG is therefore ambiguous and, in order for this ambiguity to be resolved, the underlying tense on PROG (PAST or PAST and PRES) must be accessed. The constant across these three cases (i.e. with or without PROG) is some relation between the PAST and the PRES. Some linguists have noted that the PAST is definite (‘‘John resigned on Monday’’), whereas the PRES PERF is indefinite (*‘‘John has resigned on Monday’’ but ‘‘John has already resigned’’ and ‘‘John has resigned twice’’). Hofmann (1993) explains this phenomenon by saying that the speaker using the PRES PERF either does not know the definite time of the PAST or does know the definite time of the PAST but does not consider this definite time to be relevant to the communication. However, the PRES PERF prohibits the definite PAST even if it is known to the speaker and is considered relevant to the communication. The restriction on PRES PERF + DEFINITES can be explained in the following way. The adverbials that cooccur with the PRES PERF are either boundary adverbials (since, yet, already, for two years, etc.) or frequency adverbials (often, n(ever), frequently, seldom, twice, ten times, etc.). Reichenbach (1947) defines the PRES PERF as E_S, R, i.e. as a tense in which E precedes S, and S is simultaneous with R. However, boundary and frequency adverbials do not modify S in PRES PERF E_S,R, as Reichenbach suggests. Rather, they qualify the occurrence (already, since, before, etc.) or non-occurrence (yet, since, before, etc.) of E2 in the timespan between E2 and El, or they qualify the length of the time span (for two years, since 1984, etc.) between E2 and El, or they qualify the frequency of occurrence of E2 (often, n(ever), seldom, etc.) during the timespan between E2 and El. On one level, the restriction on PRES PERF + DEFINITES is explained by the fact that PRES PERF + BOUNDARIES/FREQUENTATIVES, answering the questions ‘‘Since when?,’’ ‘‘How long?,’’ and ‘‘How often?’’. On another level, the restriction on PRES PERF + DEFINITES is explained by the timespan between E2 and El in E2_E1, which is the definition of PERF as given in (4)b.. If PERF (i.e. E2_E1) prohibits definiteness in the PRES PERF, then it should also prohibit definiteness in the FUT PERF and the PAST PERF, as in ‘‘John will

 Philip A. Luelsdorff

have finished work by (*at) sunset’’ and ‘‘John had finished work by (*at) sunset.’’ This appears to be the case, with the result that the restriction on PRES PERF + DEFINITES is generalizable to a restriction on PERF + DEFINITES. This discussion of PERFs and DEFINITENESS bears on the natural order of tense (NOT), aspect (NOA), and tense-aspect (NOTA). First, we distinguish between the order of events in production and the order of events in comprehension. In production, the simple tenses (FUT, PRES, PAST, GEN) are anterior to their corresponding respective aspects (TO FUT, PROG, PERF, PASS). For example, we need to know that ‘‘John resigned’’ before we can say ‘‘John has resigned,’’ and we need to know that ‘‘John will resign’’ before we say say ‘‘John will have resigned by sunset.’’ Evidently, any given event must be assigned a tense before it is assigned an aspect. In comprehension, the aspects are anterior to their corresponding simple tenses. For example, given ‘‘John has resigned,’’ we know that ‘‘John resigned’’ before the time at which ‘‘John has resigned’’ is uttered. Second, in order to be able to modify PERF (E2_E1) by PERF (i.e. boundary, frequency) adverbials, it is necessary to know all of E2, El, and E2_E1, in addition to the semantics of the adverbials themselves. Within the Reichenbachian (1947) theory of tense, there is no reference point (R) corresponding to _ in E2_E1, to say nothing of all E2_El. In the present opinion, the PASS (E2_E1, where E2=MV and El=BE) is a resultative,15 which expresses the result or effect of the activity or cause expressed by its corresponding ACT(IVE). Since products (results) follow processes (activities), and effects follow causes, PASSs follow ACTs. On this resultative theory of the PASS, there is a shift in focus from the ACT process to the PASS product, and this shift in focus causes a shift in time from earlier (the process) to later (the product). For example, ‘‘Cézanne painted that still life’’ (completed process) and, as a result, ‘‘That still life was painted by Cézanne’’ (later product of earlier process). Note that both PERF and PASS are defined as E2_E1 in (4b)., but El=have in PERF and El=be in PASS. On the theory of natural order under development, not only verbs are tensed and aspected, but also nouns, adjectives, and adverbs, i.e. all major-class items. For example, in ‘‘John intends to resign’’ (TO FUT), ‘‘John’’ is the subject of both ‘‘intends’’ and ‘‘resign.’’ However, since ‘‘intends’’ is in the PRES, and ‘‘resign’’ is in the FUT, ‘‘John’’ in ‘‘John intends’’ is also in the PRES, and ‘‘John’’ in ‘‘John resign’’ is also in the FUT. In ‘‘John is resigning,’’ ‘‘John’’ is the subject of both ‘‘is’’ and ‘‘resign.’’ Since both ‘‘is’’ and ‘‘resigning’’ are in the PRES, ‘‘John’’ in ‘‘John is’’ and ‘‘John’’ in ‘‘John resign’’ are also in the PRES. In the PRES PERF ‘‘John has resigned,’’ ‘‘John’’ is the subject of both ‘‘has’’ and ‘‘resigned.’’ Since ‘‘has’’ is in the PRES, ‘‘John’’ in ‘‘John has’’ is also in the PRES, and since ‘‘resigned’’ is in the PAST, ‘‘John’’ in ‘‘John resigned’’ is also in the PAST. In general, the temporal relations (TR) (anteriority, simultaneity, posteriority) between events

The natural order of cognitive events 

(TR(E1,E2)) determine the TR between tenses (TR(T1,T2)) and the temporal relations between participants (TR(P1, P2)). Aspects are either simplex or complex. Simplex aspects are of the form El TR E2, where El and E2 are events, and TR is either anterior, simultaneous, or posterior. As noted in (4), the natural order of simplex aspects is (1) TO FUT, (2) PROG, (3) PERF, and (4) PASS. Complex aspects are of the form El TR E2, where El and/or E2 are aspects. The natural order of complex aspects is also (1) TO FUT, (2) PROG, (3) PERF, and (4) PASS. By way of example of complex aspecting, consider the case of the TO FUT, i.e. E1T1 TO E2T2, where El is a verb, noun, or adjective, and E2 is its infinitive complement. According to NOA, the simplex TO FUT ‘‘John has to go’’ can combine with PROG to form the complex TO FUT PROG ‘‘John has to be going,’’ with PERF to form the complex TO FUT PERF ‘‘John has to have gone,’’ and with PROG PERF to form the complex TO FUT PROG PERF ‘‘John has to have been going’’ (home at the time of the crime). In these examples, El (have) is constant, while E2 (go) varies from PROG to PERF to PROG PERF. In other examples, E2 (go) remains constant, while El (have) varies from PROG to PERF to PROG PERF, e.g. ‘‘John has to go’’ can combine with PROG to form the complex ‘‘John is always having to go,’’ with PERF to form the complex TO FUT PERF ‘‘John has had to go,’’ and with PROG PERF to form the complex TO FUT PROG PERF ‘‘John has always been having to go.’’ Note that PASS can combine with TO FUT (‘‘John was persuaded to resign’’), with PROG (‘‘John is being mobbed’’), and with PERF (‘‘John has been persuaded to resign’’), but not with both PROG and PERF (*’’John has been being mobbed,’’ *’’John has been being persuaded to resign’’). This cooccurrence restriction on *PROG PERF PASS, coupled with the aspectual analysis of both the PERF and the PASS as E2_E1, suggest that the structure of the English auxiliary is (9), where M=modal, BE1=be/have BE2=be, and MV=main verb, and the TO FUT is omitted. (9) English Auxiliary Structure M T BE1 T BE2 T MV T On this analysis. T=FUT, PRES, PAST, or GEN, and PROG=(BE1, BE2) or (BE1, MV) (where BE1=be), PERF=(B1, BE2) or (Bl, MV) (where BE1=have), and PASS=(BE2, MV) (where BE2=be) A determinacy grammar (DG) consists of three components: (1) a dictionary of variables (DV), (2) a data matrix (DM), and (3) a determinacy analysis (DA) (Chesnokov and Luelsdorff 1991; Luelsdorff and Chesnokov 1996). One advantage of the natural orders of tense (NOT), aspect (NOA), and tense-aspect (NOTA) proposed is that they predict the correct temporal alignment of complex auxiliary data in the determinacy grammar of English.

 Philip A. Luelsdorff

The tenses and aspects can be viewed as formed and ordered by movements. According to Priest (1995) (cf. Honderich 1995), ‘‘x moves iff x is at some place, P1, at time t1, and x is at some numerically distant place, P2, at some later time t2, and x exists at some juxtaposed set of places between P1 and P2 and at all times between t1 and t2.’’ On this definition of ‘‘movement,’16’ movement requires a space, positions, times, and at least one object changing places, place by place, through time. Tense and aspect movements are isomorphic in that both are of the form X < Y → X=Y → X > Y, whereby later tenses and aspects occur in earlier positions. For example, PRES is later than FUT, and PAST is later than PRES in time, but PRES is earlier than FUT, and PAST is earlier than PRES in space. Consider the tense and aspect movements in the two-dimensional matrix-space in (10). Here, the tenses and aspects are formed and ordered by the same movements of different elements, S/E in the case of the tenses and E1/E2 in the case of the aspects. (10) Tense and aspect determinacy movements N P1 P2 P3 P4 P5 T MOVEMENTS 1 0 0 0 0 0 t1 NONE 2 0 0 0 0 Y t2 ADD E/E2 3 0 0 X 0 0 t3 ADD S/E1 4 0 0 X 0 Y t4 FUT, TO FUT 5 0 0 X Y 0 t5 PRES, PROG 6 Y 0 X 0 0 t6 PAST, PERF/PASS Jackendoff (1983) defines a ‘‘temporal field’’ in the following terms: (1) (EVENTS) and (STATES) appear as theme; (2) (TIMES) appear as reference object; (3) Time of occurrence plays the role of location. On this conception, ‘‘. . . temporal expressions preserve much of the force of lexically parallel spatial expressions . . .’’ The movement model of time and space in (10) extends the parallel between the expression of time and space in prepositional phrases (e.g. ‘‘on my birthday, on the table’’) to the expression of tense and aspect (e.g. ‘‘in the future (FUT), at present (PRES), in the past (PAST))’’ In general, if X=S, X is in front of Y in the FUT, at Y in the PRES, and behind Y in the PAST. Similarly, if X=E1 and Y=E2 (PERF), then E1 is after (follows) E2. By way of summary, on this movement account of tense and aspect, both tenses and aspects move from right-to-left, first forming, then ordering, tensed and aspected events by the same movements. According to Palmer (1989), English verb forms are either finite (tensed) or non-finite (infinitivals, participials), and, according to Chomsky (1981), INFL (inflection) is an element ‘‘indicating in particular whether the clause is finite or infinitival.’’ If INFL is (+Tense), the clause is finite; if INFL is (-Tense), the clause

The natural order of cognitive events 

is infinitival. The inventory of English infinitivals and participials (cf. Palmer 1989) is given in (11). (11) Infinitivals and participials Infinitivals Participials a. to take a. taking b. to be taking b. *being taking c. to have taken c. having taken d. to have been taking d. having been taking e. to be taken e. being taken f. to be being taken f. *being being taken g. to have been taken g. having been taken Infinitivals and participials are underlyingly tensed.17 In support of this claim, we argue from (1) tense movement (identity), (2) temporal adverbial modification, and (3) discourse context, as illustrated by the examples in (12). (12) Non-finites as finites a. Identities 1. Infinitivals Given XYZ, if X
 Philip A. Luelsdorff

For example, ‘‘take’’ in ‘‘John wants to take a break’’ is timed and tensed as FUT, the FUT tense expressed by TO. Moreover, in conformity with the temporal constraints on TO FUT (E1<E2), the time-tense on ‘‘wants’’ (PRES) precedes the time-tense on ‘‘take’’ (FUT). The assumption that infinitivals and participials are underlyingly tensed explains why temporal adverbials agree with them in time and vice versa. That is, if the infinitival or participial is FUT, PRES, PAST, or GEN, then the temporal adverbial is also FUT, PRES, PAST, or GEN, leading to the determinacy of the latter by the former in production and the determinacy of the former by the latter in comprehension. The discourse contexts in (12) (c) are designed to show that infinitivals and participials are underlyingly tensed. In the responses B1 and B2 to (1) a. the speaker assumes that ‘‘please’’ in ‘‘to please’’ in (1) a. is in the PAST because PAST is marked on ‘‘was’’. In response (B1) to (2) a., the speaker uses the PAST because the main clause in (2) a. is PAST, even though the participial clause is PAST PERF (‘‘After he had tenured his resignation, John left on an extended vacation’’). It is widely believed that English has finite and non-finite verbs, and that each clause can contain maximally one finite verb. On the basis of the above, this largely morphologically-based view bears reexamination. On this view, all verbs are tensed, some overtly, some covertly. Moreover, to the extent that things and attributes change, these are timed and tensed as well. According to Palmer (1979), the modals can have epistemic, deontic, and dynamic meanings. The dynamic sense of CAN/COULD is ‘‘It is possible for x to y,’’ the dynamic sense of MUST/HAVE (GOT) TO is ‘‘It is necessary for x to y,’’ the dynamic sense of SHOULD/OUGHT TO is ‘‘It is advisable for x to y,’’ and the dynamic sense of MAY/MIGHT is ‘‘It is possible for x to y.’’ Other authors (e.g. Coates 1983) call dynamic possibility ‘‘root possibility’’ and dynamic necessity ‘‘root necessity.’’ Clearly, modal expressions must be bound by context to modal meanings and vice versa. Some ‘‘syntactic’’ binders of epistemic and deontic modality (cf. Downing and Locke 1990) are given in (13). (13) Syntactic Binders of English:Modality a. Epistemic modality 1. Existential subject There might be enough. 2. be + ing He might be waiting. 3. Stative verbs He might be twenty. 4. Dynamic verbs He might leave early. 5. Passive voice It might be rebuilt. 6. Lexical auxiliary He might have to resign. 7. have + en He might have left by now.

The natural order of cognitive events 

b. Deontic modality 1. Past time expressed by forms other than have+ en He had to leave. He was allowed to leave. 2. The subjects of active verbs are agents, i.e. the subject carries out the action by his or her own energy. You must finish writing. Will you wait here? 3. Existential subjects are virtually excluded. 4. Stative verbs are equally uncommon. In addition, for Coates (1995), subjectivity (cf. Lyons 1997) is an integral component of the expression of epistemic possibility. ‘‘Subjectivity’’ is defined as ‘‘devices whereby the speaker, in making an utterance, simultaneously comments upon that utterance and expresses his attitude to what he is saying.’’ Heine (1995) presents a set of criteria for distinguishing root (agent-oriented, deontic, dynamic) modality and epistemic modality in German. These are reproduced in somewhat abbreviated form in (14). Heine maintains that ‘‘While prototypical instances of agent-oriented modality are characterized by the presence of the properties (i.e. F, A, D, L, and P), prototypical instances of epistemic modality lack all the properties except P.’’ In this case, P must be added to the list of characteristic features of epistemic modality given in (13) a.. (14) Syntactic binders of deontic modality (German) a. There is a force (F) with an interest in the occurrence or non-occurrence of an event. b. The event has an agent a.. c. The event is dynamic (D). d. The event occurs, if it occurs at all, later (L) than the reference time. e. The event is non-factual, but more or less probable (P). Modal meanings are expressed by short forms (SF) (modals) and long forms ( semi-modals, modal periphrastics), SFs consisting of one word, LFs consisting of two or three words. Some modal meanings, modal SFs, and modal LFs are given in (15). (15) Modal short forms and modal long forms Modal meaning Short form Long form a. PERMIT may be permitted to b. MAKE POSSIBLE can be possible for c. INTEND will be going to d. OBLIGE should be obliged to e. NECESSITATE must be necessary to f. ENABLE can be able to

 Philip A. Luelsdorff

Modalization — the process of modal cognition — is a causative process consisting of modalizing causes and modalized effects. For example, if circumstances ENABLE Boris TO play tennis, then Boris IS ENABLED TO play tennis, Boris IS ABLE TO play tennis, and Boris CAN play tennis. Moreover, prior to Boris’s ENABLEMENT TO play tennis, Boris was UNABLE TO play tennis, and, subsequent to Boris’s ENABLEMENT TO play tennis, it is POSSIBLE FOR Boris TO play tennis and POSSIBLE THAT he does. This suggests the typology of modal cognition given in (16), in which MC stands for modal cause and ME stands for modal effect. (16) Typology of modal cognition a. b. c. d.

Type I: MC: (−Modal), ME: (−Modal) Type II: MC: (−Modal), ME: (+ Modal) Type III: MC: (+ Modal), ME: (−Modal) Type IV: MC: (+ Modal), ME: (+ Modal)

Examples of the types of modal cognition given in (16) are given in (17). (17) Modal types (examples) a. b. c. d.

John left the party early, because he was tired. John should leave the party, because he is bored. John works because he has to. If John could quit work, he would.

These and countless other examples suggest that it is necessary to distinguish modality that is overt from modality that is covert in both modal production and modal comprehension, Consider the examples in (18). (18) Covert and Overt Modality a. Johnny’s mother is worried, because he should have been home by dark. Assume the following scenario. Johnny usually comes home by dark, because it is POSSIBLE (P) for him to do so, he is able a. to do so, and he is obliged (O) to do so. He must have (N) lost track of the time or he might have been detained or he could (P) still be at his friend Richie’s. b. Johnny is going to the movies, because his mother said he could. Assume the following scenario. Johnny intends to go the movies because he wants to, his mother gave him her PERMISSION and the money, thereby making it materially POSSIBLE FOR him to do so. As a first approximation, all overt modalities are also covert, but only some covert modalities are also overt. Those modalities that are covert but not overt can be elicited by interrogation in the form of requests for justification and explanation.

The natural order of cognitive events 

Some such cases are evident (Fish swim → Fish are able to swim → It is possible for fish to swim), while others are more or less context-sensitive, but, for being context-sensitive, are nonetheless linguistically relevant. If this analysis of modal cognition is pursued in sufficient depth, implicit modalities become explicit, and all the modals appear in one or another mood, tense, aspect, and voice. Again, by way of first approximation, modalization as causation is given as the Modal form (MF) in (19). (19) Modal form (MF) a. B not C b. A modalizes B to C c. B is modalized (by A) to C d. B modal C e. B C The advantages of the causal theory of modality have been discussed elsewhere (Luelsdorff 1996a, 1996b, 1997b). Moreover, all modals can be tensed (FUT, PRES, PAST, HAB) and aspected (TO FUT, PROG, PERF, PASS), as suggested for tense by the examples in (20) (cf. Luelsdorff 1997, 1998). (20) Modality and Tense a. Gertrude could swim the Channel b. next year. (FUT) c. right now. (PRES) d. last year. (PAST) Once subconscious tensed and aspected modalities are made conscious, however, additional modal tense and aspect structure emerges. For example, if ‘‘Gertrude could swim the Channel next year’’ (FUT), then (1) ‘‘She will be able to swim the Channel next year’’ (FUT), and (2) ‘‘It will be possible for Gertrude to swim the Channel next year’’ (FUT). Moreover, if ‘‘Gertrude will swim the Channel next year’’ (FUT), after swimming the Channel ‘‘Gertrude will have swum the Channel (FUT PERF), and ‘‘It will have been possible for her to have swum the Channel’’ (FUT PERF). Similarly, if ‘‘Gertrude could swim the Channel right now’’ (PRES), then (1) ‘‘Gertrude is able to swim the Channel right now’’ (PRES) and (2) ‘‘It is possible for Gertrude to swim the Channel right now’’ (PRES). Finally, if ‘‘Gertrude could swim the Channel last year’’ (PAST), then Gertrude was able to swim the Channel last year’’ (PAST), and (2) ‘‘It was possible for Gertrude to swim the Channel last year’’ (PAST). Therefore, in the examples in (20), both the overt modal could and its covert modal periphrastics be able to and be possible for . . . to can be tensed FUT, PRES, or PAST. We take it, then, that modals are tensed (and aspected). Deeper investigation of tense and aspect suggests that a distinction be drawn between propositional tense, on the one hand, and modal tense, on the other. This

 Philip A. Luelsdorff

distinction is motivated by examples in which there is a difference between the tense on the proposition and the tense on the modal, as in ‘‘Gertrude could (it seems to me now) (PRES) swim the Channel next year (FUT)’’ and ‘‘Gertrude could (I now infer) (PRES) swim the Channel last year (PAST).’’ Some further examples are given in (21). (21) Modal tense vs. Propositional tense a. John might (PRES) have quit (PAST). b. John might (PRES) quit (FUT). c. John was (PAST) able to be present (PRES). d. Next week John can (FUT) swim (HAB). Apparently, modal tense and propositional tense need not coincide. Coincide or not, therefore, modalities are temporally situated judgements about temporally situated states and events. Hornstein (1990) argues that the modals can bear tense, i.e. FUT (imperative will), PRES, and PAST (Modal + Have). For Givón (1993), Modal + Have is Modal + PERF, and for Okamura (1996), Modal + Have is Modal + PERF infinitive. Consider the examples in (22), which support the claims that (1) modals are tensed, (2) aspects (here PERF) are tensed, and (3) modal tensing and aspect tensing are in principle independent of one another. (22) Modals + Have a. John should have left by tomorrow. (FUT PERF) b. John should have left by now. (PRES PERF) c. John should have left by yesterday. (PAST PERF) In these examples, the modal is should, the aspect is PERF (have left), and both the modal and the aspect are three-ways tense-ambiguous among FUT, PRES, PAST. Moreover, should itself is ambiguous between deontic (O(BLIGATION)) and epistemic (P(OSSIBILITY)) readings, and each of these readings can be tensed FUT, PRES, or PAST and aspected for PERF (but not PROG). For example, ‘‘John will be/is/was obliged’’ and ‘‘It will be/is/was possible’’ quite independently of the tense and aspect on the proposition (main predication). The observation (Coates 1983) that deontics tend to take main predications in the FUT and that epistemics tend to take main predications that are statives does not invalidate the claim that modal tense and aspect and propositional tense and aspect are independent, i.e. ((T (A (MODAL)) (T (A (PROP)))), as in (It is (PRES) possible) ((that) John retired (PAST)) → (John might have (PRES PERF) retired). Thus, event tenseaspect is one thing, modal tense-aspect another. The simple tenses (FUT, PRES, PAST) take temporal adverbs that are punctual (e.g. at dawn, at five o’clock, etc.), whereas the PERFs take temporal adverbs that are boundary (since, since then, before, already, etc.) or frequentative

The natural order of cognitive events 

(seldom, often, once, twice, frequently, etc.). Curiously, Modal + Have constructions, although ex hypothsi PERFs, take both. Consider the examples in (23). (23) Modals+ PERF+ Adverbials a. John left at dawn. b. John has left already. c. John should have left at dawn. d. John should have left already. e. *John has left at dawn. In a., at dawn modifies left, and, in b., already modifies has left, i.e. the interim between the PAST (left) and the PRES PERF (has left). (c) should be ungrammatical, because it contains a PERF (have left) modified by a punctual temporal adverbial (at dawn). Something permits punctuals to modify PERFs. Since the only difference between ungrammatical (e) and grammatical (c) is that (e) is non-modalized whereas (c) is modalized (by should), modals can be said to license the modification of PERFs by punctuals. Evidently, modals can have ambiguous scopes. Consider the examples in (24). (24) Modal scope a. John should have left at dawn. 1. John did not leave. 2. John left, but he did not leave at dawn. b. John should have left already. 1. John has not left yet. If sentence a. is read as (1), the scope of should is have left, i.e. should (have left), i.e. MODAL(PERF). However, if sentence a. is read as (2), the scope of should is left at dawn, i.e. should (left at dawn), i.e. MODAL((PAST)ADV(T))). If sentence b. is read as (1), the scope of should is have left, i.e. MODAL(PERF), which, with already, is MODAL((PERF)ADVa.)). This scope analysis of modality explains the grammaticality of (23)(c) by requiring that ADV(T) (here at dawn) modify tense (here PAST), whereby both ADV(T) and PAST are in the scope of MODAL (here should) Note that on the same analysis (23)(e) is ungrammatical because ADV(T) (at dawn) is not in the scope of MODAL. Modality exhibits headship, whereby Mn is the head of Mn+1. Some examples are given in (25), and more are given in Luelsdorff (1997b). (25) Modal headship a. A + E=A b. N + A=N c. O + A=O d. O + N=O

f. g. h. i.

I + O=I R + E=R P + E=P P + A=P

 Philip A. Luelsdorff

In the natural order of modalized events, it seems clear that Es (actions, processes, states) are not only determined by As and Ps but by all the other modalities. For example, ‘‘Maria pushed a cart across the square’’ entails that she was able a. to do so, that it was possible (P) in general to do so, that it was possible (P) for her to do so, that it is certain (P) that she did so, that she intended (I) to do so, that it may (P) have been necessary (N) for her to do so, and that she may (P) have been experiencing a sense of obligation (O) to do so, etc. Evidently, the deeper one probes, the more covert modalities become overt. Contemporary studies in English modality are concerned with the specification and classification of modal forms and modal meanings and the contexts in which modal forms are bound to modal meanings (cf. above), sometimes based on large corpuses (Brown, Lancaster, Survey of English Usage) and sometimes involving one or another statistic. Such studies (although they go under a different name) are the beginning of a determinacy analysis of English modality. After the meanings of the modals have been isolated (Coates 1983; Suzuki 1986; Downing and Locke 1990), the semantic relations within and across the modals, being determinacies, should undergo a determinacy analysis. For example, Suzuki (1986) notes that the modal can can mean (1) Epistemic Possibility, (2) Root Possibility, (3) Ability, and (4) Permission, but the deterministic relationships between and among these meanings, and the deterministic relationships between and among these meanings of can and those of the other modal meanings remain unexplored terrain. In the example given, Epistemic Possibility determines Root Possibility (and vice versa) with greater or lesser measures if determinacy accuracy (I) and completeness (C). Similarly, across modalities. Obligation (ought) determines all of (1) Epistemic Possibility, (2) Root Possibility, and (3) Ability (and vice versa) with varying measures of I and C. Therefore, the analysis of modality ought to extend to the (clearly situation specific) deterministic investigation of all the modal meanings, whether affirmed or negated, whether overt or covert, requiring individual corpuses that are elicited rather than recorded. Downing and Locke’s (1990) functional conception of modality fosters our understanding of modal temporal organization. In modal structures of the type Epistemic + Event, the Event is the Topic, and the Epistemic is the Comment (Focus); in modal structures of the type Deontic + Event, the Event is the Topic, and the Deontic is the Comment (Focus). Generalizing over the remaining possibilities, i.e. (1) Epistemic + Epistemic (‘‘It has to be possible’’), (2) Epistemic + Deontic (‘‘may have to’’), (3) Deontic + Epistemic (‘‘ought to be possible’’), and (4) Deontic + Deontic (‘‘should be necessary’’), the terms on the right are always in the scope of the terms on the left. Moreover, if X(Y), i.e. if Y is in the scope of X, then X + Y, i.e. X precedes Y. First the speaker intervenes in the Event — this intervention is Deontic. Then, the speaker expresses confidence (Epistemic) in the intervention (Deontic). On this functional approach (Topic — Focus, Functional

The natural order of cognitive events 

Sentence Perspective) to the natural order of modalization, PROP (the Topic) temporally precedes MODAL(PROP) (the Focus) in modal cognition, whereas PROP (the Topic) temporally succeeds MODAL(PROP) (the Focus) in modal serialization. Modal judgements are, as it were, first suffixed and second prefixed to propositions, as shown in (26). (26) Natural Order of Modality (NOM) a. Modal Cognition Event < Modal1 < Modal2 < Modal3 b. Modal Serialization Modal3 < Modal2 < Modal1 < Event Several refinements are possible. For example, within the modals, Epistemics (e.g. ‘‘It is possible (THAT) . . .’’) follow Deontics (e.g. ‘‘It is possible (FOR) . . . TO . . .’’); within the Deontics, Volition (V) and Intention (I) follow Obligation (0) and Necessity (N), and each of these follows Ability a.; within the Epistemics, Possibility (P) follows Certainty (C), Probability (W) follows Possibility (P), and Probability (W) follows Certainty (C). To repeat, the order of modal cognition appears to be inversely related to the order of modal serialization. Tenses can shift from direct to indirect (reported) speech, from direct to indirect questions, and from assertions to conditions. Some examples are given in (27). (27) Tense shifting a. Indirect speech 1. ‘‘It is going to rain.’’ 2. John said (that) it was/is going to rain. b. Indirect questions 1. ‘‘Is it going to rain?’’ 2. John asked if it was/is going to rain. c. Conditions 1. It didn’t rain. 2. If it had rained/rained . . . Tense-shifting phenomena are inventoried in (28). Tense shifts in modals WILL (CAN, MAY) → WOULD (COULD/MIGHT) have been omitted because they are otherwise represented on the list. (28) Tense-shifting phenomena 1. FUT → FUT 4. PRES → PAST 2. FUT → PAST 5. PAST →PAST 3. PRES → 6. PAST → PAST PERF PRES

 Philip A. Luelsdorff

The tense-shifts in (28) are determinacies of the general form x,z → y, I=m and C=n. If left unbound, each of these determinacies has an accuracy (I) and/or completeness (C) that is less than 1.00. That is, these determinacies as they stand, are ambiguous. For example, I(PRES → PRES)=0.50, C(PRES → PRES)=1.00, and I(PRES → PAST)=0.50, C(PRES → PAST)=0.25. This state of affairs gives rise to the following question: Which binder(s), if any, contribute to the resolution of tense-shifting determinacy ambiguities? Several functional answers to this central question have been given. Proceeding from the observation that automatic sequence-of-tense analyses of backshifting phenomena are inadequate, attempts are being made to isolate the semantic, syntactic, pragmatic, communicative, and other contextual factors which favor or disfavor the choice of one or another shift. In the sequel, several of these attempts are reviewed. According to Hoffmann (1976), temporal, spatial, and demonstrative deictics are shifted if such shifts are required by a discrepancy between a ‘‘direct’’ speech event (Sprechsituation) and a reported event (Berichtsituation). For example, direct speech ‘‘I will be coming again the day after tomorrow’’ is indirect speech ‘‘She said she would be coming again the day after tomorrow,’’ if reported on the same day, but indirect speech ‘‘She said she would be coming again two days later,’’ if reported on a later day. As noted, Hoffmann’s constraint on shifting pertains to temporal, spatial, demonstrative and, we might add, pronominal deictics, but it can also be construed to apply to shifts in tense. The importance of the role of discrepancy resolution in binding tense-shifting (although it does not go under that name) is stressed in the insightful work of Palmer (1989) and Comrie (1986). Palmer maintains that there is a choice for the second (reporting) speaker between retaining the original present tense or shifting to the past and that this choice depends on whether the statement being reported continues to be true for the second speaker. In particular, if the second speaker believes that the statement being reported continues to be true, then ‘‘. . . he may (but is not required to) retain the present tense.’’ For example, ‘‘John said he still likes chocolates’ implies that the second speaker believes that John still likes chocolates. Similarly, Comrie also observes that the formal sequence-of-tense rules need not apply if the second speaker believes that the reported statement continues to be true. Therefore, on both Palmer’s and Comrie’s accounts, continuing applicability is (in our terms) a binder that favors the retention of the original tense. Conversely, discontinuing applicability is (in our terms) a binder disfavoring the retention of the original tense. More succinctly stated, belief retention favors identity (X → Y, where X=Y), whereas belief suspension favors difference (X → Y, X ≠ Y). Dirven and Radden (1977) subject tense-shifting to a ‘‘Distance Principle’’ according to which a speaker rendering direct speech by indirect speech ‘‘dis-

The natural order of cognitive events

tances’’ himself by not adopting an attitude towards (taking a stance on) the truth value of the direct utterance. However, while both direct and indirect speech is true or false (i.e. either John did or did not say ‘X’ (direct) or (that) X indirect), indirect speech, in addition, does license the expression of second-speaker attitudes towards first-speaker utterances in ways in which direct speech does not. For example, ‘‘John said he is/was ill, and I believe he is / but I believe he isn’t’’ (indirect), but not ‘‘John said ‘I am ill’, and I believe he is / but I believe he isn’t’’ (direct). Since indirect speech admits the focusing of a large number of attitudes which direct speech does not, including attitudes towards both speech events and narrated events, attitudes must be parameters in the theory of tense-shifting. Thus, tense choice in reported speech is not automatic, but subject to complex constraints. In a topical study of tense choice in reported speech, Declerck and Tanaka (1996) report that the use of PRES is hampered by a number of factors (our binders), including temporal domains, temporal focus, syntax, semantics, pragmatics, communicative function, and context. For example, these authors maintain that, in the unmarked case, head clauses and that-clauses agree in tense, with the result that ‘‘John said that he was sick’’ is understood as John’s having said ‘‘I was sick’’ and not ‘‘I am sick.’’ To the extent that this description is correct, the corresponding determinacy is PAST (said), PAST (was) → PAST (was), with higher values of I and C than the determinacy PAST (said), PAST (was) → PRES (is). Backshifting can be understood in a number of ways. According to van Eynde (1991), transposition (i.e. back-shifting) can be treated as a syntactic transformation or as an irregularity in the relation between form and meaning. In the present opinion (cf. Luelsdorff and Chesnokov 1996), there is nothing ‘‘irregular’’ about the same content being expressed by different forms and vice versa; on the contrary, such content-expression ambiguities are the rule (the more frequent) rather than the exception (the less frequent). Within the Praguian tradition (Jakobson 1957; Leška 1994), the code (C) and the message (M) are the basic conditions on communication, and the classification of verbal grammatical meanings requires the notions of speech event (Es), narrated event (En), participants in the speech event (Ps) and participants in the narrated event (Pn). ‘‘Shifters’’ are categories defined by their relationship to Es or Pn, e.g. person is understood as Pn/Ps and time (tense) as En/Es. These notions bear on the classification of original, direct, and indirect speech. Using X/Y to mean ‘X is about Y’, original speech is a speech event about a narrated event (Es/En); direct speech is a speech event about a speech event about a narrated event (Es/Es/En); and indirect (reported) speech is a speech event about a speech event about a speech event about a narrated event (Es/Es/Es/En). Cast in these terms, each type of speech is an intentional structure serving one or another intentional function, e.g. original speech — narration, direct speech —



 Philip A. Luelsdorff

exact rendition, indirect speech — inexact rendition. Moreover, in original speech, the first speaker and the subject of the reporting verb are the same, i.e. Ps=Pn, with the result that the first speaker sees the world from his or her own perspective; in direct speech, Ps=Pn or Ps ≠ Pn, with the result that Ps sees the world through the eyes of Pn; and, in indirect speech, Ps=Pn or Ps ≠ Pn, but Ps (the second speaker) sees the world of Pn’’ through his or her own eyes, the eyes of Ps. Evidently, Ps’s perspective on P (i.e. Ps/Pn) is one of the binders of the determinacies from Original to Direct and from Direct to Indirect, together with the second speaker’s attitudes towards the first speaker’s utterances and their contexts of occurrence. Declerck and Tanaka (1996) explain the PAST/PRES difference in ‘‘John said that Bill was/is ill’’ in terms of relative (PAST: was) and absolute (PRES: is) tense. In the authors’ words, ‘‘The relative past tense form relates the time of its situation (i.e. the time of Bill’s being ill) to the time of the situation referred to in the head clause (i.e. the time of John’s utterance) in terms of simultaneity. The absolute tense form, by contrast, relates the time of its situation directly to the time of utterance.’’ Some reported speech, however, cannot be understood in these relative-absolute terms, e.g. ‘‘John has often said that Bill was/is/*has been ill’’ (relative) and ‘‘John said that Bill *will be ill’’ (which cannot be absolute, because the time of the utterance is PRES, not FUT). By way of alternative, there is one shift to PAST, and this shift is responsible for both head clause (reporting verb) PAST and that-clause (reported verb) PAST. In other words, SHIFT is a bound determinacy, which, like all determinacies, is quantified for I and C. In summary, the tense shifts in (28) are bound determinacies of the form x, z → y, I=m and C=n, where x=y (identity shift) or x ≠ y (non-identity shift), depending upon the value (s) of z that bind x to y. For example, PRES in direct speech ‘‘Bill said ‘John is (PRES) sick’’’ determines, when bound by reported speech, either PRES or PAST ‘‘Bill said (that) John is/was (PRES/PAST) sick.’’ Here, as elsewhere, the natural order of cognitive events is (1) x, (2) y, (3) z, (4) xz, (5) x, z → y. Complementizers function to variously introduce subordinate clauses. Of the several complementizers posited for English (Rosenbaum 1967; Chomsky and Lasnik 1977; Luelsdorff and Norrick 1977; Kayne 1980; Chomsky 1981; Watts 1983), we are interested here in the temporal relations (i.e. the anteriority, simultaneity, and post-teriority) among the complements of the complementizers listed in (29). (29) English complementizers (incomplete) a. IF/WHETHER . . . OR NOT b. (FOR) . . . TO c. (POSS) . . . ING

The natural order of cognitive events 

d. (THAT) The complementizers in (29) can be illustrated by the sentences in (30). (30) English complementizers (examples) a. b. c. d.

Steffi knows if it is possible to win. It is possible for Steffi to win. Steffi’s winning is possible. It is possible that Steffi will win.

Two cases of IF/WHETHER complementation, one negative, the other affirmative, can be distinguished: (1) Steffi does not know if A or not-A and (2) Steffi knows if A or not-A. In the first case, IF/WHETHER antedates (THAT), because not knowing A precedes knowing A, if A is learned at all. In the second case, IF/WHETHER is simultaneous with (THAT), because knowing if A (or if not-A) is also knowing that A (or that not-A). Therefore, matrix sentence negation or affirmation determine whether IF/WHETHER complements express events anterior to, or simultaneous with, the events expressed by (THAT) complements. (FOR) . . . TO and (POSS) . . . ING, and their complements, appear to paraphrase one another. Following Suzuki (1986), example (30b), ‘‘It is possible for Steffi to win,’’ ((FOR) . . . TO), does not mean ‘‘Steffi has the ability to win,’’ but (30c), ‘‘Steffi’s winning is possible’’ ((POSS) . . . ING). That is, ‘‘It is possible for . . . to . . .’’ means ‘‘an event/action is possible. On this (root possibility) reading, the complements of (FOR) . . . TO and (POSS) . . . ING, as noted, are cognitively synonymous. (FOR) . . . TO=(POSS) . . . ING follow IF/WHETHER and precede (THAT), because the possibility of Steffi’s winning (30b, c) entails the possibility of winning general (affirmative (30a), and the epistemic possibility reading in (30d) entails the root possibility readings in (30a–c). Complements can be embedded in complements, as shown by the examples in (31), and, in such cases, the embedding clause is the head. For example, (31) a. expresses a doubtful root possibility, and (31b) expresses a doubtful epistemic possibility. (31) Embedded complements a. b. c. d.

Mary doubts if it is possible for John to win. Mary doubts if it is possible that John will win. Mary doubts that it is possible for John to win. Mary doubts that it is possible that John will win.

Complementizer temporalization and serialization seem to be governed by the same iconic, semantic principle, namely, the epistemic (or epistemological) principle that knowledge that is less certain precedes knowledge that is more

 Philip A. Luelsdorff

certain in both time and space, or ‘‘doubt precedes certainty.’’ Note that this Epistemological Principle may also explain the natural orders of tense and aspect (e.g. FUT is less certain than PRES, and PRES is less certain than PAST) , as well as the natural order of modality (e.g. epistemic (less certain) precedes deontic (root, agent-oriented) (more certain)). A tentative account of complementation temporalization and linearization is given in (32). (32) Natural order of complementation (NOC) a. Temporalization IF/WHETHER < (FOR) . . . TO=(POSS) . . . ING < (THAT) b. Linearization IF/WHETHER < (THAT) < (FOR) . . . TO < (POSS) . . . ING Just as there are natural orders of tense, aspect, tense-aspect, modality, and complementation, there is a natural order of moods (NOMO). English has four moods, (1) interrogative (Q), (2) declarative (D), (3) imperative (I), and (4) conditional (C), exemplified by the dialogue in (33). (33) Natural order of moods (examples) 1. A: ‘‘Is the door closed?’’ 2. B: ‘‘No, it isn’t.’’ 3. A: ‘‘Please close it!’’ 4. A: ‘‘If the door is open, close it!’’

(Q) (D) (I) (C)

Here, Q precedes D, D precedes I, and D and I precede C (C entails D and I). The dialogue could have begun with (4) (= C), but, in silent speech, it still would have entailed (1, 2, 3) (= Q, D, I), whereby speaker A models his/her (1, 3) and speaker B’s (2), i.e. A does not know if the door is closed or not (1) and if B says that the door is not closed (2), then A will tell speaker B to close it (3, 4). Evidently, the natural order of moods (NOMO) is the order given in (34). (34) Natural Order of Moods (NOMO) Q E and that Future < Present < Past.

The natural order of cognitive events 

Within the theory of determinacy grammar (DG), all (cognitive) linguistic relations enjoy determinacy form (DF) (Chesnokov and Luelsdorff 1991; Luelsdorff 1996a, 1996b, Luelsdorff and Chesnokov 1996). DF, as given in (4) a., consists of six parts: 1. a subject (x) 2. a predicate (y) 3. a binder (s) (z) 4. an oriented determinacy arrow (→) 5. an accuracy measurement (I) 6. a completeness measurement (C) Since a determinacy consists of six parts, it is possible for one determinacy to differ from another only in a limited number of ways. If there is one difference, the difference is simplex, and if there is more than one difference, the difference is complex. In this fashion, it is possible to compare and contrast determinacies and measure their similarities and differences (Luelsdorff and Chesnokov 1995). Numbering among determinacy differences are individual and/or communal norm observations and norm violations. For example, there are individual differences in language knowledge and use, including individual differences in language acquisition and attrition in both the normal and the pathological case. The model of temporal relations proposed has implications for investigations in these domains, including the clinical neuropsychological study of aphasia, agnosia, and amnesia. While neuropsychology handbooks (e.g. Poeck 1982) define and discuss such classical agnosias as visual object agnosia, acoustic and tactile agnosia, color agnosia simultaneity agnosia, spatial agnosia, and prosopagnosia, conspicuously absent, from the present perspective is what might be termed ‘‘temporal agnosia.’’ Error mechanisms can be opposed to error processing strategies (Luelsdorff 1986). The main types of error mechanism are defined in (35). (35) Typology of error mechanisms a. XAY → X0Y (Deletion) b. X0Y → XAY (Addition) c. XAY → XBY (Substitution) d. XAYBZ → XBYAZ (Left-displacement) e. XAYBZ → XBYAZ (Right-displacement) f. XAYBZ → XBYAZ (Transposition) g. XAYBZ → XBYBZ (Anticipation) h. XAYBZ → (Perseveration) XAYAZ Note that Y=0 in displacements (d, e), that Y ≠ 0 in transpositions (f), and that

 Philip A. Luelsdorff

Y=0 or Y ≠ 0 in anticipations and perseverations (g, h). Note further, that all the error mechanisms in (c-h) can be reduced to the error mechanisms in (a b), i.e. deletions and additions. Additions and deletions, in turn are movements into and out of the three components of a Determinacy Grammar — a Dictionary of Variables, a Data Matrix, and a Determinacy Analysis. Now, the model of tense and aspect presented consists of four tenses (FUT, PRES, PAST, GEN) and four aspects (TO FUT, PROG, PERF, PASS). Each tense is composed of the elements (relata) S (speech time) and E (event time), and each aspect is composed of the elements (relata) E1 (one event time) and E2 (another event time). Moreover, the elements composing each tense and each aspect are related by one of the three relative temporal relations < (anterior),=(simultaneous), and > (posterior). Finally, as we argue, there are natural temporal orderings among and across the several tenses and aspects themselves, e.g. FUT < PRES < PAST < GEN, and if El=E2 (PROG), then T1=T2. Summarizing, the English tenses and aspects are defined on the set (S, E, El, E2, <, =, >). Restricting the discussion to the (‘‘simple’’) tenses (i.e. FUT, PRES, PAST, GEN), we now ask how this model of tense (and, by extension, model of aspect and tense-aspect) is dissociable (or lesionable) in terms of the established error mechanisms listed in (35), in particular, the error mechanisms of deletion and addition in (35) (a, b). These are cases in which one or more of (S, E, <, =, >) can be omitted or omitted and some other element or relation added. (36) Temporal error mechanisms a.

Unbound (x z → y, where z=0) 1. S → 0 (Speech-time error) 2. E → 0 (Event-time error) 3. < → 0 (Anteriority error) 4. = → 0 (Simultaneity error) 5. > → 0 (Posteriority error)

b. Bound (x, z → y, where z ≠ 0) 1. S < E → 0 < E (Speech-time error) 2. S < E → S < 0 (Event-time error) 3. S < E → S 0 E (Anteriority error) 4. S 0 E → S=E (PRES/FUT error) 5. S 0 E → S > E (PAST/FUT error) ... ... Briefly, the temporal errors in (36) a. are determinacies that are unbound (context-free), while the temporal errors in (36) b. are determinacies that are bound (context-sensitive The former describe general dissociations of the elements (S, E) and relations (<, =, >) of tense; the latter describe dissociations of the

The natural order of cognitive events 

elements and relations of tense that are specific. A type-token analysis of (S, E, <, =, >) would disclose temporal errors that are specific to one or another domain or speech event. Of the examples given in (36), all are deletions except (36) b. (4, 5). These are additions whose input is the deletion in (36) b. (3). The determinacy form (DF) is 0, DEL) → =, I=m and C=n (for (36) b. (4)) and 0, DEL → >, I=m and C=n (for (36) b. (5)). As always, the analyst must discriminate between norms (whether individual or communal) and deviations and, among the deviations between errors of competence, errors of performance, and errors of both competence and performance. Moreover, distinctions must be drawn between errors in production and comprehension and between errors in writing and speaking. Just as the neuropsychological disorder of prosopagnosia (i.e. the inability to recognize otherwise familiar faces) can be meaningfully treated by the systematic introduction of just those binders (binding precisors) needed to transform ‘‘nonovert’’ recognition into ‘‘overt’’ recognition (Luelsdorff 1992), temporal agnosia (i.e. the inability to produce or recognize otherwise familiar temporal relations) might be meaningfully treated by the systematic introduction of just those binders (e.g. temporal adverbials and the names of persons, and events) needed to transform non-overt temporal production and recognition into temporal production and recognition that are overt.

.

Conclusion

Reichenbach’s Elements of Symbolic Logic (1947) contains an analysis of conversational language, including tenses, and sequence of tenses, with special (although not exclusive) reference to English. It is safe to say that Reichenbach’s excellent analysis has been the most influential in modern English linguistics. Reichenbach represents time by a line and the direction of time by the direction of the line from left-to-right. Different times are represented by different relations among three points — the point of the event (E), the point of reference (R), and the point of speech. Simultaneous points are separated by a comma (,), and sequential elements are separated by a space (_). In order to represent the time extension of an event, Reichenbach introduces the auxiliary cover symbol ( ). These elements (S, E, R) and relations (,, _, ) enable Reichenbach to define twelve English tenses (non-extended and extended), the non-extended of which are given in (37). (37) The English Tenses (Reichenbach 1947) a. Past Perfect d. Present

 Philip A. Luelsdorff

b. Simple Past c. Present Perfect

e. Simple Future f. Future Perfect

Our analysis of the English tenses and aspects deviates from the Reichenbachian tradition in several central respects. First, according to Reichenbach, English has tenses (cf. the tenses in (37) and their corresponding extended forms); for us, English has both tenses and aspects. The tenses are FUT, PRES, and PAST, and the aspects are TO-FUT, PROG, and PERF. The tenses are defined in terms of temporal relations between S and E, and the aspects are defined in terms of temporal relations between El and E2. Second, according to Reichenbach, the English tenses are defined in terms of the temporal relations between S, E, and R (the R-ful theory of tense), whereas, according to us, the tenses and aspects are defined in terms of the temporal relations between S and E (or El and E2) without R (the R-less theory of tense and aspect). Of these two theories of tense (or tense and aspect), the R-less theory, dispensing as it does with R, is clearly the simpler. Third, Reichenbach is concerned with defining tenses (the relevant section of his book is called ‘‘The Tenses of Verbs’’), but we are concerned with defining tenses and aspects and establishing the natural orders of tensed and aspected events, including the natural orders of tensed events that are aspected (and vice versa). The order of tenses Reichenbach gives, reproduced in (37), i.e. PAST < PRES < FUT, cannot be the correct order of tenses, requiring as it does that an event is tensed PAST before it is tensed PRES and tensed PRES before it is tensed FUT. Finally, Reichenbach introduces the ad hoc symbol ‘ ’ in order to represent extended (PROG) events and makes no mention of tense in voice. As we have argued, both PROG and PASS are aspects which can be tensed FUT, PRES, or PAST. Tenses and aspects are composed and ordered by mental movements. The central feature of the model of movement presented is that the tense/aspect pairs FUT/TO FUT, PRES/PROG, PAST/PERF, PAST/PASS, and their strict natural orders, flow from the same cognitive movements. This movement model is iconic in that tense/aspect movements directly depict tense/aspect meanings. On the natural order of modality proposed, it must be possible for (P) (= root possibility) agents to be able a. to act (E), i.e. P > A > E. Moreover, the possibility (likelihood) that (P) (= epistemic possibility) the agent will (FUT) act is greater if the agent has permission (R) (in R-situations) to act, wants (V(olition)) or intends (I), and is obliged (0) or necessitated (N) to act. All modalities have causes (= modal causes) and effects (= modal effects) — e.g. x obliges y to z and, as a result, y is obliged to z and, as a result, y should/ought/must z and, as a result, y will z. Surface modals are therefore terms in a chain of cause and effect. Finally, modalities are overt (surface) and/or covert (deep), as are tenses and aspects. For example, in ‘‘Boris castled on his queen’s side’’ no modalities are overt, and the only overt tense is PAST. However, there are several modalities and tense/aspects that are covert,

The natural order of cognitive events 

such as A, P (root), P (epistemic), V, I, R (the rules of chess), (possibly) N, among the modalities, and FUT, PRES, PROG, and PERF, among the tense/aspects. In the section on (back)shifting we attempt to lay the foundations of a determinacy analysis of shifters in indirect (reported) speech. For one, original, direct, and indirect speech are differentiated in terms of the participants in the speech event (Ps) and the participants in the narrated event (Pn), whereby Ps can adopt his/her own perspective or the perspective of Pn. Shifters arise as the result of movements (changes) in the perspectives (i.a. spatial and temporal coordinates) of Ps’s and Pn’s. For example, the shift from PRES Es to PAST Es determines the shift from PRES Es to PAST Es in reporting verbs, as in ‘‘Boris says: ‘Anatolij will push a pawn’’’ (original), ‘‘Boris said: ‘Anatolij will push a pawn’’’ (direct), and ‘‘Boris said (that) Anatolij will/would push a pawn’’ (indirect). Declerck and Tanaka (1996) document a larger number of factors which hamper the use of PRES in reported speech. These factors are argument binders in determinacies of the abstract form preshifted, factors → preshifted and preshifted, factors → postshifted. Finally, in view of the fact that preshifted and postshifted elements in original, direct, and indirect speech are coreferential, backshifting phenomena appear to be anaphora which are bound (to a greater or lesser extent) by those factors which hamper the use of preshifted phenomena, on the one hand, and promote the use of postshifted phenomena, on the other. The discussion proceeds with an inconclusive proposal for natural orders of complementizers and moods. It is suggested that IF/WHETHER-complements precede (THAT)-complements, and that (FOR) . . . TO=(POSS) . . . ING-complements are situated someplace in between. If [±WH], zero, and null complementizers were also taken into consideration, the posited natural order of complementation might have to be revised. Among the moods, questions are found to precede answers, answers to precede imperatives, and imperatives to precede conditionals. The paper concludes with a pilot investigation of a variety of conjectured neuropsychological temporal disorders — symptoms and syndromes which are subsumed under the designation ‘‘temporal agnosia.’’ A typology of established error mechanisms is presented (including omissions, deletions, substitutions, (backwards and forwards) displacements, transpositions, anticipations, and perseverations) and used to predict a number of possible dissociations (disconnections, lesions) in some of the temporal relata and relations proposed. In retrospect, the natural orders of cognitive events share the relative temporal relations of anteriority, simultaneity, and posteriority. Corresponding to these relative temporal relations are the epistemic judgements less certain, more certain, most certain. To the extent that this is correct, certainty and temporality are directly related.

 Philip A. Luelsdorff

Notes . Vendler (1967) recognizes four categories of verbs (STATES, ACTIVITIES, ACCOMPLISHMENTS, and ACHIEVEMENTS), and Prior (1968) recognizes (THINGS) (which change) and (EVENTS) (which happen or occur). A functional approach to the question of events is Hansen et al. (1982), which is an attempt to relate predicate types (STATES, PROCESSES, ACTIONS) to role types (PATIENT, STATIVE, PROCEDENT, PROCESSIVE, AGENTIVE, OBJECTIVE, FACTITIVE, DATIVE, INSTRUMENTAL). . A further distinction can be drawn between GEN (ERIC) (‘‘The earth revolves around the sun’’), on the one hand, and HAB (ITUAL) (‘‘John is (always/sometimes) late’’), on the other. . Also, in PROG, as opposed to TO FUT, PERF, and PASS, the participants in the narrated event are contemporaneous with BE and the MAIN VERB, e.g. in ‘‘Johann was writing a letter,’’ ‘‘Johann’’ and ‘‘letter’’ are contemporaneous with ‘‘was’’ and ‘‘writing.’’ . On this conception, by extension, PASS expresses a state posterior to the completion of an action rather than the mere result of an action. . These representations are not intended to suggest that tenses and aspects ‘‘flow’’ in lines. On the contrary, tenses and aspects result from movements in two-dimensional space, as shown in (10). . Tense-aspect data does not always support one or another hypothesis in (8). For example, ‘‘John wants (PRES) to leave (FUT) tomorrow’’ supports H1, but ‘‘John wants (PRES) to leave (PRES) now’’ does not, suggesting that the latter sentence is an example of H2 rather than H1. . Cf. Note §4 in which it is maintained that PASS, by analogy to PERF, expresses a state incumbent upon an action rather than the result of an action per se. . On this definition of ‘‘movement,’’ generative ‘‘alpha-movement’’ is disqualified from the domain of movement. . In the present opinion, tense (FUT, PRES, PAST, GEN) in infinitivals and participials is neutralized, and it is neutralized because it is fully determinate.

References Bühler, K. (1934). Sprachtheorie: Die Darstellungsfunktion der Sprache. Jena: Gustav Fischer. Chesnokov, S. V. and P. A. Luelsdorff (1991). ‘‘Determinacy analysis and theoretical orthography,’’ Theoretical linguistics 17: 1–3, 231–62. Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris. Chomsky, N. and H. Lasnik (1977). ‘‘Filters and control,’’ Linguistic inquiry 8: 3, 425– 504. Coates, J. (1983). The semantics of the modal auxiliaries. London: Croom Helm. Coates, J. (1995). ‘‘The expression of root and epistemic possibility in English.’’ In: J. Bybee and S. Fleischman (eds.), Modality in grammar and discourse. Amsterdam/Philadelphia:

The natural order of cognitive events 

John Benjamins, 55–66. Comrie, B. (1986). ‘‘Tense in indirect speech,’’ Folia linguistica 20, 265–96. Declerck, R. and K. Tanaka (1996). ‘‘Constraints on tense choice in reported speech,’’ Studia linguistica 50: 3, 283–301. Dirven, R. and G. Radden (1977). Semantische Syntax des Englischen. Wiesbaden: Akademische Verlagsgesellschaft Athenaion. Downing, P. and P. Locke (1990). A university course in English grammar. New York: Prentice Hall. Dressler, W. U. (1994). ‘‘Functional sentence perspective within a model of natural textlinguistics.’’ In: S. Cˇmejrková and F. Štícha (eds.), The syntax of sentence and text: A Festschrift for František Daneš. Amsterdam/Philadelphia: John Benjamins, 89–104. Enç, M. (1996). ‘‘Tense and modality.’’ In: Shalom Lappin (ed.), The handbook of contemporary semantic theory. Oxford/Cambridge, Mass.: Blackwell, 345–58. Garvin, P. L. (1994). ‘‘Karl Bühler’s field theory in the light of the current interest in pragmatics.’’ In: S. Cˇmejrková and F. Štícha (eds.). The syntax of sentence and text: A Festschrift for František Daneš. Amsterdam/Philadelphia: John Benjamins, 59–66. Givón, T. (1993). English grammar: A function-based introduction. Amsterdam/Philadelphia: John Benjamins. Hajicˇová, E., J. Panevová and P. Sgall (1973). ‘‘The meaning of tense and its recursive properties.’’ In: Wolfgang Klein and Armin v. Stechow (eds.), Functional generative grammar in Prague. Kronberg/Taunus: Scriptor Verlag, 173–235. Hansen, B. et al. (1982). Englische Lexikologie. Leipzig: VEB. Heine, B. (1995). ‘‘Agent-oriented vs. epistemic modality: Some observations on German modals.’’ In: J. Bybee and S. Fleischman (eds.), Modality in grammar and discourse. Amsterdam/Philadelphia: John Benjamins, 17–54. Hofmann, T. R. (1993). Realms of meaning: An introduction to semantics. London/New York: Longman. Hoffmann, H. G. (1976). Englische Mindestgrammatik. München: Max Hueber Verlag. Honderich, T. (ed.) (1995). The Oxford companion to philosophy. Oxford: Oxford University Press. Hornstein, N. (1990). As time goes by. Cambridge, Mass.: The MIT Press. Jackendoff, R. (1983). Semantics and cognition. Cambridge, Mass.: The MIT Press. Jakobson, R. O. (1957). ‘‘Shifters, verbal categories, and the Russian verb.’’ Russian Language Project. Cambridge, Mass.: Harvard University. Kayne, R. (1980). ‘‘On certain differences between French and English,’’ Linguistic inquiry 12: 3, 349–71. Leška, O. (1994). ‘‘Morphological categories in the Prague School.’’ In: Philip A. Luelsdorff (ed.), The Prague School of structural and functional linguistics. Amsterdam/Philadelphia: Benjamins, 73–121. Luelsdorff, P. A. (1984). ‘‘Einstellung als ein Faktor des erweiterten Organon-Modells der Sprache.’’ In: Achim Eschbach (ed.), Bühler-Studien. Vol. I. Frankfurt/ Main: Suhrkamp, 317–42. Luelsdorff, P. A. (1986). Constraints on error variables in grammar: Bilingual misspelling orthographies. Amsterdam/ Philadelphia: John Benjamins.

 Philip A. Luelsdorff

Luelsdorff, P. A. (1987). ‘‘On linguistic error.’’ In: Philip A. Luelsdorff (ed.), Orthography and phonology. Amsterdam/Philadelphia: John Benjamins, 77–100. Luelsdorff, P. A. (1992). ‘‘Transitional determinacies,’’ Folia phoniatrica et logopaedica 44: 6, 282–90. Luelsdorff, P. A. (1996a). ‘‘Auxiliary movement,’’ Submitted. Luelsdorff, P. A. (1996b). ‘‘Modality revisited,’’ Submitted. Luelsdorff, P. A. (1997a). ‘‘Cognition as determinacy,’’ Prague bulletin of mathematical linguistics 68. Luelsdorff, P. A. (1997b). ‘‘Questions of modality,’’ Theoretical linguistics. Luelsdorff, P. A. (1998). ‘‘Transitional cognitive determinacy,’’ Prague bulletin of mathematical linguistics 69. Luelsdorff, P. A. and Sergej V. Chesnokov (1995). ‘‘Individual differences in similarity judgements.’’ In: R. M. Joshi and Che Kan Leong (eds.), Reading disabilities: Diagnosis and component processes. NATO Advanced Studies Institute, Series D: Behavioral and Social Sciences. Dordrecht: Kluwer, 263–85. Luelsdorff, P. A. and S. V. Chesnokov (1996). ‘‘Determinacy form as the essence of language.’’ In: E. Hajicˇová et al. (eds.), Travaux du cercle linguistique de Prague n.s./Prague Linguistic School papers 2. Amsterdam/Philadelphia: John Benjamins, 205–34. Luelsdorff, P. A. and N. R. Norrick (1979). ‘‘On if/whether complementation,’’ Linguistische Berichte 62: 25–47. Lyons, John (1977). Semantics. Cambridge: Cambridge University Press. Okamura, Y. (1996). ‘‘The grammatical status of pure future ‘will’ and the category of future form,’’ Studia linguistica 50: 1, 35–49. Palmer, F. R. (1979). Modality and the English modals. Cambridge: Cambridge University Press. Palmer, F. R. (1989). The English verb. Second edition. London/New York: Longman. Poeck, K. (ed.) (1982). Klinische neuropsychologie. Stuttgart/New York: Georg Thieme Verlag. Prior, A. N. (1968). Papers on time and tense. Oxford: Clarendon Press. Reichenbach, H. (1947). The elements of symbolic logic. New York: The Free Press. Rosenbaum, P. (1967). The grammar of English predicate complement constructions. Cambridge, Mass.: The MIT Press. Suzuki, N. Y. (1986). ‘‘The modals in American English: It is their background that matters,’’ Paper A156, L. A. U. D. T., University of Duisburg. van Eynde, F. (1991). ‘‘The semantics of tense and aspect.’’ In: M. Filgueiras et al. (eds.), Lecture notes in artificial intelligence 476, 97–121. Vendler, Z. (1989) ‘‘Verbs and times.’’ In: Linguistics and philosophy. Ithaca, New York: Cornell University Press, 97–121. Watts, R. J. (1983). ‘‘On infinitival complement clauses,’’ Studia anglica posnaniensia 16, 45–70.

section V

Poetics

The principle of free verse Miroslav Cˇervenka Prague

0. Introduction In the European literary culture, in its national and social subcultures, during different stages of their development verse formations are created which are rhythmically ‘‘looser’’ than it would be assumed by the broadly accepted standard of the respective versification. The degrees of this looseness can differ widely both in the synchronic and in the diachronic dimensions. In any case, the looseness consists either in (a) the standing norms being applied in the given texts with a weaker commitment than in standard ones, or (b) some or all of the norms loosing their validity altogether in a specified group of texts, not controlling any more the choice and the combination of language units used; the general verse norm, however, is applied even when such texts are created: the whole text is evaluated as ‘‘deviant’’, be it in a positive or in a negative sense, rather than for every position in the text to be judged independently from the viewpoint of this norm (and then often to be classified as ‘‘unusual’’, ‘‘deviant’’ or ‘‘mistaken’’). (c) Instead of standard norms, which are backgrounded, new norms, as the case may be, emerge that have not yet been present in the standing form of the literary culture, and that have a broader or narrower range of validity. And also these norms are applied in the way stated under a., i.e. more according to a probabilistic principle (it is or is not recommended to . . .) than to a deterministic one (it is prescribed to . . . or prohibited to . . .).

.

The tradition of free forms and a perceiver

It is symptomatic for modern literature that the above mentioned forms constitute their own tradition: they are not created and perceived only on the background of standard forms (and therefore always as ‘‘deviant’’ or similarly), but also on the background of previous formations of their own type, the older ‘‘free’’ formations. This is an immensely important factor, because this opens the way both for a

 Miroslav Cˇervenka

steady increase of the use of these verse forms (which lose their original character of deviance), and for their always progressing looseness. The domestication of the free forms of every period constitutes a basis for the acceptability of even looser forms in the next period. Together with the incorporation of free forms into their own tradition, their inner structure reveals a shift between the relevance of inherent (textually manifested) rhythmical features and the relevance of the orientation (Einstellung) and mental activities of the participants of the literary communication for the acceptance of the given text as a versified form. The latter factor gains substantially more significance. While it is hardly possible to imagine a reader of Cˇech’s Ve stínu lípy who would doubt to be perceiving verses, with Holan’s Prˇíbeˇhy this image is quite well conceivable. This has a great importance also for the methodology of inquiries: it gives ground to an implication of the necessity of a persistent regard to the pragmatic layer of literary communication; any ‘‘objectively’’ describable textual feature must be evaluated from the point of view of the context and of its historically situated perceivers. The timepoint in which the developmental continuity of free rhythmic conventions emerge and gain independence in many European literatures, including the Czech, is that of the last decades of the nineteenth century. (An exception is, first of all, the German literature, in which the evolutional continuity of ‘‘free rhythms’’ can be claimed to have started in the eighteenth century.) In the sequel we are concerned with the period that then began. We refer to the object of our research as to the commonly and internationally accepted concept ‘‘free verse’’. Naturally enough, free verse shares some of its features with older analogical free formations, which, however, are not included in its tradition; this is important for us only if it influences somehow the semantics of the verse. Even so, also the older regular verse formations are significant for free verse.

.

Rhythmical intention

Different ways of taking up the contemporary or older standard verse formations and systems are of a fundamental importance for a certain part of the studied epoch in order to accept the given text as versified. Fragments, incomplete hints and secondary features of indisputable verse patterns represent intimations both for the author and the perceiver to classify the given texts as belonging to the tradition of verse art (poetry) and thus to understand their sound structuring as rhythmical. (The rhythmical structuring of speech consists in an articulation of speech into segments in which a certain configuration of speech sounds is periodically repeated, so that these segments are mutually equivalent; such an organization of speech can invoke an expectation that after

The principle of free verse 

a series of similarly organized segments another similarly organized segment will appear.) We do not want to go so far as to claim that the author thus gives the perceiver the task to mold a ‘‘complete’’ rhythm from these hints and fragments, or even to shape the verse pattern of a type suggested by the text in free verse; it would be sufficient if such a suggestion awakes the perceiver’s rhythmical orientation (or: intention), if it stimulates him to find in the text general features of similarity with texts that undoubtedly are versified, to transfer them to a common denominator and to summarize them into an overall impression of rhythmical organization. Also this general rhythmical orientation has its intended sound equivalent (see below). In this — restricted — sense the rhythmical structure of free verse can be considered as ‘‘parasitic’’, i.e. existentially dependent on other, ‘‘more authentic’’ structures. It does not seem that a free verse would be, in the framework of modern literature and of modern art, something very special in this respect: let us recall, e.g., how thin and at the same time how fully sufficient are the threads that connect some poems of early Seifert, called by the author himself ‘‘ballads’’, with traditional prototypes of the given genre; a parallel can be looked for in the socalled abstract painting, which, renouncing the presentation of shapes of objects, strengthens its appurtenance to painting by reminiscences of different epochs, trends, conceptions of space/plane and so on in the preceding development of its domain of art.

.

Phylogenetic and ontogenetic dimensions

What can be described and enumerated as a textual givenness is not uniquely decisive for the presence of versification in free verse texts, but shares this responsibility with what the author’s and the perceiver’s orientation (intention) brings into the text. Also this makes free verse a typical phenomenon, rather than an exception, in the framework of modern art: compare the visual and literary readymades, the music composed from non-tonal sounds, etc.; after all, these very phenomena have led modern aesthetics to prefer functional specification of arts to a characterization based on certain qualities, substantial features. With Tynjanov (1924/1987: 450) this can be also formulated so that free verse bases its rhythmical character on a minimum of rhythmical features. This minimum, of course, is subject to developmental changes: what suffices for the evocation of rhythmical orientation today did not suffice fifty years ago, or even a century ago, when free verse just started to develop in Czech poetry. This was noticed already then by F. X. Šalda (1898/1951: 16,32), who denounced the amorphy of some verselibristic texts, because he did not find in them a sufficient number of inherent rhythmical features capable to evoke the contemporary

 Miroslav Cˇervenka

reader’s rhythmical orientation. Today we accept some of these texts without problems as being versified. A potential preparedness to accept more and more free texts is substantiated by the independent tradition of free verse, which was mentioned in Section 2 above. Apart from this phylogenetic aspect of the prerequisites for the acceptance of free verse, there also exists an ontogenetic dimension–the perceiver’s preparedness that is restricted in time by an individual’s life and stratified according to the social scale; the latter dimension is determined by the perceiver’s experience with (modern) poetry. The degree of the possible freedom of free verse corresponds in principle with the vertical differentiation of literature. Until the present time, free verse is not a phenomenon of poetry waiting for a mass response (cf. the abandonment of free verse in the poetry of ‘‘socialistic realism’’), neither a phenomenon of text designed for special circles of readers (verses for children), or for extra-aesthetic functions (advertisements), etc. Regularly, the most free verse formations belong, in their time, to the most exclusive ones.

. A list of evokers The rhythmical intention naturally is both the basis and one of the conditions of the production and perception of any verse formations. With a regular, standardized verse this concerns the acceptance of the text in the context and on the background of the accepted norm and of the experience with the realization of this norm in the previous analogical formations. Only in such a case the stimuli arising from the relevant inherent features of the text can be summarized into integrated rhythmical impressions, which constitute one component of the semantic structure of the piece of art (we do not assume a rational articulation of these impressions). In case of free verse, the existence of verse as such is concerned. A more important role is played by factors which can be approximately denoted as ‘‘external’’ (from the viewpoint of rhythm, not always from that of the piece of art as a whole). First of all the following factors, ordered from the most external ones are at play: 1. the appurtenance to the typical communicative situations in which usually versified speech is used, e.g. publishing of a text in a book series of poetry, recitation from the platform of a poetic café, etc.; 2. the authorship of the text: its author is a writer known as a poet; 3. the title (subtitle) of the text, especially, of course, metatextual indications in it, and also its stylistic and thematic features; 4. symptomatic features of poetic genres in which verse is generally used; 5. themes usually appearing in poetry;

The principle of free verse 

6. ‘‘poetically’’ marked elements of the language and style from language layers other than the sound patterns — tropes, syntactic figures, parallelisms and inversions, etc.; 7. configurations of elements of the sound layer of language which themselves are not bearers of rhythm, but are often connected with it, or, as the case may be, represent its accompanying, optional features: sound sequences and responses, patterning of intonational segments from the point of view of their placement, scope and melody, the features of regularity in changes of tempo and of switching of voice timbres, etc.; 8. paralinguistic phenomena in the narrow sense: voice and other (esp. visual) elements which are imposed on the poem and on the impression from it by the production of the performer and — within the framework of the most usual way of communication of poetry, silent reading — the graphic arrangement of the text. A most significant feature, of course, is the characteristic articulation of the text into lines, which do not reach the right-hand side margin of the page and thus are not, as for their extent, mechanically determined by the length of the line. This arrangement indicates a special articulating intention — we assume, on the basis of the analogy with standard versified texts that also in case of free verse this intention concerns rhythm. The above enumeration purposefully does not include the previously mentioned fragments of traditional rhythmic structures, synecdochic allusions which make certain free verses similar to linguistic structures of classical verse. These are phenomena of a different range, which themselves and immediately — though in an incomplete shape, calling for the reader’s integrating activity — are bearers of rhythm. On the other hand, we would like to emphasize once more that the factors 1. — 8. are not bearers of rhythm in the sense in which the distribution of stressed and non-stressed syllables in syllabotonic verse bears rhythm. What we have enumerated is, sit venia verbo, a list of evokers of a rhythmical orientation. They do not create rhythm themselves, but they exhort the participants of communication to accept the text as rhythmical; this call, however, contains also a promise that the participants of the communication create in the text, or find there, supports for rhythm, inherent linguistic features which confirm such an acceptance. Nevertheless, even if this activity is not successful, the given text is brought into the domain of versified texts and should be — in the only sphere where the rhythm of literary works is realized, namely in the sound sphere — transformed in the sense of the demonstrated rhythmical intention, perhaps from the outside and by force. It is possible that the last part of the preceding sentence holds only for that development stage in which the free verse has already been fully established as verse within the rhythmic consciousness. In the initial stages a successful search for at least fragmentary inherent sound configurations in the text is a condition for free

 Miroslav Cˇervenka

verse being accepted as verse. Thus there is a big difference between a situation in which the background of a free verse text is constituted by other free verse texts and the initial situation, in which this background consists of texts in traditional verse.

.

The graphic arrangement

Among the factors of rhythmical orientation, an extraordinary significance is displayed by the graphic arrangement of the text. In contrast to the sound and other signals supplied by recitation, the graphic arrangement has the advantage that it is connected with the communicative situation that is most typical of modern poetry, i.e. with a silent reading of a book. It is a marker without which free verse cannot exist: it is difficult to imagine such free verses presented in continuo that would be distinguishable (even for a cultivated perceiver) from ‘‘rhythmisized prose’’ or from similar transition formations. The only exception are rhymed free verses, with which rhyme would play the most important role of the graphic arrangement, i.e. marking the boundaries between verses. Free-verse texts share the presence of inequal right border of the typed lines (as a rule placed far before the physically conditioned right margin of the page) with poems of classical metric verse; and this feature, in contrast to other, optional markers, is by the poets of free verse realized completely, not only as an allusion and fragment. With the incompleteness and optionality of other rhythmical features, the graphic marking of boundaries seems to be a substantial condition of rhythm in free verse. In classical verse, such graphic marking is more or less redundant, because the boundaries between verses are sufficiently marked by the means of the sound rhythm itself. In free verse, the informational impact of this graphic arrangement is twofold: first, it is a message implied by the overall visual impression, evoked by the contours of the printed text on the page; its contents is a statement similar to ‘‘this is a poem’’ (and ‘‘this is versified speech’’), thus also a call to take on an appropriate attitude to the text as a whole. The graphic arrangement shares this information value with other factors of rhythmic orientation. Second, the individual endings of the lines mark the beginning and the end of each individual rhythmic unit, and thus also its scope (this has been one of the most important semantic characteristics of verse since long ago), the distribution of denominations in its range, the immediate relation between the verse articulation and the syntactic structure, and so on. The graphics thus supplies a rhythmical skeleton that is a basis for parallelisms, relations that connect words in accordance with their positions in the line, or, as the case may be, constitutes also higher compositional wholes (strophes). At the same time, the graphic arrangement belongs to those factors of rhythmic orientation which, as we have said, are not bearers of the rhythmic

The principle of free verse 

organization proper. Also the graphic arrangement is an ‘‘evoker’’, an initiator. It functions in the text as an appeal to apply also to this new text the accumulated experience with texts the rhythmical arrangement of which is indisputable (i.e. experience with metric texts and with free verse texts that have already been accepted). Thus, the rhythm in free verse is constituted, among other factors, by a rhythmical impulse (an expectation of periodicity), as determined by the whole history of poetry, internalized in the literary competence of the speaker and the perceiver. When reading a metric poem, its previous lines (the rhythmical impulse evoked by them) make us understand e.g. Brˇezina’s verse škubnutí Blížení po drátech éterných as an alexandrine, rather than a dactylic tetrametre, although the inherent distribution of stresses supports the latter solution; in a similar vein, all poems read up to then make us understand Kolárˇ’ s lines Naši využívají prˇerušení a strˇídají. Nastupuje první útocˇná . . . as (free) verses. There is still another condition, of a more outspoken pragmatic nature. The experience with the previous kinds of poetry in specific situations and for a certain type of reader can of course have an impact also in a quite opposite direction, namely it may lead to a conviction that ‘‘this is not a verse’’. In this way Brˇezina and Sova certainly were read by a number of those used to the poetry of Vrchlický. The orientation of verse towards a minimum of rhythmic features may be shared only by a speaker and perceiver with a certain attitude to poetic tradition. It is an attitude of openness to new things, an attitude for which tradition is a support of change. This attitude concentrates attention to the remaining minimal set of features, so that even the yet unknown, but expected possibility of rhythm is welcome and even with a radical reduction of features, a differing formation remains faithful to its principle.

. The sound nature of literary artifact Our careful formulation of the characterizations of the role of graphic in the constitution of free verse has good reasons. The operational space of our considerations is not large. On the one hand, we are aware that by means of a special marking of verse boundaries supraindividual factors enter into the poetic communication, making a rhythmic orientation possible; without this orientation free verse cannot exist. On the other hand, we refuse to include graphic into proper bearers of rhythm. This would, in fact, mean to separate free verse from another, more

 Miroslav Cˇervenka

regular arrangement of versified speech. It hardly can be doubted that each rhythm in literary art has a sound nature (even if it were supported by potential, intended sounds, rather than by actually sounding ones). This statement is connected both with the view that if a piece of literary art addresses directly some sense organ, it is the ear (in more detail, see Cˇervenka 1992/1996), and with the commonly accepted impression that rhythm in arts is altogether a sound phenomenon. And even if we get reconciled with a visually perceived ‘‘rhythm’’ (e.g. a colonnade) as with a descriptive, rather than metaphorical denomination, it is difficult to imagine that from a viewpoint of artistic effect there would be some analogy between this visual rhythm and a sensually very scanty impression from the switching of printed and blank parts of lines in the printed poem. Apart from special cases (as the contemporary experimental poetry) placed on the boundary between the art of word and the visual arts, we would in any case have to classify the effects of such a kind rather with the visual art of book outline, which, in our culture, necessarily accompanies a literary work presented in print or in writing, although it is not a proper component part of this piece of art. Different printings of the same text are a trivially simple example of this: it would be rather strange to believe that Brˇezina’s very long free verses have a different rhythm in each of the visually so different presentations as if placed once in a wide page of the Mánes edition, where they mostly fit into one line in spite of their length, and at another time in a small book in the edition of Kveˇty poezie, where almost every verse must be divided into two lines. Graphic is a conventional signal where to direct the strength of our accumulated experience with verse, to be able to feel the presence of verse even in a line not corresponding to the standard norms of verse, and to modify it in sound accordingly.

.

Intonation

The question what is the bearer of rhythm in free verse has got more particular, but the answer seems to be more and more paradoxical. We have seen that the graphic arrangement of text initiates the activity leading to a constitution of the rhythm of free verse, and this may be only a sound rhythm. The nature of the literary art implies that these rhythmically arranged and shaped sounds may be only sounds of speech. At the same time, the ever lasting loosening of free verse during its development leads us to conclude that its rhythm cannot be fatally connected with any determined and positively quite generally specifiable configuration of speech sounds as we know this from the classical verse in different prosodic systems. The question of a minimal rhythmical mark is the question concerning the barest and most primitive ‘‘essence’’ of verse rhythm: we ask what ‘‘is left’’ if we remove all hitherto known features and norms of rhythmisation.

The principle of free verse 

If we are interested in a historically concrete form of free verse — and this will be our objective elsewhere — our task looks quite differently: we can base our judgments on a positive description of quite specific configurations of linguistic elements. This does not help us very much from a general point of view, because neither of such configurations is omnipresent or generally obligatory. It might have been, or may be, eliminated by further development. We are in a similar position as the specialists in aesthetics of the Prague school were when they were looking for a definition of literary art. In this situation, Mukarˇovský and the early Jakobson (but not the Jakobson of his study Linguistics and poetics) turned from positive (qualitative) specifications towards a functional definition, which places into the center the intentions and the attitudes of subjects involved in the artistic communication — and we do the same from the very beginning. This intention is the starting point, and in individual cases it is possible to assume that this call will evoke an offer of a certain tangible support by which the texts respond. Our strategy concerning the rhythm of free verse may be analogous, but we should not forget that with verse we have arrived at a lower degree in the hierarchy than when the ‘‘poeticalness’’ as such was at play. Also in our case the intention precedes a specific textual answer, but it is determined a priori that we should reach a sound rhythm and its linguistic incarnation. Similarly as Mukarˇovský (1933/1982) more than six decades ago, we also believe today that the only component of the sound layer of language that always can constitute a reply to this call is intonation. What is meant here are changes of the melody of the voice, suprasegmental changes of its strength and, as the case may be, pauses and instant changes of the rate that are connected first of all with the utterance articulation. These sound phenomena, or at least their more abstract schemes, may be assumed to be present also in the case of silent reading. The intonation is a polyfunctional and thus a manysided component of language. Along with a number of functions it carries out in everyday speech, it can receive the rhythmic function, which is realized by specifying speech segments in which an abstract intonation figure is repeated. The forms of intonation in different texts are manifold: intonation of course carries out its other functions also in the case of this new load consisting of the rhythmic function; the rhythmic layer of intonation means gets the shape of special modifications of the intonation schemes connected with the common functions. We are not going to examine here the nature of these modifications. It is sufficient to be aware, as the usual recipients of poetry are, that verses are (actually and virtually) pronounced with ‘‘another tone of voice’’ than the same text would be pronounced if presented without rhythmic intention. Viewed this way, verse intonation is and is not a linguistic phenomenon: it is a projection of the rhythmical intention into speech, a phenomenon brought into

 Miroslav Cˇervenka

speech from outside. It is a characteristics of a certain type of speech, but this does not mean that it would presuppose any specific means which would always be present as its bearers. In concrete cases it, of course, finds its supports in individual language means which are potential markers of the rhythmic arrangement. Outside the sphere of poetry they remain afunctional and thus unconscious, i.e. actually non-existent. An example of such means can be seen in statistical shifts in the patterns of stressed and unstressed syllables, or in phenomena more closely connected to intonation, such as non-motivated inversions of words. A regressively foregrounded supporting means for the confirmation of a given rhythmic intention consists in intertextual relations to prototexts that indisputably have been incorporated into poetry, displaying characteristic sound patterns. When, at the beginning of the 1930s, Mukarˇovský was writing his study Intonace jako cˇinitel básnického rytmu (1933/1982), he had in mind the same idea that we are attempting, following him, to express here. The then held views of pragmatic factors of speech patterns easily could bring about that claims on a linguistic phenomenon present independently of any positively describable language pattern did not appear to be convincing. However, Mukarˇovský’s orientation towards the attitudes and intentions of the participants of communication can clearly be traced in the fact that he proposed his theory of verse intonation as a parallel to Karcevskij’s (1931) theory of the intonation of the sentence (utterance). Also for the great Geneva linguist, intonation is a primary sign of utterance, of utterance intention, which precedes any of its grammatical shapes. The intention to constitute a sentence, signalized by intonation, can be confirmed by the actually installed grammatical relations between the constituents (and at least in writing this in most cases really comes about), but this intention is not refuted by their inadequacy or even absence. In any case, the standing analyses of actually instantiated (spoken) discourse prove that we normally accept language creations that from a grammatical viewpoint are incomplete or imperfect. The presence of the intention to create and to perceive an utterance, mediated by intonation, thus often is decisive, even though the grammatical equipment of the created formation is incomplete in any aspect. It may be assumed that this also holds in what concerns both the intention towards verse, signalized by verse intonation, and a further fulfillment of this intention in an actual sound pattern of a line (text).

. The interplay free-metric From the presented views on the substance of free verse it follows that it would not be reasonable to consider this formation as constituting a further specific prosodic system on the level of syllabic, syllabotonic, tonic and other systems, a

The principle of free verse 

system which would be based on intonation or even on graphics, similarly as those systems are based on the number of syllables, on the distribution of stresses, etc. A specific kind of intonation, characteristic of every versified speech, is a component of rhythm also with verses based on classical prosodic factors, connected with word phonology; already Mukarˇovský concluded that after these factors of rhythmization are set aside or radically reduced, intonation remains as a last support for free verse concerning its rhythm, deprived of its standard bearers. If in the classical verse intonation was a redundant accompaniment of the rhythmic organization based on other factors, in free verse it alone is foregrounded. The dependence of free verse on tradition and on a specific, historically determined literary consciousness, moreover, is reflected in the fact that free verses always somehow appear on the background of those prosodic systems from which they emerged, loosened by their freedom; they also retain fragments and indications of those systems, at least in the first stages of their existence. The French free verse was formed as ‘‘syllabic’’ (using fragments or whole series of classical syllabic metres), and the Czech free verse, which in its beginnings certainly was inspired by French models, quite independently on them took a way of ‘‘syllabotonism’’, which intensively patterns the distribution of stresses and of their configurations. And this connection has been repeatedly constituted: along with the continuous gradation of freedom (i.e. with the separation from metric verse), in the development of modern poetry, border formations always reappear through which free verse ‘‘returns’’ to some of the features of its classical background. And, on the other hand, metric verse gets unsteady and relativized, accepting some freedoms typical of its unruly descendant. However, also there where the two competing forms fully retain their integrity, the development of modern verse acquires a shape different from that epoch in which the whole of its contents was filled by changes of a single form, the metric one. The choice of metric verse itself, which once was the only possibility and thus was semantically neutral, now gains a more and more distinct stylistic colouring. Free verse retains its interesting qualities and its artistic intensity only as long as it represents a choice of freedom, reflected on the background of other possibilities. Wiping out metric verse would mean a loss for the possibilities of the artistic function of poetic rhythm in general.

. Seeking the confirmation Relationships to metric verse have to be also projected into the acts of creation and perception of free verse. The preceding considerations can be undertaken from the opposite end: Yes, we decide — and we also can refuse to do so — to ascribe the verse intonation to every speech that allows us, even on the basis of its

 Miroslav Cˇervenka

external markers, to distinguish the intention to constitute a verse. From our experience with verses, we are, at the same time, used to expect that the corresponding text in some way meets halfway this act of sound patterning: that we find in its inherent features some kind of confirmation or the justification of our good will to accept it as rhythmically organized. From this viewpoint, even in free verse the linguistic shape of the text is not irrelevant for its rhythmic character. The claim of the basic and decisive role of the rhythmic intention does not exclude specific linguistic issues of rhythm in free verse. However, at the same time, a new installment of the question of inherent rhythmic features, if it occurs in the theoretical frame outlined above, does not mean a mere return to the traditional approach, describing only the inherent, palpable markers. It can be understood that now, after the intentional frame of creating and perceiving this verse formation has been put into the light of clear consciousness, the activity of a researcher in the domain of theory and history of verse cannot consist in anything else than, again, coming back to the analysis of the qualitative of rhythm, which enable the text to meet its perceiver as a result of intentional linguistic shaping. It is certain that he can carry out his analyses, without which he could achieve neither a typology of free verse, nor an identification of its individual realizations, being aware that it is not necessary for him to expect their eventual goals to contain an unambiguous specification of completed forms that would not depend on most different contexts.

References Cˇervenka, M. 1992/1996. ‘‘Literární artefakt’’ [Artifact in Literature] In Obléhání zevnitrˇ, M. Cˇervenka, 40–78, Praha: Torst. Karcevskij, S. 1931. ‘‘Sur la phonologie de la phrase’’. In Travaux du Cercle linguistique de Prague 4: 188–223. Mukarˇovský, J. 1933/1982. ‘‘Intonace jako cˇinitel básnického rytmu’’ [Intonation as a Factor of the Poetic Rhythm]. In Studie z poetiky, J. Mukarˇovský. 191–204, Praha: Odeon. Šalda, F. X. 1898/1951. Kritické projevy 4 [Critical texts 4], Praha: Melantrich. Tynjanov, J. 1924/1987. ‘‘Problém básnického jazyka’’ [The Problem of Verse Language]. In Literární fakt, J. Tynjanov, 423–534, Praha: Odeon.