The Grammaticalization of ‘Give’ ⫹ Infinitive
Trends in Linguistics Studies and Monographs 256
Editor
Volker Gast Fo...
459 downloads
980 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
The Grammaticalization of ‘Give’ ⫹ Infinitive
Trends in Linguistics Studies and Monographs 256
Editor
Volker Gast Founding Editor
Werner Winter Editorial Board
Walter Bisang Hans Henrich Hock Heiko Narrog Matthias Schlesewsky Niina Ning Zhang Editor responsible for this volume
Volker Gast
De Gruyter Mouton
The Grammaticalization of ‘Give’ ⫹ Infinitive A Comparative Study of Russian, Polish, and Czech
by
Ruprecht von Waldenfels
De Gruyter Mouton
Doctoral thesis – University of Regensburg 2009 (D355)
ISBN 978-3-11-029369-2 e-ISBN 978-3-11-029377-7 ISSN 1861-4302 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. ” 2012 Walter de Gruyter GmbH & Co. KG, Berlin/Boston Printing: Hubert & Co. GmbH & Co. KG, Göttingen 앝 Printed on acid-free paper 앪 Printed in Germany www.degruyter.com
Acknowledgements
This study is a heavily revised version of a dissertation defended at Regensburg University in 2009. There are probably more people to thank than I can list here, and I apologize to those who go unmentioned. In no particular order, I would like to thank the people in ulica celetná, who have been colleagues, friends and informants, especially Saša Rosen; my not quite related aunt Libuška Badewitz, who spent two days interpreting corpus examples; Dagmar Divjak, for her support by Skype; Roland Meyer, for his critical and always sympathetic support from the other side of the aisle, and for being a friend; Adam Przepiórkowski (who might not remember); Andrej Jašlavskij, Andrej Kožanov, Natalia Kožanova, Irina Kiseleva, Denis and Anja Usvjat and others for answering make-shift questionnaires and ever new variations of what seemed to be one and the same question; Sergej Saj, Vladimir Plungjan, Marina Rusakova, who I will not forget, and Aleksandr Rusakov for shelter, support and advice; Elena Uryson, for her friendly interest and calm discussion of the semantic conception; Gary Toops, who almost became my advisor; Lenka Nerlich, who impressed me with her thorough judgements; incredible Matti and Małgorzata Jokipii, Łukasz Sommer, who is really a linguist, and his wife Marta, Artur Kolasi´nski, who first taught me Polish pronunciation, and many more which I have pestered with my questions at odd and even times; the people at Sail Labs, who first got me started with lassen, especially Hansi Völkel; Björn Hansen, who gave me this topic, which I only appreciated much later (and who was probably surprised at what I made of it); Daniel Weiss and Björn Wiemer, and an anonymous referee for reviews and valuable comments; and, of course, Petr Karlík, who helped in so many ways. Thanks to Yannis Kakridis, for support and proverbs; Bernhard Wälchli, who encouraged me to submit the book to the Trends in Linguistics series; the editors of the series for accepting and the team from De Gruyter for their friendly help. And, of course, I have to thank Charlotte, Anouk and Nike, all of whom contributed in their own way; and especially I have to thank my wife Ariane - no way I could have written this book without her.
Conventions
Throughout this book, da(va)t in small capital letters is used as a cover term to refer to ‘give’ in the Slavic languages, i.e. Russian pf. dat’, ipf. davat’, Polish pf. da´c, ipf. dawa´c, Czech pf. dát, ipf. dávat, and so forth.; da(va)t+inf relates to the constructions these verbs form with an infinitive. The abbreviations of grammatical categories loosely follow the Leipzig Glossing Rules: ACC ADV AOR AUX COMP COND COP CVB DAT DU FUT GEN IMP IMPF INF INS IPF IPS IRR
accusative adverb aorist auxiliary complementizer conditional copula verbal noun dative dual future genitive imperative imperfect (tense) infinitive instrumental (case) imperfective (aspect) impersonal (verb form) irrealis
LOC M N NEG NOM PF PL PPP PREP PRF PRS PRT PST PTCP REL RFX SG VBN
locative (case) masculine neuter negation nominative perfective (aspect) plural past passive participle preposition perfect (tense) present particle past participle relative pronoun reflexive singular verbal noun
In order to keep the text accessible and reduce redundancy, per-wordform glossing is kept to a minimum and primarily employed where the syntactic structure is important. In most places, a more focused and economical type of glossing is employed: the main structure (usually matrix verb, main actant of the complement clause and complement infinitive) is made explicit by providing grammatical glosses in the original; lexical information is represented by highlighting the appropriate word forms in the translation, as in the following Russian example:
viii Conventions (1)
i
On dalPS T :S G PetruDAT zakonˇcit’INF . ‘He let Peter finish.’
Glosses generally designate tense, person and case, sometimes also number (which can generally be inferred from the translation). Gender and verbal aspect are only given where relevant for the discussion. If the order of the word forms in the original and the translation differs, subscripts are given to link the highlighted word forms in the original and the original: (2)
ii
PetruDAT nikto ne dalPS T :S G zakonˇcit’INF . ‘Nobody let2 Peter1 finish3 .’
If relevant to the discussion, translated material that does not have a counterpart in the original, e.g., because of ellipsis, is given in parentheses: (3)
iii
Petr ne dalPS T :S G projtiINF . ‘Peter didn’t let (anybody / us) pass.’
Ungrammaticality is signalled by an asterix, question marks are used to signal degrees of acceptability. # is used if only a different from the intended reading is possible. Translation does not primarily aim at English idiomaticity, but tries to faithfully render both semantic and syntactic structures of the source text without loss of intelligibility. Most examples are attestations from large reference corpora. Their sources are given in footnotes with roman numbers that start anew on every page. Arabic numbers, in contrast, refer to endnotes given at the end of the book.
i References are given in footnotes with Roman numbering third reference
ii Another
reference
iii A
Contents Acknowledgements
v
Conventions
vii
1 1.1 1.1.1 1.1.2 1.2 1.3 1.3.1 1.3.2 1.3.3 1.3.4 1.3.5 1.4 1.4.1 1.4.2 1.4.2.1 1.4.2.2 1.4.3
Introduction, overview and theoretical framework Introduction . . . . . . . . . . . . . . . . . . . . . . Give with an infinitive in Russian, Polish and Czech . Overview and structure of the study . . . . . . . . . . Grammaticalization studies . . . . . . . . . . . . . . Causatives . . . . . . . . . . . . . . . . . . . . . . . Definitions and basic notions . . . . . . . . . . . . . Permissive (‘letting’) causation . . . . . . . . . . . . Curative factitive (‘having’) causation . . . . . . . . Causative domains: from interpersonal to cognitive . . Causee coding: Type I and Type II causatives . . . . Modals and causatives . . . . . . . . . . . . . . . . Modal domains in causation . . . . . . . . . . . . . . Negation with modals and causatives . . . . . . . . . Types of negation and informativity . . . . . . . . . . Modals and causatives on the square of oppositions . Conclusions . . . . . . . . . . . . . . . . . . . . . .
1 1 1 3 5 14 14 15 18 19 21 27 29 32 32 33 39
2 2.1 2.1.1 2.1.2 2.1.3 2.2 2.2.1 2.2.2 2.2.2.1 2.2.2.2 2.2.2.3 2.2.3 2.2.3.1
da(va)t+inf in Russian Introduction . . . . . . . . . . . . . . . . . . . . . . . . . Syntactic constructions . . . . . . . . . . . . . . . . . . . Material & Method . . . . . . . . . . . . . . . . . . . . . Prior research . . . . . . . . . . . . . . . . . . . . . . . . Semasiological description of permissive da(va)t in Russian Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . Interpersonal causation . . . . . . . . . . . . . . . . . . . Profiles of usage . . . . . . . . . . . . . . . . . . . . . . . Semantic description . . . . . . . . . . . . . . . . . . . . . Polarity related meanings: negative causation . . . . . . . . Manipulative causation . . . . . . . . . . . . . . . . . . . Profiles of usage . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . .
41 41 41 43 46 52 52 54 54 58 65 71 71
x Contents 2.2.3.2 2.2.4 2.2.5 2.2.6 2.2.6.1 2.2.6.2 2.2.6.3 2.2.7 2.2.7.1 2.2.7.2 2.2.7.3 2.2.8
Semantic analyis . . . . . . . . . . . . . . . . . . . . Impersonal causation . . . . . . . . . . . . . . . . . Conclusions: permissive da(va)t in three domains . . Specific issues in the analysis of the permissive . . . . Why is the perfective imperative daj! so frequent? . . Reflexive permissives . . . . . . . . . . . . . . . . . Evidence for auxiliarization as event integration . . . Further constructions types . . . . . . . . . . . . . . ‘Letting you know’: cognitive and perception verbs . Impersonal passive da(va)t plus infinitive . . . . . . . Secondary predication and referential use of infinitives Summary . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
3 3.1 3.1.1 3.1.2 3.1.3 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.2.4.1 3.2.4.2 3.2.4.3 3.2.4.4 3.2.5 3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.4 3.5 3.6 3.6.1 3.6.2
da(va)t+inf in Polish Introduction . . . . . . . . . . . . . . . . . . . . . . . . Syntactic types . . . . . . . . . . . . . . . . . . . . . . . Prior research . . . . . . . . . . . . . . . . . . . . . . . Material & Method . . . . . . . . . . . . . . . . . . . . Permissive da(va)t in Polish . . . . . . . . . . . . . . . . Overview: reflexive and non-reflexive . . . . . . . . . . Usage profiles . . . . . . . . . . . . . . . . . . . . . . . Non-reflexive permissive . . . . . . . . . . . . . . . . . Reflexive permissive . . . . . . . . . . . . . . . . . . . The causee phrase . . . . . . . . . . . . . . . . . . . . . Prepositional (type II) causees . . . . . . . . . . . . . . The causee with dative reflexives . . . . . . . . . . . . . Conclusions: form and function of the causee phrase . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . Factitive da(va)t . . . . . . . . . . . . . . . . . . . . . . Curative causation . . . . . . . . . . . . . . . . . . . . . Other types of factitive causation . . . . . . . . . . . . . Ambiguous permissive/factitive causation . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . Passive da(va)t plus infinitive . . . . . . . . . . . . . . . Cognitive and perception verbs . . . . . . . . . . . . . . The modal passive . . . . . . . . . . . . . . . . . . . . . Types of the modal passive . . . . . . . . . . . . . . . . Modal passive constructions and the reflexive permissive.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
74 78 84 85 85 93 100 103 103 109 110 112
. . . . . . . . . . . . . . . . . . . . . . . .
113 113 113 114 114 119 119 120 125 129 134 137 140 143 144 144 144 146 147 149 149 150 153 153 162
Contents
xi
3.6.3 3.6.4 3.6.4.1 3.6.4.2 3.6.4.3 3.6.5 3.7
Usage profiles . . . . . . . . . . . . . . . . . . . . The aspect of da´c si˛e in subjectless MPs . . . . . . Morphological criteria. . . . . . . . . . . . . . . . Contextual criteria . . . . . . . . . . . . . . . . . . Conclusions concerning the aspect of da´c si˛e . . . . Conclusions: different degrees of grammaticalization Summary . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
167 171 174 178 182 184 185
4 4.1 4.1.1 4.1.2 4.1.3 4.1.4 4.2 4.2.1 4.2.2 4.2.3 4.2.3.1 4.2.3.2 4.2.3.3 4.2.3.4 4.2.4 4.2.4.1 4.2.4.2 4.2.4.3 4.2.5 4.2.6 4.3 4.3.1 4.3.2 4.3.3 4.3.4 4.4 4.5
da(va)t+inf in Czech Introduction . . . . . . . . . . . . . . . . . . . . . Overview and prior research . . . . . . . . . . . . . Competition in causation: nech(áv)at with infinitive Syntactic types . . . . . . . . . . . . . . . . . . . . Data & annotation . . . . . . . . . . . . . . . . . . Causative . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . Non-reflexive permissive . . . . . . . . . . . . . . Reflexive permissive . . . . . . . . . . . . . . . . . Accusative reflexive . . . . . . . . . . . . . . . . . Dative and other reflexives . . . . . . . . . . . . . . Comparison to nech(áv)at . . . . . . . . . . . . . . Conclusions: permissive causation . . . . . . . . . Factitive . . . . . . . . . . . . . . . . . . . . . . . Curative and other interpersonal factitive . . . . . . Manipulative and impersonal . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . Cognitive causation . . . . . . . . . . . . . . . . . Conclusions: causative dá(va)t . . . . . . . . . . . Modal passive . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . Agreeing and non-agreeing construction . . . . . . Further characteristics . . . . . . . . . . . . . . . . Conclusions: modal passive . . . . . . . . . . . . . Residual types . . . . . . . . . . . . . . . . . . . . Summary and conclusions . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
187 187 187 188 189 190 193 193 193 196 196 201 204 207 207 208 216 217 218 221 222 222 225 228 230 231 232
xii Contents 5 5.1 5.1.1 5.1.2 5.1.3 5.1.4
Czech, Polish and Russian in parallel Introduction . . . . . . . . . . . . . . The permissive . . . . . . . . . . . . . The factitive . . . . . . . . . . . . . . The modal passive . . . . . . . . . . . Conclusions . . . . . . . . . . . . . .
. . . . .
233 233 235 238 239 241
6 6.1 6.2 6.3 6.4 6.4.1 6.4.2 6.4.3 6.5 6.5.1 6.5.2 6.5.3 6.6
da(va)t+inf in OCS and earlier stages of Polish and Czech Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . da(va)t+inf in Old Curch Slavonic . . . . . . . . . . . . . . A side glance at da(va)t+inf in Russian . . . . . . . . . . . . da(va)t+inf in Czech . . . . . . . . . . . . . . . . . . . . . . Permissive and factitive . . . . . . . . . . . . . . . . . . . . The development of type II causees . . . . . . . . . . . . . . The modal passive . . . . . . . . . . . . . . . . . . . . . . . da(va)t+inf in Polish . . . . . . . . . . . . . . . . . . . . . Permissive and reflexive permissive . . . . . . . . . . . . . . Factitive causation . . . . . . . . . . . . . . . . . . . . . . . The modal passive . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . .
243 243 243 248 249 251 257 261 265 270 271 272 277
7 7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 7.2.6 7.2.7 7.3
Conclusions and directions for further research Overview . . . . . . . . . . . . . . . . . . . . . . . Developments of da(va)t+inf . . . . . . . . . . . . General remarks . . . . . . . . . . . . . . . . . . . From give to permissive . . . . . . . . . . . . . . . From permissive to reflexive permissive . . . . . . . Factitive: from give or from the permissive? . . . . By-phrases: causatives as diathesis constructions . . From reflexive permissive to agreeing modal passive From agreeing to subjectless modal passives . . . . Directions for further research . . . . . . . . . . . .
279 279 280 280 281 284 286 289 296 301 303
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . . . . . . .
. . . . .
. . . . . . . . . .
. . . . .
. . . . . . . . . .
. . . . .
. . . . . . . . . .
. . . . . . . . . .
Notes
305
Bibliography
313
Index
330
Chapter 1 Introduction, overview and theoretical framework
1.1.
Introduction
1.1.1. Give with an infinitive in Russian, Polish and Czech Give is one of the basic lexemes of human language and the prototypical ditransitive verb. Cross-linguistically, the functional extension of lexemes denoting a canonical transfer (to give in English) towards other, more grammatical functions, i.e., its grammaticalization, is a fairly widespread phenomenon. The development of ‘giving’ into items denoting permission, causation, dative, benefactive, among others, has been described for genetically, geographically and typologically distant languages (Lord et al. 2002; Heine & Kuteva 2002). The present study deals with the grammaticalization of give in conjunction with an infinitive in a subset of the Slavic languages, namely in Polish, Czech and Russian. Here, this development involves the following broad construction types in the causative, modal and imperative domain: • causative: permissive (i.e. ‘letting’ causation), e.g. in Russian: (4)
On ne dal emu dogovorit’. He NEG give.PST:SG him.DAT finish.speaking.INF ‘He did not let him finish speaking.’
reflexive permissive (‘letting oneself’), e.g. in Polish: (5)
Nie dał si˛e przekona´c. NEG give.PST:SG RFX:ACC persuade.INF ‘He did not let himself be persuaded.’
factitive (‘making’ causation) e.g. in Czech: (6)
Dal zahradu ohradit. Give.PST.3SG garden.ACC enclose.by.fence.INF ‘He had the garden fenced in.’
2 Introduction, overview and theoretical framework • modal: modal passive, e.g. in Polish: (7)
W ko´ncu da si˛e pracowa´c. in end give.3SG RFX.ACC work.INF ‘Finally it’s possible to work (lit. it lets itself work)’
• imperative: Inclusive 1st person plural imper., ‘hortative’, only Russian: (8)
Davajte rabotat’! give.IMP.2PL work.INF ‘Let’s work!’
All three of the above domains involve modal meaning: modals are dedicated constructions for the expression of modality; causatives can be seen to be the transitive counterparts to modals, as will be discussed in more detail below; imperatives arguably involve modality in the interhuman sphere of communication (go! means either you may go or you must go). These three domains divide into two groups. On the one hand, we have the causative formed by give plus infinitive, present in all Slavic languages and evidently inherited from an earlier common stage of Slavic. In this group, several subdivisions are in order, most importantly, that between permissive (letting) and factitive (having, making) causation. This construction has furthermore given rise to modal constructions in all Western Slavic and in the Western South Slavic languages, quite possibly under the influence of a similar development in German. On the other hand, Russian is exceptional among the languages considered here concerning the grammaticalization of a hortative construction involving the imperative of imperfective give, as in (8). The development of the particle davaj(te) does not seem to be linked to the causative construction; it stands apart and its evolution evidently involved quite different mechanisms of development than the causative and modal constructions. There are similar constructions in South Slavic languages, such as Serbian dajte da otvorimo radnju ‘let’s open a shop’ or Lower Sorbian daj nas hy´s ‘let’s go’; these languages, however, remain outside the scope of this study. The first two domains, causatives and modals, thus involve a complex group of related constructions relevant in all three languages and attested in texts since a very early stage. The evolution of the hortative in Russian, on
Introduction
3
the other hand, involves a construction not shared with the other languages in the sample and belonging to the domain of spoken language, much less well accessible in texts that have come down to us. This study primarily focuses on the first group; the development of the imperative will be discussed only in general and primarily to establish the differences from the first group. Many of the relevant constructions have, as a whole, not been investigated in detail before. It is therefore an aim of this study to provide a comprehensive description of their meaning and use from a diachronic, synchronic and comparative point of view. This is done in the context of grammaticalization studies, a line of research that takes both both synchronic relations and diachronic processes into focus. From the more general perspective, this study therefore adds to the research on language-specific attestations of widely attested grammaticalization paths. I use the term construction in this study mostly informally to refer to specific morphosyntactic configurations in a general sense. On a theoretical level, however, I subscribe to the view that constructions in the sense of regular configurations of morphemes or word forms may express specific functions beyond their compositional make-up on all levels of linguistic structure, as posited by by various branches of construction grammar (see e.g. Fillmore et al. 1988; Goldberg 1995; Croft 2001). Grammaticalization in my understanding involves an increase in the regularity of a constructional pattern and the development of specific senses associated with it in a gradual development. Since this study aims at modelling both established and incipient constructions that have various degrees of compositional transparency, a corpus-based, multilingual perspective is taken to arrive at quantitative and comparative evidence for my analysis.
1.1.2.
Overview and structure of the study
This study is organized in seven chapters. The present introduction is followed by a synchronic part that consists of three separate synchronic chapters on Russian, Polish and Czech and a comparative chapter on all three languages based on parallel corpus data. A diachronic chapter follows with studies on Polish and Czech and additionally including Old Church Slavonic. The study closes with a chapter containing conclusions and directions for further research.
4 Introduction, overview and theoretical framework In each empirical chapter, the language in question is analyzed separately. The approach is corpus-driven: In each case, the analysis takes a large sample of da(va)t+inf from a monolingual corpus as point of departure; this sample is evaluated in full. The instances in each sample are categorized according to a common scheme with only minor adaptation for language specific characteristics, since the analysis aims to cover all uses of da(va)t+inf in each language. These categorizations reflect the tertium comparationis of the comparative analysis. At the highest level, all examples are analyzed according to their semantic/syntactic type, distinguishing causative (all languages), modal passive (Polish and Czech) and hortative (only Russian) uses, and further minor types. Causative and modal construction are further subcategorized. In the causative domain the subcategorization is the most complex, distinguishing permissive and factitive causation, different configurations of control of the participants and singling out a class of causative constructions formed with verbs of perception and cognition. These are top-down distinctions largely arrived at by theoretical considerations and founded on the existing research on causative constructions in general; they are relevant in all languages. Complementing this deductive categorization, these types are analyzed according to the forms used, distinguishing tense, negation, aspect of da(va)t and the infinitive, the set of infinitives used in the sample and reflexive constructions. In this way, differences and common traits of the function of da(va)t+inf across the three languages are rendered comparable. The subcategorization of modal passive constructions, in contrast, is centered on their syntax and based on the argument structure of the infinitive and the expression of arguments in the modal passive construction. The empirical part starts with an analysis of da(va)t+inf in Russian in chapter two. In this chapter I focus on the various shades of permissive meaning, the core use of causative davat’. Only here do I use explications to clarify different semantic subtypes that arise in context; these are referred to in later chapters. Two morphosyntactic configurations of permissive da(va)t+inf stand out in comparison: the imperative daj and use of the permissive under negation in general. They are looked at in more detail, relating them to other functions of daj in Russian and to polarity-based restrictions in usage of the permissive construction. The following chapter three deals with Polish; here, reflexive permissive constructions as in da´c si˛e przekona´c ‘let oneself be persuaded’ move into
Grammaticalization studies
5
the center of attention and are shown to be more grammaticalized than their non-reflexive counterparts. Additionally to the causative, the modal passive use of da(va)t as in da si˛e zrobi´c ‘it can be done’, absent in Russian, comes into play. Two competing constructions, one involving a subject and another, subjectless variant are examined; the analysis focuses on their differences in terms of their grammaticalization and concludes with showing that at least the latter variant can be said to have lost aspectual distinction. Section four concludes the language-specific sections with Czech. Here, a further causative type, namely factitive (have, make) causation is important, while permissive causation, prominent in Polish and Russian, is more restricted. The most important function of da(va)t+inf in Czech is the modal passive, which is shown to be more grammaticalized than its Polish counterpart. Each language is thus analyzed separately, but from a common perspective, forming the core of the empirical basis of the present study. This perspective is supplemented by a comparative study in chapter five based on data from the parallel corpus ParaSol. Here, results of the analysis of each language in isolation are validated; use of parallel data also allows for a controlled widening of scope to examine competing formants of da(va)t+inf in its respective domains. These synchronic studies are supplemented from a diachronic perspective in chapter six. The use of da(va)t+inf in Old Church Slavonic is investigated in order to account for the earliest stage in Slavic; diachronic corpora of Polish and Czech are used to trace the developments up to the present stage. Here, specific questions of the change of causative and modal constructions are addressed. Chapter seven is devoted to the further discussion of some selected issues concerning the grammaticalization of da(va)t+inf in Russian, Polish and Czech. It provides a summary and the conclusions.
1.2.
Grammaticalization studies
The past 30 years have seen intense interest in grammaticalization as a special mode of language change whereby lexical items become grammatical, or less grammatical items become more grammatical items. The term has received a plethora of definitions and has been criticized extensively from several points of view (see Campbell 2000). Definitions as well as criticism crucially rely
6 Introduction, overview and theoretical framework on different concepts of grammaticality (in the sense of ‘belonging to the grammar of a language’) as well as theoretic inclinations. Most basically grammaticalization as a process may be defined as the move of more lexical to more grammatical items or constructions. Such a definition begs many questions: what exactly is meant by lexical and grammatical, by item, by construction? Depending on what concepts are understood by theses terms, different conceptions of grammaticalization are used; conversely, work in grammaticalization has also had the aim to clarify these terms. In its most theory-free usage the term grammaticalization encompasses language change as defined by a certain type of source and target construction or item in language change, and as such it might be used in any work on language change that has some conception of what is grammatical and what is not. For example, it is uncontroversial to attest that word order in English has grammaticalized in this sense, or that going as a grammatical future marker in English derived from the lexical verb go must have grammaticalized at some point. However, grammaticalization in a narrow sense as a mode of language change has found much interest from researchers within the functional paradigm working on both synchronic as well as in diachronic aspects of language. In this research tradition it is contended that grammaticalization is a type of change that has its own characteristics, distinct from other modes of linguistic change. Most importantly, it is claimed to be inherently directed and gradual, setting it apart from other processes of change such as analogy, reanalysis, lexical expansion or contraction and semantic change. On the other hand, this has also been contested, claiming that grammaticalization does not exist in the sense of a distinct mode of change, but rather constitutes an epiphenomenon of other, more basic types of language change such as those just mentioned (see, e.g., Joseph 2000). The characteristics of this particular kind of language change is seen to carry implications for a synchronic theory of language for many researchers working within a functional, and especially usage-based theory of grammar. Bybee et al. (1994: 23f.) make this point very clearly: "the study of grammaticalization [. . . ] provides a new approach to the understanding of grammar: by studying the pathways and mechanisms of the creation of grammatical morphemes, we hope to get closer to an understanding of why language has grammar at all and why grammar takes the particular form and meaning that it does."
Grammaticalization studies
7
Figure 1. Parameters and processes in grammaticalization in Lehmann (20022 : 146)
In his influential Thoughts on grammaticalization, first published in 1982, Christian Lehmann provides an operationalization of the term with six parameters that apply synchronically as well as diachronically. While they describe a specific mode of languages change, these parameters are also seen to be indicative of the grammaticality (in the sense of belonging to the grammar of a language) of linguistic items. These parameters are applications of the superordinate notion of the loss of the integrity of the sign in relation to both syntagmatic and paradigmatic levels and relate to both its content and its expression; in a later paper, Lehmann (2002) extends the scope of these parameters to constructions in general. Figure 1 provides an overview over these parameters and the associated subprocesses of grammaticalization. These highly abstract parameters are relevant to both content and form of the linguistic sign in question. Much work has been done to apply, refine as well as critically assess these parameters. Other parameter sets with slightly different emphasis have been developed by other researchers (see, e.g., Heine et al. (1991); Hopper & Traugott (1993); Bybee et al. (1994)), while Lehmann’s parameters have been called into question specifically in the area of Slavic linguistics in the realm of modal and passive analytic constructions, an area immediately relevant to the present study (Hansen 2004;
8 Introduction, overview and theoretical framework Wiemer 2004); still, they remain the most influential and detailed coherent theoretical exposition of grammaticalization. It is a crucial claim of grammaticalization studies that these individual processes take place in parallel, constituting a macro process of grammaticalization with inherent directionality. However, as Hansen (2004) and Wiemer (2004) have shown, in the realm of passive and modal auxiliaries in Slavic not all of these parameters apply to the same extent; especially the processes of fixation in terms of word order, phonological attrition and coalescence do not play an important role. Obligatorification, lastly, is hard to apply to generally optional categories such as modals and passives. What is in the foreground of the above mentioned processes in Slavic are semantic bleaching or attrition, the formation of tightly integrated paradigms as well as formal attrition in the sense of the loss of verbal categories of passive and modal auxiliaries. Similar results are arrived at in the present study. Structural scope of a grammaticalizing formant, finally, is perhaps the most difficult parameter. Lehmann defines it as "the structural size of the construction which it helps to form (Lehmann 20022 : 128)" and makes it clear that structural, not semantic criteria should be taken into account: even if, for example, a tense suffix may semantically modify the whole situation, structurally, it modifies only the predicate it is affixed to: it has small structural scope. The question arising in the context of the present study is how structure in more complex syntactic environments is to be modelled, since the syntax of, e.g., passive, modal or causative constructions is itself far from agreed upon. As the analogy to tense markers suggests, what is referred to in general as reduction of structural scope may be framed as the more specific process of auxiliarization; again, the nature of auxiliaries and the theoretical status of the term is not agreed upon either (see the overview in Heine 1993). One way to relate auxiliarization to the parameter of reduced scope is to view the process as a reduction in the number of arguments that are governed or selected by the grammaticalized formant: as, e.g., modal elements grammaticalize, they cease to select their own arguments and selection becomes solely dependent on the lexical verb it modifies. Let us compare these parameters to a smaller set of parameters arrived at in the tradition of Bernd Heine and his colleagues (e.g Heine 1993, 2003) who adduce only four parameters: desemanticization, decategorialization, cliticization and erosion. Without going into details, we can see that three of these parameters are found on a similar level in Lehmann’s set: desemanticization and erosion
Grammaticalization studies
9
relate to the parameter attrition (of form and content), cliticization to the parameters of bondedness and syntagmatic variability. The fourth parameter, decategorialization, is perhaps best related to what Lehmann calls "morphological degeneration", a process Lehmann generally sees as an instance of attrition on the one hand, but also as a general symptom of the change in status occurring as a "grammaticalized sign moves down the grammatical levels, from phrase via word form to morpheme.(Lehmann 20022 : 118)" Generally, Lehmann’s model is more abstract and complex than Heine’s, as it aims to encompass a maximally wide range of phenomena that can be thought of as grammaticalization. Its main interest is in accounting for form and content in language from a structural point of view, rather than from a cognitive perspective more prominent in Heine’s work. In the main part of this study, only a subset of phenomena will be focused on. As we will see, processes of phonological change or fixation of word order are generally of minor relevance, if they may be diagnosed at all. Rather, what is in the focus of attention are syntactic and semantic characterization as well as the associated functional load of the constructions in question, diagnosed in terms of frequency. As such, desemanticization or semantic bleaching are most prominent; some of the structural changes, such as the rise of by-phrases with causatives in Polish and Czech, are less easily related to the parameters adduced above. This will be discussed in the concluding section. In this study, I assume grammaticalization to be a specific mode of language change relying on parallel changes in form, function and frequency of an item or construction, broadly relying on the work of Lehmann, Bybee, Heine, Hopper and Traugott as mentioned above. Grammaticalization as a development and a resulting state is characterized, as far as our object of investigation is concerned, by the development to and of • highly abstract, schematic and synsemantic meaning (‘semantic bleaching’; ‘semantic bondedness’) • small class membership (both paradigmatically and syntagmatically; opposed to the morphology and syntax of large, open word classes; ‘decategorization’). • clause integration, that is, structurally, the downgrading of an arguably biclausal predicate to a unequivocal uni-clausal complex predicate; on the semantic side, this corresponds to event integration, that is, a situation that potentially involves two events is conceptualized as one, e.g. with a single truth value • high usage frequency with a variety of types
10 Introduction, overview and theoretical framework This represents a selective generalization in the spirit of Lehmann’s and Heine’s as well as Bybee’s and other researchers’ more frequency-based approaches to grammaticalization. The first two characteristics are standard generalizations in grammaticalization studies. The third is more complex, and several characteristics relevant in this context will be examined in the course of this study. Clause integration is not per se a standard process of grammaticalization; rather, it is a complex process involving several parameters, such as semantic bleaching, loss of inflectional categories and scope reduction, leading eventually to a complex predicate where the grammaticalized formant becomes an auxiliary that does not open its own argument slots, but serves to modify the predicate in some way. The last characteristic, high frequency, is less standard. The role of frequency is emphasized by different scholars in the field, and it is generally acknowledged that a hallmark of grammaticalized items is high frequency; but usually, it is not seen as a symptom of grammaticalization per se, but rather a phenomenon conditioned by other characteristics such as synsemantic and abstract meaning. At the same time, frequency is regarded as one of the main driving forces of grammaticalization by many researchers; see, specifically, Bybee (2003). High frequency may be characteristic of both grammaticalized and lexicalized linguistic formants. The crucial point of difference is productivity: lexicalized formants are not productive, that is, they combine with a limited, potentially closed class of co-formants, with each combination occurring very often. Grammaticalized formants in contrast are productive, that is, they are also used frequently, but with an open class of co-formants and in many types, typically with a high proportion of combinations being attested only once. This may be approached from a quantitative point of view in terms of the relationship of type and token frequency in a corpus (see, e.g., Evert & Lüdeling 2001; Baayen 2002). For such purposes, type is defined as the unique combination of some putative synsemantic formant and its more lexical counterpart, such as a derivational suffix and the base it attaches to, or — as in in our case – as an auxiliary and the infinitive it governs. Tokens, in turn, are defined as actual attestations of these types in a corpus. A productive formant will have only few types that occur very frequently, yielding a low type/token ratio; an unproductive, lexicalized formant will occur in only a few, frequent combinations and therefore with a high type/token ratio. In the present study, this
Grammaticalization studies
11
observation will be used to compare different uses of da(va)t+inf in respect to their productivity. Intermediate stages in productivity are possible and lexicalized patterns need not be stable. For example, we will see that the Russian construction of da(va)t with verbs of perception and cognition is very frequently used, albeit with only a few verbs and clearly not productively; nevertheless, we see that the set of verbs entering this construction has been subject to change, with the combination davat’ zametit’ gone out of use in the 19th century and the aspectual variability of the verbs entering the construction having changed. I give a short characterization of grammaticalization and lexicalization as I understand them, broadly relying on the conception outlined by Lehmann (2002, 2005). With Lehmann, I assume that utterances and subutterances in language may be processed both holistically and analytically. If approached holistically, they are given a meaning as a whole, and the compositional dimension of this unit is deemphasized. If approached analytically, the compositional dimension is in the foreground and its overall meaning construed according to rules of grammar and lexical meaning. In general, both approaches are possible in regard to the same utterance. Lexicalization and grammaticalization involve accentuating either of these two approaches, as Lehmann (2002: 3f.) elaborates: Accessing a collocation XY holistically means treating it as an entry of the inventory, as a lexical item. If this mode of access to XY gets more prominent in language activity, it is the initial step of the lexicalization of this sequence. Accessing a collocation XY analytically means treating it as a grammatical construction in which the structural properties of either X or Y or both matter and make a regular contribution to the pattern. If this mode of access gets more prominent in language activity, it is the initial step of the grammaticalization of XY.
Lexicon and grammar are thus situated on a continuum, as suggested by figure 2 (following page). In this sense, grammaticalization and lexicalization are orthogonal processes – part of a freely formed construction might lose compositionality while the resulting new formant at the same time gains in productivity, grammaticalizing to form compositional constructions where its contribution is synsemantic rather than autosemantic. We will see that such a case is found with the development of reflexive da(va)t, where the reflexive element develops away from making an independent contribution to what is an increasingly grammaticalized modal construction.
12 Introduction, overview and theoretical framework
Figure 2. Lexicon and Grammar (Lehmann 2002: 3)
Further aspects of grammaticalization study relevant in this study An important finding of grammaticalization studies is the formulation of grammaticalization paths or channels. A grammaticalization path describes a generalization about the development of markers expressing specific, basic lexical content into formants of constructions expressing grammatical content. The comparison of homonymy patterns in large numbers of languages has identified recurrent paths of grammaticalization that show that certain basic lexical concepts regularly give rise to certain grammatical constructions. These findings are often represented in semantic maps that allow generalizations about what is, and what is not expected in grammaticalization (Haspelmath 2003; de Haan 2005). Majsak (2005), for example, shows that in a large sample of languages, go has grammaticalized into markers of future, past tense and other, more peripheral meanings. Results such as these are important because they empirically investigate what is common in human language; see Heine & Kuteva (2002) for a compilation of a large number of grammaticalization paths. The theoretical relevance of these paths is viewed by many scholars particularily interested in the cognitive foundation of language to lie in that they tell us something about the origin of language and about basic cognitive processes behind the formation and use of grammar. In the present study, no conclusions of a general kind can be drawn; however, the grammaticalization of give can be reconstructed with better evidence in Slavic than in other language families with less documented history, thereby complementing more general studies, such as those mentioned in the introduction above. Language change starts with variation, and only if this variation is propagated does variation become change in relation to a language seen as a sys-
Grammaticalization studies
13
tem. An important issue in grammaticalization studies is therefore the study of the micro-processes of language change, focusing on how variation comes about and what the reasons are for the directionality in grammaticalization. A process prominently discussed in the literature is known as the conventionalization of implicature, whereby meaning elements arising pragmatically in the context become conventionalized and finally part of the regular meaning of a formant (see Hopper & Traugott 1993; Eckardt 2006 for a formalization). A different process is structural reanalysis, arguably a mechanism distinct from, but often involved in grammaticalization (Haspelmath 1998). A further perspective that has to be mentioned here is that of usage-based models of language. Such models share the assumption that linguistic structure evolves out of usage; put simply, it holds that what are first behavioral facts in discourse evolve to become coding properties in grammar. This view is very prominent in grammaticalization studies, and frequency is often seen as a driving force in grammaticalization (see e.g., Haspelmath 1999; Bybee 2003; Hoffmann 2004 for specific discussions). Perhaps the most important point for many linguists working in functional directions is that grammaticalization theory gives a principled way of accounting for heterogeneity in grammar, and explains it in the sense that it provides an integrated theory that encompasses a large area of inquiry, namely diachrony and synchrony, language use and language system. Heterogeneity is expected, since items are not categorized as either lexical or grammatical, but rather situated in a continuum between two extreme points. Formants in different stages of grammaticalization may compete in the expression of certain meanings; this is called layering. At the same time, some lexical item may have retained its identity as a lexical item in one context and acquired a new, more grammatical function in another: an example is going as an inflected form of go in I’m going home or as a future marker in I’m going to call you. Such a development is called divergence. Lastly, the meaning and use of grammaticalized items may still involve elements that are inherited from less grammaticalized stages, explaining cross-linguistic differences in grammatical meaning; this is referred to as persistence. Persistence as well as variation plays an important role in this study, especially regarding Polish, which is analyzed to be in a state of transition.
14 Introduction, overview and theoretical framework 1.3.
Causatives
1.3.1.
Definitions and basic notions
In their seminal article on causative constructions, Nedjalkov & Sil’nickij (1969) define a causative construction on the basis of the notion of a causative situation. Semantically, a causative situation is a situation composed of a causing and a caused (micro)situation: T he sun (= causer) shines. ⇒ T he ice (= causee) melts. | {z } | {z } causing microsituation
|
caused microsituation
{z
}
Causative macrosituation: T he sun melts the ice
Semantically, the main participant of the caused situation will be called causee, the main participant of the causing situation a causer. Note that in such a situation-based treatment of causatives, neither causer nor causee needs to be a prototypical agent, such as in the above example. I will refer to these situations and their linguistic expression as causing and caused as well as (in the context of analytic constructions) matrix and complement events and situations, treating the two largely as synonyms. There are many ways to express such a situation: (9)
a. b. c. d. e. f.
The sun was shining and the ice melted. (causation implicit) The ice melted in the sunshine. (causation implicit) The ice melted because the sun was shining. (cause - effect). Because the sun shone, the air became warmer, and therefore the ice melted. (several causes in a chain) The sun made the ice melt. (analytical verbal construction) The sun melted the ice. (lexical causative; labile verb)
The last two examples are the most compressed and at the same time the most specific in terms of marking causation. Both are causative constructions in the narrow sense: they are specialized lexemes or constructions. In many languages, causatives are formed by specialized morphemes, as for example in Finnish: (10)
a.
b.
Jää sula-a. ice melt-3SG The ice melts. Aurinko sula-tta-a jää-tä. Sun melt-CAUS-3SG ice-PART The sun melts the ice.
Causatives
15
Such morphemes are often highly grammatical: they denote the abstract semantics of causation, synsemantically forming different senses depending on argument properties and semantics of the base verb; for example, in the above example involving sulaa ‘to melt’, the causative morpheme denotes physical causation; with tappaa ‘to kill’ the same morpheme forms the verb tapaattaa ‘to let/have kill’, causation most probably grounded in human interaction. As a rule, such morphemes are situated somewhere between derivational and inflectional morphemes, since they may be more or less restricted in their productivity. Lexical causatives, such as command, force, allow are more specific than causative morphemes: they encode specific causative actions and, consequently, have strong selectional restrictions. Grammaticalized causative auxiliaries such as make, have, let are arguably situated between those two classes; they relate to the focus of this study. Numerous studies have shown that crosslinguistically, causatives are of two main types: permissive (as in let, allow) and factitive causation (as in make, force, have); see for example Nedjalkov & Sil’nickij (1969); Shibatani (1976b); Wachowicz (1976); Talmy (1988); Dixon (2000); Shibatani & Pardeshi (2002). A third category mentioned by Nedjalkov & Sil’nickij (1969) is assistive causation, found much less frequently; this is very close to what Shibatani & Pardeshi (2002) call associative causation and situate on a continuum between direct and indirect causation. This category will not play a role here. Often, these functions are expressed by underspecified markers (Nedjalkov & Sil’nickij 1969; Kulikov 2001). The most familiar examples are perhaps the cognates of English let in the Germanic languages such as German lassen, Dutch laten and Swedish låta. Morphological causatives are typically either ambiguously permissive and factitive, or dedicated factitive causative markers; dedicated permissive morphemes seem to be unusual (Wachowicz 1976). This might be a reason why the bulk of the literature on causatives focuses on factitive causation, sometimes called ’causation proper’ (Comrie 1985). With the analytic causative constructions formed by give in the Slavic languages, the permissive function, is, however, dominant, and in this study, more space will be devoted to the specifics of permissive causatives. I refer to permissive and factitive as the modes of causation.
16 Introduction, overview and theoretical framework 1.3.2.
Permissive (‘letting’) causation
Nedjalkov and Sil’nickij (1969) define permissive causation as causation where the ’causer’ is not the sole and most important agent of causation, as in factitive causation, but rather one that makes a contribution to the coming about of the caused event by “not impeding or permitting” the caused situation that is nonetheless perceived to be more saliently related to some other, more immediate cause. It is therefore by definition indirect and, one can add, more complex than factitive causation. Talmy’s (1988, 2001) equally influential definition is at first sight very similar: starting from the concept of two central participants with some dynamic inherent energy, he views causation as their interaction: factitive causation takes place when the agonist imposes his energy on the antagonist, rendering it irrelevant; ‘letting’ causation takes place when the antagonist’s inherent tendency is either (1) not interfered with by the agonist’s action (‘extended letting’) or (2) some preexistent obstacle is removed by the agonist (‘onset letting’). These relations are given in figure 3 in terms of Talmy’s force dynamic schemata. Talmy uses the same schemata to capture the modal notions of possibility and necessity; the relation of causatives and modals will be taken up below. In a study devoted to permissive causation in Romance, based on Talmy, (da Silva 2007: 3f.) stresses: “In fact, the ‘letting’ category covers a continuum of non-opposition and agent involvement in the event, which goes from non-reflected passivity (due to indifference, carelessness or negligence) through a strong committed sense of granting permission to a voluntary ceasing of opposition and even to cases of affectedness (imposition of force).” Rather than adopting Talmy’s distinction of ‘onset’ and ‘extended letting’, I will distinguish two focal types, closely related to Nedjalkov’s "noninterference" vs. "permission". The difference between the two lies in the activity in the subject: in the one case, labeled non-intervention, the causer does not interfere with the situation, and therefore bears responsibility for it. In the other, the causer enables the causee to do something and thereby causes the completion of the event; this is dubbed enablement. These two will be taken up in more detail in the remainder of the theoretical part below as well as in the analysis of the Russian part, where the focus is on the semantics of permissive da(va)t+inf. Generally, causative constructions involve a conceptualization of causing and caused microsituation according to which the former is seen to be de-
Causatives
factitive
17
permissive a.
b.
Figure 3. Force dynamics of causation, following Talmy (2000: 418,420). In factitive causation, an agonist (left) with intrinsic motion is introduced to transfer energy to the antagonist (circle). The antagonist changes from rest to motion (arrow below). There are two principle modes of permissive causation depicted: (a) is ‘extended letting’, where the agonist is a potential barrier and stays out of the way of an antagonist with intrinsic movement; no change occurs; (b) schematizes ‘onset letting’, where the barrier is removed, causing the antagonist to change from rest to motion.
cisive among the set of real-world causes and neccessary conditions for the latter. Ontologically, the causing situation is never the only one that can be conceptualized as contributing; rather, it is foregrounded as the crucial part of a causal chain. One has to keep in mind that there are no events that occur unicausally. To switch on the light does not describe all the necessary conditions that cause a bulb to illuminate - structural conditions such as circuitry, power infrastructure, make-up of switch, bulb and socket as well as contributing dynamic processes such as the flow of specific electrons, subatomic processes, light waves etc. can all be, generally speaking, foregrounded as ‘causes’, ‘necessary conditions’ for the situation denoted by he turned on the light. This leads us to the dichotomy of direct vs. indirect causation. This distinction has been discussed for quite some time, under quite different headings (see Wierzbicka 1998: 117, Wolff 2003 for overviews; Shibatani & Pardeshi (2002) for a principled typological assessment of some of these approaches and a new proposal). Directness / indirectness of causation is a question of degree; languages differ in whether or not they make grammat-
18 Introduction, overview and theoretical framework icalized distinctions along this line as well as where they make it. Comrie (1985: 332f) summarizes as follows: [. . . a ] relevant parameter is the degree of closeness between the cause (i.e. the causer’s action) and the effect (resultant situation). [. . . e.g.] John broke the stick implies an immediate connection between John’s action and the breaking of the stick [. . . ], wheras John caused the stick to break suggests rather a mediated chain of events [. . . ] One often finds that, where a language has both analytic and morphological or lexical constructions, the former implies less direct causation than the latter [. . . ]
In permissive causation, the causer makes the caused situation possible, and, all other things being equal, thereby makes it happen. It presupposes some second cause that is not completely defocused; it is therefore per se indirect in nature.
1.3.3.
Curative factitive (‘having’) causation
A different type of indirect causation is what may be called curative causation, from Latin curare ‘to have done’ (Pennanen 1986; Toops 1989). Curative causation is a subtype of factitive causation where the causer acts in order to accomplish something by way of an intermediate animate agent, the causee. In curative causation, this accomplishment is in focus, rather than the interaction with a causee that is usually unexpressed and either irrelevant, generic or contextually known (see Loewenthal 2003 for a corpus based study on Dutch). In English, have as in I had the car repaired expresses this type of causation; most examples below for type II causatives relate to this subtype. Nedjalkov & Sil’nickij (1969) relate this to the indirect-direct distinction (in their terms contact vs. distant causation – kontaktnaja / distantnaja kauzacija) and note that many languages tend to not overtly mark such causation. A language in point is Russian: Ona sšila sebe plat’e, lit. ‘she sewed herself a dress’ may also mean ‘She had herself a dress sewn’ (example from Toops 1987). This phenomenon is well known in respect to reflexive grooming verbs such as postriˇc’sja ‘cut one’s hair / have one’s hair cut’ in the context of the polysemy of the reflexive postfix -sja; however, already Lötzsch (1972) has shown very clearly that this is part of a wider phenomenon, presenting the quite drastic example of rasstreljaju lit. ‘I’ll shot you’ in the meaning I’ll have you shot in Babl’s Konarmija (see also Babby 1983; Bulygina & Shmelev 1999).
Causatives
19
The phenomenon of zero marking of curative causation or contextual causatives (Nedjalkov 1976: 29) is not restricted to languages lacking a productive factitive causative: Like Russian, Polish and Latin, Japanese (Hashimoto 1989: 73f.) likewise tends to not mark such causation at all. Other languages such as Lithuanian (Toops 1989) possess specialized morphologic markers for this type. 1.3.4.
Causative domains: from interpersonal to cognitive
Another level of distinctions may be drawn according to the degree of control (Lehmann 1991) the causer / the causee has in the resulting macrosituation. Positing control as the relevant category is compatible with approaches taking semantic roles or the distinction between unaccusative, unergative and transitive verbs (Shibatani & Pardeshi 2002), as it represents an abstraction over semantic roles as well as verb classes. Intuitively, it is clear that causation by a human on a thing is something quite different from a human telling another human to do something, or a thing being the reason for something to happen. Elaborating on these parameters partly based on Koo (1997) I posit three principle domains of causation given below. A fourth domain obtains with inattentive perception and cognition verbs. In terms of control the causee exerts in the caused situation, verbs of perception and other cognitive processes constitute a special class since they denote actions that are in a sense involuntary as they happen, but cannot be said to be beyond the control of their experiencer. It is a wellknown fact in typology that these verbs behave differently from both prototypical control verbs such as transitive causative action verbs and from prototypical non-control (unaccusative, inactive, intransitive) verbs. The causation of these states are lexicalized in Stimulus-Experiencer or Experiencer-Stimulus verbs (En. to frighten; to show - to see, Ger. erschrecken (trans./intr.), Rs. ispugat’ - ispugat’sja) (see also Loewenthal 2003; Chappell & Peyraube 2006 for similar typologies of analytical causatives). • Manipulative causation – only the causer has control: A human agent acts volitionally on either – a non-volitional patient: Peter flies a kite or – a volitional patient that acts without control: Peter kills Paul or Peter lets Paul die
20 Introduction, overview and theoretical framework • Interpersonal causation – causer and causee have control: A human agent causes another human agent that is thus also a patient to do something volitionally: Peter makes/has/lets Paul do the dishes. • Impersonal causation – the causer does not have control: Something (a force or an event) is the cause for – an animate to do something, as in What made you decide to leave me? or This allowed me to take a shower. – for something to happen to an animate or inanimate patient without control: That made the bubble burst or The plug’s coming loose let the water run through or That made him laugh • Cognitive causation – the causee is an experiencer: The causer represents a stimulus for the experiencer-causee, as in He made me think of Paul, He surprised me, That let me see the point One can see that this tentative typology presents a selective combination of the parameters of the causer’s animacy and the causee’s control. In the annotation of my corpus this categorization was approximated by annotating whether causee and causer were animate or not, since control is a gradient concept and there are numerous cases where this fact leads to intermediate types. Consider sleeping: it is unaccusative like to die, but unlike to die, both imperative and attempt are grammatical (albeit with a slight semantic shift). However, clearly unaccusative verbs were marked and evaluated qualitatively, rather than quantitatively.
Reflexive causative constructions In the empirical part, special emphasis is given to the development of reflexive causative constructions. Reflexive causative constructions are those where some participant of the caused event is at the same time also the causer, as in he let himself be cheated. Constructions such as these have found some attention in the literature concerning the evolution of passive constructions (von der Gabelentz 1861; Knott 1995; Haspelmath 1990) because they might help in explaining the transition of causatives to passive, observed in many languages. They do not have an apriori status: I expect any language with causative and reflexive constructions to freely combine them, just as other
Causatives
21
constructions combine in usage. However, these constructions have grammaticalized further in Czech and Polish, but not in Russian. This will be discussed in detail in the empirical part of this study.
Summary I now summarize the distinctions adopted and used in the empirical part of the study as the semantic tertium comparationis, before focusing on specific questions concerning the syntax of causatives and the interaction of negation, modality and causativity in the following sections. Distinctions are drawn on two levels: first, I distinguish two modes of causation, factitive and permissive. Where appropriate, I further distinguish the factitive subtype curative as well as two permissive subtypes: nonintervention and enablement. Second, these modes are distinguished in the four domains of manipulative, interpersonal, impersonal and cognitive causation. This categorization is based on participant properties such as animacy and control. These categories form the grid in terms of which da(va)t+inf is analyzed across languages. Some further distinctions are also drawn where appropriate: this includes the interaction with reflexive constructions as well as the syntactic coding of the causee and the distinction of deontic and dynamic causation, both introduced below.
1.3.5.
Causee coding: Type I and Type II causatives
An important issue in the grammaticalization of causative da(va)t+inf in Polish and Czech is the form of the causee. In several Czech constructions, in particular, the only way to introduce the causee is as an instrumental or in a prepositional phrase, while in earlier stages, only the dative is possible. This section introduces the main points of the literature regarding the syntactic coding of the causee. Starting with a descriptive generalization by Comrie (1976) with a rich literature that aims to refine, explain and syntactically model this generalization, the form that the causee takes crosslinguistically has been an object of intense scrutiny. Causatives fall, syntactically, into two broad types (for illustration I adduce Spanish data from Bordelois 1988: 57f., adapting emphasis and parentheses):
22 Introduction, overview and theoretical framework • Type I: (11)
Hicieron destruir la ciudad a los soldados. (they) made destroy the city the soldiers. ‘They made the soldiers destroy the city.’
(12)
Hizo venir a Juan. (he) made to come to John. ‘He made John come.’
• Type II: (13)
Hicieron destruir la ciudad (por los soldados.) (they) made destroy the city by the soldiers. ‘They had the city destroyed by the soldiers.’
(14) *Hizo venir por Juan. (he) made to come by John. ‘He made John come.’ In Type I causatives, the causee is given canonical object case marking. In Type II causatives, it is represented as an optional argument-adjunct in nonobject coding similar or identical to optional argument-adjuncts in passive constructions. This class is restricted to transitive verbs, as Type II causatives with intransitive verbs (14) are ungrammatical. While these are the basic facts, the discussion soon revealed that the issue is more complex: not all transitive complement predicates allow type II causatives, cf. another example taken from (Bordelois 1988: 89): (15) *Hicieron ver la ciudad por los turistas. (they) made see the city by the tourists. intended: ‘They had the city seen by the tourists.’ One of the points of debate is what factors exactly play a role in determining the distribution of complement predicates between Type I and Type II causatives (see below). The similarity of the lexical and structural restrictions concerning Type II causatives and those regarding passivizability, as well as the fact that in many languages, the passive and the causative by-phrase coincide in form, has received much attention in linguistic research. Without claiming to do justice to the immense literature, I wish to point out in broad terms some approaches concerning this question.
Causatives
23
The first is the line initiated by Comrie (1976), according to which the causee in a causative construction assumes the highest-ranking case in a case hierarchy that ranges from the core cases nominative and accusative down to oblique and prepositional cases. The transitivity restriction concerning Type II is thus motivated by an avoidance of double accusatives: In the causative of intransitives, the causee may be assigned accusative case in view of the absence of an object of the complement verb; in the case of transitive complement verbs, the causee is assigned oblique (dative) case. This was an approximation the shortcomings of which Comrie was well aware at the time. It did explain the fact that oblique case causees tend to appear with transitive verbs, but it failed to adequately address the observation that in many languages, accusative doubling is possible (as in German, for example; see Kozinsky & Polinsky 1993 for a thorough analysis of this issue in Dutch and Korean), and to give motivation to the particular form in which the causee appears. Later works along case-semantically oriented lines include Saksena (1982) and Cole (1983), who finds that agency of the causee is the decisive factor for the causee to be expressed as an optional by-phrase and who claims that counterexamples should be explained by grammaticalization of agency to the more general structural feature of transitivity. Analyses in this approach are, in my view, united in that they give precedence to explanations in terms of semantic roles, discourse prominence and other functional properties of sentence participants. Apparently following Christian Lehmann’s conception, his student Koo (1997) explains the connection between the expression of the causee as prepositional phrase and factitive vs. permissive interpretation of German lassen with the property of control that the causee is said to have. Kemmer and Verhagens 1994 influential paper in the framework of cognitive grammar is akin to the aforementioned in that it looks for semantic regularities in the expression of the causee. It provides a thorough summary of the functionalist and some of the generative literature on this issue. Starting with conceptual considerations, Kemmer and Verhagen emphasize the similarities between the argument coding of transitive and ditransitive verbs on the one hand, and causatives of intransitives and transitives, on the other. Essentially, they argue, case assignment in causatives is driven by the same semantic and pragmatic factors as elsewhere in language, rather than by a formal hierarchy or, one may add, purely syntactic factors. The following account of the occurence of von-phrases in German provides a representative picture of the factors viewed as relevant in this and many other approaches (the example referred to is Er ließ seinen Sohn / von seinem Sohn den Brief abtippen):
24 Introduction, overview and theoretical framework Presumably, any of these factors (lack of conceived importance in effecting the letter’s typing, autonomy, physical distance, lack of direct verbal contact) holding of a given conceived situation can contribute to the use of the von construction to code the causee rather than accusative case. We might predict that the more of such factors that hold, the more likely or felicitous will be the coding with von. (Kemmer & Verhagen 1994: 133)
This quotation shows the wealth of factors that, as most linguists at least in the functional paradigm will probably agree, influence the case marking of the causee in causative constructions. A rather different, generative, tradition focuses on the crosslinguistic similarity of the passive and the causative by-phrase. In an important paper on German, Reis (1976) raises many issues concerning the generative treatment of lassen; Suchsland (1987) later posits passive of the infinitive complement without passive morphology to account for von/durch-phrases. The main problem with such approaches is that there is no passive morphology to support the claim that the by-phrase is licensed by a passive complement. Gunkel (1998), for example, makes a proposal to have it licensed by the matrix verb in an HPSG analysis. In his account, lassen is an auxiliary verb that may, like werden, the formant of the canonical passive in German, assign an agentive by-phrase, albeit optionally and without causing the lexical verb to take participle form. A more recent discussion is concerned with solutions according to which the causative formant in question may either take a propositional argument and a causee NP or simply a propositional argument. Expression of the causee as an optional prepositional phrase is taken to be one of the characteristics of the latter (see Guasti 1996 and Folli & Harley 2007 for an overview of the discussion). Broadly speaking, the property of causatives to licence both argumentadjunct by-phrases and unequivocal objects has been an important issue in generative frameworks. Other issues include the precise modelling of monoclausal and bi-clausal analyses of causatives in various languages, the question of whether we are dealing with auxiliary, object control, or ECM verbs, more specifically the number of crosslinguistic types apart from Type I and II introduced above, and the status of the by-phrase as argument or adjunct. As one can see, many issues in the functionalist and more generativist approaches are similar, but solutions tend to differ with respect to their more syntactic or more functional nature. Needless to say, there are many intermediate cases where work from many frameworks involving syntax, semantics
Causatives
25
and pragmatics is acknowledged and accommodated; it is therefore perhaps an exaggeration to speak of two paradigms. On the other hand, in many cases, there is no mention of the work done on the other side of this apparent divide. Summary, the following are the main points emerging from the discussion: • in the languages of the world, there is a diverse but constrained picture of causee argument coding • broadly speaking, the causee may be coded as object or as optional argument-adjunct, with many languages having two or more choices with the same formant • semantically, transitivity of the complement predicate, as well as functional participant properties of the causer, causee, and (if applicable) complement predicate’s object – including agency, affectedness, animacy and discourse prominence – are factors relevant to causee coding. • permissive vs. factitive reading of causatives, and more or less direct causation in both the sense of strength of obligation and of absense/presence of mediation have been linked to these factors of agency, if less prominently. • there is a connection to passive by-phrase and/or instrumental coding in many languages. I will now discuss one approach which has been influential in the formal literature, albeit less so in the functional paradigm.In Alsina (1992), work done in the framework of Lexical Functional Grammar (LFG), the major parameter of typological variation relating to the expression of the causee is both semantically and syntactically motivated and is tied to the assumption that causatives subcategorize three arguments: the causer, the caused proposition and a patientive object that is acted upon: this patient is said to be ‘fused’ with an argument of the complement clause. Alsina (1992: 522) posits two variants of meanings, which constrain the types of arguments that the patient of causation may fuse with: Variant 1: The causer, in order to bring about an event, acts on an individual who is the participant most in control of that event. [=the causee, RvW] Variant 2: The causer acts on an individual by causing an event that affects that individual. [= the patient of the complement predicate, RvW]
26 Introduction, overview and theoretical framework Type I causatives are thus instances where the matrix predicate’s patient ‘fuses’ with the causee, while in Type II causatives, the matrix verb’s patient ‘fuses’ with a patient of the complement predicate; in the latter case, the causee ends up as an optional by-phrase. Since such a choice between causee and complement patient exists only if the complement verb has an object, it follows that the complement predicate of Type II causatives has to be transitive; the fact that not all transitive verbs combine with a Type II causative is captured with the further condition that the object has to be a patient. Much of the rest of Alsina’s argumentation is rather technical. Alsina provides syntactic and pragmatic evidence showing that in Chichiwa Type II causatives, the object of the matrix verb is a patientive object of the complement, while in Type I causatives, it is the causee. Thus, a good answer to the question what did they do to the soldiers? is (11), not (13); it is not a good answer to the question what did they do to the city?, which, in turn, is an appropriate question to be answered with (13), but not with (11). He adduces more evidence to show that only those complement arguments that are patients may fuse with the matrix argument, which links this approach to those mainly concerned with binding. Alsina also deals with the question why the argument-adjunct in Type II causatives should take the same form as the passive argument adjunct in so many languages. His syntactic reasoning roughly amounts to the following: If the external argument of a verb cannot be expressed, it becomes an optional argument-adjunct. In the passive, suppression is triggered by a passive marker; this is termed specified suppression. In the case of the causative, as well as in the case of verb nominalizations, suppression is unspecified and follows from the fact that there simply is no slot available for the external argument (the sentence subject position is occupied by the matrix verb subject, the causer). If the causee argument does not fuse with the patient argument slot of the matrix verb, the causee appears in an agentive by-phrase; this is the Type II case. If, in contrast, not the causee, but the patient of the complement fuses, we get a Type I causative. If the verb is intransitive, there is no choice: the causee must occupy the patient position of the matrix verb, thus yielding only Type I causatives of intransitive verbs. Like Gunkel (1998) in the framework of HPSG, and similar to various functional proposals, Alsina thus sees an identical syntactic component at work in passives and causatives rather than one to underlying the other. One difference of his approach as opposed to others (including functional approaches), lies in the fact that it takes into account not only the semantic
Modals and causatives
27
role of the causee, but also that of the complement patient. This is no contradiction, since these two are not independent; there are presumably few verbs with strong affected patients and weakly agentive agents1 (in other words, we are dealing with highly transitive verbs (Hopper & Thompson 1980) in both cases). Nevertheless, such a shift in focus is important for an understanding of the development of Type II causatives in Polish and Czech. In this study, I focus on a rather different similarity between causative and passive that explains specific restrictions of the occurence of the by-phrase in reflexive permissive and factitive causatives. It is semantic and emphasizes the discourse structure and saliency of arguments in these specific configurations of the causative, arguing that the agentive by-phrase in causatives is identical to the by-phrase in passives because it serves the same purpose: identification of an argument demoted for semantic and discourse-specific reasons. It follows that, diachronically, the thing first to happen is omission of the agent in language use. Only at a second stage, the possibility of reintroducing an agent appears. This succession of stages is described in the diachronic part and discussed in the concluding part.
1.4.
Modals and causatives
Adopting a narrow working definition, modals in this study are to be understood as verbs or verb-like elements forming or co-forming the main predicate of a cause, thereby qualifying the semantics of the clause in relation to possibility, necessity or volitionality. Classical exponents of such elements are können, müssen, sollen or wollen in German or can, must, should or want in English; in these languages, modals form coherent classes with highly grammaticalized exponents. Hansen (2001) shows that the corresponding verbal elements in the Slavic languages form more heterogeneous classes composed of more or less grammaticalized elements extending from more lexical verbs to modal auxiliaries. As a whole these classes are not as grammaticalized as their Germanic counterparts. The present study touches upon the topic of the grammaticalization of modals in as much da(va)t+inf has developed into a modal in several Slavic languages. This development involves the transition from a permissive causative to a modal of possibility with varying degrees of grammaticalization. The present section therefore provides background on the relationship of modals and causatives in general. Another issue relating to both modal
28 Introduction, overview and theoretical framework and causative formants derived from da(va)t is their (varying) tendency to be used under negation; the relationship of negation to causatives and modals is therefore addressed below. I will first review some statements in the literature concerning the relationship of causatives and modals. In the following subsection I then suggest that the same domains generally associated with modal meaning - dynamic, deontic and epistemic modality - are also relevant in the analysis of causatives. Subsequently I focus on negation of causatives and modals from a common perspective, concluding with a summary. Modality and causation are implicitly treated as one phenomenon in the works of von Wright and other more philosophical or logical approaches where causatives have been used to exemplify or define modality; note that causative verbs such as permit, command or force are often used in definitions of modals such as can or must (e.g., von Wright 1963; Döhmann 1974). Generally, it is a trivial observation that intransitivized causatives may have the same meaning as modals (‘to be permitted’ →‘may’). More difficult is the question if the reverse is true, that is, whether and in what sense causatives represent transitivizations of modals (Nedjalkov 1971). Obviously, there is more information in causatives than in modals, since causatives introduce a further participant. On the other hand, there clearly is a modal component in causatives, evident from these examples. This connection has not figured prominently in the linguistic literature. (da Silva 2007), for example, uses notions from modal semantics to describe permissive causatives without, however, addressing the issue directly. In some older treatments of the German causative lassen, it is included in the class of modal verbs (or ‘near modal verbs’). In German there is an obvious syntactic connection: only modal verbs, perception verbs and lassen take the bare infinitive in the perfect. Talmy (2000: 442ff.) explains modals in force-dynamic terms, including modals and causatives as well as ‘help’ in a “Greater Modal System”, defined by force dynamic semantics and the subcategorization for a bare infinitive. He gives a list: (16)
He can/may/must/should/would not/need not/dare not/had better I made him/let him/had him/helped (him) –push the car to the garage. (Talmy 2000: 444)
In his discussion of Turkish morphological causatives, Kural (1996, 2000) explicitly links modality and causatives in terms of possible-world seman-
Modals and causatives
29
tics. He introduces an abstract operator cause that denotes a superset of both modes of causation and relates, broadly, to responsibility on the part of the causer for what is caused. Factitive and permissive causative meanings are defined over a relation in the set of causally possible worlds and universal and existential quantification, such that (17)
a. b.
A let P = 1 iff ∃ wC , wC a causally possible world, such that A cause P = 1 in wC A make P = 1 iff ∀ wC , wC a causally possible world, A cause P = 1 in wC .
He arrives thus at a relation parallel to that of possible-world treatments of modal semantics, in short defining (18)
a. b.
let P = cause 3C P make P = cause 2C P
However, such a treatment is problematic, since causing something to be the case in at least one possible world is expressed by make possible and to enable, typically expressed by lexical verbs; let, however, means make possible and make happen. I will show below that this is a crucial difference; the more grammaticalized permissive markers investigated in this study imply the (non)occurrence of the caused situation. The question of the relation of causatives to modals cannot be investigated in sufficient detail here. In the remainder of this section, I will focus on two areas: first, the relevance of the distinction of dynamic, deontic and epistemic domains seem to be important in the grammaticalization of modals (Hansen 2001); secondly, the relationship of negation and modals and causatives as dual operators.
1.4.1.
Modal domains in causation
Following Palmer (20012 ); Hansen (2001) and many others, I distinguish three domains of modal meaning: Dynamic, deontic and epistemic meaning. In this subsection, I examine to what extent these domains are also relevant for causatives.
30 Introduction, overview and theoretical framework Dynamic domain Dynamic modality is modality restricted by objective reality: something is possible, impossible or necessary by virtue of intrinsic characteristics of a participant or of objective circumstances: You have to leave now in order to be on time is an example. Clearly, this can be easily extended to causation in terms of objective circumstances or intrinsic characteristics. Illustrative examples: (19)
a. b. c. d.
The timetable forces me to leave now. Peter made Paul laugh. Peter let Paul drown. Peter (unintentionally) let Paul take the last piece of cake.
All these examples do not involve norms, but objective possibility and necessity as resulting from causation. Causatives thus make the source of modality that is left implicit with modals explicit. The paramater of control that the participants have over the situation may vary in these constructions, as may be seen in (c) where the causee and in (e) where the causer does not consciously control the situation. In all cases, the caused event involves possibility or necessity as modal notions: I had to leave, because of the timetable; Paul had to laugh, because of something Peter did or say; Paul could drown, because Peter did not rescue him; Paul could take the cake, because Peter did not pay attention.
Deontic domain The deontic domain is identified in the literature as the domain of obligation and permission: both terms are derived from transitive causative verbs. The link to modality is thus most obvious here. Verbs denoting factitive causation such as to oblige or to command mean make deontically necessary in relation to some system of deontic norms. Similarily, verbs expressing permissive causation such as to permit, allow entail a resultant state where something is deontically possible in relation to a certain deontic souce. Both types of causatives are often expressed by speech act verbs that can be used performatively.
Modals and causatives
31
Epistemic causation The epistemic domain is the domain of the speaker’s commitment or certainty in relation to what is expressed. Grammaticalized modals typically also encompass senses in this domain: Paul must/should/could be sitting in the park may denote that the speaker only assumes for some reason that Paul is sitting in the park. With the above two domains, we have seen that the causative makes the source of modality explicit (e.g., Paul could take the cake because Peter didn’t pay attention). No straightforward analogy in the epistemic domain is apparent; e.g., we do not say The time of day makes Paul sit in the Park to denote Peter must be sitting in the park (since it is lunch time). However, in some cases permissive causatives seem to have evolved to denote meanings belonging to the epistemic domain, such as to assume. One such extension I would like to mention involves expressions diachronically or synchronically derived from Russian pustit’ ‘to let, discharge’ such as dopustit’ ‘admit, concede, allow’ as in dopuskat’ vozmožnost’ ‘admit the possibility’ and the optative marker pust’ / puskaj as in the following examples (from the RNC): (20)
Puskaj, naprimer, Javlinskij sdelaet oficial’noe javlenie [. . . ] ‘Let’s assume, for example, that Javlinskij makes an official statement [. . . ].’
In English or German, similar extensions are found. The imperative of lassen and let may denote a proposal to consider something, as in Let it be as it may; Nedjalkov (1976: 209) calls this meaning of lassen ‘hypothetical permissivity’ and adduces Laß ihn dreimal ein Gentleman sein. . . — das war Vernichtung (Arnold Zweig) as an example. These uses may be considered extensions of causation to the epistemic domain of assumption and certainty.
Summary In both causatives and modals, the distinction of modal and deontic domain has shown to be of linguistic relevance; however, in contradistinction to these, no clear analogous paradigmatic relation is found in the epistemic domain, even though some use of causatives relating to this domain may be observed.
32 Introduction, overview and theoretical framework 1.4.2.
Negation with modals and causatives
This section is concerned with the interaction of causatives and modals with negation and has the aim to provide the ground for the explanation of an empirical finding, namely, that da(va)t+inf in these functions has a tendency to be used foremost in negative polarity. My point of departure is the assumption that informativity is a factor influencing usage frequency as analyzed in the empirical part. I first analyze negation in terms of informativity and then extend the discussion to dual operators, a class both modals and causatives belong to. Finally, I examine internal negation in the semantics of permissives and relate this to tests distinguishing permissive from factitive usage. Note that this treatment is not meant to be exhaustive and glosses over many complicated issues.
1.4.2.1.
Types of negation and informativity
A basic notion concerning negation is the distinction of contrary and contradictory opposite terms, ultimately going back to Aristotle’s metaphysics. Contrary terms obey the Law of Non-Contradiction (henceforth LNC): One cannot at the same time say of a piece of paper that it is all blue and all red; these color terms are contrary in the sense that they cannot both be true of one object at the same time. It may, however, be neither, for example, it may be black. Contradictory terms obey both the LNC as well as the Law of the Excluded Middle (henceforth LEM) that states that some predication is either true or not true.2 Predicates that stand in a contradictory relationship cannot both be false at the same time: one of them has to be true, tertium non datur. A number is odd or even; a person is dead or alive; ice on a lake may hold or may not hold. To negate one is thus to affirm the other. This directly relates to informativeness. Affirmative statements tell us something about the world; for example, the statement Walt Whitman lived in New Jersey tells us something about the state of the world. Negation of contraries do not tell us much, and will, in order to be informative, somehow refer to an assumption in the context that the predicate might be true: To say that Walt Whitman did not live in London will not tell us much about the world, unless we have reason to suspect he did.
Modals and causatives
33
The negation of contradictories, conversely, is informative: to say Walt Whitman is not among the living is informative, since it is equivalent to the affirmation that he is dead.3 All this is, of course, a simplification. First of all, saying that a certain utterance needs context in order to be informative is in a certain sense another way of saying that the hearer will try to supply some context that will make the utterance informative (by Gricean principles). On the other hand, affirmative information is also in need of context in order to be informative; for example stating that Walt Whitman is dead in this study is only informative if it serves as an illustration to some point relevant to its subject matter. This links up to the difficult question of what is to be counted as affirmative (Horn 1989). I will not discuss this here; note that in any case, negative, not positive polarity is the marked category in language and that I suggest one of the reasons for this is inherent in their differences with respect to informativity. As we move from the realm of logical semantics and turn to natural language, the role of pragmatics becomes more important. For example, as Horn (1989) suggests, what is commonly known as neg-raising of verbs such as think, assume, believe, etc., can be understood as a contradictory reading of a logically contrary negation, triggered by Gricean principles of cooperation. I do not think Walt Whitman is dead is, prima facie, not a very informative statement, since it does not tell us much about the speaker’s beliefs, only about his non-beliefs. Horn argues that under the assumption of cooperation, such a statement is interpreted as a member of a contradictory opposition: one may either think that p, or one may think that not-p. Under the assumption that a third option is not given, that is, that these two propositions stand in a contradictory relationship, not think that p is interpreted as a polite understatement equivalent to think that not p. Contradictory oppositions in language may thus be grounded in discourse rather than in logic.
1.4.2.2.
Modals and causatives on the square of oppositions
A special class of contradictory predicates is known as dual operators. This class includes partial and total quantifiers, modal auxiliaries expressing possibility and necessity and factitive and permissive causatives – the predicates we are interested in here. All these operators are dual in the sense that each pair is interdefinable by negation (see Löbner 1990 for discussion of a large
34 Introduction, overview and theoretical framework set of dual operators). They involve two places of negation: the operator itself and its argument may be independently negated. Wide-scope negation over predicate and argument is equivalent to its contradictory term and narrow scope of negation. Thus, not possible is equivalent to necessarily not; not anybody speaks is equivalent to all do not speak; not to let somebody do something is equivalent to make somebody not do something. In general, with dual operators a and b, the following set of relations holds: (21)
a. b.
¬ a (p) = b (¬ p) a (p) = ¬ b (¬ p)
If we consider all combinations of predicates and negation in terms of contradictory and contrary opposition, we arrive at the Square of Oppositions derived from Aristotle and his medieval commenter Boethius and shown in figure 4. This last section has served to introduce the distinction between contradictory and contrary negation and to suggest that negated contradictories are as such more informative than negated contraries. In the following, I relate this idea to modals and causatives. I again start from the assumption that affirming something about a state of affairs is more informative than negating something about it. With causatives and modals, this question becomes more complicated because we are dealing with two potential negations: the modal or causative operator itself as well as the proposition it modifies may open a place for a negating element. This links up to the property of veridicality of an operator, that is, whether or not the truth value of that operator determines the truth value of its proposition. I will next evaluate all four corners of the square in respect to their veridicality and informativeness. English let is used to exemplify a permissive marker in this discussion. To enable is used to exemplify a straightforward transitivization of possibility cause to be possible. An aim of this section is to highlight the difference between these two types of predicates in order to shed light on the nature of the relation of permissive causatives to modals of necessity. It can be easily seen that only the top two corners of the square of oppositions contain veridical predicates. I will start with the upper left (A) corner. If something is necessarily the case, this implies that it indeed is the case. Likewise, if somebody is forced to do something, it follows that it is the case that they do it. The dual operator counterparts behave likewise, although they seem to be more unusual and, generally, harder to understand4 : if something is impossible not to be the case (not poss not p), it is indeed the case; the ana-
Modals and causatives
A nec (p) = ¬ poss (¬p)
contrary: cannot both be true, but both may be false
fact (p) = ¬ perm (¬p)
35
E nec (¬p ) = ¬ poss (p ) fact (¬p ) = ¬ perm (p)
contradictory: either one is true or false
implies I poss (p) = ¬ nec (¬p)
implies O poss (¬p ) subcontrary: both may be true at = ¬ nec (p ) the same time, but not both false
perm (p) = ¬ fact (¬p)
perm (¬p ) = ¬ fact (p)
Figure 4. Square of oppositions for dual operators adapted to modals and permissives. See van der Auwera (1985); Löbner (1990); Horn (1989); in relation to modals, see van der Auwera (1996, 2000); de Haan (1997).
logue applies in the case somebody does not let somebody not do something (not perm not p); the latter is implied to do it. The A-corner of the square thus contains veridical modal and causative predicates, which I consider informative by themselves, that is, without strong contextual anchoring. In the upper right (E) corner, the same applies. In saying that something is impossible, one implies that it is not done / not the case. Likewise, if somebody does not let somebody else do something, this implies that the latter does not do it. Not to enable, conversely, does not mean the same as make impossible: it does not entail anything. Both permissive let and modals denoting impossibility are thus veridical: they imply a truth value of the term over which they have scope, albeit a negative truth value. Again, the dual counterparts show the same characteristic, but are not nearly as natural: to say that something was necessary not to be the case implies that it was not the case (nec not p); if somebody is forced not to
36 Introduction, overview and theoretical framework do something (fact not p), he does not do it. The E-corner of the square thus contains veridical operators; however, they are not as informative as the upper left square, as the implied truth value is negative: they entail that something is not the case, which itself needs to be salient information. Even less informative without context is the lower left (I) corner of the square. To say that something is possible is not to say anything about whether or not it is the case; logically, this position is therefore not veridical. However, modal operators often do imply, perhaps pragmatically, the truth value of their propositional argument: note that a sentence such as He was able to walk away, but didn’t is odd, if not ungrammatical; he would have been able to walk away, but didn’t is clearly preferable. In Russian, the perfective smoˇc’ expresses both possibility and realization of the event. The picture is more categorial with permissive causatives: he let him go implies that he did go; he enabled him to go does not (this difference will be taken up below). The dual counterparts not necessary that not and not make somebody not do something are highly unusual and very elaborate; logically, they are clearly non-veridical. Least informative of all is the lower right (O) corner . That something is not necessary does not logically entail anything about whether it is the case; also, no unequivocal pragmatic effects seem to obtain: he didn’t need to go there, but did / and did not are both felicitous. The same evidently applies to the causative case he did not make him do it. The dual counterpart possible not to seems to behave basically the same; it is logically nonveridical. The O-corner is hard to express with many formants. It is infelicitous with the grammaticalized modal can as in ?he can (not go there); he is able not to do it is possible, but unusual at best. Internal negation of permissives is very unusual, but possible with many formants. With let the proposition is veridical by virtue of the auxiliary (see below): he let her not do it, *but she did. Enable somebody to not do something is non-veridical, but marginal at best. Horn (1989: 260ff.) has proposed that lexicalizations of this corner are unusual, with counterexamples later adduced including German nicht brauchen (not nec p) and others (see van der Auwera 1996: 186). In any case, it seems reasonable to presume that in natural language, this corner has the least functional load of the four. Overall, we have seen that factitives behave very similar to modals of necessity while some differences mainly concern permissives and modals of possibility.
Modals and causatives
37
Only the upper two corners of the square, that is, neccesity / factitive causation as well as negated possibility / negated permissive causation are logically veridical, in as much as both upper corners can be logically framed to involve wide scope affirmation of necessity / factitives. The right corner contains negation over the predicated proposition; the truth value of the predicated proposition is implied to be false. As such, the informativity of this negated proposition itself relies on the saliency in context. The lower left (I) corner is less clear, and there is an analogous difference in the behavior of expressions of modality and causation with respect to veridicality. Logically, these operators are non-veridical, but in natural language, modals of possibility do seem to have a certain inclination towards veridicality, depending on specific lexicalizations or certain morphosyntactic surroundings, the indicative past tense in particular. It is plausible to attribute this to an implicature to the effect that if one says (in the past) that something was possible and does not state whether it actually took place, there is an assumption that it was in fact done which increases informativity and relevance; thus, we are dealing with a quantitative implicature that may become conventionalized. With permissives, a similar picture obtains. There are two options: General permissive markers, such as English let, French laisser, Portuguese deixar, German lassen, zulassen, Finnish antaa and Slavic da(va)t as well as cognates of pozvolit’ are, as far as one can tell, generally veridical. There are also more narrow permissive lexical items such as (most clearly) umo˙zliwi´c or to enable, and less categorically, to permit, that are non-veridical or may show veridicality only in certain pragmatic or morphosyntactic surroundings. On first sight, this difference seems to be a clear difference to possibility modals, but since modals of possibility also show a certain inclination to veridicality, this is arguably a difference of degree. The lower right (O) corner, finally, is not veridical and generally of lower functional load in language.
Internal and external negation So far, we have seen that causatives, like modals, are interdefinable by negation, and that the corners of the square of opposition, representing all combinations of wide and narrow scope negation, differ in their veridicality, that
38 Introduction, overview and theoretical framework is, by extension, in their informativity. Disregarding internal negation (which seems to be a dispreferred option at least in the languages covered here; see table 1 for a corpus count), the two squares may be simplified as follows, with veridical corners marked in bold: modals
causatives
A nec (p)
E ¬ pos (p)
A fact (p)
E ¬ perm (p)
I pos (p)
O ¬ nec (p)
I perm (p)
O ¬ fact (p)
Modals and causatives thus differ in their veridicality. Modals of possibility are not, generally, veridical, unless they are negated. Permissive causatives such as let are veridical regardless of polarity; however, items that encode the causation of possibility, such as to enable behave like modals of possibility. Modals of necessity, as well as factitives, behave alike: they are veridical only if affirmed. The relevance of the above for the analysis of da(va)t in Slavic is that these considerations supply an explanation for permissive markers as well as markers of possibility to have a certain inclination towards negative polarity. Whether or not this is reflected in actual usage frequency of a given marker will, however, rely on the overall system of causative and modal formants: I do not predict that all permissive or possibility markers are necessarily more prone to be used in negative contexts. However, I do predict that negated necessity or factitivity is overall less frequent than negated possibility or permissivity. A preliminary search in the Polish IPI PAN Sample for two grammaticalized predicative modals and two modal adjectives in combination with negation was employed to test this suggestion. The results for the predicatives (nie) mo˙zna (nie) ‘(not) poss (not)’ and (nie) trzeba (nie) ‘(not) nec (not)’ as well as the adjectives (nie)potrzebny ‘(un)necessary’ and (nie)mo˙zliwy ‘(im)possible’ are shown in table 1 and support the suggestion; the proportion of negated instances is higher with possibility than with necessity, most clearly with the predicative modals (p<.0001, V=.15). It also shows the marginality of internal negation with these modals.
Modals and causatives
mo˙zna (poss) trzeba (nec) mo˙zliwy (poss) potrzebny (nec)
M(p) 20,086 (84%) 13,153 (94%) 3,943 (77%) 3,727 (85%)
¬M(p) 3,824 (16%) 795 (6%) 1,152 (23%) 645 (15%)
M(¬p) 88 8 -
¬M(¬p) 24 0 -
39
all 24,022 13,956 5,095 4,372
Table 1. Corpus counts for Polish modal predicatives and adjectives
The above hierarchy of veridicality and informativity explains preferred interpretations for ambiguous markers. I shall give two examples. The German underspecified causative marker lassen encodes permissive and factitive causation alike, that is, all four corners of causation may be expressed by this formant. In terms of usage, however, negation serves to denote predominantly the E corner, rather than the O corner: summarizing findings from his extensive corpus study on the German causative auxiliary, Nedjalkov (1976: 50) states that “the percentage of negated phrases [. . . ] ist to a large extent directly proportional to the percentage of permissive phrases5 (my translation)”. This can be explained in the light of the above discussion: lassen, like other underspecified causative markers, mainly encodes three functions: factitive (the A corner), permissive (the I corner), and negated permissive (the E corner). A similar case in point is the modal infinitive construction in German, as in das Buch ist zu lesen, which can be interpreted as expressing either possibility or necessity. As a native speaker, my first interpretation of a negative instance like das Buch ist nicht zu lesen will be that of negated possibility: the book cannot be read (with the second option not excluded). However, I don’t have any data whether this intuition finds a parallel in actual usage.
1.4.3.
Conclusions
As a whole, I have suggested that permissive causatives are polarity sensitive items, i.e., they are informative under negation as well as when affirmed. Furthermore, based on the distinction to straightforward transitivizations of modals, we have seen that permissive causatives involve modal semantics, but that this is clearly not sufficient. Rather, they involve the effectuation of possibility as well as the effection of the caused situation. In the discussion so far, I have explored an analysis of causatives and their relation to modal operators from a logical standpoint, linking it to the notions
40 Introduction, overview and theoretical framework of veridicality and informativity. I have rendered plausible the notion that negated permissives (less than negated factitives) potentially play an important functional role in language by providing an explanation for their polarity sensitivity. While modals and causatives have much in common, a clear difference between predicates denoting cause to be possible and permissives like let has emerged. Both clearly involve modality, since their resultant state is that of possibility; however, permissives encode not only possibility, but also the realization of the predicated element. We may thus say that make possible and therefore happen seems to be lexicalized together in permissive causatives. Moreover, permissive causation may also express non-intervention: not make impossible and therefore make happen. These two are not independent: to make possible implies not to make impossible, but not vice versa. In the deontic domain, whether or not these are coextensive depends on questions of a general framework of norms (von Wright 1963). I take these two to be focal senses of permissive meaning and to be testable in a general way. If we can paraphrase a given utterance involving a permissive marker such that it can be said because the causer did something, the causee could do p, we are dealing with positive enablement. If we can paraphrase with because the causer did not do something, the causee could do p, we are dealing with non-intervention. Note that both are veridical: • nonintervention: perm (p) = ¬cause (¬poss (p)) and cause (p) • enablement: perm (p) = cause (poss (p)) and cause (p) These two focal senses will be distinguished in the following, empirical chapters and conclude the introduction of the conceptual framework that will used below.
Chapter 2 da(va)t+inf in Russian 2.1.
Introduction
In Russian I concentrate on the analysis of da(va)t+inf in causative function. The semantics of the causative function is analyzed in greater detail than is done concerning the other languages; this section thus provides background for the sections on Polish and Czech. The second important function of da(va)t+inf, the hortative, will not be investigated in detail except for where it is relevant to the permissive construction. This section is organized as follows. After adducing broad syntactic types I provide details on the material and method of the empirical analysis and give an account of prior research. The analysis of permissive da(va)t+inf in terms of domains of causation follows. I then turn to specific questions concerning the analysis of permissive constructions formed with the imperative daj as well as reflexive constructions, closing with an analysis of the residual types and a conclusion. 2.1.1.
Syntactic constructions
There are three basic syntactic constructions formed with da(va)t that are relevant to this study, only one of them pertaining to its focus on causative constructions. • In its causative use, da(va)t+inf forms what may be analyzed as an object control construction, where the dative object referent is also the understood subject referent of the infinitive6 . This is referred to as the permissive or object control construction. (22)
On dal emui otdoxnut’i / zakonˇcit’i He.NOM let.PST.SG him.DAT.SG relax.INF / finish.INF rabotu. work.ACC.SG ‘He let himi rest / finish hisi work’
This construction is also formed by a set of other verbs such as prosit’ ‘ask for’, pozvolit’ ‘allow’, velet’ ‘command’; they can be called dative control
42 da(va)t+inf in Russian verbs, since the matrix verb subcategorizes a dative phrase that specifies the notional subject of the complement clause. • The second person singular or plural imperative of da(va)t’, namely perfective daj(te) and imperfective davaj(te) are frequently used with a finite predicate. This is referred to as the finite clause construction, cf. (23)
Daj/davaj (ja) sdelaju / Maša sdelaet. Let(PF/IPF).IMP.2SG (I) do.PF.1SG / Maša.NOM do.PF.3SG ‘Daj/davaj I’ll do that / Maša will do that ’
(24)
Daj/davaj pošli / idi sjuda. Let(PF/IPF).IMP.2SG go.PST.PL / come.IMP.2SG here ‘Daj/davaj let’s go / come here!’
In this construction there is no clear hierarchical structural relationship with the finite predicate: daj/davaj is not strictly necessary for this construction and the subject pronoun may or may not be omitted. This construction is testimony to the general development of daj/davaj to pragmatic markers, similar to, e.g., English come on. Such a development, however, it is not specific to the imperative of da(va)t, which in (23) may be replaced by several other verbal forms such as the imperatives pozvol’ ‘permit’, razreši ‘allow’ or the question xoˇceš’ ‘do you want’ (Barentsen 2006). This is, however, a rather heterogeneous class: such a replacement is for example not possible in (24) which is also lexically more restricted. Daj and davaj have partly different functions in this construction; most notably, davaj, but not daj has developed to form a dedicated hortative construction (see next type). • Davaj(te) with infinitive: (25)
Davaj (*my/ #nam) pet’! Davaj we.NOM us.DAT sing.INF! ‘Let’s sing!’
This construction unambiguously expresses a hortative, that is, an imperative of the first person plural. It is not formed with daj or any other particle and takes an infinite complement. As a difference to the object control construction, no dative pronoun may be introduced without blatant change of meaning and change of reference of the agent.
Introduction
43
This construction enters a paradigmatic relationship with a specific subtype of the second construction. It is used only with imperfective verbs; its counterpart with perfective verbs is a finite clause construction, an option also given for imperfective verbs in the analytical future: (26)
Davajte pet’. — Davajte (my) budem pet’. Davajte sing(IPF).INF — davajte (we) AUX.1PL sing(IPF).INF ‘Let’s sing (imperfective aspect).’
(27)
Davajte spoem. Davajte sing(PF).1PL ‘Let’s sing (perfective aspect).’
There are further uses of the hortative construction that are neither straightforward first person plural imperatives nor seem to follow from nonimperative uses of imperative forms, cf. (28): (28)
Zasel s ruž’em okolo okna i davaj vo vse Sit.down.PST.SG with gun at window and davaj in all storony palit’. directions shoot.INF. ‘He sat down with his gun at the window and started to shoot in all directions.’i
Here, davaj plus infinitive (no plural davajte is possible here) expresses sudden action, hence the translation. Such uses are beyond the scope of this study.
2.1.2.
Material & Method
As material basis for the analysis of da(va)t in Russian, I took a snapshot of all instances of the verb followed by an infinitive in the Russian National Corpus on Jan. 12, 2006.7 The distance between da(va)t and the infinitive was constrained to a maximum of 5. The results were then filtered to exclude irrelevant instances8 and the remaining 9,639 hits were input into a database and categorized on a per-item level. 7 main types were distinguished: i Andrej
Žitkov. Supermarket (2000)
44 da(va)t+inf in Russian • Permissive construction, as in (29)
On ne dalPS T :S G emuDAT dogovorit’INF . ‘He did not let him finish speaking.’
• Factitive constructions with cognitive verbs, such as (30)
On dalPS T :S G emuDAT e˙ toACC ponjat’INF . ‘He let1 him2 understand4 that3 (i.e., told him).’
• Other infinitive constructions involving da(va)t, with either secondary predicates or infinitives used referentially (typically with consumptive verbs) (31)
On dalPS T :S G emuDAT pal’toACC nosit’INF / on dalPS T :S G emuDAT poest’INF . ‘He gave him the coat to wear / he gave him to eat.’
• Constructions with dano, the neuter form of the past passive participle, as in (32)
Ne danoPPP:S G:N emuDAT bylo sud’boj stat’INF xorošim pisatelem. ‘It was not given to him by fate to become a good writer.’
• 1. person plural imperative (hortative) with davaj(te), as in (33)
DavajteI MP:2PL rabotat’INF ! ‘Let’s work!’
• Various phraseologic types where the infinitive is dependent on da(va)t, as in (34)
Kuda skaˇceš’, ja teDAT dam1S G udirat’INF ! ‘Where are you going, I’ll teach you to run way (lit.: I’ll give2 you1 to run away3 ).’
• Double entries and other irrelevant hits of various types, such as ni dat’INF ni vzjat’INF ‘give or take’. The morphological analysis of da(va)t as well as aspect of the infinitive and negation were extracted from the initial corpus data. The resulting hits were further split in relation to the latest date given in the metadata of the document, yielding a subcorpus spanning from the 18th
Introduction
type permissive cognitive dano other inf. hortative phras. irrelevant
post-1945 abs rel 3,364 42% 1,334 17% 267 3% 757 9% 1,035 13% 264 3% 1,015 13% 8,036 100%
45
pre-1945 abs rel 695 43% 244 15% 23 1% 149 9% 166 10% 51 3% 273 17% 1,603 100%
Table 2. Breakdown into raw types for the older and newer subcorpus.
century to 1945 and a post-1945 subcorpus. Except as concerns cognitive causation, only the latter is used in this chapter. The quantitative breakdown into these types is given in table 2. These types were further subcategorized to different degrees, with the following annotation: • permissive and cognitive – animacy of matrix and complement subject, yielding four types – negation – morph. categories of the matrix verb – infinitive lexeme and aspect – reflexive construction ∗ type of reflexive (clause internal/external sebja, dative, instrumental, prepositional phrase) ∗ presence/absence of causee – unaccusative complement verbs were marked This extensive corpus makes a quantitative, usage-based analysis of semantic types and causative domains possible. Besides, it was compiled to be able to ensure that infrequent cases would also be captured and hypotheses could be validated/invalidated in a mixture of quantitative and qualitative approaches.
46 da(va)t+inf in Russian An evaluation set of 300 randomly chosen examples was further annotated semantically and used in specific comparisons with finer grained distinctions. Only here was the presence of an overt causee consistently marked and further qualified. This is referred to as evaluation set in the analysis.
2.1.3.
Prior research
There has been quite some research on causative constructions in Russian; works to be mentioned here include Lötzsch (1972); Babby (1983); Toops (1985, 1987, 1991); Paduˇceva (1994); Bulygina & Shmelev (1999); Rozina (2000). A recurrent topic have been contextual causatives, that is, zero marking of curative causation as discussed in the introduction. I will focus on two accounts specifically addressing the grammaticalization of da(va)t in Russian.
Toops’s analysis. Toops (1991) explicitly adresses the auxiliary status of da(va)t with infinitive. He discusses two causative functions: permissive, and factitive, rejecting the latter as only pragmatically inferred in sentences such as (35)
Byla telegrafistkoj, teper’ služu v Gorodskoj uprave i nenavižu, i preziraju vsë, cˇ to mne tol’ko dajut3PL delat’INF . (Toops 1991: 47) ‘I used to be a telegraph operator, now I work at the city administration and I hate and despise everything they give me to do.’ (my translation, rvw)
Toops shows that such usage of da(va)t is not evidence of causative factitive usage, but of ‘give’. He finds factitive causation to be (incompletely) grammaticalized with the prefixed cognate otdavat’ ‘give up, hand over’ as in (36), as well as in “contextual causatives” (Toops 1987) where causation is left implicit, cf. (37),: (36)
On otdal mašinku v remont. He give.up.PST.3SG machine.ACC to repair.ACC.SG ‘He had the machine repaired.’
(37)
On otremontiroval mašinu. He repair.PST.3SG car.ACC.SG ‘He repaired the car’ or ‘He had the car repaired.’
Introduction
47
Toops discusses several properties of permissive da(va)t, including the impossibility of complement negation (38), adduced already in Nedjalkov & Nikitina (1965) and the fact that it is veridical, carrying the implication that the caused event has taken places, as in (39): (38)
Otec pozvoljaet3S G /*daet3S G doˇceriDAT neNEG xodit’INF v školu. ‘The father allows / *lets the daughter not go to school.’
(39) *Otec dalPS T :3S G doˇceriDAT otdoxnut’INF , no ona ne otdyxala (Toops 1991: 42) ‘The father let the daughter rest, but she didn’t.’ Toops also notes the limits of productivity of da(va)t+inf in sentences such as (40) below and attributes this to a transitional phase in grammaticalization of this formant (40) ?Otec dalPS T :3S G doˇceriDAT postroit’INF novyj dom. intended: ‘The father let (his) daughter build a new house.’ He concludes that grammaticalization of permissive da(va)t is incomplete, at least partly conditioned by the competition to da(va)t in the transfer sense. Podlesskaya’s analysis The status of da(va)t+inf as a permissive auxiliary in Russian has been explicitly addressed in a series of articles by Vera Podlesskaya (2004, 2005b, a, 2006), who argues for far-reaching grammaticalization of this formant. Podlesskaya (2005) introduces the terms locutive for the speech act of permission, encompassing what is called deontic permissive in the present study, and favoritive for what is equivalent to dynamic causation in my terms. She points out that verb razrešat’ ‘permit’ reflects a speech act with the content that the causer does not have objections to what the causee does and that the verb davat’ in contrast expresses volitional action or refrainment of action in order to facilitate or not hamper the causee’s actions (p.126). Pozvolit’ stands between the two as it expresses both readings. Podlesskaya points out that the observation originally made by Nedjalkov & Nikitina (1965) that the complement verb of da(va)t cannot be independently negated also extends to pozvolit’ in dynamic reading: pozvolili ne govorit’ ‘allowed not to talk’ can only be understood as referring to a speech act, not to dynamic letting. Podlesskaya concludes from this that the impossibil-
48 da(va)t+inf in Russian ity of negating the complement verb should be taken as a characteristic of dynamic permissive meaning, rather than of grammaticalization of da(va)t. (Note here that in the corpus, a single token of double negation of both matrix and complement verb was found which was, however, acceptable not to all informants). She goes on to show (2005:127) that only razrešat’ and pozvoljat’ in deontic reading allow the impersonal passive, while davat’ and pozvoljat’ in dynamic reading do not: (41)
Po pravilam bezopasnosti imDAT ne bylo razrešenoPPP:S G:N / pozvolenoPPP:S G:N / *danoPPP:S G:N vyxodit’INF . ‘According to safety regulations, they were not permitted (allowed) to get out / *let get out.’
Podlesskaya (p.128) concludes that the impersonal passive thus shows different behavior in relation to the deontic and dynamic permissive with the latter “being more grammaticalized, i.e. being more limited in grammatical variation.” She further shows that inanimate causers as well as resumptive pronouns may form the causer only under negation: (42)
On mnogo zanimalsja i e˙ to pozvoliloPS T :3S G /*daloPS T :3S G emuDAT poluˇcit’INF diplom. ‘He studied a lot and that allowed / let him earn his diploma.’
(43)
MneDAT *(neNEG ) davaliPS T :3PL šumyNOM:PL na ulice spat’INF . The noise4 on the street did *(not2 ) let3 me1 sleep5 .
Podlesskaya gives a semantic interpretation: while the positive form denotes deliberate action, the negated form denotes the absence of such action; only the latter is therefore applicable to inanimate agents as well. Note that the above generalization does not hold up empirically: in section 2.2.4 I adduce several corpus examples with inanimate causers in affirmative sentences, which, however, should be counted marginal and semantically and stylistically awkward. More importantly, however, it is not clear why such a restriction should be a symptom of grammaticalization: grammaticalization is usually seen as involving the loosening of selectional restrictions, as for example with respect to inanimate causees in manipulative causation, as pointed out by Podlesskaya herself (see below); strong selectional restrictions would be an argument for more lexical status, i.e., wide scope.
Introduction
49
Lastly, Podlesskaya points out that unlike razrešat’ ‘permit’ and pozvolit’ ‘allow’, davat’ cannot be used as a performative verb: (44)
Ja (ne) razrešaju1S G / #daju1S G tebeDAT sjuda voitiINF ! I do not allow / *let you (to) come in here!
Summarizing these characteristics, Podlesskaya concludes that the constructions involving da(va)t are "more integrated: they are subject to restrictions in the scope of negation, in the animacy of the Agent, in passivisation. [. . . ] The symptoms of grammaticalization in these constructions can be viewed as resulting from their favorative [=dynamic permissive, RvW] meaning. (p.130)" In a later paper, Podlesskaya (2006: 317) additionally stresses that davat’, unlike pozvolit’, may take an inanimate causee, as in (45)
On dalPS T :3S G / *pozvolilPS T :3S G supuDAT zakipet’INF , i tol’ko togda posolil. ‘He let / *allowed the soup (to) start to boil, and only then added salt.’ (my transl., rvw)
In the present study, this type of use is subsumed under manipulative causation and may be seen as an instance of less restrictions on the causee, as such a clear symptom of grammaticalization. Podlesskaya thus shows that apart from the question of selectional restrictions with manipulative causation, there are in fact little structural grounds that independently argue for grammaticalization of da(va)t; rather, at the foreground of this analysis is the semantics of permissive causation, also shared by pozvolit’ in favorative (dynamic) reading. The present analysis, by and large, supports such an analysis; in the case of Polish da(va)t, as we will see, clearer structural symptoms of grammaticalization are found.
Hybrid constructions Podlesskaya further investigates structural symptoms of grammaticalization in constructions that allow both transfer and permissive meaning dubbed ‘hybrid constructions’, such as (46)
Ja dalaPS T :3S G ValeDAT svoe pal’toACC ponosit’INF Lit. ‘I gave [to] Valja my coat to wear’ (Podlesskaya 2005a: 131; glossing added)
In (46), the infinitive and da(va)t shares an object (pal’to). Podlesskaya strives to show that this construction shows ‘controversial syntactic behav-
50 da(va)t+inf in Russian ior’ (p.131), as the object cannot be attributed to either predicate alone. The claim that the behavior of the hybrid construction in fact reflects grammaticalization of da(va)t is of relevance for the present thesis, as the evolution of permissive da(va)t from a construction involving a secondary predicate, as in (47), is one of the hypotheses discussed in the concluding section. However, I argue the arguments Podlesskaya forwards do not point in this direction. Rather, they reflect characteristics of secondary predication with a purposedenoting infinitive, not characteristics of da(va)t. I will therefore review her points and comment on them in turn: Claiming this to be evidence that the patient in hybrid constructions does not meet ‘necessary requirements for objecthood’, Podlesskaya shows that in this configuration, da(va)t cannot be passivized and the patient cannot promoted: (47) *Pal’to byloAUX danoPPP:N ValeDAT ponosit’INF . ‘The coat was given [to] Valja to wear.’ However, this cannot be taken as evidence for the grammaticalization of the construction involving da(va)t, since other verbs show the same characteristics with secondary predicates, e.g. prinosit’ and peredavat’: (48)
a. b.
Ja Vale dala/ prinesla/ otdalaPS T :3S G pal’to ponosit’INF . I gave/ brought / handed over Valja my coat to wear. Pal’to bylo ??dano /??prineseno /??otdanoPPP:N Vale ponosit’INF . ‘The coat was given/brought/handed over to Valja to wear.’
As evidence of the hybrid status of the object of da(va)t in relation to the infinitive, Podlesskaya adduces an example from a child speech corpus where the object appears in accusative case, showing it to be governed by da(va)t rather than by the infinitive; on the other hand, it is positioned after the infinitive, the position typical for the patient of the infinitive: (49)
DajI MP mneDAT poigrat’INF mašinkuACC . lit.: ‘Give me to play [your] toy-car!’ (p.132)
While I do not take issue with this example, I contend it should not be taken to show special properties of da(va)t in this context. The same structure with other verbs seems no more and no less felicitous, as in (50)
PrinesiI MP:S G mneDAT poigrat’INF mašinkuACC . Bring me to play your toy-car!
Introduction
51
The issue in question is word order and perhaps a discontinuous constituent; it is otherwise expected behavior for secondary predicates to omit the object argument if present in the main clause. Informants prefer infinitive-final word order in both cases: (51)
Prinesi/dajI MP:S G mneDAT mašinkuACC poigrat’INF . ‘Bring / give me your toy-car (for me to) play (with it).’
As further evidence for idiosyncratic syntax, Podlesskaya shows that the object can be omitted in certain instances, as in (52)
DajI MP popit’ / poprobovat’ / poest’INF ! ‘Give (me) to drink / to try / to eat!’
Again, I take issue with this and analyze such constructions to dependent on special properties of consumptive and other complement verbs, not of da(va)t, as seen in sentences like (53)
Popit’INF est’COP / ostalos’PS T :3S G ? Is2 there anything (left3 ) to drink1 ?
Lastly, Podlesskaya claims that control relations in these constructions are exceptional. According to Testelec (2001), in structures like (54), the nonexpressed agent/subject of the complement clause is expected to be controlled by main clause participants according to the hierarchy Patient > Recipient > Agent. However, this is not the case here: (54)
Petja dalPS T :3S G mneDAT dvux pomošˇcnikovACC kontrol’nye proverjat’INF . ‘Peter gave me two assistants to check [the students’] tests.’
In (54), three interpretations are possible for the unexpressed complement subject: Either the speaker (the recipient), the assistents (the patient) or both are understood to check the tests; Podlesskaya explains this by agentive properties of the patient. While I do not take issue with this assertion, this phenomenon cannot be attributed to davat: replacement by prislat’ ‘send’ leads to identical control relations: (55)
Petja prislalPS T :3S G mneDAT dvux pomošˇcnikovACC kontrol’nye proverjat’INF . ‘Peter sent mei two assistants j to checki, j [the students’] tests.’
Summarizing, Podlesskaya opts for a hybrid nature of the constructions involving da(va)t with donatory object and purposive infinitive and asserts that they exhibit “high integrity”, which she views to be “a symptom of grammat-
52 da(va)t+inf in Russian domain Interpersonal Manipulative inan. causer / anim. causee both inanim.
n
rel.
2,423 421 398 122 3,364
72% 13% 12% 4% 100%
negated abs . rel. 1,508 62% 224 53% 369 93% 118 97% 2,219 66%
Table 3. Distribution of permissive causative da(va)t across domains and proportion of negated instances.
icalization”. However, the linguistic facts she adduces for this assessment in relation to da(va)t alone is questionable, since they also hold for other verbs, as I have shown. While the set of these verbs was not systematically evaluated, it seems to be the case that all of them belong to the class of verbs of giving and receiving, pointing to a more general, semantic basis of the phenomena described by Podlesskaya.
2.2.
Semasiological description of permissive da(va)t in Russian
In this subsection, I shall be concerned with the meaning and use of permissive da(va)t in Russian. Factors that influence the meaning include animacy, control, negation and mood. Table 3 shows an overview of the distribution of permissive causatives across domains in the corpus. As we can see, the main use of permissive da(va)t+inf is in the interpersonal domain; it is used in all domains, however. The table moreover shows that it is used predominantly with negation in all domains without being a Negative Polarity Item stricto sensu. This will be discussed in detail below. In the following analysis it will be shown that in some cases, qualitative difference of grammaticality obtain between positive and negative usage. For each domain, I discuss specific readings of the permissive, showing them to be linked to the domain in question as well as to negation. Only in this section relating to Russian will the focus thus be on semantics, and, correspondingly, only in this section will I make extensive use of explication in order to clarify these readings.
Semasiological description of permissive da(va)t in Russian
2.2.1.
53
Overview
Aspect use with da(va)t+inf involves two parameters: the aspect of the infinitive and the aspect of da(va)t itself. In the course of the present study, I compare the use of da(va)t+inf across languages and constructions and I show that there are differences in terms of aspect usage with da(va)t+inf in the various languages. However, aspect choice in infinitive constructions is a subject that is in general not well described (but see for example Fielder 1984). This subject would need to be approached from a broader point of view, involving a comparison of da(va)t+inf to several different matrix verbs. An additional complicating factor for the present study is that the aspectual systems in the three languages involved differ (see Dickey 2000). An exhaustive analysis of the use of aspect with da(va)t+inf in all languages examined is therefore not attempted in this study. With some exceptions, aspect choice in the use of da(va)t+inf will be examined only from a bird eye’s view, examining profiles of usage rather than individual cases. However, I would like to point out that, by and large, aspect choice of the aspect of da(va)t in permissive function in Russian seems to be subject to the general rules of the Russian aspectual system. As (56) illustrates, in canonical contexts such as the actual present or iterative/habitual, perfective dat’ may not be used, while the infinitive may take both aspects. (56)
On vsegda *dal / daval mne He always *let(PF).PST.SG / let(IPF).PST.SG me.DAT dogovorit’ / dogavarivat’. finish(PF).INF / finish(IPF).INF ‘He always let (*pf/ipf) me finish (pf/ipf).’
I assume that like its lexical counterpart denoting to give, dat’ in permissive function denotes a bounded event by lexical default, and will not further analyze this matter. As in the remainder of this study, detailed usage profiles of da(va)t and its infinitive as well as frequency lists of the verbs found in the corpus are presented. These usage profiles help single out differences between the languages, and their format is held consistent across them; they also serve documentational ends enabling the reader to evaluate claims made in the texts. Verb lists serve to document combinations of verbs and infinitives in the corpus, allowing the reader to assess their variety not only of the most frequent combinations; secondly, they allow an evaluation of productivity. As
54 da(va)t+inf in Russian pointed out in the introduction, productive formations may be distinguished from non-productive, lexicalized formations from a corpus-based perspective by evaluating the proportion of seldom used combinations to the overall use of the formation: the higher that proportion is, the higher the productivity of such formations. In this section, I will first focus on a semasiological analysis of permissive da(va)t+inf in Russian on a per-domain basis and single out different focal senses obtaining. This is done for positive and negative polarity separately. I then turn to specific questions, namely the use of the imperative form daj; in section 2.2.6.2 I discuss the use of the reflexive permissive construction, important for the comparison to the other languages, where it has further grammaticalized into modals. After that, cognitive causation with da(va)t is looked into in more detail before covering residual types and reaching my conclusion regarding the grammaticalization of da(va)t+inf in Russian. 2.2.2. 2.2.2.1.
Interpersonal causation Profiles of usage
In the interpersonal domain, both causer and causee are animate and agentive; both act intentionally and possess control. Tables 4 to 5 give overviews of various combinations of aspect, negation as well as frequency of forms in interpersonal permissive causation, that is, the most frequent type of permissive da(va)t. Table 4 offers an overviews over the morphological forms of da(va)t used in the corpus and their respective use in (a) negative contexts and (b) with perfective complement infinitives. It shows that negative polarity dominates with both aspects of da(va)t and all forms except the perfective imperative and infinitive. Note that throughout this study, numbers for polarity take only overt negation of da(va)t itself into account; the polarity of superordinate predicates such as modals or functors involving covert negation, such as adverbials like vrjad li ‘hardly’, is disregarded. Since infinitives occur typically with such superordinate predicates, the figures concerning the polarity of the infinitive are skewed toward of positive polarity. I therefore consider the low number of negated infinitives an artefact of my approach.
Semasiological description of permissive da(va)t in Russian
forms infinitive participle imperative morph. present past all w/o inf, imp
count 274 113 356 245 529 1,517 887
perfective dat’ neg. perf. compl. 131 (48%) 258 (94%) 91 (81%) 111 (98%) 34 (10%) 326 (92%) 161 (66%) 192 (78%) 356 (67%) 487 (92%) 773 (51%) 1,374 (91%) 608 (69%) =65% of all
count 32 187 31 312 266 828 765
55
imperfective davat’ neg. perf. compl. 25 (78%) 13 (41%) 158 (84%) 156 (83%) 26 (84%) 11 (35%) 261 (84%) 175 (56%) 217 (82%) 124 (47%) 687 (83%) 479 (58%) 636 (83%) =35% of all
500
67% pos neg
300
10% 84%
48%
82%
66% 84%
100
81% 78%
84%
0
absolute number of tokens
Forms and polarity
inf.
part. imper. non−past perfective DAT
past
inf.
part. imper. non−past imperfective DAVAT
past
500
92% ipf pf
300
92% 56%
94%
47%
78% 83%
100
98% 41%
35%
0
absolute number of tokens
Forms and complement aspect
inf.
part. imper. non−past perfective DAT
past
inf.
part. imper. non−past imperfective DAVAT
past
Table 4. Forms of interpersonal permissive da(va)t and negation and complement aspect. In general, negation dominates, uniformly so across imperfective forms of da(va)t (p = .86), but not across forms of dat’ (p < 0.0005), where infinitives and imperatives deviate. Complement aspect is more varied, but generally correlates with matrix aspect (see also next table).
56 da(va)t+inf in Russian
matrix perf perf imperf. imperf.
complement perf imperf. perf imperf.
abs 1,374 143 479 349 2,345
rel 59% 6% 20% 15% 100%
of which pos neg 682 (50%) 692 (50%) 62 (43%) 81 (57%) 83 (17%) 396 (83%) 58 (17%) 291 (83%) 885 (38%) 1,460 (62%)
DAVAT Aspect dat 81
ipf
davat
62
ipf pf Infinitive Aspect
682
pf
neg
291
692
Polarity
396
pos
58
83
Table 5. Matrix and complement aspect and negation, combined view. The mosaic plot clearly shows the association of imperfective aspect and polarity (right and left half of upper and lower part, respectively) and the covariance of matrix and complement aspect (horizontal split in the four quarters).
Semasiological description of permissive da(va)t in Russian
57
However, figures for the perfective imperative are not likely to be distorted. This form is exceptional in two respects: first, in contrast to the other forms, the imperative is not linked to negative polarity: it clearly departs from the pattern of the other forms (excluding infinitives) in respect to negation (same distribution with p < 0.0005 and an effect size of Cramer’s V = 0.5). Second, it is very frequent, with 356 of 1,517 amounting to 23% of all perfective instances and to 15% of all interpersonal instances. Note that this is remarkable also in comparison to the Polish construction, where the imperative is not as frequent and patterns with the other forms. Use of the perfective imperative will therefore be analyzed in detail in section 2.2.6.1. As concerns aspect of da(va)t, the perfective form is more frequently used than its imperfective counterpart (65 to 35%). The examination of the combinations of tense and aspect shows that perfective dat’ is more frequent in the past tense than in non-past with 529 to 245 cases, while the imperfective is more often used in the non-past with 261 to 312 cases (p<.001; Cramer’s V =.22). This can be expected from the basic functions of these categories9 . Imperfective aspect is linked to negation, as table 5 and the mosaic plot given there show. Imperfective aspect in positive sentences is rare (17% vs. ca. 50% of negative instances; cf. upper and lower halfs of the mosaic plot; p value for same distribution <.00001; Cramers’s V =.32). Since numbers for negation are skewed to be too low, this correlation ist really even stronger. Such a link of imperfective aspect and negation is expected for predicates that have a perfective default, such as telic verbs, in view of the observation that negation is linked to temporal delocalization of a proposition and that delocalized propositions are generally expressed by imperfective aspect (see e.g. V.Lehmann 1999; Bogusławski 2000). The table shows that the complement infinitive is most often perfective (79% of all cases). Closer inspection shows that there is a clear tendency of matrix and complement aspect to agree: if da(va)t is perfective, so is the infinitive, and vice versa (p=.18, V=.03 for same distribution of complement aspect in negative and positive instances with perfective da(va)t and p=.86, V=0.01 for imperfective da(va)t; compare the composition of the quarters of the mosaic plot for a visualization). The discussion so far has served to show two points, which gain more significance later in comparison. The first point is that there is a clear domination of perfective matrix and complement aspect; we will see that this applies to other Slavic languages to even greater degrees. The second point is that aspect assignment of complement and matrix predicate seems to be variable and
58 da(va)t+inf in Russian in line with general usage, rather than grammatically fixed or idiosyncratic; we will see that this variability is given to a lesser degree in other Slavic languages and in the modal passive function of da(va)t+inf. Verbs used in the corpus In list 1 all complement verbs with a frequency above 10 are given for reference. Since over 900 verbs are involved, no further categorization will be presented. The main point that this list shows is that of productivity: there is a pyramid of forms, with little verbs of high frequency, and gradually more verbs as frequency decreases. Three verbs have frequency over 50 occurrences; about 600 verbs appear once. Some of these combinations are collocations and it might be thought that specific, frequent collocations are responsible for the high overall proportion of negative instances. However, a closer inspection of those verbs that occur only once reveals that about two thirds of them, that is, 396 of 605, are used under negation. This indicates that negation is linked to the use of da(va)t, not to the use of individual collocations, adding to the analysis given above. n=68: opomnit’sja ‘come to one’s senses’; n=61: dogovorit’ ‘finish speaking’; n=59: spat’ ‘sleep’; n=49: skazat’ ‘say’; n=43: ujti ‘leave’; n=42: žit’ ‘live’; n=41: otdoxnut’ ‘relax’; n=37: rabotat’ ‘work’; n=28: umeret’ ‘kill’; n=27: sdelat’ ‘do’, zakonˇcit’ ‘finish’; n=24: govorit’ ‘speak’; n=23: projti ‘pass (intr.)’; n=22: vyskazat’sja ‘have one’s say’, pogovorit’ ‘speak a bit’; n=20: propast’ ‘disappear’; n=19: podumat’ ‘think a bit’, pogibnut’ ‘perish’, porulit’; n=18: raskryt’ ‘expose’; n=17: pospat’ ‘sleep a bit’, delat’ ‘do’; n=15: zabyt’ ‘forget’, razvernut’sja ‘unroll’; n=14: upast’ ‘fall over’; n=13: posmotret’ ‘look at’, pomeret’ ‘die’, otkryt’ ‘open’; n=12: podnjat’sja ‘rise’, peredoxnut’ ‘take rest’; n=11: vzdoxnut’ ‘take breath’, vyspat’sja ‘have a good sleep’, otvetit’ ‘answer’; n=10: sosredotoˇcit’sja ‘concentrate’, snjat’ ‘take off’, zasnut’ ‘fall asleep’, vtjanut’ ‘pull into’, otdyšat’sja ‘recover one’s breath’ , vstavit’ ‘put into’, vyrvat’sja ‘escape’, vygovorit’sja ‘have one’s say’. 145 verbs with frequency >2; 127 with frequency =2; 605 hapax legomena verbs. Verb list 1: Verbs in interpersonal causation, ordered by frequency.
Semasiological description of permissive da(va)t in Russian
2.2.2.2.
59
Semantic description
I distinguish four semantically related senses of permissive da(va)t which arise in context. I will develop these senses in explications, examining specific types according to domains and for positive and negative polarity separately, before putting these senses into a common perspective. The first two senses belong into the dynamic domain: causing by not interfering vs. causing by actively supplying a possibility, by enabling. The third, more peripheral meaning is the extension of dynamic enablement in the deontic domain: causing by giving consent. The fourth meaning, where da(va)t denotes simple waiting rather than causation is clearly an extension based on the first.
Non-intervention The clearly most central and, according to my corpus analysis, most frequently encountered positive use of permissive da(va)t is non-intervention, exemplified in the following: i
(57)
- Možet, dadite2PL rasskazat’INF , kak bylo delo? ‘Perhaps you’ll let (me) tell you, what happened?’
(58)
Inymi slovami, "‘SOK"’ daet3S G ljudjamDAT rabotat’INF tak, kak ii oni privykli, vnosja liš’ neznaˇcitel’nye korrektivy. ‘In other words, "‘SOK"’ lets people work the way they are used to, introducing only minor corrections.’
(59)
Vyryvat’sja-to on vyryvalsja, no, vyrvavšis’, nikuda ne bežal, a kak by zastyval i davalPS T :S G policejskimDAT sebjaRFX:ACC sxvatit’INF . iii ‘As concerns breaking free, he did break free, but he then did not run away, but so to speak froze and let1 the police2 catch4 him3 .’
(60)
Sporit’ ne ljubil, vsegda gotov byl zamolˇcat’ i dat’INF iv vyskazat’sjaINF drugomuDAT . ‘He didn’t like to fight, was always ready to fall silent and let1 the other3 have his say2 .’
i Dar’ja
Doncova. Uxa iz zolotoj rybki (2004) ii Gleb Stoljarov. Po zakonam voennogo vremeni // Delo, 2002 iii Vladimir Orlov. Al’tist Danilov (1980) iv Andrej Sedyx. M. A. Aldanov (1957)
60 da(va)t+inf in Russian I give an explication of this sense below. In this and the following explications, X refers to the causer, Y to the causee and p to the proposition that is permitted / caused and whose main participant is Y. Expl. 1: non-intervention (interpersonal) a. presupposition: (i) Y wants and intends to do p. (ii) There is a set of possible actions a such that (iii) If X does a, Y cannot, and therefore will not, do pp. (iv) X knows about the above. b. assertion (i) X does not do some a that would make p impossible. (ii) Y (does) p. Firstly, note that not only could X hinder Y from doing p, his not doing so is sufficient for Y to do p: Y’s action is highlighted as the decisive precondition of X(p). The explication above stresses negative refrainment from an action on the side of the causer to be the necessary condition for Y to do p. The causer’s action referred to is negative, but it is more or less specific in the context what he could do. The causer is passive, but acts intentionally. We can say in these cases: Y can do p because Y does not do something. This can be used in a test: An adequate paraphrase of the relevant part of (60) may be other’s had their say, because he fell silent, didn’t say anything.
Enablement A similar paraphrase is clearly not appropriate in the following example, where the causer provides a possibility by doing something actively: (61)
Otpravljat’ rebenka na otdyx za rubež nemnožko bojazno, zato e˙ to otliˇcnyj sposob dat’INF emuDAT uvidet’INF mirACC , poˇcuvstvi ovat’INF sebjaRFX:ACC vzroslymINS . ‘To send a child abroad for vacation is a bit scary, but on the other side it’s a perfect way of letting her see the world, feel (lit. herself) grown-up. ’
i Rebenok
i leto: 5 vozmožnostej otdoxnut’ // Semejnyj doktor, 2002
Semasiological description of permissive da(va)t in Russian
61
Expl. 2: dynamic enablement (interpersonal) a. presupposition: (i) Y wants and intends to do p / doing p is in his/her perceived interest. (ii) There is some a such that if X does a, Y can do p. (iii) X knows about the above. b. assertion (i) X does some a that makes p possible. (ii) Y does p. Examples include: (62)
[. . . ] - cˇ toby uˇcenikamDAT davaliINF pol’zovat’sjaINF komp’juteri amiINS , a ne pokazyvali ix na kartinkax. ‘[. . . ] - so that they would let2 students1 use3 computers4 (=give them the actual possibility), rather than show them to them on photographs.’
(63)
Ego vzjali v sbornuju, davaliPS T :PL igrat’INF - i on ne podvel koii mandu. ‘They took him into the team, let (him) play - and he did not disappoint them.’
(64)
Kogda mne bylo 7 let, otec s gordost’ju [. . . ] sažal menja k sebe na koleni i za gorodom davalPS T :S G mneDAT nemnogo porulit’INF . iii ‘When I was 7 years old, my father proudly sat me on on his lap and, out of town, let me drive (the car) a bit.’
(65)
Kogda letela k vam, mneDAT daliPS T :PL postreljat’INF iz pulemeta iv ‘While flying here, people let2 me1 shoot3 with the machine gun.’
Note that, as far as refrainment from action can itself be an action if it has consequences, non-intervention is a specific type of dynamic enablement and can always be understood as the latter. The above definitions make clear that the difference between these two focal meanings lies in the activity of the causer: we can say for enablement: Y can do p because X does something. To paraphrase (65) as ja mog postreljat’ iz pulemeta potomu cˇ to oni mne ne mešali ‘I could shoot because they did not hinder me’ would not be appropriate in this context; rather, what is expressed is ja smog postreljat’ iz pulemeta i Škol’nikov naˇ cinajut puskat’ v internet so vtorogo klassa // Veˇcernjaja Moskva, ii Šaxmatnoe obozrenie, 2004 iii Bogatej (Saratov), 2003.04.10, 2003 2002.08.08 iv Aleksieviˇ c. Cinkovye mal’ˇciki (1984-1994)
62 da(va)t+inf in Russian potomu cˇ to oni soglasilis’, pustili k pulemetu, pokazyvali kak ‘I could shoot because they agreed to it, allowed me to approach the weapon, instructed me how to use it’. As the last example shows, in both senses the type of action referred to may also belong to the deontic domain, extensionally involving an act of giving permission or not forbidding. However, this usage is rather marked and is only present where the dynamic and deontic domain overlap in such a way that the de facto possibility of doing the caused action is dependent on the causer’s consent. Whatever is permitted is construed to be impossible to do without the consent of the causer, as in (65) or in (66): (66)
Vy znali cˇ etko, cˇ to, možet byt’, raz v tri goda vamDAT dadut3PL i s’ezdit’INF v socstranu [. . . ] ‘You knew very well that, perhaps, they would let2 you1 visit3 a socialist country every three years.’
It is not appropriate to use da(va)t in instances of everyday, expected permission where something might be granted or not without much ado. A situation where permission is seen more or less as a formality will rather not be referred to using da(va)t. A situation in which, for example, a child is allowed to play with other children can well be expressed using let in English: The mother let her play with the neighbors’ children. The corresponding phrase using da(va)t (67)
Mat’ dalaPS T :S G ejDAT poigrat’INF s det’mi iz sosednego doma. ‘Mom let her play with the neighbors’ children.’
is quite different in the relevant sense, since it presupposes that the child tried very hard to convince and finally succeeded; that this permission is usually not given; that there is such a relationship that the child can play with these children only in case permission is given. In the ordinary, every-day situation referred to here pozvoljat’ / ‘allow’ razrešat’ ‘permit’ is preferred as lacking these specifications. In the deontic domain, da(va)t is therefore quite specific and marked. An attempt at an explication of this usage is given below: i Vasilenko,
Poˇcinok. // Oteˇcestvennye zapiski, 2003
Semasiological description of permissive da(va)t in Russian
63
Expl. 3: deontic enablement (interpersonal) a. presupposition: (i) Y wants to do p. (ii) Y is dependent on X, and X might be expected to forbid / not allow Y to do p. (iii) Y cannot do p if X does not allow it / if X forbids it. (iv) X knows about the above. b. assertion (i) X allows / does not forbid Y to do p. (ii) Y does p.
Waiting The fourth use, waiting, foregrounds deliberate inaction and backgrounds the element of permissive causation; it can therefore be called post-causative. In (68) and (69), it is hardly appropriate to say that the most salient reason the causee could do p is the inaction of the subject; rather, the subject of da(va)t wants the causee to p undisturbed and to this end delays some action. (68)
Okazyvaetsja, vse ravno proverjali, tol’ko inaˇce: davaliPS T :PL i priblizit’sjaINF k liftu i tut proiznosili: ‘It turns out, they checked anyway, but differently: they let( them) approach the lift and there they said:’
(69)
Ot zabora ja ne otlipal, cˇ toby ne obnaružit’ sebja, kogda, ispuganno povernuvšis’, ona gljadela v temnotu. Ja xotel dat’INF ejDAT ujtiINF za ugol. [. . . ] Kogda ona došla do ugla, ja ot zabora otdelilsja [. . . ] ii ‘I kept close to the fence, so as not to reveal myself, when she looked into the darkness. I wanted to let her go around the corner. [. . . ] When she reached the corner, I came away from the fence [. . . ].’
In (69), there is no causal relationship between the causee turning the corner and the causer’s movement: we cannot say that because the ‘causer’ didn’t interfere or made it possible, the ‘causee’ could reach the corner. Rather, what is highlighted is the deliberate inactivity of the ‘causer’ (usually directed at preparing the ground for some subsequent action) and the temporal relationship of the causer’s action and the causee’s action. When used in this i Irina Murav’eva. Dokumental’nye s’emki (1997-1998) tunnelja (1994)
ii Jurij
Nagibin. T’ma v konce
64 da(va)t+inf in Russian way, such a clause is typically part of a chain of events, e.g. constructed with coordinated clauses, converb constructions or subordinate clauses. Expl. 4: waiting (interpersonal) a. presupposition: (i) Y about to do / finish p. (ii) X is about to do some other action a. (iii) If Y does a at this point of time, Y will not have done p. b. assertion (i) X does not do a before Y has finished p (he refrains from action, does not interrupt). (ii) Y (does) p. This type is encountered only with telic or perfective verbs, since it presupposes some internal boundary to be reached by the complement subject. This function can be considered an instance of semantic bleaching, where a part of the original causation meaning is extended to other functions. Summary The main divide between the senses isolated above runs between active and inactive subject. The difference in meaning relevant here is captured by the notions of enabling and not intervening. The dividing line can be found to lie in the relevance of the presence of the causer: while in non-intervention, the absence of the causer would not make a difference, with enablement, the causer (and whatever he does) is seen as crucial, so that in the causer’s absence, the causee could not proceed with Y. From this perspective, non-intervention and waiting belong together. The four senses isolated do not exclude each other; da(va)t has to be considered underspecified: in (70), for example, all are applicable: (70)
[...] poˇcti ežednevno pojavljalsja jastreb [..., kotoryj ...] podxvatyval ˙ bylo zaxvatyvajušˇcee zrelišˇce, ja inogda naroˇcno cyplënka [...]. Eto davalPS T :S G emuDAT ujtiINF i tol’ko togda kriˇcal dlja oˇcistki sovestii ‘Almost every day a hawk appeared [. . . ] that captured a chicken [. . . ]. It was a fascinating spectacle, sometimes I let it get away on purpose and only then, to ease my conscience, started shouting.’
i Fazil’
Iskander. Petux (1962)
Semasiological description of permissive da(va)t in Russian
65
In this example, the causer does not interfere with the bird getting away, thereby giving him the dynamic (perhaps even deontic) possibility, which at the same time involves waiting in inaction before proceeding with something else. Cases like these are quite characteristic of davat’.
2.2.2.3.
Polarity related meanings: negative causation
Negated interpersonal is by far the single largest attested group in the corpus, comprising over 40% of all occurrences of permissive da(va)t across all domains. This resonates with findings in Wierzbicka (2002), von Waldenfels (2004, forthc.), and with Topoli´nska (1993), who prefers to define Polish da(va)t in negative form and claims that its use presupposes some negating element to be present in the context. As we will see in the section on Polish, this claim cannot be upheld; however, in all three languages we observe a high frequency of negated instances. A clue to its high frequency is that the negated permissive is semantically positive in the sense that it involves an action on the part of the causer that makes something the causee is doing or planning to do impossible. Besides by not let, this can be expressed by a wealth of constructions in English such as keep from doing something, stop from doing, etc. The following two translations from J.K. Rowling’s Harry Potter, taken from ParaSol (see chapter 5) illustrate this: (71)
a. b.
(72)
a. b.
So Dobby stopped us from getting on the train and broke your arm[.] Znaˇcit , e˙ to Dobbi neNEG dalPS T :S G namDAT sest’INF v poezd i slomal tebe ruku ... Literally: ‘it was Dobby who did not let us take the train and broke your arm ’ I have to fly around our hoops and stop the other team from scoring. Ja dolžen letat’ vokrug svoix kolec i neNEG davat’INF drugojDAT komandeDAT zabit’INF tuda mjaˇc. Literally: ‘I have to ... not let the other team score a goal’
It seems that da(va)t in this case is in fact the only possible adequate rendition of stop from. A full English-Russian comparison would be beyond the scope of this study. The following explication represents this function:
66 da(va)t+inf in Russian Expl. 5: Intervention (interpersonal) a. presupposition: (i) Y wants and intends to do p. (ii) There is a set of possible actions a such that (iii) If X does a, Y cannot do p. (iv) X knows about the above. b. assertion (i) X does some a. (ii) Y cannot do p. Expl. 5 represents the negated counterpart to non-intervention, however, it is simpler. While in the positive version, the realization of the caused event needs to be specified explicitly, in the negative version, this follows from impossibility itself. The negative equivalent to Y will do p becomes redundant. Since the causer actively does something to interfere with the causee’s action, an appropriate paraphrase is: Because X did something, Y could not do p. Examples include: (73)
- Ub’et, oj, ub’et, - nadryvalas’ babka, neNEG davajaCV B nikomuDAT i vstavit’INF slovo. ‘He’ll kill her, oh, he’ll kill her - the old lady shouted, not letting anybody insert a word.’
(74)
Den’ roždenija cˇ elovekuDAT neNEG dadut3S G spokojno otprazdnovat’INF , - dosadlivo proiznesla Aleksandra Prokof’evna. ii ‘They won’t2 let3 a person1 celebrate4 their birthday in peace, Aleksandra Prokofjevna sadly pronounced.’
The negated version of enablement is hardly attested. In (75), for example, the paraphrase because the causer does not do something, Y can’t do X could be constructed to hold; however, the active version because the causer does something, Y can’t do X is always also possible and seems more appropriate: (75)
Kartin emuDAT snimat’INF neNEG davaliPS T :PL , daže korotkometražnyx, izredka razrešali montirovat’ kinožurnal "Novosti dnja". iii ‘They didn’t3 let4 him1 make2 any movies, not even short films; only seldom did they allow him to edit "news of the day"’
(76)
NamDAT daže neNEG daliPS T :PL oznakomit’sjaINF s proektom programmy, - govorit zam predsedatelja komiteta Gosdumy po agrarnoj
i Vlada Valeeva. Skoraja pomošˇ ii Maksim Milovanov. Estestvennyj otbor c’ (2002) (2000) iii Vasilij Katanjan. Prikosnovenie k idolam (1998)
Semasiological description of permissive da(va)t in Russian
67
iv politike Aleksandr Fomin. ‘They did not2 even let3 us1 familiarize4 ourselves with the draft program - says the deputy chairman of the Duma’s committee on agricultural policy, Aleksandr Fomin.’
(77)
Ix ne peˇcatajut, imDAT neNEG dajut3PL snimat’INF i vystavljat’sjaINF ; a možet, prošˇce: oni uže davno niˇcego ne i pišut, ne snimajut i ne vystavljajut. ‘They (=authorities) don’t print their works, don’t2 let3 them1 shoot4 movies or make5 exhibitions; or perhaps, it’s simpler: they (=the artists) stopped writing, shooting films and making exhibitions long ago.’
In (78), it is explicitly stated that the causer did something actively, although its constructed positive version clearly could be understood in terms of deontic or dynamic enablement. (78)
Soedinennymi usilijami universitetskix fizikov oˇcen’ staroj i samoj molodoj gvardii TammuDAT neNEG daliPS T :PL vernut’sjaINF na kafedru teoretiˇceskoj fiziki, kotoruju on vozglavljal do e˙ vakuacii. ii ‘With the combined efforts of the physicists of the old and of the very youngest generation they didn’t2 let3 Tamm1 return4 to the chair for theoretical physics, which he had held during the evacuation.’
(78)’ TammuDAT daliPS T :PL vernut’sjaINF na kafedru teoretiˇceskoj fiziki [. . . ]. ‘They let2 Tamm1 return3 to the chair for theoretical physics [. . . ].’ Likewise, (79), an example adapted from an example given for enablement on page 60, takes an intervention reading under negation: (79)
a.
b.
RebenkuDAT daliPS T :PL uvidet’INF mir, poˇcuvstvovat’INF sebja vzroslym. →enablement ‘They let2 the child1 see3 the world, feel4 grown-up.’ RebenkuDAT neNEG daliPS T :PL uvidet’INF mir, poˇcuvstvovat’INF sebja vzroslym. →intervention ‘They didn’t2 let3 the child1 see4 the world, feel5 grown-up.’
Candidates for negated waiting can be found. In these cases, the causer does not wait for the causee to do what he intends to do, but interrupts. In general iv Marina Smovž. Krestovyj poxod otkladyvaetsja. Programma zernovyx intervencij do six i Edvard ˙ ne utverždena // "Izvestija", 2002.10.02 Radzinskij. Naš Dekameron (1980-1990) ii Kapica, Tamm, Semenov: sbornik vospominanij i materialov (1998)
68 da(va)t+inf in Russian it is not possible to draw a clear line to intervention, since in all these cases it is possible to paraphrase with because the causer did something, the causee couldn’t do p. Examples include: (80)
[. . . ] i tut že, vdoxnovlennyj, neNEG davCV B ZabelinuDAT iii vozrazit’INF , prinjalsja stavit’ zadaˇcu: ‘[. . . ] and, inspired by this, he immediately, not letting Zabelin object, started setting the problem:’
(81)
NeNEG davCV B emuDAT daže pozdorovat’sjaINF , Tregubec zakriˇcal i v trubku: ‘Not even letting him (=leaving him time to) say a greeting, Tregubez shouted into the telefone:’
Conclusions: negated permissive Summarizing, instances of negated enablement or waiting are rare. Generally, it seems that if positive corpus examples are negated, their interpretation changes to intervention. Intervention thus denotes positive action, that is, it refers, like an interpretative verb, to some concrete undertaking by the causer. As such it is informative. This seems to be the default reading of negated permissive davat’. Restrictions on positive polarity In some cases, negated instances may not be changed to positive polarity without loss of grammaticality. In this section, we turn to cases where this restriction is systematic, namely to uses of permissive da(va)t with animate causees that lack control over the complement event. This type is situated between manipulative causation and interpersonal causation: on the one hand, the causee is animate, on the other, it does not have control. Negated da(va)t is used freely in this context. Consider (82), where the causee does not act as an agent; zamerznut’ ‘to freeze to death’ is unaccusative: (82)
Bezdomnyx, naprimer, e˙ vakuirujut s ulic do nastuplenija xolodov v special’nye prijuty, tem samym neNEG davajaCV B imDAT ii zamerznut’INF .
iii Vsevolod
i Petr Galickij. Opasnaja kollekcija (2000) ii Marija Danilov. Bank (2000) Beloklokova, Irina Podlesova. Ne moroz’ menja. Smert’ ot pereoxlaždenija - otliˇcitel’naja cˇ erta megapolisov // Izvestija, 2001.12.19
Semasiological description of permissive da(va)t in Russian
69
‘They take the homeless, for example, from the streets into special shelters before the night cold sets in, thereby not leting them freeze to death.’ However, its positive counterpart is, to say the least, strange (unlike its English translation): (83) ?Bezdomnyx ne e˙ vakuirujut s ulic v special’nye prijuty, tem samym davajaCV B imDAT zamerznut’INF . ‘They do not evacuate the homeless into special shelters, granting them a death of cold (lit.: letting them freeze to death.)’ The intensional content that renders this sentence semantically strange is twofold: firstly, the causer (some government authority) is said to act intentionally, purposefully: this collides with world knowledge telling us that what is referred to is indifference, not homicide. Secondly, crucially, it collides with the part of the explication of non-intervention on page 59 that states that the causee wants and intends to do p: the homeless clearly do not wish to freeze to death, but this is part of the intensional content of da(va)t in this construction. Accordingly, no difficulties of interpretation arise with non-agentive verbs if they denote something that is in the interest of the causee: (84)
On [=Stalin] ne tronul i Stanislavskogo, davCV B emuDAT umeret’INF i svoej smert’ju. ‘He [=Stalin] didn’t touch Stanislavski, letting him die in peace (lit.: die his own death).’
The intensional content of non-intervention in (84) is fully compatible with conceptual and situational knowledge: obviously, it is in the interest of Stanislavsky to die ‘his own’, undisturbed death. The sentence becomes inadequate if the complement is reduced to umeret’ ‘die’ alone, since it is hardly in anybody’s interest to die: (85) ?Stalin ne tronul Stanislavskogo, davCV B emuDAT umeret’INF . Stalin did not touch Stanislavski, letting him die. Again, such an example is fully acceptable under negation: Stalin ne dal Stanislavskomu umeret’ ‘Stalin didn’t let Stanislavski die’ is felicitous and denotes that Stalin did something that prevented Stanislavsky’s death, which would have been imminent had he not done something; what exactly, is not specified by this sentence. i Georgij
Burkov. Xronika serdca (1990)
70 da(va)t+inf in Russian Such negative use of da(va)t with unaccusative verbs is most often met where the infinitive denotes something that is undesirable to the causee, as illustrated in (86) and (87) below; however, this is not necessarily the case, as the following examples (88) and (89) show: ii
(86)
Klementi neNEG davalPS T :S G FilduDAT obrastiINF žirkom. ‘Klementi didn’t let Fild get fat.’
(87)
[. . . ] sejˇcas postavlena zadaˇca prervat’ e˙ pidemiologiˇceskuju cepoˇcku, ne dopustit’ vtoriˇcnogo zaraženija, to est’ neNEG dat’INF komu-toDAT i zarazit’sjaINF [. . . ]. ‘[. . . ] now the task ist to break the epidemiological chain and not to permit secondary infection, i.e., not to let anybody contract (the disease) [. . . ].’
(88)
[. . . ] - o mužickom proisxoždenii mneDAT neNEG davaliPS T :PL ii zabyvat’INF okružajušˇcie. ‘[. . . ] the others didn’t2 let3 me1 forget4 my lowly descent’
(89)
Dalee, dlja raznoobrazija, možno obvinit’ instruktora v tom, cˇ to on iii zažimaet informaciju i e˙ tim neNEG daet3S G vamDAT rastiINF . ‘Aside from that, just for the change, you can accuse the instructor of holding back information and thereby not letting you grow / keeping you from growing.’
Generally, negated da(va)t meaning intervention may thus be used irrespective of whether the causee wants to do p (expressed by the infinitive). I conclude that while the explication of non-intervention correctly involves the statement that the causee wants to do p, this element should not be present in the explication of intervention. This leads us to the following, modified explication of intervention: Expl. 6: Intervention (interpersonal, modified) a. presupposition: (i) Y is about to p (for some independent reason). (ii) If X does something, Y will not p (at this point of time). (iii) If X does not do this, Y will p. (iv) X knows about the above ii Jurij
i Boris Bronštejn. Kazan’ vystoit. V stoDavydov. Žemˇcužiny Filda (1993) ii Matrena lice Tatarstana pytajutsja lokalizovat’ vspyšku xolery // Izvestija, 2001.07.31 iii Rasputina. Rasputin. Poˇcemu? (1960) Dmitrij Medvedev. Mastera. Uˇceniki i stili // Boevoe iskusstvo planety, 2003
Semasiological description of permissive da(va)t in Russian
b.
71
assertion (i) X wants that Y do not do p and therefore does what will make p impossible. (ii) Y does not p.
Overall conclusions concerning polarity Under negation and with an animate causee that lacks control, da(va)t may be used to express intervention, that is, a positive action on the part of the causer to keep the caused event from happening. Parallel cases without negation are more restricted, as they may only be used if the complement action is intended by the causee, or in his perceived interest. The explication for nonintervention is therefore more specific than the explication for intervention. We can thus diagnose a lower degree of semantic specification of Russian da(va)t under negation. 2.2.3.
Manipulative causation
I now turn to permissives in the manipulative domain proper, with animate, controlling causer and inanimate, non-controlling causee.
2.2.3.1.
Profiles of usage
Tables 6 and 7 give an overview of use in terms of morphosyntactic configurations analagous to the interpersonal case. Table 6 with associated visualizations shows that the tendency to be used under negation is less pronounced in manipulative than in interpersonal use with only about half the instances negated (p<0.0003 for same distributions, with, however, a small effect size of V=.07; all morph. forms taken into account). As to the use of individual form categories, the imperative is not as exceptional for polarity as in interpersonal causation, but likewise quite frequent. The exceptionally high use of the infinitive seen in table 6 seems to be due to its typical function in instructions, as well as to its use in the scope of modal or other predicative operators:
72 da(va)t+inf in Russian
forms infinitive participle imperative morph. present past all
count 139 23 32 17 48 259
perfective dat’ neg. perf. compl. 59 (42%) 134 (96%) 9 (39%) 23 (100%) 6 (19%) 27 (84%) 12 (71%) 15 (88%) 26 (54%) 48 (100%) 112 (43%) 247 (95%) =64% of all
count 7 41 6 56 38 148
imperfective davat’ neg. perf. compl. 6 (86%) 2 (29%) 33 (80%) 33 (80%) 6 (100%) 1 (17%) 29 (52%) 46 (82%) 29 (76%) 28 (74%) 103 (70%) 110 (74%) =36% of all
pos neg
100
150
42%
52%
50
54% 39%
80%
19% 71%
86%
76% 100%
0
absolute number of tokens
Forms and polarity
inf.
part. imper. non−past perfective DAT
past
inf.
part. imper. non−past imperfective DAVAT
past
ipf pf
100
150
96%
82%
50
100% 100%
80%
84% 88%
29%
74% 17%
0
absolute number of tokens
Forms and complement aspect
inf.
part. imper. non−past perfective DAT
past
inf.
part. imper. non−past imperfective DAVAT
past
Table 6. Forms of manipulative permissive da(va)t with negation and complement aspect in detail.
(90)
Slegka posolit’INF každyj vnutri, zatem perevernut’ i dat’INF i steˇc’INF židkostiDAT v teˇcenie poluˇcasa.
i Najman, Narinskaja. Process edy i besedy. 100 kulinarnyx i intellektual’nyx receptov // Oktjabr’, No.3, 2003
Semasiological description of permissive da(va)t in Russian
matrix perf perf imperf. imperf.
complement perf imperf. imperf. perf
abs 247 12 38 110 407
rel 61% 3% 9% 27% 100%
73
of which pos neg 138 (56%) 109 (44%) 9 (75%) 3 (25%) 6 (16%) 32 (84%) 39 (35%) 71 (65%) 192 (47%) 215 (53%)
dat 3
ipf
DAVAT Aspect davat
109
pf
neg
32
9
39
138
Table 7. Matrix and complement aspect and negation. Over half the cases involve perfective matrix and complement. Imperfective complement is used in less than 15% of all cases.
‘Lightly salt1 each from the inside, then turn and let2 the liquid4 drain3 for half an hour.’ Tables 6 and 7 show that the generalizations concerning aspect in the interpersonal domain also hold for the manipulative domain: perfective aspect is more frequent in both matrix and complement. As the matrix plot of table
ipf Infinitive Aspect
6
pf
pos
Polarity
71
74 da(va)t+inf in Russian 7 illustrates, use of the imperfective aspect of da(va)t is significantly higher under negation (p<.0001, V=.25). Infinitive aspect, in turn, correlates with matrix aspect in the sense that imperfective aspect is significantly more often used if da(va)t is imperfective, too (p<.0001, V=.31). This is what we also found for interpersonal causation. A list of complement verbs is given in list 2. It is clearly an open set of verbs that enters this construction. Of 156 verbs used only once, 97, that is, about two thirds, were used under negation, confirming that the core of usage is non-affirmative. n=14: otstojat’sja ‘lettle’, nastojat’sja ‘draw’; n=12: steˇc’ ‘drain’; n=10: vyjti ‘leave’; n=9: postojat’ ‘stand’, razvivat’sja ‘develop’, ostyt’ ‘cool’, otdoxnut’ ‘rest’; ‘n=6:’ pogibnut’ ‘perish’, razgoret’sja ‘flare up’, razvernut’sja ‘develop’; n=5: prosoxnut’ ‘dry up’ , privyknut’ ‘get used to’, vozniknut’ ‘arise’, vysoxnut’ ‘dry up’, upast’ ‘fall down’, zakipet’ ‘begin to boil’, sozret’ ‘ripen’; n=4: razvalit’sja ‘collapse’, teˇc’ ‘flow, leak’, ugasnut’ ‘die down’, umeret’ ‘die’, prorvat’sja ‘burst open’, razguljat’sja ‘break loose’, kipet’ ‘boil’, razvit’sja ‘develop’; n=3: prevratit’sja ‘turn into’, rasti ‘grow’, pobedit’ ‘win’, ostynut’ ‘cool’, zaržavet’ ‘become rusty’, vzojti ‘rise’, podnjat’sja ‘rise’, soskol’znut’ ‘slip’, somknut’sja ‘close’; n=2: isportit’ ‘break’, vyrasti ‘grow’, nalit’sja ‘fill’, nabuxnut’ ‘swell’, soveršit’sja ‘happen’, peredoxnut’ ‘rest’, porabotat’ ‘work’, sostojat’sja ‘happen’, projavit’sja ‘show’, bankrotit’ ‘go bankrupt’, propast’ ‘vanish’, proniknut’ ‘penetrate’, zagloxnut’ ‘fail’, proexat’ ‘miss’, prervat’sja ‘ be interrupted’, zaležat’sja ‘lie idle’, zatverdet’ ‘harden’, uspokoit’sja ‘quiet down’, poselit’sja ‘settle’, prostaivat’ ‘stand’, rascvesti ‘bloom’, sorvat’sja ‘break’, osušˇcestvit’sja ‘come to be’ , skatit’sja ‘roll down’, svalit’sja ‘collapse’, vskipet’ ‘come to the boil’, realizovat’sja ‘come to be’, sveršit’sja ‘come to be’, vspyxnut’ ‘break out, flare up’, vzorvat’sja ‘burst, explode’, ostanovit’sja ‘come to a stop’, ševelit’sja ‘stir’, obrušit’sja ‘collapse’; 156 verbs once. Verb list 2: Verbs in manipulative causation, ordered by frequency. Note that verbs apparently taking an animate subject referent such as umeret’ ‘die’ or otdoxnut’ ‘rest’ were used metaphorically or metonymically as in ne dat’ nadežde umeret’ ‘not let the hope die’ or dat’ nogam otdoxnut’ ‘let the legs rest’.
Semasiological description of permissive da(va)t in Russian
2.2.3.2.
75
Semantic analyis
Generally, the senses encountered here may be considered extensions of the meanings found in interpersonal causation, but without any relation to the volitionality of the causee (since it is inanimate). The most central are nonintervention (91) and waiting (92), which are hard to distinguish extensionally: (91)
Non-Intervention Malo togo, suxie semena ne “poˇcuvstvujut” zimu, ix nužno neprei menno zamoˇcit’ v vode i dat’INF imDAT nabuxnut’INF . ‘Besides, dry seeds don’t “feel” the winter; one has to make sure to soak them in water and let them swell up.’
(92)
Waiting Ja daju1S G vygoret’INF suxojDAT traveDAT na polmetra vokrug osii novnogo ogon’ka, potom pristrunivaju ognennuju konnicu [. . . ] ‘I let1 the dry3 grass4 burn down2 in a distance of about half a meter around the main fire, then I attach the fire cavalry [. . . ]’
As Podlesskaya (2006:290) points out, da(va)t may only be used if the causer can in principle influence the caused event; wholly unrelated waiting may not be expressed, cf. (93) *On dalPS T :S G lužamDAT vysoxnut’INF i tol’ko togda pošel guljat’. ‘He let the puddles dry up and only then went for a walk.’ No unequivocal examples of enablement were found. The constructed example below shows that use of da(va)t in this sense is at least strongly restricted: in (94), da(va)t is appropriate for (a), where the desired effect (the drying of the paint) is accomplished by non-intervention/waiting, but not for (b), where this involves enablement by providing certain circumstances: (94)
a.
Nužno ostavit’ kartinu na noˇc’, . . . ‘One needs to leave the painting for a night. . . b. ?Nužno povesit’ kartinu v teploe mesto, . . . ‘One needs to hang the painting in a warm place . . . (a/b) . . . cˇ toby dat’INF kraskeDAT vysoxnut’INF . . . . in order to let the paint dry.’ Golubovskij. Anagrammnye variacii: sostukivanie slov // Vestnik SŠA, 2003 Solouxin. Tret’ja oxota (1967)
i Mixail ii Vladimir
76 da(va)t+inf in Russian The central component seems to be inaction, by way of which the causer allows some independent process to take place. Note that this example is slightly awkward, possibly because dat’ kraske vysoxnut’ ‘letting the paint dry’ functions as a goal by itself here, rather than as a stage in a sequence of events, as it would be typically the case. I will only explicate Nonintervention modeled on the interpersonal explication: Expl. 7: Non-intervention (manipulative) a. presupposition: (i) Y will do p / undergo p for some independent reason. (ii) There is a set of possible actions a such that (iii) If X does a, Y cannot, and therefore will not do/undergo p. (iv) X knows about the above. b. assertion: (i) X does not do some a that would make p impossible (waits). (ii) Y does/undergoes p. It is remarkable that accidental causation by negligence is not expressed by da(va)t. This is in contrast to German lassen or English let, as in die Milch sauer werden lassen / to let the milk spoil (by accident). Instead, a deagentive construction is in Russian: (95)
a. *On dalPS T :S G molokuDAT isportit’sjaINF . Intend./lit.: ‘He let the milk spoil’ b. Uby negoGEN molokoNOM isportilos’PS T :S G . ‘He let the milk spoil’ lit. ‘with him the milk spoiled.’
There do not seem to exist systematic restrictions concerning the causee other than that the process has to be perceived to be more or less autonomous, governed by forces outside the immediate control of the causer, yet at the same time under his influence. Causees may be abstract (96) or concrete (97). Often it is a personalization (97) with or without metonymic reference to an animate entity. i
(96)
Dat’INF zastojat’sjaINF javlenijuDAT bylo opasno. ‘It was dangerous to let the phenomenon stagnate.’
(97)
Verxnjuju cˇ ast’ l’da sloem v 2-3 mm sleduet vybrosit’, zatem dat’INF ii osnovnojDAT masseDAT vodyGEN zamerznut’INF .
i Nel’zja odomašnivat’ monstrov // Professional, 1998.07.01 zalog zdorov’ja // Sad svoimi rukami, 2002.12.15
ii Marina
ˇ Lacis. Cistota -
Semasiological description of permissive da(va)t in Russian
77
One needs to throw away the upper, 2-3 mm thick layer of ice, afterwards one lets the main_body of water freeze. (98)
Dajut3PL nervamDAT otdoxnut’INF , cˇ toby potom uže vzjat’sja za iii stimuljator i pustit’ tok. ‘One lets the nerves rest, in order to then take to the stimulator and turn on the current.’
Negative polarity Under negation, the meaning of da(va)t may denote the failure to wait, as in (99), or active hindering, as in (100). We can say because the causer did not wait, the complement event didn’t take place for (99) and because the causer did something, the complement event couldn’t take place for (100): (99)
[. . . ] bifšteksy iz file byli nemnogo žestki, možet byt’ potomu, cˇ to i imDAT ne daliPS T :PL vyležat’sjaINF . ‘. . . the beefsteaks were a bit tough, perhaps because they didn’t let2 them1 age3 .’
(100)
V objazannosti mužˇcin, poxože, vxodilo neNEG davat’INF kostruDAT potuxnut’INF - i oni po oˇceredi podbrasyvali v ii ogon’ xvorost, ležavšij tut [. . . ] ‘It was apparently part of the duties of the men to not let the fire go out - and they took turns feeding it with brushwood that was lying around.’
In (101), a modal element is not as clearly justified as above. The aim of action is not to not let the ammunition explode, but to keep it from exploding. It is not an autonomous event that would take place in any case; rather, it is an eventuality that would happen for an independent reason which can, however, be hindered. We can say: because the causer did something, the complement event does not take place. (101)
V moi objazannosti i vxodilo zatopit’ po prikazaniju artpogreb [. . . ], cˇ toby neNEG dat’INF vzorvat’sjaINF boepripasamDAT i ne pogubit’ iii korabl’[. . . ]. ‘One of my duties was to flood the magazine, if commanded to do
i Vladimir Obruˇ Markosjan-Kasper. Kariatidy // Zvezda, No.8, 2003 cev. Zemlja ii Sannikova Gromov Vadim. Kompromat dlja oligarxa (2000) iii Boris Dubinin. V tom dalekom sorok cˇ etvertom // Naš sovremennik, 2004 iii Goar
78 da(va)t+inf in Russian so, [. . . ] in order to keep the ammunition from exploding (lit.: not1 let2 the ammunition4 explode3 ) and save the ship.’ In contrast to what we have seen in the interpersonal domain, it seems that no restrictions apply concerning affirmative vs. negative use. An explication of intervention might look like this: Expl. 8: Intervention (manipulative) a. presupposition: (i) Y will do p / undergo p for some independent reason. (ii) There is a set of possible actions a such that (iii) If X does a, Y cannot, and therefore will not, do or undergo p. (iv) X knows about the above. b. assertion (i) X wants X to not p and does what will preclude p / make p impossible. (ii) Y (does) not p.
Summary da(va)t extends well into the manipulative domain with a very similar set of meaning as in the interpersonal domain; however, enablement is a subtype that is even more restricted than in the interpersonal domain. With intervention, no provision must be made concerning the volitionality of the causee, and the positive and negative uses of da(va)t in the manipulative domain should be considered completely parallel. I conclude that this extension into the manipulative domain, constitutes a high degree of synsemanticity of da(va)t. While this may seem an almost trivial, natural extension, note that already in closely related Polish, such an extension is not found.
2.2.4.
Impersonal causation
I will now turn to inanimate causers. Tables 8 and 9 provide profiles for impersonal causation, separately for animate and inanimate causee. In view of the small number of instances, no visualization is provided for the breakup into morphological forms.
Semasiological description of permissive da(va)t in Russian
forms infinitive participle non-past past all
n 6 1 12 7 26
perfective dat’ neg. perf. compl. 6 (100%) 6(100%) 1 (100%) 1(100%) 12(100%) 9 (75%) 7 (100%) 6 (86%) 26(100%) 22(85%) =21% of all
count 1 26 58 11 96
79
imperfective davat’ neg. perf. compl. 1 (100%) 0 26(100%) 21(81%) 55 (95%) 30(52%) 10 (91%) 8 (73%) 92 (96%) 59(61%) =79% of all
Table 8. Impersonal causation: Negation and aspect (causer and causee inanimate).
forms infinitive participle non-past past all
n 4 2 27 34 67
perfective dat’ neg. perf. compl. 3 (75%) 4 (100%) 2 (100%) 2 (100%) 24 (89%) 23 (85%) 28 (82%) 30 (88%) 57 (85%) 59 (88%) =17% of all
count 1 53 128 149 331
imperfective davat’ neg. perf. compl. 1 (100%) 1 (1%) 47 (89%) 31 (58%) 118(92%) 74 (58%) 146(98%) 92 (62%) 312(94%) 198(60%) =83% of all
Table 9. Impersonal causation: Negation and aspect (causer inanimate, causee animate).
The negative polarity affinity of da(va)t is most evident with impersonal causation, be it with animate or inanimate causee: more than 90 percent of all corpus examples are under the scope of negation. However, contrary to what has been claimed in the literature (see 2.1.3 above), it cannot be upheld that positive da(va)t with inanimate causees is in general ungrammatical. Markedly different in contrast to the other domains, imperfective davat’ is much more frequent than perfective dat’, even in past tense. An explanation for this is that this type often expresses delocalized, general statements. The tendency to use perfective complement aspect is likewise less pronounced.
Positive polarity I will start with those cases where both causer and causee are inanimate. Positive corpus examples (n=4) are either under the scope of a negated modal or other operator, or they are of dubious stylistic quality, as (102):
80 da(va)t+inf in Russian (102)
Reguljarnye razryvy v takoj zastrojke dajut3PL ulicamDAT spustit’sjaINF k naberežnoj, otkryvajut vidy na reku iz rajona i proi puskajut svet v kvartiry. ‘Regular gaps in the line of buildings let the streets descend to the embankment, open views of the river and lets light pass into the apartments.’
We can see that (102) would be interpreted as enablement: Because there are regular gaps in the line of buildings, there is possibility (space) for streets to lead down to the river side. However, this is clearly creative usage. Positive examples with both inanimate causer and causee must be considered marginal. With animate causee, positive instances do occur without stylistic difficulty, if also not very frequently (n=29): (103)
Demokratija, koneˇcno, daet3S G otdel’nomuDAT cˇ elovekuDAT i razvernut’sjaINF . ‘Democracy, of course, lets (enables) the individual person (to) develop.’
In these cases, the initiative comes from the causee’s side; the causer represents circumstances that enable the causee’s action. We are therefore generally dealing with enablement. In these cases, it is often possible to expand da(va)t to a light verb construction involving dat’ vozmožnost’ ‘give the possibility’ without any clear semantic difference, e.g. daet3S G otdel’nomu cˇ eloveku (103)’ Demokratija, koneˇcno, vozmožnost’ACC razvernut’sjaINF . Lit.: ‘Democracy, of course, gives the individual person the possibility to develop.’ More examples include: (104)
Zato polnyj privod i bol’šie kolesa dadut3PL zaexat’INF v ponraii vivšijsja lesok ili podobrat’sja k prigljanuvšemusja vodoemu. ‘But four-wheel-drive and large wheels let (one/you) drive into that little forest that caught your fancy or get to that beautiful lake you saw driving by.’
i Elitnoe ˙ žil’e - približenie k idealu // Mir - Dom. City, 2004
i okrestnosti (1998)
ii Sergej
i Aleksandr Genis. Dovlatov
Mišin. Po svjatym mestam // Za rulem, 2004
Semasiological description of permissive da(va)t in Russian
(105)
81
No vsë vremja soxranjal Žukovu zvanie cˇ lena Stavki - i e˙ to daloPS T :S G emuDAT mnogo nauˇcit’sjaINF u voenno obrazovannyx iii Šapošnikova, Vasilevskogo i Vatutina. ‘But all the time Žukov remained a member of general command and that let him learn a lot from the military experts Šapošnikov, Vasilevskij and Vatutin.’
In view of the clearly peripheral status of da(va)t in this domain, I will not give explications, which would need to reflect different properties of the causee (as event (105), instrument (103), abstract entity, and others). It should be clear that the meaning denoted by positive da(va)t is roughly in line with the explication given for enablement in the interpersonal domain; however, the causer is not an agent or force; rather, it denotes some circumstance that is decisive for the causer’s action to be possible. I hold this usage to be a regular extension of interpersonal da(va)t that is, however, weakly grammaticalized and should be regarded rather marked usage. Negative polarity Negated da(va)t with inanimate causer is more frequent and conveys intervention in the sense of hindering circumstances. There do not seem to be substantial semantic differences concerning animacy of the causee in this respect. Examples with inanimate causees (n=118) involve, for example: (106)
[. . . ] ežegodnaja paxota neNEG dalaPS T :S G drevesnojDAT i porosliDAT vyrvat’sjaINF naružu iz ovražka. ‘. . . annual ploughing did’t let the wooden shoots break through out of the ravine.’
It is generally used more frequently (n=369) with animate causee: (107)
Poslednee vremja on spal ploxo, golod neNEG davalPS T :S G xorošo ii spat’INF . ‘Lately he didn’t sleep well, hunger didn’t let (him) sleep well.’
(108)
Gody proxodjat i prinosjat zabvenie, no est’ genetiˇceskaja pamjat’ nacii, kotoraja naveˇcno živet, kotoraja neNEG daet3S G zabyt’INF . iii ‘Years go by and bring oblivion, but there is such a thing as the genetic memory of a nation, that lives forever and does not let (you/anybody) forget.’
iii Aleksandr
i Aleksandr Solženicyn. Željabugskie Solženicyn. Na krajax (1994-1995) ii Vyselki (1998) Varlam Šalamov. Kolymskie rasskazy (1954-1961) iii Irina Cypina. Svoi i cˇ užie // Lebed’ (Boston), 2003
82 da(va)t+inf in Russian (109)
Massirovannyj udar vitaminov, nanesennyj po infekcii v moment ee iv zaroždenija, neNEG dast3S G vamDAT razbolet’sjaINF . ‘A massive blow of vitamines, administered just when the infection starts to form, does not let you get sick.’
As one can see, the last two examples involve inagentive verbs (zabyt’ ‘forget’, razbolet’sja ‘get sick’). In general, both agentive and inagentive verbs are met frequently. In verb list 3, the list of complement verbs with a frequency above 2 is given. The vast majority of these verbs were used only negatively; the only exceptions are given separately. Very similar to what we have seen with interpersonal and manipulative negated da(va)t, we are dealing with permissive negation, where a modal element is appropriate: because the causer exists or acts in some way, the causee cannot do what he or she would like to do or is expected to do. In some cases, however, the justification of the modal element becomes weaker. Consider the following examples: ˙ ad! [. . . ] Ostaetsja liš’ gadat’, kakie sily, mysli i cˇ uvstva neNEG (110) Eto i dajut3PL emuDAT slomat’sjaINF uže na pervyx e˙ tapax. ‘It is hell. [. . . ] It can only be guessed what forces, thoughts and emotions do not let him break (=keep him from breaking) already in the first parts of the trip.’ ˙ neNEG daet3S G staret’INF (111) Družit’ nado s temi, kto molože tebja. Eto ii — vo-pervyx, vo-vtoryx, legko postupat’ po-svoemu. ‘It’s better to befriend those younger than you. That does not let you get older (=keeps you from getting older) - besides, it’s then easy to act on your own. ’ (112)
(113)
Odnako nekotorye meloˇci neNEG dajut3PL e˙ tojDAT knigeDAT iii stat’INF klassiˇceskoj. ‘However, a number of small problems do not let this book become a classic.’ ˙ dovol’no složnaja programma, kotoroj ja dolžen strogo prideržiEto vat’sja v teˇcenie dlitel’nogo perioda. Ona deržit moj tonus i, kstati, iv neNEG daet3S G tolstet’INF .
iv Alla Družinina. Dieta pri prostude // Semejnyj doktor, 2002
i Bratstvo krapovyx beretov
// Soldat udaˇci, 2004 Krupin. Vybrannye mesta iz dnevnikov 70-x godov // Naš sovremennik, 2004 iii Vladimir Kaganskij. Krest’janovedenie Rossii vernulos’ na Rodinu // Oteˇcestvennye zapiski, 2003 iv Vjaˇceslav Fetisov. Overtajm (1997) ii Vladimir
Semasiological description of permissive da(va)t in Russian
83
• Verbs used with inanimate causer and animate causee: n=24: spat’ ‘sleep’; n=18: usnut’ ‘fall asleep’, dyšat’ ‘breath’; n=13: zabyt’ ‘forget’; n=12: rasslabit’sja ‘relax’; n=11: sosredotoˇcit’sja ‘concentrate’; n=9: zasnut’ ‘fall asleep’; n=7: rabotat’ ‘work’, žit’ ‘live’; n=6: sdelat’ ‘do, make’, upast’ ‘fall’, rassmotret’ ‘examine’; n=5: uvidet’ ‘see’, umeret’ ‘die’; n=4: byt’ ‘be’; n=3: sdvinut’sja ‘move’, otdat’sja ‘surrender’, ostanovit’sja ‘stop’, propast’ ‘disappear’, dumat’ ‘think’, popast’ ‘hit’, igrat’ ‘play’, stat’ ‘become’, uspokoit’sja ‘calm down’; n=2: skuˇcat’ ‘be bored’, videt’ ‘see’, umirat’ ‘die’, podnjat’sja ‘ascend’, opustit’sja ‘descend’, sobrat’sja ‘get ready’, vpadat’ ‘hit’, poverit’ ‘believe’, zadumat’sja ‘become thoughtful’, idti ‘go’, razvernut’sja ‘unfold’, radovat’sja ‘be glad’, pet’ ‘sing’, projavit’ ‘display’, proxodit’ ‘pass’, prosnut’sja ‘wake up’, povernut’ ‘turn’, osušˇcestvit’ ‘accomplish’, ujti ‘go away’, otvetit’ ‘answer’, pol’zovat’sja ‘use’, govorit’ ‘speak’, trudit’sja ‘labour’, otvleˇc’sja ‘get distracted’, podnjat’ ‘rise’, somknut’ ‘close’, dogovorit’ ‘finish speaking’, zaskuˇcat’ ‘get bored’ (163 hapax legomena) • Of which also used in positive polarity, n=1: dumat’ ‘think’, byt’ ‘be’, dyšat’ ‘breathe’, razvernut’sja ‘unfold’, uvidet’ ’see’ • Verbs used with inanimate causer and inanimate causee: n=5: razvivat’sja ‘develop’, stat’ ‘become’; n=3: osušˇcestvit’sja ‘come to be’, razvit’sja ‘develop’, isˇceznut’ ‘disappear’; n=2: prevratit’sja ‘turn into’, rasti ‘grow’, obrazovyvat’sja ‘arise’, skol’zit’ ‘slip’, ostyt’ ‘cool down’, zamerznut’ ‘freeze’, upast’ ‘fall’, absoljutizirovat’sja ‘become absolute’, uspokoit’sja ‘calm down’ (84 hapax legomena). Verb list 3: Verbs in impersonal causation ‘That’s a pretty complicated scheme I am supposed to strictly adhere during a longer period of time. It will help sustain my "tone" and, by the way, will not let (me) gain weight.’ (114)
To, cˇ to e˙ to ljudi - moi sovremenniki, neNEG daet3S G mneDAT s takoj v siloj poˇcuvstvovat’INF svoe odinoˇcestvo. ‘The fact that these people are my contemporaries does not let me feel (=keeps me from feeling) my solitude so intensely.’
v Georgij
Sviridov. Iz knigi Muzyka kak sud’ba // Naš sovremennik, 2003
84 da(va)t+inf in Russian In general, these are unaccusative verbs without control by the causee. It is here that the equation of not let = cause not to seems to be most appropriate not only logically, but in that it captures the full content of what is expressed. Again, consider an example from Harry Potter: (115)
a. b. c.
“A stone that makes gold and stops you from ever dying!” said Harry . Kamen’, kotoryj delaet zoloto i neNEG daet3S G umeret’INF ,skazal Garri (lit: ’does not let (you) die’) . Kamie´n , który wytwarza złoto i czyni3S G nie´smiertelnymINS ! (lit: ‘makes immortal’)
In this, and the above cases (110) to (114), ne davat’ is a translational equivalent of English ‘stop from’ rather than of let, which does not seem to be used as easily in this meaning. Taking these cases into account, I arrive at the following unified explication for negated impersonal da(va)t: Expl. 9: Stop/Keep from a. presupposition: (i) Y will do or undergo p: as a matter of natural or other consequence, possibly, but not necessarily, by his own will. (ii) X is part of a situation that influences Y and p b. assertion (i) Because of X, Y does not p. As one can see, this is semantically the most unspecific explication so far; as such it is not very restrictive and may be considered to encompass all definitions of negative intervention above. Note that cases like these are not met in Polish and Czech. This usage exist alongside more clearly permissive cases such as ne davat’ spat’ ‘not let someone sleep’ which are easily expressed using da(va)t in the other Slavic languages, as well as by let or lassen.
2.2.5.
Conclusions: permissive da(va)t in three domains
I have presented a semasiological analysis of permissive da(va)t in three domains, interpersonal, manipulative and impersonal. While there are differences, the degree of semantic overlap is considerable, most clearly in the case of negative causation.
Semasiological description of permissive da(va)t in Russian
85
Russian da(va)t is most commonly used, most synsemantic and least specific in negative usage. In negative usage, it denotes foremost intervention. This meaning is present in all domains; it should be considered its semantic core and other meanings such as waiting as extensions of intervention. In positive use it predominately denotes non-intervention rather than enablement. It is restricted in use and is subject to further semantic or pragmatic conditions of usage that make it semantically more specific than its negated version. Russian negated da(va)t thus can be seen of limited compositionality: in the morphosyntactic configuration with negation it is to a certain degree more grammaticalized. In the following chapters, we will see that a certain disposition to negative polarity is characteristic of all the constructions involving da(va)t in the languages investigated; however, Russian is exceptional in the degree of this disposition and the qualitative differences between positive and negative polarity.
2.2.6. 2.2.6.1.
Specific issues in the analysis of the permissive Why is the perfective imperative daj! so frequent?
The present section is concerned with the use of the perfective imperative in the permissive construction, exemplified below: i
(116)
Pogodi, pogodi, Tašen’ka, [. . . ] DajI MP:2S G umyt’sjaINF . ‘Wait a second, Tašenka, [. . . ] Let (me) wash up.’
(117)
Papa! - isteriˇceski kriknula Tat’jana. DajI MP:2S G ejDAT odet’sjaINF ! ii ‘Papa! - Tatjana cried out hysterically. Let her get dressed!’
In the quantitative overview in section 2.2.2.1, we have seen that use of the perfective imperative is both exceptionally frequent and, unlike the other forms, not linked to negative polarity. The use of the imperfective imperative davaj(te), in contrast, is not exceptional. It is neither extraordinarily frequent, nor does it depart from the usual polarity sensitive pattern of da(va)t+inf in Russian: of only 31 cases, 26 are negated. The perfective imperative daj(te), in contrast, appears in 365 instances, only 34 of which are negated. In this section I suggest that this extraordinary usage pattern can be explained as due to the grammaticalization of daj in yet another construction, namely the one with finite complement: i Ljudmila Ulickaja. Medeja i ee deti (1996)
ii Irina Murav’eva. Filemon i Bavkida (1995)
86 da(va)t+inf in Russian (118)
I tut, smotrju, on ee pod ruku vzjal i govorit: «Ja vas provožu. . . i DajteI MP:2PL jaNOM vas provožu1S G ». ‘An then he takes her by the hand and says: “I’ll accompany you. . . , let me accompany you (lit.: let’s I accompany you).”’
Both constructions, I argue, converge in their grammaticalization in the direction of a general imperative construction. The high frequency of daj in the permissive is linked to the use of this imperative form in both constructions. After reviewing the literature, I will adduce a number of arguments showing, on the one hand, that daj fulfills functions that are different from its nearest equivalent, its imperfective counterpart davaj, and, on the other hand, that daj in both the finite and the permissive construction converges in several of these functions. I then examine differences between the two constructions before concluding. Evidence for my assessment comes from different sources. First of all, note that the functional difference between the imperative of the permissive and the finite clause construction is not always clear. The next example shows such a case where the two constructions are equivalent: (119)
DajteI MP:2PL mneDAT , – skazal vdrug Lesnik. – DajteI MP:2PL jaNOM emu skažu1S G ... MneDAT daj-teI MP:2PL ... ‘Let me, Lesnik suddenly said, let’s I tell him ... Let me...’ (Strugackie, Barentsen 2003b: 25, my translation)
Here, the speaker starts with the permissive construction and then changes to the finite clause. The full permissive Dajte mne emu skazat’ ‘let me tell him’, not finished by the speaker, is in this case semantically virtually indistinguishable to the finite clause construction. A link between daj in the permissive and in the finite construction is, at first, unexpected and counterintuitive: the two constructions differ strongly in respect to both the paradigmatic and syntagmatic axis and would first and foremost be considered different constructions. In their study of Russian imperative constructions, Xrakovskij & Volodin (1986: 130f) make this point very clear as they focus on the structual difference between daj(te)I MP:2S G(2PL) (mneDAT ) spet’INF ‘let me sing’ and daj(te)I MP:2S G(2PL) (jaDAT ) spoju1S G ‘let’s I’ll sing’. First of all, the syntactic structure is different: one is a control construction, with the personal pronoun in the dative governed by daj/te, the other involves a particle and a finite clause, with a nomi Mixail
Anˇcarov. Kak Ptica Garuda (1989)
Semasiological description of permissive da(va)t in Russian
87
inative personal pronoun. Furthermore, only in the finite clause construction can the adressee be made explicit: dajPF:I MP:2S G tyNOM (mneDAT ) spet’INF ‘you2 let1 me3 sing4 ’ but *dajPF:I MP:2S G tyNOM (jaNOM ) spoju1S G ‘daj *you I’ll sing’. While these are undoubtedly important structural differences, I will argue that the facts of usage show that drawing a sharp border between daj as a particle in the finite clause construction and daj as an inflectional form in the paradigm of permissive da(va)t is not adequate. The special status of daj in the permissive construction has, however, not gone unnoticed in Xrakovskij & Volodin (1986). They remark that “if the expression daj/te+infinitive is used without pronominal actant, then the missing actant is practically always the first person singular, exceptionally also first person plural (Xrakovskij & Volodin 1986: 130, my translation10 )”. Concerning the finite construction, Xrakovskij & Volodin (1986: 121-130) single out a regular imperative of the 1.sg., formed from the imperative of da(va)t in both aspects, an optional particle and a finite predicate: (120)
Daj/davaj (-ka) (ja) spoju / budu pet’ Give(PF/IPF).IMP PRT I sing.PF.1SG / AUX.1SG sing.IPF.INF ‘Let’s I sing (pf./ipf.)!’
Concerning other than first person finite forms they do not find a similar equivalence of daj and davaj. Summarizing their analysis they state: “the particle davaj/te (and to a lesser degree, daj/te) so to speak advances into the role of a standard imperative marker for the whole paradigm of finite forms (Xrakovskij & Volodin 1986: 130, my translation11 )”. Podlesskaya (2006) draws a slightly different picture. She posits a common paradigm that unites the permissive, object control use of da(va)t with its uses with finite complement in a single ‘permissive’ class and opposes it to a ‘hortative’ class formed with davaj and finite complement or bare infinitive. She refers to the following examples as to a minimal pair (p.293, my glossing): (121)
a. b.
Davaj detiNOM:PL poxodjat3PL bosikom. (hortative) Daj detiNOM:PL poxodjat3PL bosikom. (permissive) ‘Let (daj/davaj) the children take a walk barefoot.’
She writes: “The two constructions differ in the character of reaction the speaker is applying for: in hortatives, the speaker suggests that an action is performed and asks if the hearer has any objections, while in permissives with finite lexical verbs, the hearer is asked to allow or simply not to prevent the action (Podlesskaya 2006: 294)”. In contrast to Xrakovskij & Volodin (1986),
88 da(va)t+inf in Russian person first third
causee implicit overt implicit overt
daj(te) 157 100 6 98
davaj(te) 1 2 11 17
Table 10. Reference of the causee in imperative permissive constructions. With the perfective imperative, the single most important referent is the first person (singular or plural), and it is typically left implicit. With the imperfective imperative, first person reference is the exception.
she thus stresses the opposition of davaj and daj12 in the finite construction as a whole without granting the first person singular special status. The use and functions of the first person singular imperative as in daj/davaj spoju ‘let’s (pf./ipf) I sing’ as singled out by Xrakovskij & Volodin (1986) were investigated in detail in Barentsen (2006, 2003b). Barentsen shows that the mechanisms of aspect selection with da(va)t are in line with general functions of aspect in the imperative. According to his corpus study, both perfective and imperfective imperative appear approximately at the same frequency in this construction; davajte is in fact slightly more frequent. However, there is a functional difference: only daj, but not davaj is used in internal selfincitement as in: (122)
[. . . ] ja rešil – dajI MP:2S G ja imDAT poˇcitaju1S G vslux – ne možet byt’, cˇ toby oni ne zametili, kak ja xorošo cˇ itaju (Iskander). (quoted from Barentsen (2003b: 19), omission mine) ‘[. . . ] I made up my mind: why don’t I try reading it to them (lit.: give1 I read3 (it) to them2 ) - it’s impossible that they wouldn’t notice how well I can read.’
Xrakovskij & Volodin (1986) and Barentsen (2006, 2003b), in contrast to Podlesskaya (2006), thus grant the first person imperative formed with daj/davaj in the finite construction special status and do not see a clear parallel to davaj in the other persons. As concerns the link between the use of the permissive and the finite construction, only Podlesskaya (2006), but not Xrakovskij & Volodin (1986), stress a functional connection; Barentsen’s (2006) analysis of the contrast between the two in the first person will be focused on below.
Semasiological description of permissive da(va)t in Russian
first singular second singular third singular
daj(te) with pron. 45 0 1
w/o 182 0 0
89
davaj(te) with pron. w/o 89 193 12 37 0 65
Table 11. RNC corpus query on combinations of imperative of da(va)t followed by optional nominative pronoun and finite verb.
Evidence in the corpus clearly confirms the special status of daj with the first person singular for both the finite construction and the permissive construction. Use with an omitted first person causee in the permissive construction (daj projti ‘let (me) pass’) is generally typical for daj, but not for davaj. Table 10 summarizes the relationship of causee reference and omission in our corpus. All permissive imperative constructions were annotated as to causee reference and omission. It shows that first person causees are very frequent (250 of 350 cases) with daj and very often implicit; in contrast, third person causees (as in daj emu spat’!) are much less frequently used (100 of 350 cases) and they are almost always overt. The imperfect imperative davaj(te) shows a completely different picture: here, first person causees are the clear exception with 3 of 31 cases. We have thus pinpointed an clear characteristic of the perfective imperative that has to do with its function: in contrast to davaj, dajis used with omitted first person causees most of the time. In order to compare these findings to the patterns of use of daj and davaj in the finite construction, I searched the full RNC corpus for combinations of daj or davaj followed by first, second and third person singular predicates with and without personal pronoun intervening.13 The results are given in table 11: first person use is characteristic for daj with finite complement, but to a much lesser degree for its imperfective counterpart davaj. With both daj(te) and davaj(te), first person use clearly dominates; the number of usages is comparable. This confirms the findings in Barentsen (2003b). Second person usage is less frequent with davaj(te), but clearly possible; not a single example was found with daj(te), however. As the suppletion test shows, it is unfelicitous in many cases:
90 da(va)t+inf in Russian (123)
Davaj/?Daj(IPF/?IPF):I MP:2S G ty zabereš’2S G ix v svoj X’juston, a ja i prileˇcu na krestiny, kogda roditsja tvoja doˇc’. ‘Davaj/?daj you’ll take them along with you to Houston, and I’ll come to the christening when your daughter is born.’
The paradigms of daj and davaj thus appear to be used differently concerning other than the first person. In the third person, like in the second person, davaj is used frequently, while daj only yielded one example from the 19th century. (124)
- Pogodi, djadja Semen, dajI MP:2S G onNOM rasskažet3S G [. . . ] ‘Wait, Uncle Semen, let him talk (lit. let he’ll talk) ’
i
While a complete analysis is beyond this study, the study confirms the analysis in Xrakovskij & Volodin (1986) according to which use as a first person imperative is by far the most important function of daj with finite complement, while davaj tends to be used with the whole imperative paradigm. The data adduced so far show that daj shares important usage characteristics regardless of the construction it appears in: with both finite verb and infinitive it is specialized towards the first person singular. Further arguments for convergence come from comparative data. Barentsen (2003b: 2) points out that in other languages, the function of both constructions is fulfilled by the same causative markers, e.g. by the Germanic family of cognates involving Eng. let, German lassen and Dutch laten. Especially enlightening is the comparison within Slavic. In Polish, the permissive construction involving da(va)t is quite similar to its Russian cognate. However, neither does daj in the permissive paradigm enjoy a similar prominence as in Russian (see below p.119ff.), nor does a construction involving daj and a finite verb, analogous to the Russian daj(te)PF:I MP:2S G(2PL) (jaDAT ) spojuINF ‘daj I’ll sing’, exist in Polish (except for a very colloquial register discussed on page 118 below). The fact that such a close cognate as the Polish permissive does not display this prominent use of daj means that we are not merely dealing with an automatic or completely compositional outcome of the combination of imperative and permissive. Moreover, the fact that both the finite construction as a whole and the special prominence of daj in the permissive is absent in Polish supports the suggestion that these two may be linked in Russian. This second point is supported by the pattern of translation into Polish: here, Russian daj in both constructions is often equivalent not to the Polish permissive, but to i Ustinova.
Podruga osobogo naznaˇcenija (2003)
i L.N.
Tolstoj. Voskresenie (1899)
Semasiological description of permissive da(va)t in Russian
91
a dedicated imperative / optative construction involving the particle niech, cf. the following examples from ParaSol: (125)
RU Prostite, šef, ja sejˇcas voobšˇce... DajteI MP:2PL otdyšat’sjaINF ... PL Przepraszam, szefie, ale ja teraz w ogóle... NiechPRT troch˛e oprzytomniej˛e1S G ... i EN Forgive me, chief, I’m a little ... let me catch my breath
(126)
RU DajteI MP:2PL podumat’INF , progovoril on. PL Zaraz, niechPRT pomy´sl˛e1S G ... EN This needs thinking about, he said.
(127)
ii
RU Podoždi, dajI MP:2S G ja tebja traxnu1S G , v šejku traxnu! PL Czekaj, niech ci˛e plasn˛e, niechPRT ci˛e plasn˛e1S G w karczek! iii ‘Wait, let me smack you, let me smack you in the neck!’
I cannot give a complete analysis here, since this would involve a thorough review of the field of imperative constructions in Russian and Polish. The facts adduced so far, however, show that the two functions converge in Russian to some degree and that Russian is clearly different from Polish here. The two constructions are not always equivalent in Russian, however. Barentsen (2003b, 2006) frames the differences of the permissive and the finite clause construction in differing degrees of autonomy of the speaker: “[. . . ] in the construction with infinitive we are dealing with a real request, with the speaker playing a rather passive role. In the finite construction the role of the speaker, conversely, is actually very active, which hinders the use of this construction with clearly controllable actions” (Barentsen 2006: 27, my translation14 ). As a typical example for the use of the reflexive construction with a non-controllable event, he quotes: (128)
DajI MP:2S G mneDAT spokojno umeret’INF . ‘Let me die in peace.’ (Barentsen 2006, my transl.)
I agree that this is an important difference between the use of the two constructions in the data: daj umru would be interpreted as ‘I propose I’ll go ahead and die now’, which is obviously very different and requires very specific context. However, rather than seeing these as invariant meanings, I prefer to see the infinitive construction as simply underspecified in this respect, rather than always expressing low autonomy. This would explain why the i Strugackie,
Piknik na ovoˇcine (Rs.), Piknik na Skraju Drogi (Pl.) Roadside Picnic (En.) Rowling: Harry Potter and the Philosopher’s Stone; Harry Potter i kamie´n filozoficzny (Pl.); Garri Potter i filosofskij kamen’ (Rs.). iii W.Gombrowicz: Ferdydurke (Pl./Rs.). ii J.K.
92 da(va)t+inf in Russian overlap of the two constructions is so large. I would like to modify Barentsens analysis and assume that we have a continuum of autonomy of the speaker with the infinitive construction, ranging from cases like dajte podumat’ ‘let me think’ (roughly equivalent to dajte podumaju ‘I propose I’ll think now’) to cases like daj mne umeret’ ‘let me die’ (not equivalent to ?daj umru ‘I propose I’ll die now (actively)’). The range of usage is different with finite clause: here, it ranges from inner dialog (daj, dumaju, pojdu ‘perhaps, I thought, I should go’ vs. ?daj, dumaju, pojti) to first person imperative signalling an imminent action by the speaker (dajte podumaju ‘I propose I’ll think’ and dajte podumat’ ‘let me think’). This does not falsify Barentsens analysis; instead of framing the distinction as categorial, I see it as a continuum with a large area of overlap15 . Recall that no such effect of exceptionally high frequency of the permissive imperative is attested in Polish, where the finite construction is likewise absent. Since Russian speakers are not forced to distinguish between these constructions functionally much of the time, I propose that dajte in both usages have influenced each other in the area of overlap, increasing the functional load of the perfective imperative as such, irrespective of whether it is used with a finite clause or an infinitive. The exact diachronic development of these formants, however, remains out of the scope of this study.
Conclusions Summarizing, the high frequency of the positive, perfective imperative in the permissive construction can be attributed to the sharing of functions fulfilled by the same formant in the finite clause construction. In both cases, a wish to perform some action which the interlocutor is asked to provide possibility for is expressed. While this seems to be in general compositionally derivable from the permissive function plus the imperative, in at least some other languages, such as Polish, a similar load of the permissive imperative is not attested. I suggest this is an extension of usage linked to the grammaticalization of a more specific construction involving the same formant, but with finite clause. From a usage based point of view the hypothesis of two functionally similar constructions with the same formant influencing each other in frequency is highly interesting, as it raises the question whether dajte should be considered a polysemous formant, two homonymous formants, or indeed a single formant, and how these senses are related in language production.
Semasiological description of permissive da(va)t in Russian
93
From a connectionist point of view, effects of mutual influence are expected. We will see that in Polish, a similar correlation between further grammaticalization of the reflexive permissives marker and the frequency of lesser grammaticalized variants can be seen to hold.
2.2.6.2.
Reflexive permissives
As we will see in later sections, in the West (and South) Slavic languages the reflexive permissive construction has grammaticalized further than in Russian. Since this development has already started in the earliest sources, we cannot with certainty reconstruct this process based on actual data. Under the assumption that contemporary Russian represents an earlier stage than the West Slavic languages, the examination of Russian reflexive permissives can provide clues for the reconstruction of this process. It is therefore discussed here in detail, introducing the distinctions relevant in the other languages as well. The discussion focuses on the interpersonal domain.
Overview Russian is a long-distance anaphora language (Rappaport 1986): anaphora in complement clauses (but not in subordinate clauses) may link to any higher subject in the sentence. This leads to potential ambiguity of the reflexive pronoun; it may link either to causer or to causee. As to its function, any argument of the complement phrase can be in principle rendered by the reflexive pronoun sebja and its inflectional forms. We can distinguish the following types found in the corpus (with frequency): • Reflexive pronoun coreferent to causer, governed by complement infinitive in accusative (n=161), dative (n=4) or instrumental (n=1) case or in a prepositional phrase (n=6) (129)
Detii dajut3PL sebjaRFX:ACC i obmanyvat’INF , priˇcem s radost’jui ‘Childreni let themselvesi be cheated (lit.: let1 (people) cheat3 them2i ), and they even like it.’
i Vladimir
Krupin. Vybrannye mesta iz dnevnikov 70-x godov // Naš sovremennik, 2004
94 da(va)t+inf in Russian (130)
[On . . . ] vse-taki ne stal idiotom, neNEG dalPS T :S G polnost’ju zabit’INF sebeRFX:DAT golovuACC ideologiˇceskim musorom. ii ‘He did not turn into an idiot, hei did not let (them) completely stuff hisi head (lit: himselfi the head) with ideological garbage.’
(131)
My neNEG davaliPS T :PL upravljat’INF sobojRFX:INS . Wei didn’t let (them) control usi .
(132)
On neNEG dalPS T :S G mneDAT ujtiINF otPREP sebjaRFX:GEN , pojmav moi nogi pod kolenjami [. . . ].i Hei didn’t let me go away from himself i , catching hold of my legs below the knees [. . . ].
iii
• Reflexive pronoun coreferent to causee, governed by complement infinitive in accusative case (n=8) or in a prepositional phrase (n=9). (133)
Poslušajte, a poˇcemu by vam ne napisat’ [. . . p]ro cˇ eloveka, kotoromuDAT neDAT daliPS T :PL polnost’ju osušˇcestvit’INF ii sebjaRFX:ACC . ‘Listen, why don’t you write about a person who wasn’t allowed to develop (lit. who (people) didn’t let develop himself fully) .’
(134)
- O cˇ em ty govoriš’, Katja, - vzbesilsja vdrug Nikolaj, - dajI MP:2S G cˇ elovekuDAT prijtiINF vPREP sebjaRFX:ACC , cˇ etyre goda cˇ eloveka iii ne bylo. What are you talking about, Katja - ..... let a personi come to his senses (lit.: come to himselfi ), he was away for four years.
• Reflexive pronoun is coreferent to causer, and denotes the causee. The causer ‘lets’ himself do something (n=23): (135)
Pora! - vdrug vykriknul Bab-Jagun, i, neNEG davajaCV B sebeRFX:DAT ispugat’sjaINF ešˇce sil’nee, Tanja bystro podnjala iv ruku s kol’com i voskliknula: It’s time - exclaimed Baba-Yaga, and, not allowing herself to become even more scared, Tanja swiftly raised her hand and cried out:
iii “Soveršenno sekretno”, 2003.07.04 ˇ Arbatov. Celovek Sistemy (2002) i Anna Tkaˇ ii Daniil Granin. Zubr (1987) iii Jurij Mamleev. Konec ceva. Privorot (1996) sveta/O cˇ udesnom (1975-1999) iv Dmitrij Emec. Tanja Grotter i magiˇceskij kontrabas (2002) ii Georgij
Semasiological description of permissive da(va)t in Russian
95
Overall, reflexive permissives are not very frequent; of 2433 examples of the interpersonal domain, only 172, that is, 7%, involve reflexive pronouns in the complement clause referring to the sentence subject and only 17 involve complement internal reflexives. It is the first type that has further grammaticalized in other Slavic languages. Henceforth, this type is meant when I speak of reflexive permissives without further qualification.
Presence of a causee In the sections on Polish and Czech, we will see that causee expression with reflexive permissives has further options in comparison to non-reflexive permissives in these languages and that one the distinguishing traits of the further grammaticalized modal passive construction is the impossibility of providing an erstwhile causee phrase. In Russian, where no such further grammaticalization is observed, there is a behavioral difference that helps explaining these development developments. While there is no coding difference in Russian, the causee is usually omitted in reflexive, but not in non-reflexive constructions. This can be seen in table 12 that compares causee presence in the set of reflexive permissives and in the evalution set, a random sample of 300 taken from the full corpus. Of 133 non-reflexive, interpersonal permissive phrases in this set, only 29 (22%) did not overtly express the causee. With reflexive permissives, in contrast, omission of the causee is the rule and the case in 85% of all cases. The difference is highly signicant (p<0,0005) with a large effect size of Cramer’s V = .64. A closer look at the 29 non-reflexive instances where the causee was not expressed shows that 9 were imperatives with first person causee reference (daj podumat’ ‘let (me) think’). We have seen in section 2.2.6.1 that this type is exceptional in several respects and that in these cases, the causee is typically omitted. Of the remaining, 20 cases, 17 were negated. Both negated and positive cases involve generic or contextually retrievable causees: (136)
A potom pribegaet iz kuxni razgorjaˇcennaja babuška i neNEG daet3S G vynosit’INF iz doma igruški, televizor i magnitofon, a takže i kurtki i šapki. ‘And then some angry grandma comes running from the kitchen and
i Petruševskaja.
Gorod Sveta (2003)
96 da(va)t+inf in Russian Reflexive permissives (All Corpus) with causee 25 15% w/o causee 147 85% all phrases 172
Non-reflexive permissives (Evaluation set) with causee 104 78% w/o causee 29 22% all phrases 133
Table 12. Distribution of causee-less phrases in interpersonal, causer-controlled reflexive permissives over the whole corpus, compared to interpersonal, nonreflexive permissives in the random evaluation set of 300.
does not let (them) carry away toys, television and tape recorder, as well as jackets and caps.’ (137)
[Oni] vse vremja vnimatel’no sledili za nami i neNEG davaliPS T :PL ii otstat’INF ni na šag. ‘They kept carefully watching us and did not let (us) fall behind a single step.’
(138)
Esli by v 92 godu Džoxar Dudaev dalPS T :S G provestiINF e˙ tot referˇ censkoj respubliki, u nas ne bylo by 95-96 gg. i endum o statuse Ceˇ ‘If Dudaev had let (people) hold the referendum on the status of the Chechen republic in 1992, we would not have witnessed (the events of) 1995/96.’
In general, these cases are exceptional and constitute about one fifth of non-reflexive permissives in the evaluation set, while those cases where the causee is mentioned often become elliptical, incomplete or change semantics if it is omitted. As the data shows, this is very different with reflexive permissives, where the causee is omitted in most of the cases. This characteristic can be attributed to the semantics of this construction: Conceptually, reflexive permissives involve the same number of participants as the lexical infinitive. The referent of the causer is at the same time the referent of the patient of the complement. In these instances, the two subevents are tightly intertwined, while the causer is defocused. Attention is centered on the subject referent in its double role as both the matrix agent (the causer), and the complement patient. The following example is typical: ii Medvedev. Barankin, bud’ cˇ elovekom! (1957) Moskvy, 2003.03.26
i Boris
˙ Nemcov. Interv’ju // Exo
Semasiological description of permissive da(va)t in Russian
(139)
97
Razumeetsja, ja nadejalsja natknut’sja na cˇ to-nibud’ interesnoe, nedarom že ja dalPS T :S G sebjaRFX:ACC vtjanut’INF - ili sam ii vtjanulsjaPS T :S G - v e˙ to predprijatie. ‘Of course, I hoped to come a across something interesting, not without reason did I let myself be pulled into this business - or perhaps went into it myself.’
In this example, two alternative descriptions are given that illustrate this point well: either, the narrator let himself be pulled into the business (dal sebja vtjanut’), or, more actively, went into it himself, the latter being expressed by the same verb but with a middle morpheme (vtjanulsja, lit. pulled-himself into it). This illustrates that the permissive construction is situated on a cline of constructions involving more or less agentive properties of the participant serving as matrix subject. A passive construction (byl vtjanut ‘was pulled’) is less agentive than the permissive (dal sebja vtjanut’ ‘let himself be pulled’), which is in turn less agentive than the reflexive middle (vtjanulsja, lit. ‘pulledhimself’). An even more agentive conceptualization of the situation would be special’no vošel ‘purposefully went into’, where the initiative is completely on the side of the subject referent.16 In the reflexive permissive construction, the causee is thus typically specific but irrelevant; the center of attention is on the behavior of the causer and how he relates to the event he is both a partial instigator of and at the same time subject to. This is in contrast to non-reflexive permissives, where the causer typically instigates a potentially autonomous event. Reflexive permissives thus so to speak modulate the tension between agency, responsibility of the causer, and its patiency in these cases. The resistance (or non-resistance) to some external force or agent that is not completely under his control is highlighted; the interaction between causer and causee is backgrounded. But this is not necessarily so in this construction; causees may be introduced, if the interaction of causer and causee is salient, as in (140)
On priznalsja, cˇ to i segodnja ne ponimaet, kak dalPS T :S G i TarasovuDAT sebjaRFX:ACC ugovorit’INF . He confessed that today he doesn’t understand, how he could let himself be persuaded by Tarasov (lit. how he could let1 Tarasov2 persuade4 him3 ).
ii Jurij Bujda. ŠCina ˇ // Znamja, No.6, 2000
(1999-2000)
i Evgenij Rubin. Pan ili propal. Žizneopisanie
98 da(va)t+inf in Russian n=10: obmanut’ ‘cheat’; n=8: vtjanut’ ‘pull into’; n=7: obnjat’ ‘embrace’; n=5: pocelovat’ ‘kiss’; n=4: ubedit’ ‘persuade’, ugovorit’ ‘talk into’, sbit’ s tolku ‘confuse’, zapugat’ ‘frighten’; n=3: pojmat’ ‘catch’, uvleˇc’ ‘lead’, ubit’ ‘kill’; n=2: rezat’ ‘cut up’, podxvatit’ ‘pick up’, snjat’ ‘remove’, muˇcit’ ‘torment’, oduraˇcit’ ‘fool’, dognat’ ‘catch up’, sxvatit’ ‘seize’, ljubit’ ‘love’, žalet’ ‘pity’, otorvat’ ‘tear off’, uložit’ ‘put into bed’, oglušit’ ‘stun’, ograbit’ ‘steal from’, obojti ‘circumvent’, uvesti ‘lead away’, ubivat’ kill; n=1: 120 hapax legomena verbs Verb list 4: Complement verbs of reflexive permissives in the corpus (interpersonal, accusative reflexive coreferent with causer) (141)
Sergej cˇ ut’ pripodnjalsja, cˇ toby dat’INF AleksandreACC ii sebjaRFX:ACC pocelovat’INF . Sergej rose a little to let1 Aleksandra2 kiss4 him3 .
The construction is productive; the set of verbs encountered in the corpus with reflexive permissives involve a clear spread that is characterized by highly transitive verbs, but not limited to these (see list 4). Typical are verbs that denote something not in the causers interest that he or she would principally avoid, but did not succeed in avoiding, as with obmanut’ ‘cheat’, vtjanut’ ‘pull into’, ubedit’ ‘persuade’, ugovorit’ ‘talk into’, zapugat’ ‘frighten’, ograbit’ ‘steal from’; but this is not true of all verbs, as pocelovat’ ‘kiss’, dognat’ ‘catch up to’, ljubit’ ‘love’, žalet’ ‘pity’, uložit’ ‘put into bed’ arguably show. Note that reflexive permissives behave atypically in yet another respect: their tendency towards negative polarity and a correlation of negation and imperfective aspect is not as pronounced as with non-reflexives, cf. table 13. I attribute this to their semantics as reviewed above; we will find a similar picture in Polish.
Manipulative domain Not covered in the above tables were those cases where the causee is inanimate. This was the case in 9 examples, two of which are cited here. Note that inanimate causees are underrepresented in this class in comparison to the ii Aleksandr
ˇ Pjatigorskij. Drevnij Celovek v Gorode // Oktjabr’, No.11, 2001
Semasiological description of permissive da(va)t in Russian
99
DAVAT Aspect dat
neg
ipf pf Infinitive Aspect
7
9
45
23 Polarity
ipf
davat
6
pf
pos
10 51 17
Table 13. Matrix and complement aspect and negation for reflexive permissives. An association of aspect and negation is not significant (p=.42); matrix and infinitive aspect tend to agree (p<.01, V=.25).
overall class of permissives. (142) involves metonymy, (143) a conceptualization of a thing as having agentive properties, that is, as a force: (142)
On byl sil’nym cˇ elovekom i ne dalPS T :S G sebjaRFX:ACC slomit’INF ni tjur’meDAT , ni ssylkeDAT [. . . ].i ‘He was a strong person and didn’t let himself be broken by jail or exile [. . . ].’
(143)
I xot’ slova zvuˇcat muzykoj, ne davajteI MP:2PL ejDAT sebjaRFX:ACC i oglušit’INF ! ‘And even if these words seem like musici , don’t let1 yourself 3 be deafened4 by it1,i .’
Overall, this type is clearly less an analogue to non-reflexive manipulative causation, where something is effected by non-interfering with a spontaneous, independent process, than rather an analogue to the reflexive interpersonal permissive construction; both types are characterized by an intermediate degree of agency of the subject referent. This assessment is supported by comparative data from Polish, where inanimate causees are practically found only with reflexive, but not with non-reflexive permissives.
i Igor’ Efimov. Šag vpravo, šag vlevo // Zvezda, No.9, 2003 ˙ 1001 vopros pro ETO (1999)
i Vladimir
Šaxidžanjan.
100 da(va)t+inf in Russian Impersonal domain While reflexive permissives with inanimate causers are not generally felicitous, a single, exceptional such case was found in the corpus: (144)
A tol’ko strannaja vešˇc’ sˇcast’e. Srazu nikogda sebjaRFX:ACC ne ii dast3S G razliˇcit’INF . ‘But happiness is a strange thing. It never lets2 itself 1 be recognized3 immediately.’
This case involves a poetic personification and constitutes clearly creative usage, not generally acceptable to my informants. Its relevance lies in exemplifying a possible starting point for further developments: examples of this type constitute bridging contexts, the fixation and expansion of which have led to the development of modal passives in other languages, notably in Polish and Czech (see below).
Conclusions In Russian, the reflexive permissive construction has not grammaticalized into a specific construction. It does not differ in its morphosyntactic properties from non-reflexive permissives, is, with less than 10% of all permissive cases, comparatively infrequent and at the same time fully productive, as the distribution of verbs shows. It should be considered a clearly compositional instantiation of the reflexive construction in combination with the reflexive. However, the typical pattern of morphosyntactic realization differs from the non-reflexive permissive: in contrast to the latter, the causee in the reflexive permissive construction is typically backgrounded and omitted most of the time. This provides the ground for the further grammaticalization of reflexive permissives to modal passives as it has occurred in other Slavic languages, notably Polish and Czech; here, the structural tendencies in Russian are reflected in categorial restrictions. The same path of development can be expected to have played a role in the crosslinguistically noted grammaticalization of passives from reflexive permissives. This question will be taken up again below in the chapters on Polish and Czech. ii Marina
Višneveckaja. Vyšel mesjac iz tumana (1997)
Semasiological description of permissive da(va)t in Russian
2.2.6.3.
101
Evidence for auxiliarization as event integration
Both Nedjalkov & Nikitina (1965) as well as Podlesskaya (2005b) quote as evidence for the auxiliarization that complement negation of da(va)t is not possible. In fact it does seem to be possible, as this only case of double negation in my corpus shows: (145)
Kogda Marinka otkryla, on razom vtisnulsja v dver’, neNEG davajaCV B ejDAT neNEG vpustit’INF , posle cˇ ego kinulsja na kuxnju i [. . . ] ‘When Marina opened, he immediately squeezed himself through the door, not letting her not let her in, after which he dashed into the kitchen [. . . ].’
There is a semantic side to this: double negation of the permissive is equivalent to the factitive. The expected (and actual) meaning is thus he forced her to let me in. Informants differ in their assessment of this sentence. In any case, such use should be counted marginal. As Podlesskaya has shown, restrictions for complement negation seem to be semantic rather than structural and apply to pozvolit’ ‘allow’ in the relevant reading, as well. It seems permissive da(va)t cannot be used to express that a non-action is permitted / not prevented: it may, however, express that a nonaction was prevented, leading to a factitive interpretation (the latter case is parallel to modals of possibility: ne mog ne expresses necessity). Temporally, permissive da(va)t expresses one situation, as I conclude from the fact that two conflicting temporal adverbials may not be used for matrix and complement event, as (146-a). With pozvolit’, this is felicitous and leads to a speech act interpretation of giving permission: (146)
Vˇcera ja ne (*dalPS T :S G /pozvolilPS T :S G ) emuDAT pojtiINF na službu v vosskresenie. ‘Yesterday I did not (*let /allow) him (to) go to work on sunday.’
In general, adverbs that relate to an agent’s relation to the event always relate to the matrix agent: (147) *Ja neoxotno dalPS T :S G emuDAT radostno pojtiINF na službu. Intended: ‘I reluctantly let him gladly go to work.’ Both neoxotno ‘reluctantly’ and radostno ‘gladly’ relate to the causer, the agent of the causing event clause, and therefore lead to a semantic conflict. i Vladimir
Makanin. Antilider (1970-1990)
102 da(va)t+inf in Russian So called ni-items such as niˇcego, nikto ‘nothing, nobody’ etc. may normally only appear under predicate negation (Pereltsvaig (2006)). These items are licensed in the infinitival clause by negated davat’ as well as negated pozvolit’ ‘allow’, but not by negated prosit’ ‘ask for, request’: (148)
a. *On menja ne poprosilPS T :S G niˇcego skazat’INF . Intended: ‘He didn’t ask me to say nothing.’ b. On mne ne (dal/ pozvolilPS T :S G ) niˇcego skazat’INF . ‘He didn’t let / allow me (to) say anything.’
While ellipsis of the infinitive is possible, the infinitival event may not be taken up with a pronoun. Again, this applies to both pozvolit’ (in the relevant reading) and davat’; see the following example and its adapted version: (149)
Mnogie pytalis’ cˇ erez more ujti, no my imDAT ne daliPS T :PL e˙ to i sdelat’INF . ‘Many attempted to flee across the sea, but we [. . . ] didn’t let2 them1 do3 that.’
(149)’ [...] no my imDAT (∅/?˙etogoGEN ) ne (dali/pozvoliliPS T :PL ). ‘Many attempted to flee across the sea, but we didn’t (let/allow3 ) them1 (∅/?that2 )’ This is a sign that da(va)t enters a strong construction with the lexical infinitive, similar to byt’ as a formant of analytical future. This may be seen as symptom of complex predicate formation, that is, grammaticalization in terms of loss of autonomy of the sign. Note that in general, no restrictions on word order are observed, e.g., there is no restriction on participants to appear between da(va)t and the infinitive. I conclude that da(va)t+inf expresses one event with high semantic and some syntactic integration. The licensing of ni-words in the complement under matrix negation and the impossibility of referring to the infinitival situation with a pronoun suggests that da(va)t and the infinitive are parts of one complex predicate; however, no further symptoms of such univerbation or coalescence are found. A more detailed assessment of the status of this predicate would need to examine other classes of predicates as well in a more elaborate theoretical syntactic framework. This is left for further research.
i A.
Platonov. "Barbero" i "Barbudos" // "Trud", 1989.01.01
Semasiological description of permissive da(va)t in Russian
forms infinitive participle imperative morph. present past all
perfective dat’ of which positive 64 (98%) 14 (100%) 8 (100%) 11 (100%) 347 (98%) 444 (98%) =57% of 797 positive: 780 = 98% of 797; count 65 14 8 11 355 453
103
imperfective davat’ count of which positive 1 1 (100%) 128 127 (99%) 1 1 (100%) 112 110 (98%) 102 97 (95%) 344 336 (89%) =43% of 797 negative: 17 = 2% of 797
Table 14. Profile of davat’ ponjat’ ‘let understand’. Positive polarity dominates, da(va)t used in both aspects. (5 reflexive/passive daetsja ponjat’ not taken into account.)
forms infinitive participle imperative morph. present past all
perfective dat’ of which positive 21 (100%) 4 (80%) 14 (100%) 32 (89%) 79 (79%) 150 (85%) =39% of 449 positive: 389 = 87% of 449; count 21 5 14 36 100 176
imperfective davat’ count of which positive 10 9 (90 %) 27 24 (89%) 1 1 (100%) 113 105 (93%) 122 100 (82%) 273 239 (88%) =61% of 449 negative: 60 = 13% of 449
Table 15. Profile of davat’ znat’ ‘let know’. Imperfective da(va)t more frequent; positive polarity prevails.
2.2.7. 2.2.7.1.
Further constructions types ‘Letting you know’: cognitive and perception verbs
A numerically important class of constructions belongs to what I have called cognitive causation: the causee is an experiencer, and the combination of da(va)t and complement verb is similar to stimulus-experiencer verbs, on the one hand, and verbs of communication, on the other. Complement verbs roughly belong to the classes of inattentive perception and cognitive verbs.
104 da(va)t+inf in Russian What sets this type apart from the constructions so far considered is the fact that factitive causation is expressed. In contrast to permissive da(va)t, positive polarity dominates and imperfective matrix aspect is used frequently. Note that cognitive verbs with permissive da(va)t were treated above with the other verbs. This type is present in all three languages. It is at least partly lexicalized, although we can see changes in use that point to the compositionality of meaning. Generally, questions of animacy of causer and polarity do not play the role they play with permissive causatives: generally, these constructions can be used with both animate and inanimate causers. The following verbs were met in the corpus (with frequency): (150)
ponjat’ ‘understand’ 802, znat’ ‘know’ 451, poˇcuvstvovat’ ‘feel (pf.)’ 65, cˇ uvstvovat’ 6 ‘feel (ipf.)’, zametit’ 1 ‘notice’
Since this is a lexically constrained type, it is both instructive and practical to compare this data to that concerning texts from before 194517 . I will restrict myself to some remarks, rather than a full comparison, as the exact make-up of the RNC was not available from the web site and results are therefore generally hard to compare. The following verbs were met in this data, spanning from around 1800 to 1945: (151)
znat’ ‘know’ 144, ponjat’ ‘understand’ 42, poˇcuvstvovat’ ‘feel (pf.)’ 26, cˇ uvstvovat’ 18 ‘feel (ipf.)’, zametit’ 8 ‘notice’
It is interesting to see that znat’ ‘know’ and ponjat’ ‘understand’ have switched places as the most frequent complement verb. I will continue the discussion focusing on the modern part, referring to the pre-1945 part only in specific places.
Dat’ ponjat’ ‘let understand’, dat’ znat’ ‘let know’ The frequencies in (150) clearly show that this type is not really productive. Instead of a spread of many infrequent verbs and gradually less verbs with higher frequency, this type is virtually restricted to two verbs that make up more than 90% of all cases. The morphosyntactic profiles of both verbs are clearly different from the permissives; they are given in tables 14 and 15. There are some qualitative characteristics that also speak against compositionality and hence for lexicalization. With animate causers, both express a notion close to to inform (indirectly). The aspectual partners of ponjat’ and znat’ do not enter these constructions; formations such as dat’ ponimat’ (ipf.)
Semasiological description of permissive da(va)t in Russian
105
and dat’ poznat’/uznat’ (pf.) can only be used as regular permissives. Dat’ znat’ may take a directional object designating the experiencer (152), both allow an intermediary introduced with cˇ erez ‘through, by’ (153). (152)
Tam oˇcen’ ispugalis’, daliPS T :PL znat’INF vPREP milicijuACC , i ego i uvezli v morg. ‘There they got afraid, informed the police (lit. let know into the police), and he was brought to the morgue.’
(153)
Aleksandr Dmitrieviˇc cˇ erezPREP svoix ljudejACC dalPS T :S G ii ponjat’INF : on vstreˇcalsja s prezidentom otdel’no [. . . ] ‘A.D. proclaimed (lit. let3 know4 ) through1 his people2 : he met with the president separately [. . . ].’
With inanimate causers, davat’ o sebe znat’ ‘let know about oneself’, as in (154), is used differently, namely as a means to introduce some more or less direct evidence for some state of affairs, e.g. (154)
Grubovato davalPS T :S G oPREP sebeRFX:LOC znat’INF zapax utrom iii raspušˇcennyx v vode drožžej [. . . ] ‘The smell of yeast dissolved in the morning in water became coarsely apparent (lit. let about itself know).’
Unlike the permissive, these constructions can be passivized, cf. danoPPP:S G:N / daetsjaPS S :3S G namDAT ponjat’/znat’INF ‘we were/are given to understand’. Rather than being derived from the permissive, they seem to be cognitively related to giving, based on the metaphorization of information as a thing (Newman 1996). Arguments in favor of such a connection come from other languages. In Polish, this type may be expressed by verbal nouns that do not form the permissive construction, as in da´cINF doPREP zrozumieniaV BN:GEN ‘give to understand’. Combinations of cognitive verbs with give appear as lexicalizations in languages that do not form a causative with giving, such as German zu erkennen, verstehen, wissen geben or English to give to understand (see also section 6.2). But there is also evidence relating these combinations to regular causatives: these verbs regularly appear with similar meanings in unambiguously causative constructions: English let know, feel; German wissen, erkennen, fühlen lassen; in Finnish, antaa ymmärtää ’let understand’ can syntactically only be related to the causative auxiliary, not to the lexical verb give i Vasilij Katanjan. Loskutnoe odejalo (1943-1999) ii Aleksandr Sadˇ cikov. Partijnyj destaˇ bilizator. // «Izvestija», 2003.02.24 iii Galina ŠCerbakova. Podrobnosti melkix cˇ uvstv (2000)
106 da(va)t+inf in Russian (see von Waldenfels 2004, forthc.). It cannot be discarded that they have developed from permissives; according to Loewenthal (2003: 102f), Dutch zien, horen, merken, weten laten evolved before laten acquired the factitive meaning with other predicates.
Less frequent types In Russian, two peripheral members of this group need to be mentioned, dat’ (po)ˇcuvstvovat’ ‘let feel’ and dat’ zametit’ ‘let notice’. Today, (po)ˇcuvstvovat’ ‘feel’ as complement of davat’ is less frequent than znat’ ‘know’ and ponjat’ ‘understand’ by an order of magnitude. While factitive usage is dominant, in contrast to dat’ znat’/ponjat’ it may also be used in a permissive sense, as the next example illustrates. These cases were categorized as permissive (see also example (114) on page 83 above): (155)
[. . . ] pytalsja usypit’ bditel’nost’ protivnika [. . . ], ne dat’INF i emuDAT poˇcuvstvovat’INF stepen’ svoej podgotovlennosti. ‘[. . . he ] tried to lull the enemy, not let him feel the degree of his preparedness’
Factitive use is illustrated in the next two examples. They also highlight another difference to dat’ znat’/ponjat’, namely, that both aspects are used: (156)
I cˇ to bylo prijatno - nikogda ne davalaPS T :S G cˇ uvstvovat’IPF:INF , ii cˇ to ona ne prosto cˇ len partii, a ešˇce i žena Molotova. ‘What was nice, she never made (lit.: let) us feel that she was not simply member of the party, but also the wife of Molotov.’
(157)
[. . . ] našlis’ by ljudi, kotorye daliPS T :PL b emuDAT poiii cˇ uvstvovat’PF:INF , cˇ to otnyne on v e˙ tom mire cˇ užoj [. . . ] ‘[. . . ] people would have been found that would have given him to understand that he was a stranger in this world from now on.’
The imperfective infinitive cˇ uvstvovat’ is much rarer than the perfective poˇcuvstvovat’. The imperfective usually appears in a reflexive construction such as the following: (158)
Sledy perežitogo davaliPS T :PL sebjaRFX:ACC poˇcuvstvovat’INF v iv samye neožidannye momenty.
i Mit’ki. Papuas iz Gondurasa (1987) Poljanskaja. Sneg idet tixo-tixo (1993)
ii Nikita iv Emma ˙
Xrušˇcev. Vospominanija (1971) iii Irina Gerštejn. Lišnjaja ljubov’ (1985-2002)
Semasiological description of permissive da(va)t in Russian
107
‘Signs of what he went through were noticable (lit: let themselves feel) at the most unexpected times.’ This type, expressing approximately ’make itself noticeable’, is often used with inanimate sentence subjects, but it isn’t restricted to these. It seems that aspectual variation of poˇcuvstvovat’ and cˇ uvstvovat’ may have diminished in the past. Perfective aspect would clearly be preferred today in the following example from the first half of the 19th century: (159)
Ona byla so mnoj neprijaznenno dobra, nikogda ne govorila mne, cˇ to ja byla ej v tjagost’, no vsjaˇceski davalaPS T e˙ to cˇ uvstvovat’INF . i ‘She was good to me in a hostile way, never said that I was a burden, but let (me) feel it in many ways.’
In the older texts before 1945, imperfective cˇ uvstvovat’ is used in 18 of 44 cases, in newer texts, in 6 of 71, suggesting a narrowing of the paradigm. However, data are too sparse to be certain. It is remarkable that a similar development seems to have also taken place with dat’ ponjat’, which was much less frequent in comparison to dat’ znat’ in the nineteenth century than today. In the older corpus we find attestations18 of its imperfective counterpart dat’ ponimat’ ‘let understand (ipf)’, e.g.: (160)
Ona pela emu ego ljubimye pesni, pokazyvala emu svoj al’bom, zastavljala ego pisat’ v nego, ne pozvoljala pominat’ emu o starom, ii davajaCV B ponimat’IPF:INF , kak prekrasno bylo novoe [. . . ]. ‘She sang him his favorite songs, showed him her album, made him write something into it, didn’t allow him to mention the past and giving (him) to understand the beauty of what was new.’
In relation to both attested cases, my informants state that only the perfective ponjat’, not the imperfective ponimat’ would be correct today.
On the way out: dat’ zametit’ ‘let notice’ Only one modern example was found with zametit’ ‘notice’ as complement: (161)
i V.A.
No esli ty po kakoj-to priˇcine gde-to ne byl, možet byt’, komu-to obešˇcal ne byt’ ili sˇcital dlja sebja nevozmožnym i otkazyvalsja ot priglašenij, dalPS T :S G vsemDAT zametit’INF , cˇ to tebja tam ne by-
Sollogub. Istorija dvux kaloš (1839)
ii L.N.
Tolstoj. Vojna i mir. (1867-1869)
108 da(va)t+inf in Russian iii vaet, kak Blok - v "Brodjaˇcej sobake" ‘But if for some reason you didn’t go some place, perhaps, you promised not to go or deemed it impossible and declined the invitation, let everybody know (lit. notice) that it is not a place you would go to, like Blok in “Stray Dog” [. . . ]’
In this case, dat’ zametit’ can be understood to mean to inform in a subtle way. This corpus example, however, is not unambiguous, and today, dat’ zametit’ is not a frequent combination (n=3 after 1945). It would generally be understood in the permissive sense, as in (162) from the impersonal domain: (162)
Esli moi prevrašˇcenija i ne vsegda byvali udaˇcny, vse že strax i posetitelej ne davalPS T :S G imDAT zametit’INF nedostatki. ‘Even if my transformations were not always successful, the fact that they were afraid kept the spectators from noticing ( lit. didn’t let them notice) their limitations.’
However, in the 19th century, especially in the first half, it was used unambiguously as a variant of dat’ ponjat’/poˇcuvstvovat’, as in the next example: (163)
Samym delikatnym obrazom dam1S G ejDAT zametit’INF , cˇ to doˇceri ii ee darom soderžat’ ne nameren. ‘I’ll let her know (lit. notice) in the most delicate fashion that I am not planning to support her daughter for free.’
Together, in the nineteenth century, 8 such uses were present in the corpus, virtually all of them in the first half of the century, as opposed to 7 permissive uses, most from the second half. In the twentieth century, only the single, doubtful example quoted in (161) above was found in factitive usage; 4 more uses of dat’ zametit’ were clearly permissive (dat’ zameˇcat’ was not found). These figures are even more significant in the light of the fact that the RNC has much less text from the 19th than from the 20th century (however, no figures for the exact corpus make-up were available). For some time in the 19th century, this collocation must have thus had some currency. It may have been a calque from French faire remarquer, but note that the latter is formed with the factitive, not the permissive auxiliary.
iii Anatolij Najman. Rasskazy o Anne Axmatovoj (1986-1987) i Aleksandr Volkov. Volšebnik Izumrudnogo goroda (1939) ii A.A. Fet. Djadjuška i dvojurodnyj bratec (1855)
Semasiological description of permissive da(va)t in Russian
109
Conclusions With cognitive verbs, da(va)t may denote factitive causation. This is limited to a small set of verbs with symptoms of varying degrees of lexicalization, that is, non-compositionality. The set of verbs and degree of lexicalization is subject to change. Dat’ ponjat’ and Dat’ znat’ are the most frequent and least variable types, although aspect variation may still have been possible with ponjat’ in the 19th century. Dat’ poˇcuvstvovat’ is less frequent and may take both aspects. In the 19th century, zametit’ was included in this set of verbs; it has since lost this function. As we will see in the analysis of the other languages, such language-internal variation is parallel to the variation across languages and points to a cognitive basis for independent lexicalization of this type, not necessarily related to permissive causation. 2.2.7.2.
Impersonal passive da(va)t plus infinitive
A construction with the subjectless singular neuter passive past participle form dano is used 267 times in the corpus. As in (164), it conveys the endowment by fate or god with some oppurtunity or gift. (164)
V cˇ isle mnogogo, cˇ ego ja lišen, mneDAT ne danoPPP:S G:N postiˇc’INF i prelest’ i smysl salonnoj žizni. ‘Among many things I am deprived of, it was not given2 to me1 to gain3 access to the beauty and meaning of Salon culture.’
n=35: znat’ ‘know’; n=24: ponjat’ ‘understand’; n=11: uvidet’ ‘see’, predugadat’ ‘foresee’; n=8: uznat’ ‘find out’; n=7: videt’ ‘see’; n=6: postiˇc’ ‘gain’; n=5: stat’ ‘become’; n=4: uslyšat’ ‘hear’, byt’ ‘be’, proniknut’ ‘transcend’, providet’ look ‘through’; n=3: 4 verbs; n=2: 17 verbs, 98 hapax legemena verbs. Verb list 5: Infinitives used with dano in the corpus. This type is productive, as verb list 5 shows. It is stylistically quite marked and usually used with negation (163 cases / 61%); I consider it only weakly linked to the permissive and rather an exponent of metaphorical giving. In any case, this has to be considered a construction of its own, linked to a particular morphosyntactic form of da(va)t. i Mixail
Veller. Nožik Sereži Dovlatova (1997)
110 da(va)t+inf in Russian 2.2.7.3.
Secondary predication and referential use of infinitives
In Russian, unlike Czech and Polish, many examples were found that were ambiguous. On the one hand, these usages contain elements of giving in its transfer sense; on the other, they can be construed to involve some shade of causative or permissive meaning. Podlesskaya calls these uses hybrid constructions and argues for their grammaticalized status; this position is not shared here (see the detailed discussion on pages 49ff. above). Syntactically, these usages should be considered pivot constructions where the infinitive functions as a secondary predicate denoting a purpose. This type is quite frequent. It is characteristic that it is underspecified: either the ‘giver’ supplies a possibility for the ‘receiver’ because the latter wants to do something with it, as in the next example: (165)
Sejˇcas ja cˇ itaju knižku francuzskogo pisatelja R. Rollana, mneDAT i eë dalPS T :S G proˇcitat’INF Gonza, i mne ona strašno nravitsja. ‘Right now I am reading a book by the french writer R. Rolland, Gonza gave2 it to me1 to read3 , and I am terribly fond of it.’
Or, conversely, the ‘giver’ gives something to the ‘receiver’ because the ‘giver’ wants the ‘receiver’ to do something with it, as in (166): (166)
Dlja proverki sposobnostej on davalPS T :S G proˇcest’INF ii každomuDAT kusoˇcek skazki. ‘In order to test their skills he gave1 each3 of them a piece of the fairy tale to read2 ’
We are thus dealing with unspecified modality in the purpose clause. In many cases, no decision regarding the type of modality is possible, nor necessary. What is focused on is giving, not causation or modality. It is not always possible to draw a clear line: is (167) a permissive or a hybrid construction? ‘Giving’ in this case implies ‘letting somebody do something with what is given’. (167)
iii Ja dalaPS T :S G ejDAT nadet’INF svoe staroe cˇ ernoe pal’to. ‘I gave her my old black coat to wear’ or ‘I let her wear my old black coat’.
During annotation, it was attempted to be as conservative as possible; in cases of doubt, the hybrid, rather than permissive interpretation was chosen. i Dnevnik devoˇ cki (1990) Pereˇcen’ obid (1997)
ii Fazil’
Iskander. Muˇceniki sceny (1989)
iii Emma ˙
Gerštejn.
Semasiological description of permissive da(va)t in Russian
111
n=85: poˇcitat’ ‘read’; n=74: cˇ itat’ ‘read’; n=60: zakurit’ ‘smoke’; n=43: vypit’ ‘drink’; n=37: poest’ ‘eat’; n=33: prikurit’ ‘light up’; n=29: proˇcitat’ ‘read’, est’ ‘eat’; n=28: posmotret’ ‘look at’; n=27: ponjuxat’ ‘take a smell’; n=23: poderžat’ ‘hold’, poprobovat’ ‘try’ (121 more verbs). Verb list 6: Secondary predicates / referring infinitives (hybrid constructions). It should be remarked that in the other languages, no such difficulties of interpretation existed; this is due to specialization of purposive or permissive construction in Polish and Czech; the same, interestingly, holds in Finnish (von Waldenfels 2004, forthc.). It is difficult to draw a line to the use of referential infinitives that metonymically denote their object. This class prototypically includes consumptive verbs (see discussion of Podlesskaya’s analysis above), such as (po)est’ ‘eat’, (po)pit’ ‘drink’, zakusyvat’/zakusit’ ‘snack’, zakurit’ ‘light up’. However, this group extends beyond these verbs and is potentially productive; e.g., poˇcitat’ ‘read’ may also be used in this way. A necessary precondition seems to be a sortal object implicated by their semantics. This use is exemplified with a series of verbs in (168): (168)
Ty (kušat’/kurit’/opoxmelit’sja/poˇcitat’INF ) prinesPS T :S G ? Did you bring2 something to eat/smoke/drink as cure for a hangover/read1 ?
Due to the difficulty in delimiting these types, they were conflated in corpus annotation. A list of the most frequent verbs is given in list 6. Note that in Polish, equivalent cases are constructed with verbal nouns, rather than with infinitives. Constructions with infinitives as secondary predicates and as referring expressions are thus not specific to da(va)t; furthermore, their equivalents in other languages investigated here involve other types of complements. A principled comparison would need to take non-infinitive constructions in Polish into account and lies outside the scope of this study.
112 da(va)t+inf in Russian 2.2.8.
Summary
In this section, I have presented a quantitative and qualitative analysis of da(va)t+inf in Russian, focusing on permissive causation. As a permissive marker, da(va)t is used primarily non-reflexive; its use spans all domains in relation to control and animacy of causer and causee. It is most often used and semantically most unrestricted in negative polarity, denoting intervention; in positive use, its use covers a continuum including both enablement and non-intervention, but with most emphasis on the latter. Semantically, we can say it is a quite underspecified marker of permissive causation with a focus on intervention and non-intervention. The second important function of da(va)t in Russian is the use of its perfective and imperfective imperative in various hortative and imperative constructions. While this function is not covered in detail, we have seen that it is not possible to draw a clear line between the use of the perfective imperative daj in the finite clause construction and in the permissive construction. There is a functional overlap, and grammaticalization of this form in one construction seems to influence the frequency of this form in the other construction, as well. Thirdly, lexicalization patterns of da(va)t with cognitive verbs have been examined, which have been shown to change over time, pointing to a compositional basis for these constructions.
Chapter 3 da(va)t+inf in Polish 3.1.
Introduction
This chapter gives an analysis of Polish da(va)t+inf, partly drawing on discussions in the preceding chapter on Russian. This analysis is primarily synchronic, but some attention is paid to recent developments as evident in the IPI PAN corpus; comparative data is adduced from ParaSol. da(va)t in Polish differs from Russian in several respects: first, it not only denotes permissive, but also factitive causation; second, the reflexive permissive is very frequent and shows symptoms of grammaticalization, while imperatives of da(va)t are unlike their Russian counterparts neither grammaticalized nor especially frequent; third, da(va)t with the reflexive has developed to form a set of modal constructions.
3.1.1.
Syntactic types
In relation to the syntactic typology of causatives in the introductory section, causatives in Polish divide into those with a dative causee, that is, type I causative constructions, as in (169), and those with a prepositional causee, that is, type II causatives as in (170): (169)
Nie dałem mui sko´nczy´ci NEG let.PST.1SG him.DAT finish.INF ‘I didn’t let him finish.’
(170)
Nie dałemi si˛e oszuka´c przez niegoi . NEG let.PST.1SG RFX.ACC cheat.INF by him.ACC ‘I didn’t let myself be cheated by him.’
The properties of these constructions form a major topic of this chapter. Moreover, da(va)t forms modal constructions with (171) or without (172) grammatical subject: (171)
To nie da si˛e zrobi´c that.NOM not let.3SG RFX.ACC do.INF ‘That (subject) cannot be done’
114 da(va)t+inf in Polish (172)
Tego nie da si˛e zrobi´c that.GEN not give.3SG RFX.ACC do.INF ‘One cannot do that that (object)’
Other grammaticalized constructions, notably a construction involving a finite complement clause19 were not found in the corpus and not taken into account.
3.1.2.
Prior research
Szlifersztejnowa (1968, 1970) covers both causative and modal da(va)t+inf constructions in the context of research on passive and reflexive constructions. As a material basis, she uses classical texts spanning the whole history of Polish; her scope involves both diachrony and synchrony, albeit in a not always clearly differentiated approach. Szliferstejnowa focuses on permissive and factitive causation, tracing the relationship of these two meanings and including other periphrastic causatives such as kaza´c and pozwala´c in her discussion. This research is most relevant in the diachronic context. Topoli´nska (1993) focuses on synchronic da(va)t plus infinitive, likewise discussing its permissive and factitive function. According to Topoli´nska (1993: 241ff.), these functions are present in two distinct constructions (terminology adapted): the factitive, as in dałam posprzata´ ˛ c mieszkanie, which does not allow mentioning of the causee and is of minor productivity; and the permissive, as in dzieci nie daja˛ Annie czyta´c which is more productive, but has to have some explicit or implicit negative element in the context. As we will see, both assertions are not borne out with the data in absolute terms, but they do hold as tendencies. Corpus analysis adds to the picture given by Topoli´nska, and in fact partly falsifies it.
3.1.3.
Material & Method
As material basis for the analysis of da(va)t in Polish, I used data from the IPI PAN corpus of Polish. The corpus is given in two versions: the full, 300 million token corpus with a clear predominance of newspaper texts, and a smaller, more balanced ‘sample’ subcorpus of 30 million tokens. The larger corpus was used only for specific, low-frequency phenomena; it will be referred to as the ‘full’ IPI PAN corpus. The 30 million sample corpus was
Introduction
115
analyzed systematically for patterns of da(va)t+inf. On the IPI PAN corpus web site20 , the make-up of the sample corpus is described as follows: • contemporary prose: over 10% • older prose: almost 10% • non-fiction: 10% • newspapers: 50% • parliamentary proceedings: 15% • law: 5%
The older prose section includes classical literature from the last quarter of the 19th to the first quarter of the 20th century. This part of the corpus was analyzed separately and will be called the older prose subcorpus. The data analyzed in this section primarily concerns the synchronic part, dating from not later than the 1970s; the older prose subcorpus is analyzed only selectively here and will be primarily used for analysis in the dedicated diachronic part. As for Russian, I extracted all instances of da(va)t followed by an infinitive within a maximum of 5 non-punctuation tokens intervening. The resulting 3,132 hits were input into a database using MS Access. Presence of the negator nie and the reflexive particle si˛e were automatically marked using perl scripts; morphosyntactic and lexical information concerning matrix and complement verb were extracted from the corpus results. It should be noted that meta information may not be easily exported from polyqarp, the IPI PAN corpus query client. Sources for Polish examples are therefore not given; note they may be easily looked up in the corpus at www.korpus.pl. The following types were distinguished during annotation: • modal passive in several types (discussed in detail below), as in (173)
i Cezary
Ideowy profilNOM:S G:M naszego bohatera dałby3S G:M:IRR si˛eRFX:ACC wówczas okre´sli´cINF trzema poj˛eciami: antykomui nista, anarchista, agnostyk. ‘The ideological profile of our hero at the time could be described (lit: let itself describe) with three concepts: anti-communist, anarchist, agnostic’ Michalski, Siła odpychania, 2002
116 da(va)t+inf in Polish • permissive (174) or factitive causative (175): (174)
W 89 min postanowił [. . . ] da´cINF odpocza´ ˛cINF PiotrowiDAT i JawieniowiDAT . ‘In the 89th minute he [. . . ] decided to let1 Piotr3 Jawien´ 4 rest2 .’
(175)
To oni na przykład poddali mu pomysł, z˙ eby dałPS T :3S G sobieRFX:DAT zamontowa´cINF w domu dwudziestoczteroii godzinna˛ inwigilacj˛e wideo. ‘So they for example gave him the idea to have 24 hour video surveillance installed at home (lit. have himself video surveillance install)’
• potentially ambiguous cases where the decision between permissive or factitive causative is not clear, as in (177): (176)
Poczekaj, syneczku, mamusia musi tylko da´cINF si˛eRFX:ACC iii sfilmowa´cINF [. . . ]! ‘Wait, son, mama only has to get her picture taken / let them take her picture (lit. let/have herself photograph)!’
• Passive matrix verb constructions involving the neuter past passive participle dane and a slightly different modal/permissive meaning: (177)
Z czasem danePPP:N mime:DAT byłoAUX:PS T :3S G:N odkry´cINF i iv doceni´cINF t˛e stron˛e jego osobowo´sci i jego z˙ ycia [. . . ] ‘With time it was3 given1 to me2 to recognize4 and appreciate5 that side of his personality and his life [. . . ]’
• Constructions with perception and cognitive verbs such as (178), including da´c zna´c and others: (178)
DałPS T :3S G si˛eRFX:ACC pozna´cINF jako wzorowy pracownik, dov bry nauczyciel, organizator oraz człowiek ‘He became known (lit.: gave himself to recognize) as an exemplary worker, a good teacher, organizer and man.’
i Dziennik
Polski - 2001-05-28 - Sport ii Max Cegielski, Apokalipso, 2004 Sieniewicz, Czwarte niebo, 2003 iv Paulas, S˛ekalski, Oni nas stworzyli, 2002 Polski - 2004-02-18 - Podkarpacie
iii Mariusz v Dziennik
Introduction
type modal passive permissive factitive ambiguous perm./fact. dane + inf. cognitive other inf. phraseologic irrelevant
abs. 1,624 889 11 15 90 279 33 10 182 3,132
117
rel. 52% 28% 0% 0% 3% 9% 1% 0% 6% 100%
Table 16. Types annotated for Polish. Modal passive and permissive are clearly the most frequent types.
• Other infinitive constructions with da(va)t in transfer meaning and an infinitive either metonymically referring to its object as in da´c je´sc´ , zje´sc´ , napi´c, pi´c ‘give to eat, drink’or as a secondary predicate (180): (179)
Mo˙zesz to sobie wzia´ ˛c – powiedziała kobieta. – Ja jeszcze mam. i Dasz2S G innymDAT przeczyta´cINF . You can take it - said the women. I have another one. Give it to others to read.
• Phraseological types that cannot be derived from the base verb: (180)
ii DajmyI MP:1PL popali´cINF oszustomDAT ! ‘Let’s show those crooks!’ lit. ‘Let’s let/give1 those crooks3 burn2 ’
• False hits including infinitives that are not dependent on da(va)t. Similar to the procedure in Russian, all occurrences of da(va)t in causative usage and with cognitive verbs were annotated concerning the following properties: presence and animacy of causer and causee; type of reflexive; form of the causee as dative or prepositional phrase. i Włodzimierz
Pawluczuk, Judasz, 2004
ii Dziennik
Polski - 2002-01-25 - Magazyn
118 da(va)t+inf in Polish The breakdown into types is given in table 16. It shows that the modal passive is the most frequently met type, with over half of all examples extracted. Next in frequency is the permissive type, with about a third of all occurrences; factitive usage is clearly marginal with less than one percent. As in Russian, the number of cases labeled as cognitive causation amounts to about one third of the permissives. It is remarkable that secondary predicates and, generally, irrelevant cases are much less frequent than in Russian21 . This seems to be due to differences in the functions of the infinitive: in Polish, purposive secondary predicates as well as predicates in light verb constructions are often formed with deverbal nouns: (181)
(182)
Dałam jego utwory do przeczytania Give.PST.1SG his work.ACC.PL for read.VBN.GEN znajomym i bardzo si˛e podobały. friends.DAT.PL and very RFX.ACC like.PST.3PL. ‘I gave his works to friends to read and they liked them a lot.’
i
Tre´sc´ [. . . ] rzeczywi´scie nie dawała Contents.NOM ... really NEG give.PST.3SG z˙ adnej podstawy do zredagowania any.GEN.SG grounds.GEN.SG for edit.VBN.GEN.SG papieskiego dokumentu. papal.GEN.SG document.GEN.SG ‘The contents really didn’t provide any grounds for an editing of the ii papal document.’
The infinitive in Polish does posess a purposive function (see Bartnicka 1982 and example (179) above), but in comparison with Russian, it seems to be much more rarely used, as, for example, reflected in a frequent equivalence of Rs. dat’ proˇcitat’ and Pl. da´c do przeczytania, both ‘give someone something to read’, in the parallel corpus. In light of prominent use of daj with finite predicate in Russian it needs to be mentioned that in a very colloquial register, a similar construction involving the first person is also notable on Polish. While no such case was found in the IPI PAN Corpus, which does not contain recordings of recorded informal speech, the National Corpus of Polish22 (balanced sample) records several such instances in its spoken part, e.g. i Maria Rostworowska, Portret za mgła˛ opowie´sc´ o Oldze Bozna´ nskiej, 2003
Szczur, Papie˙z Urban V i powstanie uniwersytetu w Krakowie w 1364 r., 1999
ii Stanisław
Permissive da(va)t in Polish
(183)
119
cukierki te˙z sa˛ dobre z tym ajerkoniakiem nie ?.. yhm dobre sa.˛ dajI MP:2S G ciDAT dolej˛e1S G jeszcze troch˛e.. nie ju˙z nie nie nie.. iii ‘those candies are great with that eggnogg, aren’t they? Hm, great. Let1 me pour3 you2 some more ... no, that’s enough, no, no... ’
However, as this example illustrates, such usage belongs to a very colloquial register, is clearly not felicitous in writing and does not seem to be generally acceptable. It is not further investigated in this study.
3.2. 3.2.1.
Permissive da(va)t in Polish Overview: reflexive and non-reflexive
In Polish, the reflexive permissive has grammaticalized further and shows characteristics of a construction of its own right that can no longer be compositionally derived from a combination of the permissive and the reflexive. In arguing for such an analysis, the comparison to Russian plays an important role, since such a developement has arguably not taken place there (see preceding chapter). The typology of reflexive relations is very similar to that in Russian and will not be repeated here23 . As in Russian, only those cases where a reflexive pronoun is bound by the causer, that is, where the causer is in some way a participant of both the causing and the caused situation, will be called reflexive without further qualification, cf. (184) as opposed to (185), where the reflexive pronoun is bound inside the complement clause: (184)
(185)
Ale mud˙zahedinii nie dadza˛ si˛ei rozbroi´c. But mujahideen.NOM.PL NEG let.3PL RFX.ACC disarm.INF ‘But the mujahideen didn’t themeselves be disarmed’ Dajmy imi si˛ei rozwija´c. Let.IMPER.1PL them.DAT.PL RFX.ACC develop ‘Let’s let them develop’ (lit.: ‘develop themselves’)
i
ii
In general, the accusative reflexive clitic si˛e may be replaced by its tonic counterpart siebie, e.g., in a contrastive context. However, in actual texts this is a rare phenomenon and no such instance was found in the corpus; cf. the constructed example in (186): iii Rozmowa przy farbowaniu włosów [conversation while dyeing hair], Łód´ z, recorded i Dziennik Polski - 2001-12-22 ii Parliamentary discussion, 1999 2002
120 da(va)t+inf in Polish type non-reflexive comp. int. refl. accusative refl. dative refl. prepositional refl.
abs 153 7 653 69 7 889
rel. 17% 1% 73% 8% 1%
Table 17. Reflexive and non-reflexive permissive da(va)t in the IPI PAN sample corpus. Comp(lement) int(ernal) refl(exive) concerns cases where the pronoun refers to the subject of the complement clause, analogous to English we let him enjoy himself ; types below the line refer to the reflexive proper, analogous to English he let himself be cheated .
(186)
Dał nie tylko siebiei , ale te˙z innych oszuka´c. Let.3SG NEG only RFX.ACC but also other.ACC.PL betray.INF ‘He let not only himself, but also others be betrayed.’
In the remainder of this section, I first establish important differences of the non-reflexive and reflexive permissive from a bird’s eye perspective, examining their profiles of use. I then analyze their semantics in separate, before focusing on the form of the causee in reflexive permissive, which I take to be an important indicator of the grammaticalization of the latter construction.
3.2.2.
Usage profiles
In clear distinction to permissive da(va)t+inf in Russian, the core of use of the Polish construction is with the reflexive: as shown in table 17, only 17% of all permissive usages do not involve a reflexive marker. This is in stark contrast to the Russian situation, where more than 90% of all permissive phrases were non-reflexive (see page 94). Note that this is not an artefact of different make-up of the IPI PAN corpus and the RNC: this contrast is also found in the parallel corpus, where the contents of the Russian and Polish part are by definition identical (see section 5 from page 233 on the parallel corpus study). Here, as table 18 shows, the same tendency is apparent: in Russian, the reflexive permissive is used in only about 30% of all cases, while in Polish, this configuration amounts to
Permissive da(va)t in Polish
type non-reflexive comp. int. refl. accusative refl. dative refl. prepositional refl.
Russian abs rel. 57 75% 16 21% 2 3% 1 1% 73
121
Polish abs. rel. 29 28% 3 3% 63 60% 6 6% 4 4% 105
Table 18. Reflexive and non-reflexive permissive da(va)t in the Russian-Polish parallel corpus (see table 17). For a description of the corpus, see section 5 below.
around 70% (difference significant at *** with a large effect size of Cramer’s V=.44). The tables under 19 show the distribution of morphological forms and the relative amount of negative phrases for reflexive and non-reflexive da(va)t separately. The special status of the perfective imperative of da(va)t we have seen in Russian is not as pronounced in Polish: while it is found less often negated in the imperative than in other perfective finite forms (p**, V=.3), it is not especially frequent, as in Russian. This is in line with the observation that the use of the imperative in Russian is exceptional and should be regarded a special development (see there), but that this development is based on a more general pattern, namely a certain potential function of the imperative of the permissive. Continuing with an eyeball assessment of morph. forms, we see that the infinitive is likewise not exceptionally frequent (recall that polarity figures here are not reliable in view of its dependence on higher-order predicates). I would like to link this to a smaller functional load of infinitives in Polish in general (viz. the role of the verbal noun in this language) and have no further comments. It may be remarked that reflexive da(va)t is mostly used in the past tense. Here and in table 20, accompanied by mosaic plots, we see that the tendency towards negative polarity is not as strong in Polish as in Russian (cf. table 4 above); only 39% of the non-reflexive, and 50% of the reflexive constructions are negated. The difference between the two types is significant, but small: p<.05, V=.09. These mosaic plots also show that a correlation of nega-
122 da(va)t+inf in Polish Non-reflexive permissive perfective da´c imperfective dawa´c forms count negative count negative infinitive 26 2 (8%) participle -6 3 (50%) imperative 18 2 (11%) morph. present 26 14 (54%) 17 8 (47%) past 58 25 (43%) 9 8 (89%) all 128 43 (34%) 32 19 (59%) =80% of all =20% of all negative: 62 of 160 (=39%) Reflexive permissive perfective da´c imperfective dawa´c forms count negative count negative infinitive 97 34 (35%) 2 participle 19 14 (74%) imperative 42 37 (88%) morph. present 76 48 (63%) 54 15 (28%) past 410 202 (49%) 29 16 (55%) all 625 321 (51%) 104 45 (43%) =86% of all =14% of all negative: 366 of 729 (=50%) Table 19. Non-reflexive permissive and reflexive permissive: in terms of aspect, forms and negation. 62 of 160 non-reflexive cases (=39%) are negated, with perfective da(va)t used in 80% of the cases. 366 of 729 reflexive instances (=50%) are negated; perfective da(va)t constitutes 86%. Nonreflexive includes complement internal reflexives.
tion and imperfective matrix aspect is significant only for the non-reflexive permissive (p<.05, V=.21); there is no correlation of negation and aspect for the reflexive permissive (p=.3 for negative and positive instances to belong to the same distribution). Recall that we found that this correllation was in Russian likewise not given for reflexiv constructions (see table 13). For reflexive and non-reflexive permissives we find, like in Russian, that the aspect of the
Permissive da(va)t in Polish
matrix aspect perf perf imperf imperf
compl. aspect imperf perf imperf perf
Non-refl. all neg. 23 9 (39%) 105 34 (32%) 13 9 (69%) 19 10 (53%) 160 62 (39%)
all 22 597 31 68 718
123
refl. neg. 14 (64%) 306 (51%) 10 (32%) 35 (51%) 365 (51%)
DAVAT Aspect
9
davat
Polarity
neg
10
35
71
DAVAT Aspect dat 14
306
8 pos
21 291
ipf pf Infinitive Aspect
14
pf
Polarity pos
4
ipf
neg
34
10
ipf pf Infinitive Aspect
9
ipf
dat
9
pf
davat
33
Table 20. Distribution of matrix and complement aspect with reflexive and nonreflexive permissives. 11 reflexive cases with biaspectual complement were not taken into account.
infinitive tends to agree with aspect of da(va)t (same distribution at p<0.001, V=.37 for reflexive and p<.05, V=.22 for non-reflexive permissives). Turning to other features which we found to be characteristic for reflexive, rather than non-reflexive permissives, we see in table 21 that the proportion of causee-less phrases is, like in Russian, sensitive to this distinction. In non-reflexive constructions the causee is omitted only in about 14% of
124 da(va)t+inf in Polish function/case non-reflexive compl.intern. accusative dative other
all 153 7 653 69 7
w/o causee 21 1 453 66 6
all causee omission 22 of 160 (14%)
525 of 729 (72%)
Table 21. Distribution of causee-less phrases with reflexive and non-reflexive permissives over the whole corpus, given in terms of reference and function of the reflexive element: difference at ***, V=.46.
cases, while in reflexive constructions it is omitted in 72% of all instances; the difference is at the same time highly significant and large (V=.46). Summarizing, we see that some characteristics of permissive da(va)t+inf coincide across Polish and Russian: a certain propensity to be used under negation that is weaker with the imperative; a tendency for matrix- und infinitive aspect to agree and perfective aspect to be more frequent in both, more clearly so in Polish, and more with reflexive than with non-reflexive permissives; a tendency for the causee in reflexive permissives constructions to be omitted; a tendency for negative polarity to be less prominent with reflexive permissive than with non-reflexive permissive and a weaker association of negation with aspect in the former. A difference to Russian seems to be the overall more marginal status of the imperfective. Recall that in Russian, the relation of imperfective and perfective matrix verb was roughly balanced, with imperfective used almost always with negation. In both Polish constructions, use of the imperfective aspect of da(va)t is less than 20% on average (table 19). In both reflexives and nonreflexives, the combination of perfective matrix and complement is by far the most frequent. We will see below that this is true for modal passive da(va)t as well. It will emerge that such a tendency is generally less pronounced with lexicalized types, and stronger with more grammaticalized types. This could be a symptom of the ongoing loss of the category of aspect of da(va)t in grammaticalization, a development that has taken place already with the modal passive in Czech and is ongoing in Polish (see below). However, at this point, we can just assert that quantitatively, perfective da(va)t is the default case for the permissive in Polish.
Permissive da(va)t in Polish
domain Interpersonal Manipulative inanim. causer, anim. causee both inanimate
n 142 6 12 1 160
rel. 88% 4% 8% 1% 100%
abs 50 1 10 1 62
125
negated rel. to domain 35% 17% 83% 100%
Table 22. Distribution of non-reflexive permissive da(va)t across domains.
That the Polish and Russian constructions are from a bird’s eye’s perspective in many respects quite similar makes their most striking difference stand out even more: the contrast in relative proportion of the constructions. While the reflexive permissive is rather marginal in Russian, in Polish, it is the single most important type. In the following sections I will show that these differences in frequency are concomitant to differences on a semantic and a syntactic level that show that in Polish, the reflexive permissive has traversed further on a scale of grammaticalization.
3.2.3.
Non-reflexive permissive
The present section presents a semantic analysis of non-reflexive permissives. Generally, since the denotative potential of Polish non-reflexive permissive da(va)t is much akin to its Russian cognate, I will not repeat all the explications given in section 2.2 in chapter two , but describe Polish da(va)t largely in reference to Russian, focusing on common and distinguishing traits. As we can see from table 22, the distribution over domains is very uneven: almost 90% of all attestations come from the interpersonal domain. In the remainder of this section, I analyze the non-reflexive permissive on a per-domain basis. Regarding positive interpersonal causation, at least two subtypes, as in Russian, can be noted: non-intervention (187) and enablement (188): (187)
[. . . ] Kwa´sniewski gło´sno ostrzega: zostawi´c rad˛e w spokoju i i da´cINF jejDAT pracowa´cINF . ‘Kwa´sniewski warns perceptively: leave the council in peace and let
i Dziennik
Polski - 2001-12-22 - Pejza˙z - Słowniczek
126 da(va)t+inf in Polish it work (= let it do it’s business).’ (188)
W tym meczu dałemPS T :1S G zagra´cINF wszystkimDAT chłopcomDAT , tak˙ze junioromDAT , którym b˛edziemy dawa´c ii szans˛e. ‘In this match I let1 all3 the boys4 play2 , also the juniors5 , who we will give a chance.
Waiting was not present in the corpus. Examples such as the constructed ˙ (189) ?Zeby mu nie przeskadza´c, dawałemPS T :1S G muDAT najpierw wyj´sc´ INF na sluzb˛e, a potem sam wstałem. ‘In order not to disturb him I let him leave to work first, and got up after that.’ (190) ?DałPS T :3S G muDAT najpierw przej´sc´ INF do stacij, zanim sam poszedł. ‘He first let him reach the station before going himself.’ seem to be dispreferred. It also seems that enablement in both deontic and dynamic domains are more often met than in Russian. However, the differences seem to be rather in usage preferences than in clear categorial differences in semantics, and neither this corpus nor the parallel corpus data provides enough data to validate such a generalization. Generally, it seems Polish da(va)t has a very similar denotative potential, with the differences apparent more in the patterns emerging from larger samples of usage than in individual, disallowed readings. Often, da´c is seen as stylistically inappropriate by informants who prefer pozwoli´c. This observation is in line with data in the parallel corpus, where Polish pozwoli´c ’allow’ rather than da(va)t is a regular translational equivalent of Russian non-reflexive davat’ (see parallel study below). n=28: odpocza´ ˛c ‘rest’; n=18: pogra´c ‘play’; n=8: z˙ y´c ‘live’, zarobi´c ‘earn money’; n=5: sko´nczy´c ‘finish’; n=4: spa´c ‘sleep’; n=4: doko´nczy´c ‘finish’; n=4: doj´sc´ ‘arrive’; n=3: powiedzie´c ‘say’; n=2: mieszka´c ‘live’, gra´c ‘play’, odpowiedzie´c ‘answer’, zje´sc´ ‘eat’, wytchna´ ˛c ‘rest’, by´c ‘be’, tkna´ ˛c ‘touch’, przej´sc´ ‘go through’; 44 hapax legomena. Verb list 7: Verbs in interpersonal causation, ordered by frequency. ii Dziennik
Polski - 2003-10-06 - O´swi˛ecim - Ku´znia talentów
Permissive da(va)t in Polish
127
Verb list 7 gives an overview of the verbs used in this domain. The number of examples is quite low; nevertheless, we can see some trends. The in comparison to Russian lower tendency to appear under negation is confirmed by the distribution of negation with hapax legomena verbs: of 44 verbs, 26 are positive and only 17 negative, amounting to about 40%. The list bears clear resemblances to the Russian list of verbs: da´c odpocza´ ˛c ‘let sbd. rest’, nie da´c sko´nczy´c, spa´c, doj´sc´ do słowa ‘not let sbd. finish, sleep, get a word in’ have direct counterparts in Russian. However, some verbs are more frequent than their counterparts in Russian: for example, the analogue to the combination da´c (po)gra´c ‘let play’ was also used in Russian, but not nearly as frequently and in different meaning, cf. (in Russian) negated ‘not let sb. play, keep him from playing’ vs. (in Polish) ‘give sb. the chance to play’. This and similar cases suggests a by and large stronger usage of enable; it also shows that generally, a comparison of the lexical infinitives without regard to the actual contexts they appear in may be misleading. Unlike Russian, it is not the case that positive polarity is restricted with unaccusative verbs. The attested example in (191) can be freely changed to positive polarity in (191)’ without the assumption arising that the causer wanted to die, as in Russian (see (81) from page 68). This is akin to German lassen and English let. (191)
Dobre sa˛ ludzie, ubogiemuDAT nie dadza˛3PL umrze´cINF z głodu.i ‘They are good people, they won’t let2 the poor1 (guy) die3 from hunger.’
(192)’ Nie sa˛ dobre ludzie, ubogiemuDAT dadza˛3PL umrze´cINF z głodu. ‘They are not good people, they’ll let2 the poor1 (guy) die3 from hunger.’ Turning to the next domain, we see that in Polish, unlike Russian, manipulative causation, that is, causation of an animate causer on an independent event involving an inanimate causee, is not regularly expressed by da(va)t. (192) *DałPS T :3S G farbieDAT wyschna´ ˛cINF / herbacieDAT wystygna´ ˛cINF . Intended: ‘He let the paint dry / the tea cool down.’ Sentences such as (192) are ungrammatical; meanings like these are verbalized with other means, e.g. with transitive verbs such as suszy´c ‘to dry (tr.)’ or studzi´c ‘cool down (tr.)’, more explicitly by using czeka´c ‘wait’, etc. Consequently, the six non-reflexive examples found in the corpus with inanimate causee mostly involved some conceptualization of the causee as i Władysław
Stanisław Reymont, Chłopi, 1909
128 da(va)t+inf in Polish animate, as in da´cINF oczomDAT odpocza´ ˛cINF ‘let the eyes take a rest’, with one exception: (193)
Przechylam butelk˛e i bursztynowy eliksir wlewa si˛e we mnie [. . . ]. Przytrzymuj˛e troch˛e w ustach i po chwili daj˛e1S G spokojnie ii spłyna´ ˛cINF do gardła. ‘I tilt the bottle and the amber elixir pours into me [. . . ]. I keep it in my mouth for a little while and after a moment I let (it) easily drain into my throat.’
In view of the other facts, I analyze this as an occasionalism that shows the construction’s potential to develop into the domain of manipulative causation, as, e.g., evident in Russian and many other languages. The other verbs encountered were (194)
z˙ y´c ‘live’, zgina´ ˛c ‘perish’, uzewn˛etrzni´c ‘externalize’, odpocza´ ˛c ‘rest’, odetchna´ ˛c ‘breathe out’
I conclude that usage in the manipulative domain is marginal. Non-reflexive permissive da(va)t does not regularly extend to the expression of manipulative causation. As in Russian, causation by an inanimate entity is rather peripheral (n=13), but possible. It typically involves an inanimate causer and an animate causee in an intervention reading under negation (the only example with an inanimate causee involved personification): (195)
B˛eda˛ si˛e tego wypiera´c, dopóki uzale˙znienieNOM:S G nie da3S G imDAT spa´cINF albo wprowadzi organizm w stan wyniszczenia. i ‘They will deny it until the addiction will not let them sleep or put their body into a state of collapse.’
The list of verbs met: (196)
n=2: spocza´ ˛c ‘rest’, zohydzi´c ‘disgust’; z˙ y´c ‘live’; n=1: zarobi´c ‘earn money’, uciec ‘run away’, spa´c ‘sleep’, prze˙zy´c ‘survive’, oglada´ ˛ c ‘watch’, nudzi´c ‘bore’.
Unlike in Russian, we cannot say, in the sense that something keeps one from aging, (197) #To nie daje starze´c. That.NOM NEG let.3SG grow.older.INF That keeps one young (see ex. (111)) ii Andrzej
Horubała, Farciarz, 2003
i Dziennik
Polski, 2003-12-31, Co nas czeka?
Permissive da(va)t in Polish
129
Summarizing this section, I conclude that categorial differences between da(va)t in Polish and Russian in the interpersonal domain relate to peripheral cases such as waiting and are otherwise hard to capture. While in general, the main two types found in Russian, non-intervention and enablement, are also present in Polish, they do not show the same affinity to negation and the relative frequency of these senses may be the most important difference between the two. Furthermore, non-reflexive permissive da(va)t is less grammaticalized in Polish than in Russian as it has stronger selectional restrictions concerning the animacy of the causee; it does not extend to the expression of manipulative causation. Negative hindering, most synsemantic and elaborate in Russian, is not as predominant; no restrictions on the use of positive as opposed to negative da(va)t are noted. 3.2.4.
Reflexive permissive
The reflexive permissive is clearly the most important morphosyntactic realization of permissive da(va)t+inf in Polish. As we will see in this and the following sections, it shows clear symptoms of grammaticalization in terms of semantic bleaching and loosening of restrictions on arguments alongside syntactic changes concerning the form of the causee. This type has attracted some attention in the literature: Wilczewska (1966: 155) notes that while this type is generally compositionally derived from permissive da´c and reflexive si˛e, its use is remarkable: “As with every construction that has been in use for centuries, also in this one certain shades have developed that are dependent on the type of noun of the subject and further complements” (my translation from the Polish).24 She discusses the peculiarities of this and the modal passive construction and concludes that permissive da´c si˛e should not be considered a lexeme in its own right in view of its principal compositionality, acknowledging, however, that it is much more frequent than similar formations with pozwoli´c ‘permit’ or other verbs that combine with a reflexive pronoun in their complement clauses. It seems that this intuition can be confirmed and, on the basis of more recent theories of language change, carried further in the present study. An overview of the reflexive types was already given in table 17 on page 120 above. The two major types involved are dative and accusative reflexive permissives, and I will be focusing on these in this section.
130 da(va)t+inf in Polish n=124: zaskoczy´c ‘surprise’; n=37: przekona´c ‘convince’; n=30: nabra´c ‘deceive’; n=23: wciagn ˛ a´ ˛c ‘pull into’, namówi´c ‘talk into’; n=22: ponie´sc´ ‘carry’; n=21: zwie´sc´ ‘delude’; n=19: wyprzedzi´c ‘overtake’; n=12: skusi´c ‘tempt’; n=11: sprowokowa´c ‘provoke’; n=10: porwa´c ‘seize’, zwariowa´c ‘drive crazy’, zepchna´ ˛c ‘push down’; n=9: uwie´sc´ ‘seduce’; n=7 : podej´sc´ ‘approach, unie´sc´ ‘raise, zabi´c ‘kill, oszuka´c ‘cheat, pokona´c ‘defeat’; n=6: nabiera´c ‘deceive’, omami´c ‘delude’, wyprowadzi´c ‘drive’; n=5: ogra´c ‘outplay’, zastraszy´c ‘intimidate’; n=4: sterroryzowa´c ‘terrorize’, zniewoli´c ‘enslave’, prowadzi´c ‘drive’, złapa´c ‘catch’; n=3: zwerbowa´c ‘enlist’, rozstrzela´c ‘shoot’, wykorzysta´c ‘exploit’, uwodzi´c ‘seduce’, zniszczy´c ‘destroy’, wprowadzi´c ‘bring’, ogrywa´c ‘outplay’, pobi´c ‘beat’, zatrzyma´c ‘stop’, op˛eta´c ‘inspire’, zje´sc´ ‘eat’, ponosi´c ‘carry’; n=2: 34 verbs; 106 hapax legomena infinitives. Verb list 8: Complement infinitive of permissive da(va)t with accusative reflexive, irrespective of animacy of the causee. With accusative reflexives we are semantically dealing mostly with dynamic (non-)intervention, similar to what we have found in Russian. Wilczewska’s (1966:155) characterisation of reflexive permissives in Polish fits remarkably well to the analysis of the Russian construction in section 2.2.6.2 from p. 93 above. She states that “the combination as a whole designates that the subject yields to an influence of subjective or independent forces [. . . ] (my translation)25 ”. Accordingly, there is a large overlap as to the complement verbs concerned as list 8 suggests. As in Russian, these verbs mostly, but not exclusively, denote undisirable actions directed towards a patient that is likely to avoid being subject to them, e.g. namówi´c ‘talk into’, przekona´c ‘persuade’, nabra´c ‘cheat’, wciagn ˛ a´ ˛c ‘pull into’, zwariowa´c ‘drive crazy’, sprowokowa´c ‘provoke’, cf. (198)
Na swoje nieszcz˛es´cie dałPS T :3S G si˛eRFX:ACC namówi´cINF i młodemuDAT m˛ez˙ czy´znieDAT , by razem poszli si˛e napi´c. ‘To his misfortune he let himself be persuaded by the young man to go for a drink together (lit: let himself persuade to the young man)’
Despite these similarities, the meaning of the construction goes beyond that of the Russian construction. In Russian, the causer possesses an immediate, i Dziennik
Polski, 2001-11-14, Tarnów, Brutalny zabójca
Permissive da(va)t in Polish
131
concrete possibility of intervention based on the interaction with the causee (which is defocused), while in Polish this is not necessarily the case. This is evident with combinations such as wyprzedzi´c ‘overtake’, podej´sc´ ‘approach’, ogrywa´c ‘overcome (in play)’, dogoni´c ‘catch up’ or zaskoczy´c ‘surprise’. Characteristic of this type are the following examples: (199)
Adamek faktycznie działał pod wpływem impulsu, ale wcze´sniej rozwa˙zał podobna˛ ewentualno´sc´ i nie dałPS T :3S G si˛eRFX:ACC i zaskoczy´cINF . ‘Adamek was in fact acting on an impulse, but he had pondered such an eventuality and did not let himself be taken by surprise (lit.: didn’t let himself surprise).’
(200)
W kategorii do dziesi˛eciu lat dałaPS T :3S G si˛eRFX:ACC wyprzedzi´cINF ii tylko starszymDAT o rok i dwa lata konkurentkomDAT [. . . ] ‘In the category up to ten years she let herself be surpassed only by competitors older by one or two years (lit: let herself surpass only older by two or three years competitors).’
Sentence (200) contains negation (by tylko ‘only’): the causer (an athlete) did not let other people get in front of her, not by hindering them, but by excelling in competition. No concrete action by the causer to preclude or not preclude some action of the (implied) causee is referred to. The causer only has a certain responsibility, indirectly, for whether or not other people are faster. The relationship to the complement action is therefore very indirect. In Russian, in contrast, the intervention referred to in the analogous phrase is construed as concretely possible. (201) is the Russian literal equivalent of Polish (200) (with negation made explicit) and expresses that she (the athlete) blocked the other’s way or did something else to intervene with her competitors: (201)
Ona neNEG dalaPS T :S G sebjaRFX:ACC obognat’/operedit’INF . (RU) ‘She didn’t let herself be overtaken / surpassed.’
A Polish positive version describes an evaluation of what happened (she was overtaken; she could theoretically have done something about it by running faster) rather than a description of the interaction of the causer and the causee in a concrete situation. In contrast, the positive version in Russian ona dala sebja obognat’ ‘she let herself be overtaken’ is less idiomatic and denotes a i Artur Baniewicz, Drzymalski przeciw Rzeczpospolitej, 2004 2000-09-20, Nowy Sacz, ˛ Sylwia najlepsza
ii Dziennik
Polski,
132 da(va)t+inf in Polish conscious decision of the causer to grant the opponent priority. Such a conscious decision of the causer to let the causee do whatever is at stake is typical for positive attestations in the Russian corpus, cf. the examples in (202) and (203): (202)
Barrikello [. . . ] bukval’no na poslednix metrax gonki snjal nogu s pedali, demonstrativno davajaCV B sebjaRFX:ACC obognat’INF "mažoru"DAT komandy. (RU)i ‘Barrikello [. . . ] literally on the laste meters of the race took his foot off the pedal, ostentatiously letting himself be overtaken by the "major" of the team.’
(203)
Napali [. . . ] na kakogo-to proxožego, tot dalPS T :S G sebjaRFX:ACC ograbit’INF i, propustiv, vystrelil v spinu. (RU)ii ‘They assaulted [. . . ] a passer-by - and that person let himself be robbed and after they left, shot them in the back.’
While this is certainly difficult to prove, I hope I have made plausible that there is a subtle difference in the use of the Russian and Polish constructions: in Russian, a more or less conscious, controlled action by the causer leads or does not lead to whatever the causee is doing; in Polish, the causee is not necessarily consciously involved - the situation is simply construed as having happened under circumstances that are generally under at least partial influence of the causer. The crucial point in Polish is that something is being done to the causer; the causee is most of the time left implicit. This is very close to a passive transformation, except that the patient retains agentive properties, which are variable and may be quite minimal. Note that this is, by and large, also true for certain combinations of reflexive permissive da(va)t in Russian, too; in Polish, however, this is the core of usage and in contrast to Russian the possibility for intervention may be very abstract; the relation expressed by the construction here is semantically thus bleached in comparison to its Russian cognate. I now turn to dative and other non-accusative reflexives where the causer is affected by the complement event. The most frequent complement verbs with dative reflexives are given in list 9. The dative is either an argument of the complement verb, as in (204), or a free dative signalling affectedness, as in example (205): i Boris Muradov. Gran Pri Avstrii: zoloto Al’p // Formula, 20010615 Iz dnevnika (1918-1919)
ii Marina Cvetaeva.
Permissive da(va)t in Polish
133
n:14: odebra´c ‘take away’; n:11: wydrze´c ‘take away’; n:10: strzeli´c ‘shoot’; n:9: narzuci´c ‘impose’; n:5: ucia´ ˛c ‘cut off’, wbi´c ‘knock in’; n:2: przerwa´c ‘interrupt’, wmówi´c ‘convince’; 11 hapax legomena Verb list 9: Complement infinitive of permissive da(va)t with dative reflexive; both animate and inanimate causees. (204)
[. . . ] nigdy nie dałemPS T :1S G sobieRFX:DAT wmówi´cINF , z˙ e infori matyk musi by´c kiepski z polskiego [. . . ] ‘[. . . ] I never let myself be persuaded that a computer scientist must be bad at Polish [. . . ]’
(205)
Pierwsza˛ rund˛e przeszli bez straty punktu, dajac ˛ CV B sobieRFX:DAT ii strzeli´cINF jedynie 5 goli. ‘They went through the first round without point loss, taking only 5 goals (lit.: letting themselves shoot only 5 goals)’
Here, the same definition as with accusative reflexives applies. (205) exemplifies the same kind of semantic bleaching as with accusative reflexives: the causer surely did not consciously let the opponent score five goals; rather, he had to take five goals and is construed to bear some responsibility for that. The next examples shows a similar relationship with reflexives in the instrumental case: (206)
(207)
Nie damy soba˛ pomiata´c[!] NEG let.1PL RFX.INS push.around.INF ‘We won’t let ourselves be pushed around!’
iii
On nic nie dał sobie powiedzie´c, nie He nothing.ACC NEG let.PST.3SG RFX.DAT say.INF NEG dał soba˛ kierowa´c let.PST.3SG RFX.INS guide.INF ‘He wouldn’t listen to anything (lit.: didn’t let himself say anything), iv didn’t let himself be guided’
Instrumental reflexives are generally less frequent, which can be expected in view of the small number of verbs taking instrumental objects in Polish. i Dziennik Polski, 2003-01-17, Podhale
ii Dziennik Polski, 2004-04-17, O´swi˛ ecim, Inau-
iii Dziennik Polski, 2003-10-17, O´swi˛ ecim, Hokej na lodzie
guracja ze zgrzytem Lupa, Podgladania, ˛ 2003
iv Krystian
134 da(va)t+inf in Polish In sum, we have seen that reflexive permissives are more grammaticalized than non-reflexive permissives in regard to their semantics: the type of causation expressed is very abstract and amounts to a general possibility of intervention. This is not dependent on the form of the reflexive, which is essentially governed by the usual syntactic rules and may appear in accusative, dative or instrumental case or in a prepositional phrase, depending on the context. In comparison to the cognate construction in Russian, which is similar in respect to its basic characteristics but which has has not grammaticalized further, the Polish construction is semantically bleached and more synsemantic. More evidence corroborating this analysis will be adduced in the next section concerned with the causee phrase.
3.2.4.1.
The causee phrase
As noted by Szlifersztejnowa (1968, 1970), the causee in constructions involving reflexive da(va)t+inf may be expressed by either a dative or a prepositional phrase formed with przez. While Wilczewska (1966) furthermore notes the alternative of framing it as an instrumental phrase, this could not be confirmed for contemporary Polish on the basis of corpus examples or in informant work26 . The expression of the causee as a prepositional phrase or in instrumental case constitutes a type II causative (see section 1.3.5 from page 21); as in many other languages, a characteristic (if not defining) trait is that this is also the form of the by-phrase in the Polish passive. Both possibilities are not reflected in the standard dictionaries and their use is empirically investigated below. I first discuss animacy restrictions in relation to the causee and then establish the status of prepositional and instrumental causees. As regards animacy of the causee, a clear difference of reflexive in contrast to non-reflexive constructions is noable. While non-reflexive permissives in Polish do not normally admit of inanimate causees (see preceding section), with reflexive permissives, such causees are both possible and comparatively frequent; this indicates fewer selectional restrictions for the latter construction. Table 23 shows the most frequent complement verbs and figures on animate, inanimate, and unexpressed causee, respectively. We can see that the
Permissive da(va)t in Polish
inf zaskoczy´c ‘surprise’ przekona´c ‘convince’ nabra´c ‘deceive’ wciagn ˛ a´ ˛c ‘pull into’ namówi´c ‘talk into’ ponie´sc´ ‘carry’ zwie´sc´ ‘delude’ wyprzedzi´c ‘overtake’ skusi´c ‘tempt’ sprowokowa´c ‘provoke’ porwa´c ‘seize’ zwariowa´c ‘drive crazy’ zepchna´ ˛c ‘push down’ uwie´sc´ ‘seduce’ podej´sc´ ‘approach’ unie´sc´ ‘raise’ zabi´c ‘kill’ oszuka´c ‘cheat’ pokona´c ‘defeat’ nabiera´c ‘deceive’
abs 124 37 30 23 23 22 21 19 12 11 10 10 10 9 7 7 7 7 7 6
none 89 30 28 22 15 2 8 6 6 8 2 9 10 3 7 0 7 6 5 5
causee anim. 32 7 2 0 8 0 1 12 3 3 0 1 0 3 0 0 0 1 2 1
135
inan. 3 0 0 1 0 20 12 1 3 0 8 0 0 3 0 7 0 0 0 0
Table 23. Most frequent complement verbs with the accusative reflexive permissive, and type of causee: none, animate, inanimate.
combinations vary. Some verbs are used almost exclusively with inanimate causees, which are mostly abstract, such as in da´c si˛e ponie´sc´ nerwom ‘get carried away’, lit.: ‘let the nerves carry oneself away’ or da´c si˛e zwie´sc´ pozorom ‘let appearances fool oneself’; however, the causee may also be an agent-like force, such as in da´c si˛e unie´sc´ pradowi ˛ ‘let the current carry oneself away’. Note that only the latter has a parallel in Russian dat’ sebja uvesti teˇceniju ‘let the current carry oneself away’. The table shows that many verbs appear with both inanimate and animate causees, such as, e.g. skusi´c ‘tempt’:
136 da(va)t+inf in Polish (208)
Dał si˛e skusi´c działaczom i podpisał Let.PST.3SG RFX.ACC tempt.INF activist.DAT and sign.PST.3SG kontrakt. contract.ACC ‘He succumbed to the activists and signed the contract’ lit.: ‘he let i himself be tempted by the activists’
(209)
[. . . N]iektóre osoby mogły si˛e da´c some people.NOM.PL can.IRR.PL RFX.ACC let.INF skusi´c ofercie oszusta. tempt.INF offer.DAT.SG crook.GEN.SG ‘Some people could fall for the crook’s offer’ (lit.: ‘could let themi selves be tempted by the offer of the crook’)
This possibility may seem unremarkable in view of the general semantics of the reflexive permissive construction, be it in Russian or in Polish. However, the possibility to introduce non-agentive causee phrases is much less restricted in Polish than in Russian. To see that, let has consider przekona´c ‘persuade’, a verb that may also be used with both inanimate and animate causee.A search of the full corpus yielded 225 hits of da´c si˛e przekona´c ‘let oneself be convinced’, including several cases of inanimate, abstract causee denoting what is convincing, as in (210), as well as animate, human causee denoting who convinces, as in (211)27 : (210)
[. . . ] społecze´nstwo ameryka´nskie daje3S G si˛eRFX:ACC przekona´cINF argumentomDAT administracji za u˙zyciem siły ii przeciwko Saddamowi Husajnowi. ‘American society believes the arguments (lit.: lets itself persuade to the arguments) of the administration speaking for the use of force against Saddam Hussein.’
(211)
Przysłuchujacy ˛ si˛e tej dyskusji telewidzowie [. . . ] daliPS T :3PL iii si˛eRFX:ACC przekona´cINF tymDAT wła´snie politykomDAT . ‘The television viewers that were listening to that discussion closely [. . . ] let themselves be persuaded by exactly those politicians’ lit.: ‘let themselves persuade to those politicians’.
i Dziennik
˙ zlowe wira˙ze Jana Krzystyniaka i , DziPolski, 2001-01-16, Podkarpacie, Zu˙ ´ ennik Polski, 2002-04-06, Pejza˙z, Ewangelia on-line ii Dziennik Polski, 1998-02-21, Swiat, ´ Wojenne nastroje iii Dziennik Polski, 2001-10-26, Slepi o kolorach
Permissive da(va)t in Polish
137
Argument is an intermediate in a causal chain and can be viewed both as an instrument of an agent or, alternatively, as an agent/force/effector (Van Valin & Wilkins 1996) itself. This is reflected in the argument structure of verbs such as Polish przekona´c or its Russian and English counterparts ubeždat’, and persuade. All these verbs behave alike in that both the argument and the person expressing the argument may form the subject: (212)
PL: ArgumentNOM:S G / politykNOM:S G goACC przekonałPS T :3S G . RS: ArgumentNOM:S G / politikNOM:S G egoACC ubedilPS T :3S G . The argument / the politician persuaded him.
There is a clear difference between the Polish and the Russian construction in relation to such verbs: In Russian, only true agents may be expressed as a dative causee in the reflexive permissive construction, while in Polish, this restriction does not apply: (213)
PL DałPS T :3S G si˛eRFX:ACC przekona´cINF ojcowiDAT . RU On dalPS T :S G sebjaRFX:ACC ubedit’INF otcuDAT . ‘He let himself be persuaded by his father (lit.: let himself persuade to father).’
(214)
PL DałPS T :3S G si˛eRFX:ACC przekona´cINF tym argumentomDAT . RU *On dalPS T :S G sebjaRFX:ACC ubedit’INF e˙ tim argumentamDAT . ‘He let himself be persuaded by these arguments (lit.: let himself persuade to these arguments).’
While both languages permit of phrases like (213) (analogous to (211) above), only in Russian the phrase in (214) entails that the argument acts in some way as an conscious agent that attempts to persuade the adressee and it is therefore infelicitous. In Polish, thus, the role of the causee is less restricted: it may not only denote an agent or an agent-like natural force, as in Russian. Rather, the restrictions of the causee in Polish seem to correspond to the structural position of the first argument of the infinitive complement. This is evidence of a more rigid, grammatical restriction that is defined in structural rather than semantic terms. While this is not specifically captured in Lehmann’s parameters, except in the general statement that grammaticalizing elements become subject to structural, rather than semantic constraints, this should clearly be seen as an increase in grammaticality.
138 da(va)t+inf in Polish
reflexive type acc. (si˛e) dat. (sobie) others
all 650 69 6 725
no causee 451 66 5 522
dative all inan. 190 76 2 0 1 0 193 76
przez all inan. 9 3 1 0 0 0 10 3
Table 24. Form and animacy of the causee in reflexive permissives (IPI PAN sample corpus). There are three options: no causee, dative causee and causee as a by-phrase introduced by przez. Inanimate causees given per type.
3.2.4.2.
Prepositional (type II) causees
In numerous corpus examples, the causee is given in a prepositional phrase formed with przez ‘by’, as in (215)
Sprawa wyszła na jaw dopiero wtedy, gdy do Makro Cash and Carry zacz˛eły zgłasza´c si˛e osoby, które dałyPS T :3PL si˛eRFX:ACC skusi´cINF i przezby akwizytorówACC . ‘The thing became public only when people who had let themselves be lured by salesmen began turning to Makro Cash and Carry.’
(216)
Uwa˙zam, z˙ e nie wolno [. . . ] da´cINF si˛eRFX:ACC tu manipulowa´cINF ii ´ a˛ACC [. . . ] przezby stron˛eACC chinsk ‘I think one should not let1 oneself 2 be manipulated3 by4 the Chinese6 side5 [. . . ].’
(217)
W ko´ncu dałemPS T :1S G si˛eRFX namówi´cINF przezby koleg˛eACC . ‘At the end I let myself by talked (into it) by a friend.’
iii
This constitutes a type II causative where the causee has the same form as the passive by-phrase used to introduce a demoted agent. In table 24, the various combinations of animacy and form of the causee as found in the sample corpus are given in detail. The number of przez-phrases is quite low, representing only about 5 percent of all mentions of the causee with si˛e. In order to take more data into account, the full corpus was searched for da(va)t followed by infinitive within 5 i Dziennik
ii Sprawozdanie Polski, 1999-08-05, Kraj, Naciagana ˛ akwizycja iii stenograficzne z obrad Sejmu RP z dnia 22.08.2001 Dziennik Polski, 2003-12-23, Kraków, P˛etla w magistracie
Permissive da(va)t in Polish
type irrelevant acc. (si˛e) dat. (sobie) modal passive
all 227 36 6 6 275
139
inan 5 0
Table 25. Results for a full IPI PAN corpus search for da(wa)´c plus infinitive and przez.
Accusative reflexive: n=2: zdominowa´c ‘dominate’, złapa´c ‘catch’, zaskoczy´c ‘surprise’; n=1: namówi´c ‘talk into’, skusi´c ‘temp/lure’, przeszkoli´c ‘reeducate’, porwa´c ‘catch’, pokona´c ‘beat’, poklepywa´c ‘clap’, oszuka´c ‘cheat’, narzuci´c ‘throw; impose’, wchłona´ ˛c ‘absorb’, manipulowa´c ‘manipulate’, kupi´c ‘buy’, gwałci´c ‘rape’, gry´zc´ ‘bite’, bi´c ‘heat’, op˛eta´c ‘overcome’, zadepta´c ‘trample’, złama´c ‘break’, zepchna´ ˛c ‘push’, zawie´sc´ ‘disappoint’, zastraszy´c ‘intimidate’, zasłoni´c ‘cover’, sterroryzowa´c ‘terrorize’, zaje´zdzi´c ‘override’, u˙zy´c ‘use’, wyssa´c ‘suck’, wysła´c ‘send out’, wyprzedzi´c ‘overtake’, wymanewrowa´c ‘outmaneuver’, wykolegowa´c, wciagn ˛ a´ ˛c ‘pull into’, zakrzycze´c ‘shout down’ Dative reflexive: odebra´c ‘take away’ (n=3), narzuci´c ‘impose’ (n=3) Verb list 10: Complement verbs with przez-causees in the full corpus. tokens and with przez ‘by’ before and after the infinitive28 . The search yielded 275 examples, 47 of which relevant przez-phrases; the results are given in table 25. The figures confirm that this possibility is marginal in terms of frequency. However, note that it was present in all genres; the list of complement verbs in verb list 10 shows that it is productive, not linked to specific verbs29 . Native speakers differ in assessment of the grammaticality of these phrases; e.g., the corpus example in (217) above was readily accepted by some and deemed ungrammatical by others. An informal questionnaire was sent to several informants and partly supplemented in discussions. As a result, it seems that dative causees are generelly preferable to all informants, while instrumental causees (see above) are always rejected. In
140 da(va)t+inf in Polish regard to przez-phrases there is wide variation. To some informants, przezphrases were not acceptable at all, while others accepted them without hesitation and judged them to be as acceptable as dative phrases. Others varied in their interpretation. It seems that the most acceptable are cases with inanimate causees, as in the next examples which were attested with dative (218) and prepositional (219) causee that seemed to be interchangable to some informants: (218)
Kiernacki nie umiał ubra´c tego w słowa, ale dałPS T :3S G si˛eRFX:ACC i oczarowa´cINF nastrojowiDAT chwili. ‘Kiernacki was not able to put it into words, but he let himself be enchanted by the spirit of the moment.’
(219)
Ona jedna z tej trójki, Justysia, nie dałaPS T :3S G si˛eRFX:ACC od razu ii porwa´cINF (przezby rwacy ˛ w tym miejscu nurtACC . She, Justysia, alone of the three did not let herself be immediately swept away by the at this place rapid current.
Least acceptable seem to be animate causees under focus, as in the following constructed example where only a dative causee was accepted: (220)
Nawet swoim dzieciomDAT / *przezby swe dzieciACC daje3S G si˛eRFX:ACC namowicINF . ‘Even by her children (dative / *prepositional) she lets herself be persuaded.’
A systematic study with a wide range of informants and environments would be necessary to assess the status of this coding option in detail and pinpoint factors relevant for its usage. In any case, I conclude that these phrases are used in Polish today and need to be accounted for in descriptions of da(va)t; however, they are clearly a secondary option in comparison to dative causees. In the diachronic study in section (444) below, we will see that such uses are attested in Polish at least since the 16th century.
3.2.4.3.
The causee with dative reflexives
In the discussion of form and function of the causee so far, I have concentrated on accusative reflexive, the most common type of reflexive permissive. I will now enlarge the scope to dative reflexives. i Artur Baniewicz, Drzymalski przeciw Rzeczpospolitej, 2004 2003-03-27, Małopolska, Pi˛etno niepotrzebnej s´mierci
ii Dziennik
Polski,
Permissive da(va)t in Polish
141
Complement verbs (sample corpus): n:14: odebra´c ‘take away’; n:11: wydrze´c ‘take away’; n:10: strzeli´c ‘shoot’; n:9: narzuci´c ‘impose’; n:5: ucia´ ˛c ‘cut off’, wbi´c ‘knock in’; n:2: przerwa´c ‘interrupt’, wmówi´c ‘convince’; 11 hapax legomena Verbs with overt causee in full corpus: Dative: n=9: odebra´c ‘take away’; n=2: narzuci´c ‘impose’; n=1: dmucha´c ‘blow’, pogrzeba´c ‘bury’, popsu´c ‘break’, dyktowa´c ‘dictate’ Prep: n=3: odebra´c ‘take away’, narzuci´c ‘impose’ Verb list 11: Dative and prepositional causees with dative reflexives in the sample corpus and in the full corpus: absolute numbers and complement verbs.
The study of the IPI PAN sample corpus alone cannot give a reliable picture of the dative reflexive construction in view of its comparatively low frequency. The sample corpus includes 69 dative reflexives, of which only three contained an overt causee (see above table 21 on page 124). It was therefore necessary to investigate the full corpus not only in regard to przez-causees, but also in regard to dative causees used with dative reflexives. Rather than extract and annotate all occurrences of dative reflexive permissives in the full corpus, a mixture of approaches was used. A manual search of the full IPI PAN corpus for dative reflexive permissives30 yielded about 1,500 hits, which were superficially inspected without extraction to a database. The vast majority in fact concerned dative reflexives, and in view of cases not covered by the query string, I take 1,300 to be a conservative estimate of the actual number of occurrences. Further searches for dative reflexive permissives and dative causee yielded 32 hits, 15 of which positive. Recall that przez-phrases in the full corpus were searched for above. In order to estimate the number of causeeless phrases I subtract these cases from the estimation of 1,300 dative reflexive phrases arrived at above. List 11 summarizes the evidence compiled, giving lists of the relevant complement verbs found here and above. It has to be noted that constructions with causee were found most frequently in sports reports, that is, in a genre generally very close to spoken, informal language; however, they were also used in parliamentary proceedings and general newspaper coverage. No inanimate causees were found. The verbs used with dative and with prepositional causee are similar in that the
142 da(va)t+inf in Polish most frequent verbs with dative causee are also the only verbs with prepositional causee, and no clear functional difference between the two options is apparent. (221)
Wi´slacy za łatwo daliPS T :PL sobieRFX:DAT narzuci´cINF styl gry i piłkarzomDAT z krainy fiordów. ‘The Wi´slacy team too easily let the soccer players from the country of fjords dictate them the style of the game (lit.: too easily let themselves dictate style to the soccer players)’
(222)
Niepotrzebnie jednak po przerwie dali´smyPS T :1PL sobieRFX:DAT narzuci´cINF warunki przezby zdesperowanegoACC rywalaACC . i ‘However, without necessity we let a desperate rival dictate us the conditions after the break (lit.: let ourselvesDAT dictate conditions by a desperate rival)’
However, consider the following transcript of a parliamentary hearing where the speaker first uses a dative, and then a prepositional causee: (223)
[Marek Wagner:] Tak z˙ e ostatecznych kompetencji pan premier nie ii dałPS T :3S G sobieRFX:DAT odebra´cINF panuDAT ProksieDAT . ‘So the prime minister did not cede ultimate powers to Mr.Proksa (lit.: did not let himself DAT take away to Mr. Proksa).’ [Zbigniew Ziobro:] Panie... Przepraszam bardzo, nie dosłyszałem. ˙ ostatecznych kompetencji nie dałPS T :3S G ... Ze ‘Mister... Excuse me, but I didn’t hear that. So the ultimate powers he didn’t let...’ [Marek Wagner:] Swoich nie dałPS T :3S G sobieRFX:DAT odebra´cINF przezby panaACC Proks˛eACC [. . . ]. ‘His own he didn’t cede to Mr. Proksa (lit.: didn’t let himself (dat. commodi) take away by Mr. Proksa)’
In this interchange, the speaker first uses a dative causee and then, when the interlocutor asks back, a prepositional causee. As an hypothesis, I attribute this to the complexity of the first construction that involves one direct object and two further dative phrases; the second construction with the prepositional phrase is used because it is more transparent because it employs more explicit coding. i Dziennik
i Dziennik Polski Polski, 2003-11-28, Wisła odpadła w karnych! ii ´ 2002-06-19, Skuteczne wyj´scie awaryjne Stenogram z 84. posiedzenia Komisji Sledczej, 20.11.2003
Permissive da(va)t in Polish
przez-causee dative causee all dat. refl.
sample 1 2 69
143
full 6 15 ≈1,300
Table 26. Causee types with dative reflexives.
In general, informant work reveals that with dative reflexives, as with accusative reflexives, there is variation concerning the acceptability of przezphrases. The picture obtained from the analysis of the full corpus confirms the tendency found in the sample corpus: with dative reflexives, causee mention is generally less frequent than with accusative causees. In the sample corpus with 69 dative reflexives, 3 mentions of the causee were found, 1 of which was introduced by przez and two of which were expressed in a dative phrase. The search of the full corpus resulted in about 1,300 dative reflexives and only 20 mentions of the causee, 6 of which were with a przez-phrase. In both cases the relation of dative-coded and preposition-coded causees was about three to one. This compares to a relation of about 20 to one with accusative reflexives (see table 26). Summarizing, we can say that with dative causees, use of a causee-phrase is very infrequent; at the same time, the comparative weight of przez-phrases is quite high. In general, both dative and prepositional causees are possible, with the second option subject to variation.
3.2.4.4.
Conclusions: form and function of the causee phrase
With reflexive permissives, selectional restrictions concerning the causee are looser both in comparison to non-reflexive permissives as well as in comparison to reflexive permissives in Russian. Both with accusative si˛e and dative sobie two alternatives for causee coding are given: as dative phrase or introduced by przez. The second possibility is not acceptable to all speakers, and possibly subject to qualitative restrictions that would need to be empirically investigated in more detail. The frequency of prepositional coding of the causee seems to be negatively correlated with the frequency of causee-mentioning as such. To see
144 da(va)t+inf in Polish this, consider the following findings: Non-reflexive causees usually contain an overt causee; no przez-phrase is possible. Accusative reflexives usually do not express the causee; przez-phrases are a minor possibility. Dative reflexives very seldom contain an overt causee; przez-phrases are relatively often used. I will forward the hypothesis that this is no coincidence, but rather due to a process of grammaticalization and a more peripheral state of the causee in those constructions that allow prepositional causees. In Czech, a similar development has led to the present state where generally only prepositional causees are possible. We will return to the question of prepositional causee coding in Polish and Czech from a comparative and diachronic point of view places below.
3.2.5.
Summary
In this section I have analyzed reflexive and non-reflexive permissives separately, showing that da(va)t+inf in the function of a reflexive permissive is more grammaticalized than in the function of a non-reflexive permissive. As reflexive permissive it imposes fewer selectional restrictions, is semantically less specific and more frequent both in comparison to the non-reflexive permissive and to its Russian reflexive counterpart. Furthermore, it differs in syntax from both: with reflexive permissives, the causee may be expressed in a prepositional phrase formed with przez, constituting a type II causative.
3.3. 3.3.1.
Factitive da(va)t Curative causation
da(va)t+inf in Polish is a productive means of denoting curative causation (see introduction): to have something done (by a service provider). However, only three unambiguous examples of curative da´c were found in the sample corpus. Since two concerned a quotation31 from the 17th century, reflecting a higher productivity of factitive da(va)t in older Polish (see section 6.5), only a single example is of relevance:
Factitive da(va)t
(224)
145
To oni na przykład poddali mu pomysł, z˙ eby dałPS T :3S G sobieRFX:DAT zamontowa´cINF w domu dwudziestoczterogodzinna˛ i inwigilacj˛e wideo. ‘So they for example gave him the idea to have 24 hour video surveillance installed at home (lit. have himself video surveillance install)’
This reflects perfectly felicitous, but evidently quite sporadic usage of da´c in this function. Generally, according to my informants other constructions are preferred to curative da´c, among them a construction involving either da´c or odda´c ’give, give away’ in conjunction with the verbal noun in a prepositional phrase (odda´c do czyszczenia lit. ‘give away to the cleaning’) or contextual causatives (see above p.18). Querying for a number of verbs adduced in the literature I found only three further examples in the full IPI PAN corpus: (225)
Akurat dostał pomidorowy materiał i z niego dałPS T :3S G i sobieRFX:DAT uszy´cINF marynark˛e. ‘At that point he got tomato-colored material and had himself a jacket made out of it.’
(226)
Wło˙zyłam te˙z troch˛e pieni˛edzy w to mieszkanie - np. pieceACC ii dałamPS T :1S G na elektryk˛e przerobi´cINF . I also invested a bit of money in that apartement - for example, I had1 the stoves1 converted3 to electric power.
(227)
. . . stara˛ bluzk˛e (...), która˛ dam1S G sobieRFX:DAT przerobi´cINF na iii wiatrówk˛e, je´sli b˛ed˛e miał pieniadze. ˛ ‘. . . old blouse (...) , which I’ll have myself altered to a wind breaker, if I’ll have the money.’
As in Czech (see below), the realization of the infinitival event is implied, cf. (228)
[. . . ] piece dałamPS T :1S G na elektryk˛e przerobi´cINF , *ale nie przerobiono. ‘[. . . ] I had the stoves converted to electric power, *but (the company) didn’t convert them.’
The curative use thus involves event integration: the commissioning and subsequent realization of some service is expressed as one complex event. Topoli´nska (1993) discusses the curative construction (without using the term) at length. According to Topoli´nska, in this construction, the causee cani Max
i Dziennik Polski, 2003-11-21, Magazyn, Symfonia Cegielski, Apokalipso, 2004 ii na urodziny Dziennik Polski, 2001-08-08, Kraków, Kto ma da´c mieszkanie? iii Dziennik Polski, 2002-03-16, Pejza˙z, Listy Hasiora
146 da(va)t+inf in Polish not be introduced as a dative phrase and the construction expresses the commissioning of professional services. As with contextual causatives, she states the causee may only be introduced by a locative phrase such as w tej firmie ‘at that company’, cf. in (229): (229)
Wła´snie w tej firmie na Marszałkowskiej Just in that.LOC.SG company.LOC.SG on Marszałkowska dałam odnowi´c nasze meble. give.PST.1SG refurbish.INF our furniture ‘Just at that company on Marszalkowska street I had our furniture refurbished.’ (Topoli´nska 1993: 244; my translation and glossing)
Dative phrases unambiguously denote the beneficiary, not the causee, cf. (230)
Dałam Jankowi uszy´c płaszcz. Give.PST.1SG Jan.DAT sew.INF coat.ACC ‘I had a coat sewn for Janek’ not ‘I had Janek sew a coat’ (Topoli´nska 1993: 241; my translation and glossing)
This finding is confirmed here. The crucial point is that in some cases, dative phrases may be understood as causee-like. However, the construction as a whole then denotes only the commissioning of some service, that is, the giving of an assignment, rather than curative causation proper. The complement action is not implied, cf. the adapted corpus example: (224)’ DałemPS T :1S G tej firmieDAT zamontowa´cINF w domu dwudziestoczterogodzinna˛ inwigilacj˛e wideo, ale nic nie zrobiono. ‘I gave the company (the assignment) to install 24-hour video surveillance, but nothing was done.’ See also Bartnicka 1982: 184 for similar cases. Unlike the causeeless curative, the construction with dative cause thus does not involve event integration: the commissioning and subsequent realization of some service are not expressed as one complex event. Informant work concerning the curative construction is not very reliable in view of the competition of dative causee and curative construction and the low frequency of these constructions. Evidently, lexical restrictions also play a role. We will return to the question of curative da(va)t+inf with a dative causee in Czech below.
Factitive da(va)t
3.3.2.
147
Other types of factitive causation
Two clearly non-curative factitive combinations of some frequency have to be mentioned, both of which have literal counterparts in German, but are not translated easily into English: da´c na siebie czeka´c, lit. ‘make wait for oneself’ (5 occurrences) and dác si˛e prosi´c, lit. ‘make ask oneself’ (3 occurences). Both are arguably phraseological, as in (231)
Kocham ci˛e, ksi˛ez˙ niczko – wymamrotał tkliwie. Jerzy nie i dałPS T :3S G si˛eRFX:ACC długo prosi´cINF . ‘I love you, little princess - he murmured tenderly. Jerzy did not make him ask for a long time (lit.: let/make himself long ask).’
This is parallel to German sich nicht lange bitten lassen, literally ‘not let/make somebody ask (the causer) a long time’. It is also used with inanimate causers: (232)
[. . . ] trudne do przewidzenia skutki nie dadza˛3PL na siebie zbyt długo czeka´cINF . ‘[. . . ] effects that are difficult to anticipate will not be long in coming’, lit.: ‘will not have (us) wait for them a long time’.
The phrase may be translated as ’keep somebody waiting’ or, with inanimate causer ‘to be a long time coming’. Dative causees were not found in the corpus. Cases like these, with a clearly non-curative factitive reading of da(va)t+inf implied, are isolated; I regard them to be lexicalizations that are ultimately derived from the permissive. To see the connection to the permissive, note that the causer in these cases refrains from action: either he remains absent (da´c si˛e czeka´c), or he does not do what he is asked to (da´c si˛e prosi´c). By inaction, the causer indirectly coerces the causee into either waiting or asking; however, as in the permissive, whatever the causee does is ultimately by his own initiative.
3.3.3.
Ambiguous permissive/factitive causation
It is often hard to decide whether one is dealing with factitive or permissive causation with reflexive permissives, since, as Szlifersztejnowa (1970: 78) i Manuela
Gretkowska; Piotr Pietucha, Sceny z z˙ ycia pozamał˙ze´nskiego, 2003
148 da(va)t+inf in Polish n=2: zrobi´c ‘do, make’; n=1: zbada´c ‘investigate’, zaprasza´c ‘invite’, zało˙zy´c ‘install’, wymalowa´c ‘paint’, wyja´sni´c ‘explain’, wpisa´c ‘inscribe’, smarowa´c ‘smear’, sku´c ‘handcuff’, sfilmowa´c ‘film’, podłaczy´ ˛ c ‘install’, pochowa´c ‘bury’, obcia´ ˛c ‘cut off’. Verb list 12: Complement verbs labeled as permissive/factitive.
points out, the “activity” and “passivity” of the subject referent depends on semantics of the infinitive and whether it expresses “a desirable or undesirable activity” 32 . This, as well as traces of more general factitive meaning were present in 23 examples of permissive/factitive meaning, as in (233)
[. . . ] kazali mi zrobi´c badanie krwi. - Dlaczego pan sobieRFX:DAT nie i dałPS T :3S G zrobi´cINF ? - Bo si˛e od dziecka boj˛e igły [. . . ] ‘[. . . ] I was told to do a blood test. - Why didn’t you have it done (lit. have/let it done to yourself)? - Because I’ve been afraid of needles since childhood [. . . ]’
A list of complement verbs is given in list 12. It is impossible to decide whether the above example should be translated as Why did you not have it done? or Why did you not let it be done?. Both readings are provided by the principal possibility of da(va)t to denote factitive and permissive causation, and extensionally, both can coincide: having oneself treated necessarily entails letting oneself be treated. In Russian, where the factitive reading is not available, these cases may felicitously be translated using da(va)t+inf because the loss of the factitive nuance under translation does not influence their extensional meaning. In the parallel corpus, 7 potentially factitive cases in Polish were found, and the translation makes the distinction between unambiguously factitive and indeterminate cases clear. In (234), the Russian (and, it be added, the Serbian) translation expresses only permissive meaning and the factitive interpretation is lost, but this does not impact on the situation described: (234)
PL Wi˛ec dasz sobie zrobi´c encefalogram? So let.2SG RFX.DAT make.INF encephalogram.ACC? ‘So you’ll have an encephalogram made / let (them) make an ii encephalogram?’
i Dziennik
Polski, 2001-12-22, Tatusiowie po godzinach
ii Stanisław
Lem, Solaris
Passive da(va)t plus infinitive
149
RU Xot’ e˙ ncefalogrammu-to daš’ sebe sdelat’? At.least encephalogram.ACC let.2SG RFX.DAT make.INF ‘At least you’ll let them do an encephalogram?’ However, this is not the case with unambiguous curative causations as in (235), where the Russian translation must use a pleosemantic verb: (235)
PL Dałe´s2S G zrobi´cINF ? ‘You had it done?’ RU Zakazal? ‘You ordered it?’
3.3.4.
Summary
i
I conclude that factitive usage of da(va)t+inf is marginal today, but clearly possible. Two subtypes are to be distinguished: curative causation, which is clearly productive, but seldom used; as we will see in section 6.5, in earlier stages of Polish this function of da(va)t+inf was more prominent. Aside from curative use, more general factitive meaning of da(va)t such as in da´c na siebie czeka´c is found with a small set of verbs; these are considered lexicalizations. Their development seems to be linked to the finding that in some cases, notably reflexive constructions, there is an overlap between the extensional meaning of permissive causation and factitive causation. Aside from these clearly causative usages, there are cases of da(va)t+inf with dative causee that are semantically close to curative factitives, but with a lower degree of event integration. These should be seen to be derived from literal giving; this issue is returned to below in the discussion of Czech.
3.4.
Passive da(va)t plus infinitive
This comparatively frequent type (n=90) already shortly described for Russian above is represented in the next example: (236)
W Urz˛edzie Stanu Cywilnego spotkało si˛e 6 par mał˙ze´nskich, którym danePPP:N byłoAUX:PS T :3S G:N prze˙zy´cINF ze soba˛ 50 lat. ‘Six couples met at the city hall to whom it was2 given1 to live3 through 50 years together.’
(237)
Wielu z nich [=kibiców] nie byłoAUX:PS T :3S G:N danePPP:N wej´sc´ INF na stadion, bo ch˛etnych było wi˛ecej ni˙z miejsc.
i Stanisław
Lem, Powrót z gwiazd
150 da(va)t+inf in Polish n=6: prze˙zy´c ‘live through’; n=4: cieszy´c ‘enjoy’; n=3: uczestniczy´c ‘take part’, spotka´c ‘meet’, prze˙zywa´c ‘live through’, obejrze´c ‘view’; n=2: wyst˛epowa´c ‘appear’, sp˛edzi´c ‘spend (time)’, do´swiadczy´c ‘witness’, wysłucha´c ‘hear out’, zadebiutowa´c ‘make a debut’, zagra´c ‘play’, doczeka´c ‘wait’, zobaczy´c ‘see’; 52 hapax legomena Verb list 13: Complement verbs of passive da(va)t
‘To a lot of the [fans] it was not given to enter the stadion, as there were more people than places.’ This use is quite similar to German es war ihm (nicht) gegeben or English it was not given to them. It expresses the occurrence or not-occurrence of something that is favorable to the ‘causee’ by virtue of uncontrolled circumstances, such as fate or god. This last component seems to be weaker than in English or German (where examples such as (236) or (237) seem ill conceived). No causer or agent may be introduced in a regular passive agent phrase. The range of verbs used in this construction and given in list 13 shows that the construction is productive. There is little tendency towards negation (22 of 90); the ‘causee’ or agent is always animate. With 90 attestations, the construction is remarkably frequent in view of the figures concerning nonreflexive permissives (160 attestations). In Russian, in contrast, there were about 300 cases of dano as opposed to over 3,000 of permissive da(va)t. I thus regard this type as a distinct type that is a specific, metaphorical extension of giving with a merely weak connection to the permissive.
3.5.
Cognitive and perception verbs
As in Russian, examples with a subset of verbs was annotated separately as they, on the one hand, denote cognitive processes or non-volitional perception and, on the other, exhibit factitive meaning. The picture is slightly more complicated than in Russian, as the permissive and this class are not as clearly distinguished. As we have seen, factitive da(va)t in Polish today is very infrequent; the verbs discussed here therefore form a special class in any case. The complement verbs subsumed under this class were (238)
zna´c ‘know’ (186), pozna´c ‘recognize’ (81), odczu´c ‘feel’ (12)
Cognitive and perception verbs
151
Note that the functionally similar da´c do zrozumienia ‘give to understand’ and da´c do my´slenia ‘give to think’, both constructed with a prepositional phrase involving a verbal noun, rather than with the infinitive, express similar notions. They were found with comparable frequencies of 207 and 58 times, respectively, in the sample corpus33 and thus seem to have a comparable functional load. This construction type is beyond of the scope of this study. With the complement verb zna´c ‘know’ and pozna´c ‘recognize, get to know’, several types have to be distinguished. First of all this concerns da´c zna´c with the meaning ‘to inform’, as in (239). Like in Russian, the combination with a directional PP is evidence of its lexicalization. This likewise applies to da´c o sobie zna´c, also met with inanimate subjects, as in (240). (239)
Sprzedawca to jednak zauwa˙zył i dał3S G zna´cINF ochronieDAT . i ‘The sales clerk however noticed and informed security (lit: let know security)’
(240)
Gro´zna infekcja [. . . ] dałaPS T :3S G oabout sobieRFX:LOC zna´cINF po ii dwóch tygodniach. ‘A dangerous infection made itself be known after two weeks (lit. let about itself know).’
Zna´c is also used as an aspectual partner of pozna´c in da´c po sobie pozna´c/zna´c ‘let see/recognize on oneself’, which has primarily permissive meaning, as in (241). Here it means (not) to let show and is used remarkably close to German sich etwas anmerken lassen: (241)
Iglione nie miała zamiaru da´c poon sobieDAT zna´c, z˙ e obserwowała iii cała˛ scen˛e [. . . ]. ‘Iglione did not plan to show (lit.: let1 it be noticed3 on herself 2 ) that she had observed the whole scene.’
(242)
[. . . ] nikt nie dawałPS T :3S G poon sobieRFX:DAT pozna´cINF zdeneriv wowania. ‘[. . . ] nobody was showing signs (lit.: didn’t let on themselves recognize) that they were being unnerved ’
Pozna´c is used in this sense also without reflexive PP: i Jarosław
Abramow-Newerly, Młyn w piekarni, 2002 ii Dziennik Polski - 2000-08-26 Pejza˙z - Babcia Fauzija iii El˙zbieta Hajnicz, Przekraczajac ˛ s´wiatło, 1996 iv Dziennik Polski - 2002-07-01 - Kultura - Wirtuozerskie popisy
152 da(va)t+inf in Polish (243)
[. . . ] ale nie spostrzegł tego, przynajmniej nie dałPS T :3S G nic v pozna´cINF . ‘but he hadn’t seen that, at least he didn’t show he did (lit.: didn’t let anything recognize).’
Da´c si˛e pozna´c almost always appears with jako with a factitive meaning, denoting ‘make oneself known’, ‘be recognized as’ (244) as well as (without jako) ‘to make oneself acquainted with somebody, introduce oneself’. (244)
Od poczatku ˛ dałPS T :3S G si˛eRFX:ACC pozna´cINF jako człowiek bardzo i tolerancyjny dla cudzych pogladów. ˛ ‘From the beginning, he showed himself (lit. let himself know) as a man very tolerant of the views of others.’
Both da´c zna´c and da´c pozna´c thus share domains of usage and can denote both permissive and factitive meanings in what appears to be an intriguing network of uses connected to both regular senses of its component parts and lexicalized patterns. The third, much less frequent, verb in this group is odczu´c ’feel, realize’. It is used with both factitive and permissive meaning, expressing ‘inform in a subtle way’ (245) or ‘not let somebody notice’ (246): (245)
DanoIPS miDAT niedwuznacznie odczu´cINF , z˙ e przeszkadzam ii urz˛edniczkom w pracy. ‘I was given to understand (lit. give me feel) in an unambiguous way that I am being in the way of the clerks’ working.’
(246)
NieNEG dałPS T :3S G miDAT odczu´cINF , z˙ e umiera, chocia˙z, jako iii lekarz, s´wietnie o tym wiedział. ‘He didn’t let me notice he was dying, although, being a doctor, he knew about it perfectly well.’
Summary With cognitive verbs, both permissive and factitive causation is expressed. It is restricted to a small set of verbs; permissive and factitive reading depends on various contextual factors, including negation. This type is thus more heterogeneous than in Russian. At the same time it is, as in Russian, to a large extent lexicalized. v Antoni
i Dziennik Polski - 2003-08-07 - Proszowice - Był Libera, Madame, 2001 ii dobrym człowiekiem... Dziennik Polski - 2002-07-24 - Podhale - Sanepid nie pomo˙ze iii Jarosław Abramow-Newerly, Młyn w piekarni, 2002
The modal passive
3.6. 3.6.1.
153
The modal passive Types of the modal passive
The reflexive permissive has further grammaticalized to a modal passive (henceforth: MP) in Polish and Czech. Two constructions with different semantic and morphosyntactic characteristics have to be distinguished, an agreeing and an non-agreeing modal passive. The agreeing modal passive is superficially alike to a reflexive permissive with inanimate causer. It contains a nominative subject triggering agreement with da(va)t, a transitive verb and an accusative reflexive. The subject denotes the patient of the complement infinitive, as in (247); the English paraphrase suggests that this construction is roughly equivalent to a passive construction with a modal verb: (247)
Ksia˙ ˛zka nie da si˛e czyta´c. Book.ACC.SG NEG let.3SG RFX.ACC read.INF ‘The book cannot be read.’ (Modal passive; agreeing)
Besides this, there is also an non-agreeing modal passive in which no nominative NP is present. Any complements of the infinitive are realized as objects (note that direct objects in Polish are marked accusative in affirmative and genitive in negated clauses). da(va)t takes default 3.Ps.Sg.Neuter marking. As I will show, this construction is better rendered with an impersonal diathesis construction that does not involve the patient as subject, but as direct object: (248)
Ksia˙ ˛zki nie da si˛e czyta´c. Book.GEN.SG NEG let.3SG RFX.ACC read.INF ‘One can’t read the book.’ (Modal passive; non-agreeing)
Beside these two focal types some more types with non-transitive infinitives will be singled out below.In the following discussion, note that I will – slightly imprecisely – use patient as cover term for anything that is expressed as subject in (247) and as object in (248), that is, the phrases corresponding to direct objects in the basic diathesis. The non-agreeing MP construction shares characteristics with modal constructions such as formed with mo˙zna, nale˙zy and others in Polish that do not involve a nominative subject, on the one hand, and with impersonal diathesis constructions such as the bezosobnik formed with -no/-to or the impersonal reflexive construction, on the other. In all these constructions, no subject is expressed and understood as neglegible, generic or contexually retrievable,
154 da(va)t+inf in Polish while the patient is not promoted to subject status and retains object marking. Note that while the non-agreeing MP construction also involves the reflexive marker ‘si˛e’, it cannot34 be compositionally derived from any other sense of da(va)t and the impersonal (or, more precisely, the indefinite personal) reflexive construction. The question of the diachronic and structural relationship of these constructions is taken up again in the concluding part. In both the agreeing and non-agreeing MP the agent is not expressed and understood to be generic or retrievable from context. In both cases, it would therefore perhaps be more correct to speak of an modal impersonal instead of a modal passive, or, following usage more established in the slavic literature on diathesis, of an modal indefinite-personal construction. However, I retain this terminology for two reasons: firstly, it is established in the literature; secondly, this nomenclature would be cumbersome and potentially confusing in light of the wide-spread use of the term impersonal modal for non-agreeing modal constructions such as mo˙zna in Polish and elsewhere (see, e.g., Hansen & de Haan 2009: 530). This is mind, I use ‘modal passive’ as a convenient cover term. Whether we are dealing with agreeing or non-agreeing MP may be diagnosed on the basis of morphosyntactic coding: In Polish, the predicate agrees in number and person with the subject, which is given in nominative case; in the past (and the analytical future), the predicate also agrees in gender. In the non-agreeing as opposed to the agreeing MP construction, da(va)t takes default marking, i.e. 3rd person singular and, in the past, neuter gender and the patient is coded as a direct object. However, it is not always clear which of the two constructions are instantiated in a given attestation. da(va)t may take 3rd person neuter marking due to default marking, but also due to agreement. There are several, hierarchically ordered criteria for deciding between default marking and agreement: First, default 3rd person singular marking is unambiguous if the subject candidate is non-singular and/or, in past tense, non-neuter. Second, if this is not the case and the candidate is singular and (in the past) neuter, subject or object status of the candidate NP may be indicated by nominative or accusative case; however, these cases are morphologically distinguishable only if the noun is animate or grammatically feminine. In negated clauses, finally, objects are marked genitive, by what is called the genitive of negation, which makes these cases also unambiguous.35 These criteria allow a decision in most, but not all cases, mainly not if the noun phrase is singular, not feminine or animate in an affirmative sentence.
The modal passive
155
All instances found in the corpus were annotated according to membership into one of five types using these criteria. If subject- or objecthood could be established, these cases were coded either as agreeing or non-agreeing modal passive. If this could not be done, they were annotated as ambiguous. Two further types were included where morphological criteria distinguishing subject- or objecthood cannot be applied: First, cases with intransitive verbs such as z˙ y´c ‘live’ , including those governing a non-accusative object such as rzadzi´ ˛ c ‘govern’ (with object in instrumental case); second, cases with sentential complement. This categorization leads to the following types: • agreeing: patient as subject (249)
. . . trociny [. . . ] daja˛ si˛e barwi´c i sawdust.NOM.PL ... let.3PL RFX.ACC color.INF and formowa´c w dowolna˛ posta´c. form.INF in arbitrary.ACC.SG shape.ACC.SG ‘saw dust can be colored and formed into any shape’
i
• non-agreeing: patient as object (250)
. . . dałoby si˛e zapewne namówi´c pana let.PST.3SG.IRR RFX.ACC surely persuade.INF mr..ACC marszałka ... marshal.ACC ‘it would certainly be possible (lit.: let itself) persuade the marii shal’
• ambiguous cases (nom = acc, singular, in the past: neuter) (251)
W dodatku da si˛e bada´c In addition.ACC let.3SG RFX.ACC research.INF rynek poszczególnych market.ACC.PL.or.NOM.PL individual.GEN.PL produktów ... product-GEN:PL ‘In addition one can analyze the market for individual products / iii the market can be analyzed’
i Dziennik Polski - 2004-04-17 - Pejza˙ z - Biesiadny savoir-vivre RP // 6 marca 2002 iii Dziennik Polski - 1999-04-19
ii 11. posiedzenie Senatu
156 da(va)t+inf in Polish type agreeing non-agreeing ambiguous intransitive sentential compl.
n 260 792 288 202 82 1,624
rel 16% 49% 18% 12% 5% 100%
Table 27. Types of modal passive in the IPI PAN sample corpus.
• intransitive verb: no or non-accusative object (252)
Tu si˛e nie da z˙ y´c. Here RFX.ACC NEG let.3SG live.INF ‘One cannot live here.’i
• sentential complement (253)
[. . . ] nigdy nie da si˛e powiedzie´c, z˙ e ju˙z ... never NEG let.3SG RFX.ACC say.INF COMP already wszystko zostało tutaj zrobione. all.NOM.SG AUX.PST.3SG here done.PPP.NOM.SG ‘... it can never be said (lit.: lets itself say) that everything here is ii already done.’
As a term jointly referring to all construction types that do not involve a subject, that is, the non-agreeing type and the types involving intransitive verbs and with sentential complement, I speak of subjectless modal passives. I will later show that diachronically as well as synchronically, these subjectless modal passive types should be considered to belong to a single construction that is opposed to the agreeing construction. Table 27 shows the distribution of these constructions in the IPI PAN sample corpus; list 14 on page 170 provides an overview of the verbs used in these constructions. As can be seen, the non-agreeing MP is by far the most common type. In section 6.5 I show that this is a relatively young construction that became i Wojciech Jagielski, Modlitwa o deszcz, 2002 Sejmu RP//19.06.2002
ii Sprawozdanie
stenograficzne z obrad
The modal passive
157
this frequent no earlier than the first quarter of the 20th century. The agreeing modal passive, in contrast, is much older; the earliest examples I have found are from the 16th century. I will show that the non-agreeing modal is more grammaticalized, as it is more general in meaning, more restricted in form and more frequent. At the same time, da´c si˛e as a formant is more lexicalized in the non-agreeing passive, most importantly, because the reflexive element can only be analyzed as part of the lexical entry of da´c si˛e, while in the agreeing construction, the reflexive pronoun still shows variability and traces of its function as an anaphora. In the following, I will first concentrate on a qualitative analysis of the agreeing and non-agreeing modal passive construction, that is, on those constructions where there is a choice between object and subject coding of the patient. I will then extend the discussion to modal passive constructions with intransitive verbs and sentential complements and examine the quantitative behavioral profiles of all construction types.
Semantic contrasts The modal passive in Polish denotes the dynamic modality of possibility. Referring to both the Czech and the Polish modal passive, Rytel (1982) points out: “These constructions, firmly connected to aletic modality, can express [. . . ] neither deontic, nor epistemic modality[.] (my translation from the Polish36 ” While both modal passive constructions may often be paraphrased with an impersonal, subjectless modal such as mo˙zna, roughly translated to English as ‘one can’, the agreeing MP differs from mo˙zna as well as from the non-agreeing MP in that it carries more specific modal semantics. This is shown in the following contrast, an example based on sentence taken from Stanisław Lem and originally containing an agreeing construction. The main clause expresses an impossibility: certain chasms on the planet’s surface cannot be passed over. The subordinate clause under (a) explicitly relates this to a characteristic of the subject referent; this is fully felicitous. The subordinate clause under (b) only implicitely refers to such characteristics (by odpowiedny ‘suited’), and under (c), reference is made to circumstances quite independent of the clefts themselves. With the agreeing MP, the variant under (a) is most appropriate, while the variants under (b) and (c) are less acceptable.
158 da(va)t+inf in Polish (254)
Rozpadliny nie daja/dadz ˛ a˛ si˛e przej´sc´ , rifts.NOM.PL NEG give.IPF.3PL/PF.3PL RFX.ACC pass.INF ‘The rifts cannot (don’t let themselves) be passed,’ a. bo sa˛ za gł˛ebokie. ‘because they are too deep.’ b. ?bo nie ma odpowiednego pojazdu. ‘because there’s no suitable vehicle.’ c. ??bo nie ma czasu. ‘because there’s no time.’
The construction thus expresses modality grounded in characteristics of the subject referent, thereby describing the referent itself. In this respect, the construction is comparable to modal adjectives (e.g., traversable in the given example). This contrasts with the non-agreeing construction. Here, all three subordinate clauses are felicitous, just as with the general impersonal modal mo˙zna: (255)
Rozpadlin nie da/daje si˛e przej´sc´ , rifts.GEN.PL NEG give.IPF.3SG/PF.3SG RFX.ACC pass.INF ‘One/we can’t pass the rifts,’ a. bo sa˛ za gł˛ebokie. ‘because they are too deep.’ b. bo nie ma odpowiednego pojazdu. ‘because there’s no suitable vehicle.’ c. bo nie ma czasu. ‘because there’s no time.’
(256)
Rozpadlin nie mo˙zna przej´sc´ , rifts.GEN.PL NEG poss pass.INF ‘One/we can’t pass the rifts,’ a. bo sa˛ za gł˛ebokie. ‘because they are too deep.’ b. bo nie ma odpowiednego pojazdu. ‘because there’s no suitable vehicle.’ c. bo nie ma czasu. ‘because there’s no time.’
We see that the agreeing, but not the non-agreeing MP correlates the source of modality with the patient, leading to a semantic mismatch of the agreeing MP in cases where the source of modality is unrelated to what is expressed as subject.
The modal passive
159
This correlation is a question of degree, since the possibility of doing something to some patient logically always depends on both qualities of the patient and on other circumstances. While the difference between agreeing and non-agreeing MP is most pronounced with concrete patients, in other cases, the connection to some intrinsic quality of the patient is less obvious. Consider the following attestation together with its adapted non-agreeing counterpart: (257)
Lecz obecno´sc´ rozumnych istot na But presence.NOM.SG intelligent.GEN.PL being.GEN.PL on Kwincie nie dawała si˛e zakwestionowa´c. Kwinta.LOC.SG NEG let.PST.3SG.F RFX.ACC question.INF ‘But the presence of intelligent being on Kwinta could not be quesi tioned (lit. didn’t let itself question).’
(257)’ Lecz nie dawało si˛e zakwestionowa´c But NEG let.PST.3SG.N RFX.ACC question.INF obecno´sc´ i rozumnych istot na presence.GEN.SG intelligent.GEN.PL being.GEN.PL on Kwincie. Kwinta.LOC.SG ‘But the presence of intelligent being on Kwinta could not be questioned (lit. didn’t let itself question).’ The sentence under (257) expresses that the presence of intelligent beings could not be questioned, e.g., because of its obvious nature. Changing this to an non-agreeing MP in (257)’ suggests it might have been for logical or other reasons. The interpretation changes to that of failed multiple attempts or general impossibility for some unspecified reason. In general, the semantic contrast described here leads to subtle changes of interpretation, rather than to clear-cut differences in acceptability. In some cases, however, such as in (258), a change from non-agreeing to agreeing MP is infelicitous: (258)
Najbardziej stresujacy ˛ był dziennik popołudniowy. NiektóreACC informacjeACC dałoPS T :3S G:N si˛eRFX:ACC przygotowa´cINF wcze´sniej i trzyma´cINF w „lodówce”, ale bie˙zacy ˛ serwis muii siał by´c opracowywany na goraco. ˛ ‘Most stressful was the afternoon paper. Some information could
i Stanisław
Lem, Fiasko
ii Jan
Paulas; Cezary S˛ekalski, Oni nas stworzyli, 2002
160 da(va)t+inf in Polish be (lit. let itself; non-agreeing) prepared earlier and held in the ‘refrigerator’, but the live service had to be handeled on the spot.’ (258)’ presents the corresponding phrase in a agreeing construction, which is hardly an option: (258)’ Niektóre informacje ?dawałyPS T :3PL / ?dałyPS T :3PL si˛e przygotowa´c wcze´sniej [. . . ]. The precise semantic contours of different modal operators are notoriously difficult to capture and these constructions do not form an exception. While a more detailed analysis is left for further research, it seems safe to conclude that the agreeing MP construction is more specific than the non-agreeing construction: some characteristic of the subject is construed to be decisive for the (im)possibility of the event expressed by the lexical verb. It thus denotes participant internally grounded dynamic possibility (see, e.g., Hansen 2001); in contrast to potrafi´c ‘to be able to’ that denotes possibility grounded in the agent, the agreeing MP expresses possibility grounded in the patient of a transitive verb. The non-agreeing construction, in contrast, is less specific; it denotes dynamic possibility proper. The same, mutatis mutandis, can be said of the other subjectless types. Modal passives with intransitive verbs and with verbs with a sententional complement pattern syntactically with the non-agreeing modal passive, since there is no patient involved that could be promoted to subject. In general, they can also be replaced by the general impersonal modal mo˙zna without loss of felicity, e.g. (259) with an intransitive verb and (260) with a verb governing an object in instrumental case. The original version is given in (a) and the replacement in (b): (259)
[. . . ] bo nie umieli´smy wykorzysta´c nawet stuprocentowych sytuacji i czy rzutów karnych. ‘... because we didn’t manage to take advantage of even completely favoring situations or penalty kicks.’ a. A tak nie da si˛e wygra´c. And so NEG let.3SG RFX.ACC win.INF ‘And that way you (simply) can’t win (lit. doesn’t let itself win)’ b. A tak nie mo˙zna wygra´c. And so NEG poss RFX.ACC win.INF ‘And that way you (simply) can’t win.’
i Dziennik
Polski - 2004-04-17
The modal passive
(260)
161
Ulic˛e przegradzała do połowy chaotyczna zbieranina barierek, gui mowych pachołków i ta´sm, ‘The street was blocked halfways by a chaotic motley of railings, rubber bollards and bands,’ a. ale prawa˛ strona˛ dało si˛e but right.INS.SG side.INS.SG let.PST.3SG.N RFX.ACC przejecha´c. pass.INF ‘but one could pass on the right side (it let itself pass)’ b. ale prawa˛ strona˛ mo˙zna było but right.INS.SG side.INS.SG poss AUX.PST.3SG.N przejecha´c. pass.INF ‘but one could pass on the right side’
The subjectless modal passive constructions should not, however, be seen as wholly equivalent to mo˙zna. First of all, they are stylistically more colloquial; secondly, they often involve some sense of successful or unsuccessful attempt, expressing a relation that is contextually grounded rather than completely generic. Depending on contextual factors not fully analyzed here, they often tend to express that something was successfuly done, rather than merely possible to do; this leads to occasional contrasts with mo˙zna: (261)
a. ?Dało si˛e je wyleczy´c w ciagu ˛ Let.PST.3SG.N RFX.ACC her.ACC.SG cure.INF in course tygodnia, ale nie wyleczono. week.GEN.SG but NEG cure.PST.IPS ‘Intended: She could have been cured in the course of one week, but they didn’t cure her.’ b. Mo˙zna było je wyleczy´c w ciagu ˛ poss AUX.PST.3SG.N her.ACC.SG cure.INF in course tygodnia, ale nie wyleczono. week.GEN.SG but NEG cure.PST.IPS ‘She could have been cured in the course of one week, but they didn’t cure her.’
Categorial restrictions apply in regard to unaccusative verbs such as zachorowa´c ‘fall sick’, umiera´c ‘die’ or upa´sc´ ‘fall’ which cannot be used with the modal passive but are felicitous with mo˙zna. i Artur
Baniewicz, Góra Trzech Szkieletów, 2003
162 da(va)t+inf in Polish Da´c si˛e seems to be infelicitous here because it gives rise to the assumption that one may attempt to get sick or die; no such assumption arises with mo˙zna. In fact, in a context where such an assumption is plausible, da´c si˛e may be used. E.g., one may say zawsze da si˛e zachorowa´c ‘one can always get sick’ in a situation where ‘being sick’ is an effective pretext for getting a day off. The unexpressed subject of the modal passive is thus implied to be an agent. We will see that with the cognate construction in Czech, combinations with unaccusative verbs are fully felicitous independent of the agent status.
Summary The modal passive expresses dynamic possibility with an optional element of successful or unsuccessful attempt. It is more restricted than the core modal mo˙zna, however, since it may not be used with unaccusative verbs; furthermore, it seems to be more prone to state the actual realization of the event besides its general possibility. The agreeing modal passive, in contrast to all subjectless types, additionally expresses that possibility is grounded in characteristics of the patient of the lexical verb; it is therefore richer semantically and more specific than the non-agreeing MP it may be contrasted with.
3.6.2.
Modal passive constructions and the reflexive permissive.
In the present subsection, I show that agreeing and non-agreeing modal passives differ in regard to traits shared with the reflexive permissive. Nonagreeing modal passives are more grammaticalized and more clearly differentiated from reflexive permissives, while the agreeing modal passive is closer to the reflexive permissive from which it has developed.
Ambiguity If the causee is not expressed, the agreeing modal passive and the reflexive permissive are identical in surface structure. Animate subjects usually lead to a default interpretation as permissive causative, cf. a modified corpus example in (262).
The modal passive
(262)
163
Tygrysy si˛e nie daja˛ zje´sc´ . Tiger.NOM.PL RFX.ACC NEG let.3PL eat.INF ‘Tigers don’t let themselves be eaten.’ (Permissive reading) ‘?Tigers are not edible /cannot be eaten.’ (Modal passive reading)
However, in some cases the modal passive is used with an animate subject that retains no agentive properties, as in (263). Here, the subject does not influence whether he can or cannot be reached, and a modal passive reading arises: (263)
Drzymalski nie da3S G si˛eRFX:ACC łatwo i szybko namierzy´cINF . i [. . . ] raczej on nas namierzy. ‘Drzymalski cannot (lit. does not let himself) be easily and quickly reached. ... instead, he will locate us.’
According to Gehrmann (1983: 12) the addition of modal adverbs such as łatwo ‘easily’ will enforce the modal passive reading even with an animate subject. This does not seem to be entirely correct, however, since such adverbs are also possible in the reflexive permissive, as (264) shows. (264)
[. . . ] instruktor nie dałPS T :3S G si˛eRFX:ACC łatwo przekona´cINF . ‘[. . . ] the instructor did not let himself be easily persuaded.’
ii
This possibility is also given with reflexive permissives in Russian, where no modal passive has evolved. This shows that the possible use of adverbs such as łatwo ‘easily’ does not depend on the availability of a modal passive reading, but on general properties of the reflexive permissive construction, which in turn provided the necessary grounds for a development of this construction into a modal. Returning the question of ambiguity, we have seen that reflexive permissives and agreeing modal passives have – as a rule – complementary distribution regarding animacy of the sentence subject, with some overlap existing. Non-agreeing modal passives, in contrast, freely combine with animate patients, as in (265)
i Artur
Syłenki nie dało si˛e obudzi´c. Sylenka.GEN.SG NEG let.PST.3SG.N RFX.ACC wake=up.INF ‘It wasn’t possible to wake Sylenka up.’
iii
ii Jan Paulas; Cezary Baniewicz, Drzymalski przeciw Rzeczpospolitej, 2004 iii S˛ekalski, Oni nas stworzyli, 2002 Artur Baniewicz, Drzymalski przeciw Rzeczpospolitej, 2004
164 da(va)t+inf in Polish Here, in the non-agreeing case, there is no relation to the reflexive permissive. The constructions are clearly separate, they do not stand in competition. This can be related to the above discussed semantic difference between the agreeing and the non-agreeing MP. In the reflexive permissive, the causer is in some way responsible for what is done by the (usually unexpressed) causee; he or she makes it possible. In the agreeing MP, the subject is also in some way responsible, in a non-agentive sense, for the (im)possibility of the potential event. There thus is some zone of overlap of the semantics of the reflexive permissive and the agreeing MP not found in regard to the nonagreeing MP.
Expression of the anonymous agent. Both agreeing and non-agreeing MP constructions imply an anonymous agent, similar to causeeless permissive reflexives. In contrast to the latter, no dative causee or agent phrase is possible, as (266) illustrates for both constructions: (266)
To / tego nie da si˛e This.NOM.SG / this.GEN.SG NEG let.3SG RFX.ACC (*mu) zrobi´c. (*he.DAT.SG) do.INF ‘This can’t be done (*by him)’ (agreeing and non-agreeing MP)
However, in the full corpus six cases of agreeing MP constructions with an agent given in a przez-phrase were found, as in: (267)
Te pierwsze [przekształcenia ...] daja˛3PL si˛eRFX:ACC całkowicie kształtowa´cINF przezby rzad ˛ ACC i parlamentACC , te drugie za´s i tylko cz˛es´ciowo. These first [transformations] may be (lit.: let themselves be) formed by government and parliament in their entirety, the other ones, however, only to a certain degree.
(268)
ProblemNOM:S G zabezpieczenia kraju w z˙ ywno´sc´ nie da3S G ii si˛eRFX:ACC rozwiaza´ ˛ cINF przezby samego pana ministraACC . ‘The problem of supplying the country with food cannot (lit. does not let itself) be solved by the minister himself.’
i Sprawozdanie stenograficzne z obrad Sejmu RP // 01.03.1996 stenograficzne z obrad Sejmu RP // 25.07.1992
ii Sprawozdanie
The modal passive
165
This is not an usual option and was only found in parliament proceedings, that is, spoken language. It was only found with with agreeing modal passives, even though this is the by far less frequent construction37 . In fact, it seems the non-agreeing MP does not allow such a possibility at all. While example (268) is accepted by various native speakers, the modified example (268)’, where the agreeing MP has been replaced with the non-agreeing MP, is only acceptable if the agent phrase is omitted: (268)’ ProblemuGEN zabezpieczenia kraju w z˙ ywno´sc´ nie da3S G si˛eRFX:ACC rozwiaza´ ˛ cINF (*przezby samego pana ministraACC ) rozwiaza´ ˛ cINF . ‘One cannot2 solve5 the problem1 of supplying the country with food (*by3 the minister4 himself)’ This suggests that there is some connection between between the agreeing MP and the reflexive permissive that is not given between the non-agreeing MP and the reflexive permissive. Variability of the pronoun The reflexive marker enjoys different degrees of freedom in the agreeing, but not in the non-agreeing MP. In neither MP construction may the reflexive clitic pronoun si˛e be changed to its tonic variant siebie: (269) *To / tego nie da siebie zrobi´c. This.NOM.SG / this.GEN.SG NEG let.3SG RFX.ACC do.INF Intended: ‘This can’t be done ’ (agreeing and non-agreeing MP) This is a clear contrast to the reflexive permissive, where, in general, the tonic variant of the reflexive pronoun is possible, if extremely infrequently met. In the agreeing modal passive, unlike the non-agreeing construction, some variation is possible. The reflexive may be given in instrumental case, cf. the constructed examples: (270)
Taki kraj nie da/daje soba˛ Such.NOM.SG country.NOM.SG NEG let.P.3SG/I.3SG RFX.INS rzadzi´ ˛ c. govern.INF ‘Such a country can’t (lit. doesn’t let itself) be governed.’
(271)
Samochód nie da/daje soba˛ kierowa´c. Car.NOM.SG NEG let.P.3SG/I.3SG RFX.INS control.INF ‘The car can’t (lit. doesn’t let itself) be controlled.’
166 da(va)t+inf in Polish However, in both cases an non-agreeing MP as in samochodem nie da si˛e kierowa´c would clearly be preferred: (272)
Samochodem nie da si˛e kierowa´c. Car.INS.SG NEG let.P.3SG/I.3SG RFX.ACC control.INF ‘The car can’t be controlled.’
This possibility, again, suggests an overlap between permissive and agreeing MP which the non-agreeing MP does not exhibit. However, the use of nonaccusative reflexive pronouns with agreeing MPs is highly unusual.
Negation of the infinitive We have already seen that permissives do not usually allow independent negation of the infinitive for semantic reasons. Non-agreeing da´c si˛e behaves comparably to the central impersonal modal mo˙zna in allowing infinitive negation, but agreeing modal passives do not. The next example is originally constructed with mo˙zna. It can be replaced by the non-agreeing, but not with the agreeing MP: (273)
a.
W Memfis mo˙zna przez całe z˙ ycie nie zobaczy´c burzy. In Memphis poss through whole life not see storm ‘In Memphis it’s possible to never in your life see a storm .’ b. W Memfis da3S G si˛eRFX:ACC przez całe z˙ ycie nieNEG zobaczy´cINF burzy. ‘In Memphis it’s possible (lit. it lets itself) to never in your life see a storm.’ c. *W Memfis burzeNOM:PL daja˛3PL si˛eRFX:ACC przez całe z˙ ycie nieNEG zobaczy´cINF . Intended: ‘In Memphis storms can (let themselves) not be observed in the course of a life time.’
However, no examples of this sort were found in the corpus. Only the negation of both predicates is attested in the sample corpus, again, only with the nonagreeing MP: (274)
[. . . ] jednak mimo to nie dało si˛e nie ... however besides that NEG let.PST.3SG.N RFX.ACC NEG zauwa˙zy´c, jak bardzo Aiohe jest przej˛eta. notice.INF how very Aiohe is worried ‘... besides, however, one cannot but notice how worried Aiohe is.’
The modal passive
167
This reflects the general possibility of double negation with modals of possibility resulting in a necessity interpretation: one cannot but notice ⇔one has to notice. Here, again, the non-agreeing MP is more like a central modal that makes use of this equation, while the agreeing modal passive is more like the reflexive permissive, where infinitive negation is hardly possible.
Conclusions The personal MP has a largely complementary distribution with reflexive permissive in regard to animacy of the subject. In contrast, the non-agreeing MP is not restricted in any way in regard to inanimate patients. Selectional restrictions apply only in respect to the unexpressed agent, which is in both cases understood to be human and agentive. The agreeing MP has meaning components as well as syntactic characteristics it shares with the reflexive permissive. The agreeing, but not the nonagreeing modal passive is sometimes used with a prepositional agent, like the reflexive permissive; it allows some variation of the reflexive marker. Finally, independent negation of the infinitive is attested only with the non-agreeing MP, but neither with the agreeing MP nor the reflexive permissive. These points advocate a less grammaticalized status of the agreeing than the non-agreeing MP in that it imposes less selectional restrictions on clausal arguments; furthermore, the former has links to the reflexive permissive which the latter lacks. This synchronic finding is corroborated from a diachronic point of view: as will be shown and further discussed in the diachronic chapter below, the agreeing construction is the initial form, developed well before non-agreeing constructions evolve in a further stage of development.
3.6.3.
Usage profiles
Table 28 repeats the distribution of MP types given in table 27 and relates these types to matrix and complement aspect as well as negation. Again, two groups emerge in respect to negation and complement aspect including all subjectless modal passive types, on the one, and the agreeing modal passive construction, on the other hand.
168 da(va)t+inf in Polish
type agreeing non-agr. ambig. intrans. snt. cmpl.
n 260 792 288 202 82 1,624
rel 16% 49% 18% 12% 5% 100%
impf. dawá´c abs. rel. 198 76% 29 4% 69 24% 11 5% 3 4% 310 19%
impf. inf. abs. rel. 45 17% 92 12% 19 7% 91 45% 2 2% 249 15%
negated abs. rel. 106 38% 626 79% 77 28% 155 77% 69 86% 1,033 64%
Table 28. Types of modal passive in the IPI PAN sample corpus. The first column gives the number of occurrences in absolute and relative terms. The next three subdivisions provide data on the absolute and relative frequency of imperfective matrix, imperfective complement and negation in relation to each type
type agreeing non-agr. ambig. intrans. snt. cmp.
forms of dawa´c part. past pres. 135 19 44 0 16 13 0 14 55 0 8 3 0 3 0 135 60 115
forms of da´c inf. past pres. 3 35 24 4 174 585 2 52 165 0 67 124 0 11 68 9 339 966
all 260 792 288 202 82 1,624
dawa´c 198 (76%) 29 (4%) 69 (24%) 11 (5%) 3 (4%) 310 (19%)
Table 29. Forms of dawa´c and da´c according to annotated types. Part. refers to the present participle, pres. to morphological present.
As to matrix aspect, imperfective matrix dawa´c is rare with all subjectless types (around 5%), but the dominant option with agreeing modal passives ( 76% of all occurrences). The proportion is intermediate for ambiguous cases (24%). The difference between agreeing and non-agreeing transtive MP is highly significant and fairly large (p<.001, V=.49). A similar contrast is found for negation. Instances of the subjectless types are negated to around 80%; the agreeing construction, in contrast, only to 38%, a difference that is highly signicant, albeit small (p<.001, V=.12)38 . In respect to infinitive aspect, no such grouping is found; the difference between the agreeing and the non-agreeing type here is barely signicant and
The modal passive
169
very small (p=.05, V=.06). Like with permissives, perfective aspect of the infinitive dominates to more than 80%, except with intransitive verbs, where the proportion is around 50%. Note this differs from the prototypical impersonal modal mo˙zna, where the proportion of perfective infinitives is about 65% overall and 56% with negation39 . The difference in distribution of infinitive aspect of the subjectless types, calculated against negated mo˙zna, is highly significant and of moderate size (p<0.001, V=.24). In respect to two of the three characteristics, two classes thus emerge: the agreeing modal passive that is mostly used in affirmation and with the imperfective form dawa´c, and, in contrast, the group of subjectless types that are predominately used under negation and almost exclusively with perfective da´c. Note that this division is not due to the influence of individual frequent lexemes, as it remains valid if we restrict the calculation to verbs that only appear once40 . A possible explanation for the exceptional pattern of agreeing modal passives (affirmative, but with imperfective aspect) is its use, in general, for the expression of some – temporally not localized – characteristic of the patient. Table 29 shows that, specifically, of 260 cases of the agreeing modal passive, 135 were as active present participle, forming syntagms functionally largely equivalent to modals!modal deverbal adjectives such as przewidywalny ’forseeable’ that assert something about the patient, as in (275)
Moim zdaniem, w dajacej ˛ si˛e przewidzie´c My.INS opinion.INS, in let.PRS.PTCP.LOC RFX.ACC forsee.INF przyszło´sci - nie. future.LOC.SG - no ‘[But is a Moscow-Berlin pact that is directed against us possible?] In my opinion, in the foreseeable (lit. letting itself foresee) future i no.’
If we discount all uses of this participle, we arrive at a proportion of 63 imperfective and 62 perfective cases, that is, an even distribution. At first sight, this is still remarkable taking into account that in all languages investigated in this study, and specifically with the reflexive permissive in Polish, perfective matrix aspect is so dominant. However, a closer look at the distribution of forms across tenses in table 29 shows that with agreeing modal passives, perfective da´c si˛e is used more often in the past and imperfective dawa´c si˛e is more often used in the present. This is what has been i Dziennik
Polski - 1998-12-11 - Magazyn
170 da(va)t+inf in Polish Complement verbs by type Agreeing: n:22: pogodzi´c ‘reconcile’; n:19: przewidzie´c ‘predict’; n:15: słysze´c ‘hear’, zauwa˙zy´c ‘notice’; n:13: odczu´c ‘feel’; n:9: usuna´ ˛c ‘remove’; n:7: wytłumaczy´c ‘explain’; n:5: opanowa´c ‘master’; n:4: porówna´c ‘compare’; n:3: okre´sli´c ‘describe’, ukry´c ‘hide’, u˙zy´c ‘use’, wykorzysta´c ‘exploit’, wyleczy´c ‘heal’; 23 verbs with n=2; 109 hapax legomena Non-agreeing: n:38: zrobi´c ‘do’; n:29: unikna´ ˛c ‘avoid’; n:24: słysze´c ‘hear’; n:18: zauwa˙zy´c ‘notice’; n:12: przeprowadzi´c ‘perform’, przewidzie´c ‘predict’, rozwiaza´ ˛ c ‘solve’, ustali´c ‘determine’; n:11: naprawi´c ‘fix’, powiedzie´c ‘say’; n:9: pogodzi´c ‘reconcile’, prowadzi´c ‘lead’, wprowadzi´c ‘introduce’, załatwi´c ‘arrange’; n:8: odwróci´c ‘reverse’, oszuka´c ‘cheat’, robi´c ‘do’; n:7: obroni´c ‘defend’, opisa´c ‘describe’, osiagn ˛ a´ ˛c ‘achieve’, porówna´c ‘compare’, ukry´c ‘hide’, zastapi´ ˛ c ‘replace’, zmieni´c ‘change’; n:6: odczu´c ‘feel’, połaczy´ ˛ c ‘connect’, przeceni´c ‘overestimate’, stworzy´c ‘create’, wyeliminowa´c ‘eliminate’, wykona´c ‘perform’, wyleczy´c ‘cure’, wytłumaczy´c ‘explain’; n:5: podzieli´c ‘divide’, pokona´c ‘defeat’, poprawi´c ‘improve’, usuna´ ˛c ‘remove’, wykluczy´c ‘exclude’, wykorzysta´c ‘exploit’; n=4: 11 verbs; n=3: 25 verbs; n=2: 47 verbs; 213 hapax legomena Ambiguous: n:31: zrobi´c ‘do’; n:23: zauwa˙zy´c ‘notice’; n:10: wyczu´c ‘perceive’; n:9: załatwi´c ‘arrange’; n:8: odczu´c ‘perceive’, powiedzie´c ‘say’; n:7: ukry´c ‘hide’; n:6: wytłumaczy´c ‘explain’; n:4: rozwiaza´ ˛ c ‘solve’; n:3: dotkna´ ˛c ‘touch’, zniszczy´c ‘destroy’, powi˛ekszy´c ‘enlarge’, słysze´c ‘hear’, zrealizowa´c ‘realize’; n=2: 26 verbs, 99 singleton verbs Intransitive verb: n:24: z˙ y´c; n:7: gra´c; n:6: dojecha´c; n:5: pracowa´c, przej´sc´ , zapanowa´c; n:4: je´zdzi´c, przejecha´c, wyj´sc´ , wytrzyma´c, zarobi´c; n:3: by´c, doj´sc´ , odpowiedzie´c, patrze´c, porozmawia´c, rzadzi´ ˛ c, skaka´c, wy˙zy´c; 21 verbs with n=2; 63 hapax legomena With sentential complement: n:28: ukry´c ‘hide’; n:9: powiedzie´c ‘say’; n:6: ustali´c ‘determine’, wykluczy´c ‘exclude’; n:5: udowodni´c ‘prove’; n:3: stwierdzi´c ‘say’, zaprzeczy´c ‘deny’, zauwa˙zy´c ‘notice’; 3 verbs with n=2; 11 hapax legomena Verb list 14: Complement verbs according to types of modal passive construction.
The modal passive
171
found for permissives in Russian and both non-reflexive and reflexive permissives in Polish above (see 19 on page 122). In view of the profile of use in the present and specifically as present participle, aspect use of the agreeing construction can thus be accounted for by the usual rules of aspect usage in Polish. However, by the same reasoning, aspect usage with the non-agreeing construction is remarkable. Even though the construction is used predominately in the morphological present, it is primarily used with perfective da´c. This is unexpected, since actual present is normally expressed by imperfective aspect and perfective aspect in Polish (as well as the other North Slavic languages) in the morphological present normally expresses future reference. Again, the same is true of the other subjectless modal passive types, which behave as one group in this respect, too. This unexpected aspectual profile is explained in the next section, where it will be shown that da´c si˛e in these constructions is no longer unequivocally perfective by several criteria; rather, its ongoing loss of the aspect category is a symptom of its grammaticalization. We will see that in Czech this process has gone even further; only the erstwhile perfective verb, dát, is used in the modal passive. One more characteristic difference deriving from the syntactic profile of the subjectless vs. the agreeing construction should be noted. In the Polish subjectless MP constructions, three forms are predominant: the perfective third person singular (neuter) in non-past, past/conditional and infinitive (da, dało(by) and da´c). Three further forms, the imperfective counterparts of these, are possible, but seldom used: daje, dawało(by) and dawa´c. The agreeing construction, in contrast, allows almost full variation in aspect, tense, number and (in the past) gender41 , giving 2*2+2*2*3=16 word forms, to which infinitive and present participle forms must be added. The non-agreeing, but not the agreeing MP thus exhibits a reduction in the paradigm of the matrix verb that is typical for grammaticalization processes. However, we will see that this particular reduction is not found in Czech, where the construction as whole is even more grammaticalized.
3.6.4.
The aspect of da´c si˛e in subjectless MPs
In the last section, we have seen that with the subjectless modal passive, the paradigm of da(va)t tends to be reduced in usage to three forms of morphologically perfective da´c. In this section, I examine the aspectual status of da´c
172 da(va)t+inf in Polish si˛e in subjectless modal passive constructions more closely. I will show that it is not unequivocally perfective anymore, and suggest that it should at least be considered biaspectual. Note that I do not attempt an exhaustive analysis; the aim is to show that according to some criteria, da´c si˛e in some functions behaves like an imperfective verb. Verbal aspect in Slavic is a matter of the stem: lexemes are of either perfective or imperfective aspect and stand in a correlative relationship of typically two members. This relationship is essentially one of word formation by way of one of two main processes: derivation with verbal prefixes leads to aspectually perfective verbs: pisa´c ‘write (ipf)’ is opposed to the prefixed verb na-pisa´c ‘write (pf)’, originally meaning ‘write down’. Suffixation with a certain set of suffixes renders imperfective verbs: da´c ‘give (pf)’ in its donatory sense is opposed to da-wa-´c ‘give (ipf)’. While the relationship between perfective and imperfective verbs in the lexicon is subject to controversy, with different opinions regarding the lexical semantic equivalence of prefixed and unprefixed verbs such as pisa´c and na-pisa´c ‘write (pf), write (ipf), write down’ and the number of possible aspectual partners in the lexicon, the basic and uncontested fact about the system is that every verb is either discernably perfective or imperfective, with just a few exceptions and with some variation across the Slavic languages. Polish verbs in their vast majority are marked for aspect, with just a small group of biaspectual verbs, most of which are loan words or homonyms that differ in aspect (see Piwowar 2002 for a classification). More important to the question of the aspectual status of da´c si˛e than loan words are modal protrafi´c ‘to be able to’ and possibly causative kaza´c ‘command, cause’, which are both biaspectual verbs in more grammatical domains; loss of the aspectual opposition is here, as in the case of da´c si˛e, a result of advanced grammaticalization, or, put more generally, a result of the move into a different semantic and syntactic class of verbs. Modals in Polish are generally not marked for aspect: they denote states and are therefore imperfectiva tantum. In some other Slavic languages, modals do have derived perfective partners; these then contain an additional element of factivity, denoting that that which is necessary or possible did in fact take place; the overall meaning then shifts to predicates such as "manage" or "succeed", "necessary and consequently done". 42 I am not in a position to ´ do justice to the literature on this complex subject; see Smiech (1971); Rytel (1982); Koschmieder (1987); Holvoet (1989); Łazi´nski (1996); Wiemer (2001); Hansen (2001) for a discussion of aspect and modals in Polish and other Slavic languages.
The modal passive
173
Since in Polish none of the prototypical modals have a synthetic aspectual partner, loss of aspect can be expected for the grammaticalization of da´c si˛e as a modal. An example of the move of an erstwhile perfective verb into the class of modals is given in the development of potrafi´c ‘to be able to’, which used to be perfective and is today marked as biaspectual in standard dictionaries. In contrast, peripheral modals, such as uda(wa)´c si˛e ‘to manage’, which involves an additional element of ’success’ besides its modal meaning, take part in the aspectual opposition. The comparison with both the erstwhile perfective, now biaspectual potrafi´c and with the peripheral uda(wa)´c si˛e that shows no sign of a reduction of the aspectual paradigm serves as a background against which the aspectual status of da´c si˛e is investigated in this section. In order to investigate the aspectual status of da´c si˛e and other lexemes, I use the following formal and functional characteristics of aspect use in Polish as diagnostic criteria: • morphological criteria – only perfective verbs form past tense adverbial participles (in -wszy) – only imperfective verbs form the present participle (in -ac ˛ / -acy) ˛ – only imperfective verbs form the analytic future (formed with an auxiliary and the infinitive or the l-participle; e.g. b˛edzi˛eAUX:3S G robi-´cINF or b˛edzi˛eAUX:3S G robi-łl−participle:M:S G , both ‘will do’) • contextual criteria – temporal adverbials: with the exception of delimitative perfective verbs, only imperfective verbs combine with a temporal adverbials of duration (e.g., robiłip f /*zrobił p f to przez godzin˛e ‘he did (ipf/*pf) that for an hour’), only perfective adverbs combine with time spans (*robiłip f /zrobił p f w godzin˛e ‘he did (*ipf/pf) that in an hour’). – phasal verbs take only imperfective complements robi´cip f /*zrobi´c p f ‘he began to do (ipf/*pf) it’)
(zaczał ˛
– perfective verb forms with non-past morphology refer to the future (with some additional options; see Perlin 2003), imperfective verbs refer to the present (robi˛eip f ‘ I’m doing it’, zrobi˛e p f ‘I’ll do it’). – subordinate temporal clauses with a˙z / dopóki ‘while/until’ covary in their meaning with the aspect of the subordinate predicate (see below).
174 da(va)t+inf in Polish I do not use historical present contexts, which do not, as in Russian, completely exclude the use of perfective verbs (see Dickey 2000 for a comparison); for similar considerations, I do not systematically discuss habitual and iterative contexts.
3.6.4.1.
Morphological criteria.
It is instructive to see that not all biaspectual verbs behave alike even according to morphological criteria. With kaza´c ‘command, cause’ , the aspectually restricted participles pf. kazawszy (n=4) and ipf. ka˙zac, ˛ ka˙zacy ˛ (n=146) as well as the imperfective analytical future (n=6) is freely formed and (if unevenly) attested in the full IPI PAN corpus; its biaspectual status is therefore uncontested. With potrafi´c ‘to be able’ the morphological evidence pointing to a biaspectual verb is more problematic, since there is little evidence for perfective use. IPI PAN does not cite any usages of pf. potrafiwszy43 , while potrafiac(y) ˛ (n=695) and analytical future forms (n=298) are used and perfectly acceptable according to my informants. However, the Nowy słownik poprawnej polszczyzny PWN (NSPP) (Markowski 2000) explicitly states the incorrectness of the analytical future of potrafi´c. This can be explained by recent language change: at some earlier point, evidently in the first half of the 20th century, the analytical future was in fact impossible, or at least substandard (since the NSPP saw it necessary to comment on such usage), but has now evolved into being acceptable. While Zygmunt Saloni (p.c.) remarks that this form was still unacceptable to the older generation after WWII, the amateur linguist Grzegorz Jagodzi´nski claims on his website that potrafi´c is in fact exclusively imperfective today since it may not have future reference in the present, observing that the prescribed Nie martw si˛e, kiedy´s potrafisz to zrobi´c ‘don’t worry, one day you’ll be able to do that’ with future reference is nowadays ungrammatical 44 . Da´c si˛e is evidently in a less advanced stage of a similar development involving aspectual change. While it cannot form the imperfective present participle *dadzac(y), ˛ the perfective participle is also very rare: IPI PAN returns only 11 hits for dawszy, all relating to literal ‘give’. Searching the web for dawszy si˛e returned 92 hits, the vast majority of which stems from reflexive permissive or factitive usage; no examples for modal passive usage were found.45 Note that both participles *dadzac(y) ˛ si˛e and dawszy si˛e could only
The modal passive
175
be felicitously formed from the personal modal passive, since they involve a coreferent subject; by definition, no subject is present in the subjectless constructions. The only form thus potentially indicative of imperfective aspect in the impersonal modal passive construction is thus the analytical future. Analytical future of perfective verbs in IPI PAN Informants differ widely on the acceptability of the analytical future of da´c si˛e, with some informants accepting it readily and without hesitation and others rejecting it outright. In order to assess the status of this form, I conducted a corpus query searching the full IPI PAN corpus for analytical future formed with the lparticiple of perfective verbs (one of two possibilit. I searched for the third person singular auxiliary b˛edzie, followed within three words by an unambiguously perfective l-participle. Between these two words interpunctation, l-participles and mo˙zna were disallowed46 . This search yielded 206 hits, of which 16 cases of da´c si˛e and 11 other analytical future occurrences of perfective verbs are relevant for the present analysis; this is shown in table 30. The 11 other occurrences were clearly occasional mistakes; an example is given in (276), where the perfective word form wypełnił ’fulfill’ is clearly not grammatical and needs to be replaced by its imperfective counterpart wypełniał: (276)
Zaproponowany przez nas przepis b˛edzieAUX:3S G wypełniłPF:PTCP:S G regulacje zawarte w tej dyrektywie. (Sejm) ‘The rule that we have proposed will fulfill the regulations contained in the directive.’
This set of real mistakes did not involve any verb twice; the relevant forms involved a heterogeneous set of verbs that does not lend itself to a coherent semantic interpretation: doprecyzowa´c ‘clarify’, dotrze´c ‘arrive’, opłaci´c ‘pay off’, podja´ ˛c ‘take’, przygotowa´c ‘prepare’, utrzyma´c ‘maintain’, wynie´sc´ ‘amount’, wypełni´c ‘fulfill’, wystarczy´c ‘be enough’, wywoła´c ‘provoke’, zapewni´c ‘assure’. As could be expected, these mistakes were predominantly from spoken language transcribed in parliament proceedings; only 3 of 11 were found in journalistic texts. Native speakers with whom I discussed these attestations clearly identified them as mistakes. The second group, comprising 16 examples of da´c si˛e, is of a different quality. The full list of attestations is cited in the notes for reference47 . Attes-
176 da(va)t+inf in Polish type aspect errors da´c si˛e other l-participles false hits
n 11 16 75 104 206
construction type of da´c si˛e impersonal intransitive ambiguous
n 6 9 1 16
Table 30. Corpus query on analytical future forms involving perfective l-participles. False hits concerns crossed clause boundaries, tagging mistakes, missing punctuation and others; other l-particliples refers to predicative uses of l-participles such as in Czeka, a˙z sytuacja b˛edzie tak nabrzmiała, z˙ e. . . ‘waits, until the situation is so exaggerated, that’ and lexicalizations such as zwykły ‘ordinary’. Cases of da´c si˛e were categorized according to type: no agreeing modal passive instances were found.
tations involving modal passive da´c si˛e stand out in term of their frequency: they form the largest single group, making it highly implausible that they should be considered ‘mistakes’ in the same sense as the other attestations of analytical future of perfective verbs48 . Moreover, they are not, like the others, but from parliamentary transcripts, but without exception from the press. Lastly, informants differed widely in their assessment, unlike the uniform rejection of the other examples. Consider the following (277)
Jego dom znajduje si˛e na zboczu góry, która si˛e osuwa i nie wiadomo czy b˛edzieAUX:3S G si˛eRFX:ACC dałoPS T :S G:N w nim mieszka´cINF . ‘His house is located on the mountain slope that is sliding and it is is not clear whether it will be possible (lit. will itself let) to live in it.’
In contrast to the above examples, there is variation in the assessment of this attestation. By some informants, it was found to be completely, notably by two translators. Other informants, however, did find fault with it and specifically pointed out that of course it is not possible to form the analytic future of perfective verbs. While the amount and nature of the variation concerning this type of use goes beyond the present study, the point here is that there is variation, in contrast to the class of real mistakes in aspect usage, where no variatin is found. All examples involved subjectless constructions, rather than the personal modal passive. This seems to be paralleled by a different degree of acceptability of the analytical future of da´c si˛e in the personal and impersonal modal
The modal passive
177
passive (according to those informants that did accept them at all), cf. a modified example from the set found in IPI PAN: (278)
(Je´sli dzi´s zostanie podj˛eta decyzja o sprzeda˙zy cz˛es´ci hali), ju˙z a. ?jej nie b˛edzie dało si˛e she.GEN NEG AUX.3SG let.PST.3SG.N RFX odwróci´c. turn.around.INF. b. *ona nie b˛edzie dała si˛e she.NOM NEG AUX.3SG let.PST.3SG.F RFX odwróci´c. turn.around.INF. ‘(If the decision to sell part of the hall today is taken today,) it will not be possible to take it back (agreeing/non-agreeing MP) anymore’
The modified example (278-b) was deemed ungrammatical also by speakers who accepted (278-a).
Summary: morphological criteria It seems clear that morphological criteria, being the most entrenched, are the last to follow suit when an item moves from the perfective to the imperfective class; also, since da´c is a very frequent verb of the achievement class, the association of modal da´c si˛e and literal da´c should make morphological behavior of these verbs even less variable. Corpus evidence, however, shows that the frequency of analytical future da´c si˛e in subjectless constructions contrasts sharply with the frequency of analytical future forms of unequivocally perfective verbs and that they cannot be regarded as ‘mistakes’ of the same quality. This corroborates the finding that native informant judgment on these sentences differs greatly. The use of the analytical future of da´c si˛e in subjectless modal passive constructions must thus be considered possible in some ideolects; this points to a transitional aspectual status that might in the future lead to the general acceptability of this form (and perhaps other morphological forms restricted to imperfective aspect). This development seems to be restricted to the subjectless variants. Moreover, note that no instance of b˛edzie si˛e dawało with infinitive was found in a similar search, pointing to an expansion of the functions of da´c si˛e.
178 da(va)t+inf in Polish 3.6.4.2.
Contextual criteria
I now proceed to the analysis of dác si˛e in terms of contexts in which aspectual variation is restricted or triggers clear semantic contrasts. Such an analysis is complicated by the fact that in MP constructions, one is dealing with two aspect values: that of the matrix predicate and that of the complement infinitive. The assignment of aspect in complex or conjoined predicates is more complex than that in single predicates; it is usually not explicitly treated in theories devoted to aspect, and, it seems, less well understood. Aspect in such conjoined predicates is a heterogeneous phenomenon: while some structures, as for example those involving communication act verbs such as obieca´c ‘to promise’, clearly involve two events, both of which can be independently characterized in time and event internal aspectual characteristics, other structures, notably those involving modal auxiliaries such as móc ‘can’, trzeba ‘necessary’ and mo˙zna ‘possible’, less readily lend themselves to a subdivision into two events. Others still, such as the various passive constructions involving past participle and the verbs by´c ‘be’ and zosta´c ‘become’, show different configurations of aspect of auxiliary and complement infinitive that may be linked to assignment of aspect in simplex predicates to some extent, but show idiosyncratic behavior nevertheless (see V.Lehmann 2004,Wiemer 2004). A thorough treatment of this problem cannot be attempted here; in any case, for the issue of aspectual classification of da´c si˛e, surroundings where complement and matrix subject have distinct aspectual values are preferable as they are the most indicative.
Aspectual adverbials Only imperfective verbs combine with durative time adverbials; however, it will be shown that putatively perfective da´c si˛e combines with such adverbials at least in some cases. The procedure is as follows: In order to make sure that the adverbial test refers to the matrix predicate, rather than the embedded infinitive, two adverbial slots are introduced, one modifying matrix and one modyfying the infinitive. In the examples below, the second adverbial slot is filled by different adverbials, showing that only the imperfective complement infinitive is possible with a durative adverbial, as in (279), and only the perfective infinitive with an adverbial of inclusive time reference, as in (280). Having thus
The modal passive
179
secured variability of infinitive aspect, the test now relies on the first adverbial that pertains to the modal matrix predicate. Since it is a durative adverbial, it only combines with imperfective matrix predicates, as, in the following examples, is demonstrated with the less central modal uda´c / udawa´c ‘to manage’. Here, only the imperfective matrix verb is possible, independent of the variation in the complement: (279)
W naszej klinice przez dziesi˛ec´ lat *udałoPF /udawałoIPF si˛e *wyleczy´cPF /leczy´cIPF chorob˛e tylko tydzie´n, bo brakowało pieni˛edzy. ‘For ten years, in our hospital (we) managed (*pf/ipf) to treat (*pf/ipf) the illness for only one week for lack of funding.’
(280)
W naszej klinice przez dziesi˛ec´ lat *udałoPF /udawałoIPF si˛e wyleczy´cPF /*leczy´cIPF chorob˛e za tydzie´n. ‘For ten years, in our hospital (we) managed (*pf/ipf) to (successfuly) treat (pf/*ipf) the illness in a week.’
However, da´c si˛e does not behave like peripheral uda´c si˛e in this test. Rather, both da´c si˛e and dawa´c si˛e are possible in the same position: (281)
W naszej klinice przez dziesi˛ec´ lat dałoPF /dawałoIPF si˛e *wyleczy´cPF /leczy´cIPF chorob˛e tylko tydzie´n, bo brakowało pieni˛edzy. ‘For ten years, in our hospital it was possible (lit. let (pf/ipf) itself) to treat (*pf/ipf) the illness for only one week for lack of funding.’
(282)
W naszej klinice przez dziesi˛ec´ lat dałoPF /dawałoIPF si˛e wyleczy´cPF /*leczy´cIPF chorob˛e za tydzie´n. ‘For ten years, in our hospital it was possible (lit. let (pf/ipf) itself) to (successfuly) treat (pf/*ipf) the illness in a week.’
This test thus shows that in terms of the combination with durative time adverbials, da´c si˛e at least in some context behaves like an imperfective verb.
Present reference It can easily be shown that da´c si˛e often has nonlocalized statal reference akin to mo˙zna in present tense. This is not expected, since the morphological present, i.e., the non-past, of perfective verbs in Polish normally cannot refer to actual or non-localized present, except with an epistemic modal shading. No such shading nor future reference is found in contexts such as (283),
180 da(va)t+inf in Polish where both aspects are possible and putatively perfective da´c is actually the more usual variant to express the present tense: (283)
W tej chwili tutaj daPF:3S G /dajeIPF:3S G si˛eRFX spokojnie rozmawia´cINF . ‘Right now it’s possible (lit.: lets itself; pf/ipf) to speak in peace here.’
In contrast, the peripheral modal uda´c si˛e ‘to manage’, cannot be used in perfective aspect in this context; the imperfective form udawa´c si˛e has to be used: (284)
W tej chwili tutaj *udaPF:3S G /udajeIPF:3S G si˛e spokojnie rozmawia´cINF . ‘Right now it can be managed (*pf/ipf) to speak in peace here.’
This is an argument for imperfective da´c si˛e. In reference to similar observations regarding da(va)t in Czech and Bosnian/Croatian/Serbian, Dickey (2000: 194, 199f.) writes that such use “fl[ies] in the face of the conventional wisdom about aspect, which is that the [perfective aspect] is not used for the actual present”. However, this can be readily explained if we consider da´c si˛e to be of imperfective aspect. Combination with phasal verbs The probably most general and most robust test for imperfectivity in Slavic is the combination with phasal verbs: only imperfective verbs combine with phasal verbs. This text fails for our purposes: there is a clear difference in acceptability between imperfective dawa´c si˛e and perfective da´c si˛e, with the latter excluded: (285)
Kiedy w reszcie zacznie3S G si˛e dawa´cIPF /*da´cPF z˙ y´cINF w tym kraju? ‘When at long last will it begin to be possible (let itself; ipf/*pf) to live in this country?’
No corpus examples concerning the combination of phasal verbs with da´c si˛e could be found. Subordinated clauses A further test for both perfectivity and imperfectivity are subordinate clauses introduced by dopóki ‘while, until’ and a˙z ‘until’ that show concurrent or
The modal passive
181
consequent temporal relationship to the main clause, and therefore enforce a specific aspect. Without negation, sentences introduced by dopóki allow only imperfective verbs and denote a concurrent state-of-affairs: (286)
Rozmawiali´smyip f , dopóki siadali´smyIPF / *siedli´smyPF . ‘We talked while sitting down (ipf/*pf).’ (concurrent)
Under negation, or if introduced by a˙z, such sentences denote a consequent event and only take perfective verbs in past and present tense or the analytical ´ future of imperfective verbs. (see Kudli´nska-St˛epie´n 1995, Smiech 1971: 75). Imperfective verbs may only be used in an iterative reading (# is used here and below to signal a shift to a different reading, in this case, to an iterative one): (287)
Wczoraj rozmawiali´smyip f , a˙z #siadali´smyip f /siedli´smy p f . ‘Yesterday, we talked until we sat down(#ipf/pf).’ (consequent)
(288)
Zazwyczaj rozmawiali´smyip f , a˙z siadali´smyip f /*siedli´smy p f . ‘Usually, we talked until we sat down(ipf/*pf).’ (consequent, iterative)
The semi-modal udawa´c si˛e ‘manage to’ again behaves in accordance to the above regularities: in the concurrent case (289), only the imperfective form is possible; in the non-iterative consequent reading, only the perfective form may be used (290). A consequent reading in an iterative context only admits of the imperfective form (291). (289)
Uspokajali´smy ich, dopóki udawałoIPF / *udało si˛e spokojnie rozmawia´c. (concurrent) We kept reassuring them while we managed (ipf/*pf) to speak in peace.
(290)
Wczoraj uspokajali´smy ich, a˙z #udawałoIPF / udałoPF si˛e spokojnie rozmawia´cINF . (consequent) Yesterday we kept reassuring them until we managed (#ipf/pf) to speak in peace.
(291)
Zazwyczaj uspokajali´smy ich, a˙z udawałoIPF / *udałoPF si˛e spokojnie rozmawia´cINF . (consequent, iterative) Usually we kept reassuring them until we managed (ipf/*pf) to speak in peace.
Again, da´c si˛e behaves differently. In contrast to uda´c si˛e, it can replace its imperfective aspectual partner with dopóki:
182 da(va)t+inf in Polish (292)
Uspokajali´smy ich, dopóki dawałoIPF / dałoPF si˛e spokojnie rozmawia´cINF . (concurrent) We kept reassuring them while we could (ipf/pf) speak in peace.
With a˙z, analogous to what we find with uda´c/udawa´c si˛e, the imperfective dawa´c si˛e receives an iterative reading, leading to a semantic contrast with da´c si˛e, which refers to a specific situation: (293)
Uspokajali´smy ich, a˙z dawałoIPF / dałoPF rozmawia´cINF . (concurrent)
(294)
Wczoraj uspokajali´smy ich, a˙z #dawałoIPF / udałoPF si˛e spokojnie rozmawia´cINF . (consequent) Yesterday we kept reassuring them until we could (#ipf/pf) speak in peace.
(295)
Zazwyczaj uspokajali´smy ich, a˙z dawałoIPF / ?dałoPF si˛e spokojnie rozmawia´cINF . (consequent, iterative) Usually we kept reassuring them until we managed (ipf/?pf) to speak in peace.
si˛e spokojnie
The crucial point for my argumentation here is that dało si˛e in (293) is used in a context where perfective verbs are in generally not possible. However, as the question mark in (295) indicates, the contrast of dawa´c si˛e and da´c si˛e is not exhaustively analyzed here.
Summary Tests for the interchangability of subjectless modal passive da´c and dawa´c si˛e in contexts that demand a particular aspectual value show that we can replace dawa´c with da´c in several contexts that only permit imperfective verbs. However, this is not possible in all contexts, notably not with phasal verbs. On the other hand, da´c si˛e appears, and cannot be replaced with dawa´c, in at least one context (see (294) above) that only allows perfective verbs; this suggests that da´c si˛e also has a perfective use.
3.6.4.3.
Conclusions concerning the aspect of da´c si˛e
Morphological and contextual criteria show that da´c si˛e in subjectless modal passive constructions exhibits characteristics of an imperfective verb: mor-
The modal passive
183
phologically, this concerns the use of the analytical future, which is attested much more frequently than with unequivocally perfective verbs, where this should be considered a mistake. However, this is not acceptable to all speakers and it is found very rarely. Tests that employ contexts that are incompatible with perfective verbs give mixed results. Da´c si˛e may be used with durative aspectual adverbials and in non-negated subordinate clauses introduced by dopóki; it may refer to a present state. However, it does not combine, it seems, with phasal verbs. While it thus seems certain that da´c si˛e is not only perfective in subjectless modal passive usage, questions remain as to its exact characterization: is it imperfective, biaspectual or ’aspectless’, that is, outside the aspectual system, as core modals may be argued to be (Björn Hansen, p.c.)? This question requires further study. Another caveat remaining is the exhaustive investigation of imperfective dawa´c si˛e in this context, and to what extent da´c si˛e also exhibits similar tendencies in permissive as well as personal modal passive use. However, there are clear indications, such as the high proportion of perfective da´c with subjectless, but not with personal modal passives and contrasts in acceptability concerning the use of the analytical future quoted above, that suggest that subjectless modal passives are more advanced in this respect. I conclude that with subjectless modal passive constructions, da´c si˛e has a transitional status in that it has lost the aspectual distinction to an advanced degree.
184 da(va)t+inf in Polish 3.6.5.
Conclusions: different degrees of grammaticalization
The section on modal passives started with a morphosynactic categorization of modal passives based on clearly definable syntactic criteria. I distinguished five types: agreeing modal passives that have an overt subject, one ambiguous type concerning unclear cases and three more subjectless constructions: with verbs that take a direct object, no or oblique object, or a sentential complement. In the discussion it emerged that from a more detailed perspective these five types fall into two groups: the agreeing modal passive and three more grammaticalized subjectless modal passive constructions. In the discussion I have shown that the agreeing modal passive should be considered a further grammaticalization of the permissive reflexive. Semantically, it can be derived from the permissive via semantic bleaching, loosening of selectional restrictions and, perhaps, metaphoric extension: instead of an animate subject that willfully allows something to be done to him, making it thereby possible, the subject denotes an inanimate patient whose characteristics are in some sense responsible for the possibility expressed by the whole construction. This is a typical grammaticalization phenomenon whereby some part of meaning that is present in one configuration is extended to be the main part in another, new configuration. The latter group arguably includes instances of a single, more grammaticalized subjectless modal passive construction involving a grammaticalizing and at the same time lexicalized formant da´c si˛e. They share several traits setting them apart from the agreeing modal passive: • they lack a subject • they denote unrestricted dynamic modality of possibility • they are mostly used with negation • they share a reducing paradigm of da(va)t converging in da, dało, da´c • in these constructions, da´c si˛e is no longer unequivocally perfective Abstract semantics, a small paradigm and lacking of verbal properties such as aspect are as such uncontroversial characteristics of grammaticalization. Moreover, seen as one rather than as a set of constructions, it can be said to be more grammaticalized than the agreeing construction since it imposes less selectional restrictions on the infinitive and its participants.
Summary
185
I conclude that these two constructions constitute a case of divergence (Hopper & Traugott 1993: 118): the older, agreeing MP construction is retained, and the newer, more grammaticalized subjectless construction has evolved from it with less specific meaning, less variance in form, a greater class of verbs it combines with (notably irrespective of valency) and much greater frequency. We will see in section 6 that diachronic data confirm this hypothesis: the expansion of the agreeing MP to the second, more grammaticalized subjectless type occurred relatively abruptly: non-agreeing, intransitive and sentential complement MPs all evolved at around the same time, while the agreeing modal passive is much older.
3.7.
Summary
I will now summarize the main points discussed concerning da(va)t+inf in Polish. As in Russian, da(va)t+inf in Polish denotes primarily permissive causation. However, in contrast to Russian, the reflexive permissive has grammaticalized further: it is much more frequent than the non-reflexive and differs from it in terms of selectional restrictions, meaning and syntax, notably in the use of a przez-phrase expressing the causee, likening it to the passive. Conversely, no grammaticalization of the imperative as in Russian is notable for Polish. Again in contrast to Russian, da(va)t+inf in Polish may generally denote both permissive and factitive causation of the curative sub type; however, the permissive function clearly predominates. The most frequent and functionally most important structure involving da(va)t+inf is the modal passive construction. It is a continuation of the reflexive permissive appearing in two variants: first, the personal modal passive with overt subject, formed from transitive verbs and restricted in meaning; second, a subjectless variant formed with all agentive and experiencer verbs. The subjectless variant is more grammaticalized and involves the development of an arguably biaspectual modal da´c si˛e, not described in the literature yet. We will return to the issue of how these two variants relate to each other in more detail in the diachronic and comparative part.
Chapter 4 da(va)t+inf in Czech 4.1. 4.1.1.
Introduction Overview and prior research
As in Polish, da(va)t+inf has two grammaticalized in two functions in Czech: as a causative and as a modal passive. In relation to the causative function of da(va)t, Czech is perhaps the most complicated to describe among the languages covered in this study due to the heterogeneity of the relevant constructions. Two fully productive constructions may be singled out: a reflexive permissive and a curative construction that both take only type II causees, a peripheral option in Polish as we have seen above. However, dative causees are preserved in several minor types of restricted productivity, most notably with cognitive and perception verbs and in the non-reflexive permissive, which has gone largely out of use. This is to be seen against the background of the competition of another formant, nech(áv)at, that may be used in almost all former and present productive functions of causative dá(va)t. This competition has been an issue in Czech prescriptive linguistics, since the extended uses of nech(áv)at have been perceived as resulting from German influence (e.g., Brus 1894: 205f.; see references in Toops 1992: 40). Some of these issues have received attention in the literature, most notably by Toops (1992) in a careful analysis of the functions of dá(va)t as opposed to nech(áv)at. Skoumalová (2003) and Medová (2006), both primarily interested in the analysis of the reflexive marker, have examined a subset of the constructions involving da(va)t+inf, namely the modal passive and the reflexive permissive construction. Generally speaking, the use of da(va)t+inf in the causative domain is functionally least important among its cognates in the languages considered, as is evident from its frequency in the monolingual as well as in the parallel corpus. The modal passive, on the other hand, is more grammaticalized than in Polish. To my knowledge this construction has not found much attention in the literature, with the exception of Rytel (1982: 152), who notes that is is
188 da(va)t+inf in Czech much more common in Czech than in Polish, where it is, in her view, confined to spoken language. As I will show, higher frequency correlates with other parameters indicating the grammaticalization of this construction.
4.1.2.
Competition in causation: nech(áv)at with infinitive
As the analysis of the development and use of dá(va)t needs to at least superficially look at the functional competition with the analogous construction with nech(áv)at, a short overview of this formant is in order. Nech(áv)at denotes both permissive and factitive causation of various subtypes. Besides uses with nominal complements, which will not be of concern here, it is used as a type I and a type II causative, much like German lassen, with which it shares its original, lexical meaning ‘let, leave, discharge’. Below, I illustrate the contrast between type I and type II causee in an ambiguous type I and a factitive type II causative; the contrast between the two may be, it seems, captured in the terms outlined in the introduction. • Type I causative accusative causee (296)
Nechali hoi podepsati ten dopis. nechat.PST.3PL he.ACC sign.INF that letter.ACC ‘They let/had him sign the letter.’
• Type II causative with instrumental/prepositional causee: (297)
Nechali podepsat ten dopis (jím / od nˇeho). nechat.PST.3PL sign.INF that letter.ACC he.INS / by him.ACC ‘They had the letter signed (by him).’
Nech(áv)at does not generally form the modal passive, as dát does, although a similar function seems to be present in some dialects. In contrast to dá(va)t, it takes an accusative, not a dative causee in the type I construction (see below). Furthermore, it has much wider usage and, as a causative, it must be seen as the generally more central auxiliary. See Grepl & Karlík (1983); Toops (1992); Panevová (2003) for more details. In a detailed comparison of the two formants, Toops (1992: 48) suggests that in those environments where both matrix verbs may be used and “both factitive and permissive semantics are equally conceivable [. . . ] dá(va)t is ‘more factitive’ and nech(áv)at ‘more permissive’.” The exact conditions for individual readings will not concern
Introduction
189
us here and a full comparison of these formants is beyond the scope of this study.
4.1.3.
Syntactic types
The syntactic structures involving da(va)t in Czech divide into the following main types, analogous to those in Polish: • Type I causative with dative causee in a dative control construction: (298)
... chtˇel oponovat Watson, ale Holmes mui ... want.PST.3SG oppose.INF Watson, but Holmes him.DAT nedal domluviti [.] NEG.let.PST.3SG finish.speaking.INF ‘. . . Watson wanted to say, but Homes did not let him finish speaki ing.’
• Type II causative with causee either in a prepositional phrase or in instrumental case: (299)
Dal jsem pokoj vymalovat (malíˇrem / Have.PST.SG AUX.1SG room.ACC paint.INF painter.INS by od malíˇre). painter.GEN ‘I had the room painted (by a painter).’
(300)
Dal se pˇresvˇedˇcit (sestrou / od Let.PST.3SG RFX.ACC persuade.INF sister.INS by sister.GEN sestry). ‘He let himself be persuaded (by his sister).’
• Modal passive constructions (with subtypes) (301)
Tomu se nedá rozumˇet. That.DAT.SG RFX.ACC NEG.let.3SG understand.INF ‘That cannot be understood’
i Cechura, ˇ
ˇ Rudolf, Cítanka pro zaˇcínající detektivy, 2003
190 da(va)t+inf in Czech In addition to these types, occasional use of an accusative causee analogous to nech(áv)at is found. In any case, however, the type II construction, formed with either an instrumental or a prepositional by-phrase is is the main type in Czech. This is in clear contrast to Polish, where this type of causee is peripheral. There is one further infinitive construction specific to Czech to be noted here that will be analyzed as an instance of literal give below (example from Skoumalová 2003): (302)
Zuza dá Aniˇcku spát do kolíbky. Zuza gives.3SG Aniˇcka.ACC sleep.INF to cradle ‘Zuza puts Aniˇcka into the cradle to sleep’
4.1.4.
Data & annotation
For Czech, the Syn2005 part of the Czech National Corpus was used. This is a 100 million token corpus consisting approximately of 40% fine literature, of 27% technical literature and of 33% journalistic texts. Journalistic texts date from the year 2000 onwards, while technical literature goes back to 1990. The fine literature section also includes translated and older texts; the Web site merely states that care was taken to include as few older texts as possible, but no exact statistics are made available. Unfortunately, only the date of publication, not of first publication or of date of writing, is made available in the metadata. The corpus was queried for forms of dát or dávat followed within a distance of 4 tokens by an infinitive, with no intervening punctuation allowed. The query string [lemma="dáv?a?t"][tag!="^.:.*"]{0,4}[tag="^Vf..*"] yielded the immense number of 44,768 results. All these results were converted for input into the data base, with matrix and complement verb tags preserved, and the presence of reflexive particles and negation prefix separately stored. This number was further reduced to 26,718 by excluding all examples from translated text. Of these, 1000 cases were randomly chosen and systematically annotated, representing 4% of the full corpus. In some cases additional analyses were conducted on all examples in the data base; these are referred to as full corpus searches49 .
Introduction
191
Annotation The characteristics distinguished during annotation are similar to what was annotated in Polish: • modal passive, with several syntactic subtypes similar to the ones in Polish (see below section 4.3): (303)
PenízeNOM:PL seRFX:ACC nedajíNEG:3PL zkrmitINF . ‘Money cannot be used as food (lit. doesn’t let itself feed)’
i
• permissive (304), factitive (305), or ambiguous causative (304)
[. . . ] chtˇel odkázat [. . . ], ale dalPS T :3S G seRFX:ACC ii pˇremluvitINF [. . . ] ‘... he wanted to refuse, but let himself be talked into it’
(305)
Jisté je, že seRFX:ACC dalPS T :3S G sám rozvéstINF a pozdˇeji se s iii Janou oženil. For certain is that he divorced (lit. had2 himself 1 divorced3 ) and later married Jana.
with further annotation concerning – semantics of the causative relation – animacy of causer and causee; clearly inagentive verbs were noted in comments – presence, type and function of reflexive pronoun / element – form of causee: unexpressed, dative, instrumental or prepositional phrase • da(va)t+inf with cognitive verbs, such as vˇedˇet ‘know’, tušit ‘notice’, slyšet ‘hear’, etc.: (306)
Kdy bude pohˇreb, tedy v kterou hodinu, dám1S G vámDAT iv vˇedˇetINF , i když se nás netýká. ‘When the funeral takes place, at what hour, I’ll let you know, even if it does not concern us.’
i Verner, Pavel, Pražské hyeny, 1994 ii Mladá fronta DNES, 6. 2. 2002 Jiˇrina, Ohlédnutí, 1994 iv Mertl, Vˇeroslav, Hˇrbitov sn˚u, 2001
iii Zábranová,
192 da(va)t+inf in Czech • phraseologic or lexicalized types: (307)
v Jaroslav Vˇetviˇcka na své vesty nedáNEG:3S G dopustitINF . ‘Jaroslav V. won’t let people say anything bad about his vests’, lit. an opaque combination of dát with dopustit ‘allow, send’; cf. Germ. ‘nichts auf etwas kommen lassen’
• Give in the literal sense with infinitive as secondary predicate or with consumptive verb (308)
i
Dává3S G miDAT jístINF - jsem jeho pes. ‘He gives me to eat - I am his dog.’
• A give plus directional complement construction meaning put: (309)
Když se vejce zaˇcnou srážet, dáme1PL omeletuACC do rozpálené ii trouby zapéciINF . When the eggs begin to coagulate, we put the omelet into the hot oven to bake.
• False hits including infinitives that are not dependent on da(va)t An overview of the frequency of these types in the corpus sample is given in table 31. As we can see, modal passive is by far the most frequent type in the sample. Permissive and factitive are of similar frequency, both representing less than 10% of the sample. This section is organized differently from the corresponding sections on Polish and Russian. In Czech, we have two main productive causative types: the reflexive accusative permissive and the factitive curative type. Both take only type II causees, and in most cases, the causee is not expressed. Besides that, we have a continuum of archaic, phraseologic and more or less productive types, including cognitive verbs. The status of all these cases are treated jointly in the following subsection 4.2. After that, modal passive constructions are considered in section 4.3, followed by an analysis of residual annotated types and conclusions. v Blesk, 14. 8. 2004 labužníky, 1998
i Ikarie
- povídky, roˇcník 1999
ii Cuhrová,
Anna, Kuchaˇrka pro
Causative
type modal passive factitive permissive ambiguous perm/fact. cognitive verbs give plus DIR phraseologic transfer plus infinitive false hits
abs 654 92 63 12 37 16 34 23 69 1000
193
rel 65% 9% 6% 1% 4% 2% 3% 2% 7%
Table 31. Broad overview of types in the 1000 random sample of syn2005. Modal passive is clearly the most frequent type, followed by factitive and permissive causative and more or less lexicalized formations with cognitive and perception verbs.
4.2. 4.2.1.
Causative Overview
Generally, the number of examples in the sample relating to causatives is, compared to what has been considered in the other languages, rather low; however, it would not be practical to acquire more data by full annotation of more random examples. To see that, consider that the extraction and annotation of 1,000 more examples would presumably yield fewer than 150 more examples of factitive and permissive usage together. Instead, specific questions are dealt with by individual corpus queries. In contrast to the other languages, this also involves the comparison with another formant, nech(áv)at, in order to assess the productivity of dá(va)t in individual functions.
4.2.2.
Non-reflexive permissive
The productive non-reflexive permissive use of da(va)t has all but disappeared in modern Czech. The corpus sample contained 5 examples from the interpersonal domain, three of which contained variations on a well-known line from the Svatováclavský chorál dating from the 12th century:
194 da(va)t+inf in Czech NedejNEG:I MP:2S G zahynoutiINF námDAT niNEG budoucímDAT , i svatý Václave! ‘Don’t let1 us3 nor4 the coming ones5 perish2 , Saint Vaclav!’
(310)
The other two examples were: (311)
"Neˇrekl bych, že Barnes je . . . ," chtˇel oponovat Watson, ale Holmes i muDAT nedalNEG:PS T :3S G domluvitINF : ‘"I wouldn’t say that Barnes ..." Watson wanted to object, but Holmed didn’t let2 him1 finish:3 ’
(312)
Pˇredstavíme vám nˇekolik muž˚u, které nedajíNEG:3PL fanynkámDAT ii spátINF a zkusíme naznaˇcit, v cˇ em TO je. ‘We’ll show you some men that keep their female fans awake (lit. don’t let their fans sleep) and we’ll try to find out what’s behind THAT.’
Four more examples were found from the impersonal domain, three containing dát spát ‘let sleep’ as in the last example and (313) below. Note that in the last example (312), technically an example of interpersonal causation, the sentence subject does not act agentively and metonymically refers to the thought of them that does not let the fans sleep; in this sense, there is no difference to the following, impersonal example. To je otázka, která nedáNEG:3S G spátINF Ameriˇcanum ˚ DAT i iii Iráˇcanum ˚ DAT . ‘That is an issue that does not let1 the Americans3 and the Iraqis4 sleep2 (i.e., keeps them awake)’
(313)
The fourth example involved vydechnout in the sense of take a rest: Poˇcasí nedaloNEG:PS T :3S G silniˇcáˇrum ˚ DAT vydechnoutINF ani v upiv lynulých dnech. ‘The weather did not let the road workers get some rest in the past days, either.’
(314)
All these cases involve complement verbs whose equivalents are very frequent with da(va)t+inf in Polish and Russian. My informants did not generally accept nedat domluvit as in (311) – those who did accept it regarded it as archaic. Further searches for typical complement verbs in Polish and Russian confirmed that some of them are also used with Czech dát, but much i Cechura, ˇ ˇ Týden, cˇ . 41/2002 Rudolf, Cítanka pro zaˇcínající detektivy, 2003 cˇ . 23/2000 iii Blesk, 17. 2. 2004 iv Hospodáˇrské noviny, 31. 12. 2001
i E.g., ii Story,
Causative
195
less frequently. This involves verbs such as vy-/oddechnout ‘take a rest’, s/odpoˇcinout ‘take a rest’, do-/promluvit ‘speak’ as well as žít ‘live’ and spát ‘sleep’. Excluding nedat spát ‘keep from sleeping’, which was found several dozen times, these other verbs appeared in the full corpus fewer than a twenty times altogether. Typically, they were negated: (315)
Pocit viny nedá3S G žítINF otcum ˚ DAT . ‘Lit: The feeling of guilt does not let1 the fathers3 live2 .’
i
(316)
NedalaNEG:PS T :3S G muDAT ani na chvíli oddechnoutINF . ‘She didn’t let him take a break, not even for a short while.’
ii
Nedát spát ‘keep from sleeping’ was considered a fixed phrase by my informants, and a full corpus search confirmed that is used much more often than the other combinations, and even slightly more often than nech(áv)at spát. In contrast to the latter, it mostly appeared under negation and with inanimate causer50 . Positive letting somebody sleep in the sense of refraining from waking somebody up is constructed with nech(áv)at, however, and is infelicitous with dá(va)t. We see that permissive, non-reflexive da(va)t with dative causee has a marginal status. The only case with dative causee found from the domain of manipulative causation is given in (317). Such use is clearly archaic; the example is from a text by Vladislav Vanˇcura (1891-1942): (317)
Pilipaninec [. . . ] dalPS T :3S G dopadnoutiINF své dlaniDAT na raiii meno nespokojence[.] ‘Pilipaninec put his hand (lit.: let1 his hand3 fall2 ) on the shoulder of the grumbler.’
Surprisingly, one use with accusative causee was also found, in the manipulative domain (318). This is analogous to a much more frequent combination with nech(áv)at, cf. (319): (318)
Zato vy jste siRFX:DAT dalPS T :3S G narust ˚ INF delšíACC nehetACC , iv Sermoyi! ‘Why did you grow (lit. let2 yourself 1 grow3 ) a long nail, Sermoyi!’
(319)
Také siRFX:DAT nechalaPS T :3S G narust ˚ INF dlouhé vlasy. ‘She also grew (lit. let2 herself 1 grow3 ) long hair’
v
i Zgustová, Monika, Peppermint frappé, 2002 ii Efeméra - erotická iii Vanˇ cura, Vladislav, Amazonský proud, Pekaˇr Jan Marhoul, Pole orná a váleˇcná, Poslední soud, 2000 iv Fischerová, Daniela, Hodina mezi psem a vlkem, 1989 v Ikarie - povídky, roˇ cník 1999
196 da(va)t+inf in Czech A corpus search yielded 4 more example with accusative causee and nar˚ust; 150 cases of nech(áv)at nar˚ust were found. This pattern seems to be evidence of a contamination of nechát and dát. I conclude that the non-reflexive permissive is marginal in Czech today; attested examples represent cases of the most common usages that da(va)t has in the other languages and may be regarded as fossilized or archaic. Dá(va)t has been almost completely replaced by nech(áv)at in this function; correspondingly, nech(áv)at is the single most frequent translation of non-reflexive Polish and Russian da(va)t in the parallel corpus (see chapter 5). 4.2.3.
Reflexive permissive
If in Polish the reflexive permissive is predominant and outnumbers nonreflexive permissives by a factor of 5 (see section 3.2.1 from page 119), in Czech, it is the only productive use of permissive da(va)t. This productivity is most clearly given with the accusative reflexive; with dative reflexives, lexicalized combinations predominate and with other forms of the reflexive pronoun, the construction is hardly attested. In all these cases, including lexicalized types, the causee can only take the form of a by-phrase, that is, of a phrase in instrumental case or of a prepositional phrase introduced by od ‘by’ (literally and originally: ‘from’). In this, the Czech construction clearly departs from the Polish one, where the byphrase is merely a marginal option and dative causees dominate. In this section, I first cover the most frequent type, accusative reflexives and then examine dative and instrumental reflexives. Finally, I compare the use of dá(va)t and nech(áv)at with these types.
4.2.3.1.
Accusative reflexive
39 cases of the permissive with an accusative reflexive were found. The list of complement verbs is given in list 15. The list suggests that da(va)t is used with a range of verbs that reflexive permissives in Polish and Russian are also typically formed with, that is, with verbs that denote some (often negative) influence or action directed at the causer that the latter does or does not succumb to. It should be remarked that other complement verbs such as zaskoˇcit ‘surprise’, whose cognates are frequent in Polish, but absent from this
Causative
197
n=3: pˇresvˇedˇcit ‘persuade’; n=2: chytit ‘catch’, odbýt ‘dismiss’, vylouˇcit ‘exclude’, vyrušit ‘disturb’, vyvést ‘bring out’, zabít ‘kill’, zastrašit ‘scare’, zlákat ‘tempt’; n:1: napovídat ‘dictate, say’, odradit ‘dissuade’, okrádat ‘steal from’, okrást ‘steal from’, omývat ‘wash’, ošálit ‘make crazy’, pobízet ‘urge’, podfouknout ‘cheat’, podvádˇet ‘cheat’, pomáhat ‘help’, potˇešit ‘comfort’, pouˇcovat ‘teach’, pˇremluvit ‘persuade’, probodnout ‘stab’, strhnout ‘carry away’, ukˇrižovat ‘crucify’, urážet ‘insult’, usmíˇrit ‘placate’, zahanbit ‘shame’, zavˇrít ‘lock up’, zmást ‘confuse’ Verb list 15: Verbs used with the accusative reflexive permissive sample, are also found in Czech in this construction in the full corpus. This suggests that this construction is productive; I will provide further evidence for this assessment shortly. The following examples illustrate causee-less use, use with both inanimate and animate causee given in instrumental case or with the preposition od: (320)
Rosenberg seRFX:ACC však nedalNEG:PS T :3S G odraditINF a zaˇcal i rozvíjet svou pˇredstavu uspoˇrádání správy obsazených území. ‘Rosenberg, however, did not let2 himself 1 be dissuaded3 and began to develop his conception of how to manage the occupied territories.’
(321)
Ale ne a ne seRFX:ACC dátINF pˇresvˇedˇcitINF , rˇval, že seRFX:ACC nedáNEG:3S G odPREP nˇejaké uˇcitelkyGEN podvádˇetINF , až jsme ii složily bedny z vozu [. . . ]. ‘But he just wouldn’t let2 himself 1 be persuaded3 , he raged on that he would not let5 himself 4 be fooled8 by6 some teacher7 , until we took the poor guy out of the cart [...]’
(322)
Ženy seRFX:ACC nedalyNEG:3PL vyrušitINF našíINS pˇrítomnostíINS iii a pokraˇcovaly ve svém polosnˇení. ‘The women did not let2 themselves1 be moved3 by our4 presence5 and went on in their trance.’
(323)
Vláda bude dál dˇelat to, co má, a nedáNEG−3S G seRFX:ACC nikýmINS iv a niˇcímINS zastrašitINF . ‘Vláda will continue to do what she has to do, and will not let1 herself 2 be scared5 by anybody4 or anything5 .’
A dative causee is in all these cases impossible, e.g. for (323): i Tejchman, iii Zeman,
Miroslav, Ve službách Tˇretí ˇríše, 1999 Jiˇrí, Afrika, 2001 iv Týden, cˇ . 37/2002
ii Zábranová,
Jiˇrina, Ohlédnutí, 1994
198 da(va)t+inf in Czech matrix asp. form dávat pres. dát inf. imp. past non-past rel.:
abs 1 5 1 18 14 39
polarity pos. neg. 0 1 3 2 0 1 8 10 5 9 16 23 41% 59%
infinitive pf ipf 0 1 5 0 1 0 15 3 11 3 32 7 82% 18%
causee zero type II 0 1 4 1 0 1 14 4 9 5 27 12 69% 31%
Table 32. Accusative reflexive permissive: occurrence in the annotated sample according to form of the matrix verb, negation, infinitive aspect and type and occurrence of causee.
(323)’ Vláda nedáNEG:3S G seRFX:ACC *nikomuDAT a *niˇcemuDAT zastrašitINF . Intended: ‘Vláda will not let1 herself 2 be scared5 by anybody4 or anything5 .’ The profile of the accusative reflexive permissive as used in the corpus is given in table 32. Both animate (n=8) and inanimate causees (n=4) are attested. There is no clear tendency in regard to polarity (negative polarity seems to dominate in this sample, but the difference to an random distribution is not significant at p=.45). The same applies for tense: the difference in frequency in the past and in the non-past is not statistically significant. Like in Polish, the imperative is not used extraordinarily often. As may be expected on the basis of the Polish and Russian data, the construction is often, but not always used without causee; in fact, the difference to an even distribution is not significant (p=.6). As with the permissive in Russian and Polish, complement aspect and matrix aspect is mostly perfective. The patterns found here are thus, by and large, no surprise; the most marked difference to the other languages is the much lower frequency of this construction (as reflected also in the parallel corpus, see section 5), and the impossibility of using a dative causee. The former characteristic is clearly linked to the use of the competing formant nech(áv)at which will be examined below; the latter I interpret as a sign of the grammaticalization of this construction as will be discussed in more depth below. Trivially, neither complement negation (*dal se nepˇresvˇedˇcit ‘let himself not be persuaded’) nor different temporal adverbials concerning matrix and
Causative
199
complement (*vˇcera dal se pˇresvˇedˇcit zítra ‘yesterday he let himself be tomorrow persuaded’) may be introduced. The caused event is implied to have taken place: (324)
Dal se pˇresvˇedˇcit, *ale nebyl pˇresvˇedˇcen. Let.PST.3SG RFX persuade.INF but NEG.3SG persuade.PPP Intended: ‘He let himself be persuaded, but wasn’t persuaded.’
As in Russian and Polish, this construction thus expresses a single event.
Additional evidence for productivity For a reliable assessment of the productivity of this construction the number of attestations in the sample is very low. Therefore, use of those complement verbs in the sample that were attested in reflexive permissive function was investigated in the full corpus. Individual instances were only cursorily inspected; the obtained figures thus also include possible modal passive uses of these infinitives51 . In order to minimize this distorting effect, verbs that also appeared in modal passive function in the corpus sample were left out; this concerned only vylouˇcit. Only cases where se was present any place before the infinitive were taken into account, and frequent light verb constructions were filtered out (see above section 4.1.4).
n=52: odradit ‘dissuade’; n=45: odbýt ‘dismiss’; n=42: zlákat ‘tempt’; n=41: pˇresvˇedˇcit ‘persuade’; n=40: chytit ‘catch’, pˇremluvit ‘persuade’; n=32: zavˇrít ‘lock up’; n=27: strhnout ‘carry away’; n=19: zmást ‘confuse’, vyvést ‘bring out’, zabít ‘kill’; n=17: pobízet ‘urge’; n=16: zastrašit ‘scare’, vyrušit ‘bring out’; n=11: zahanbit ‘shame’; n=8: ošálit ‘make crazy’; n=4: pomáhat ‘help’, omývat ‘help’; n=3: urážet ‘insult’, okrást ‘steal from’, pouˇcovat ‘teach’; n=2: okrádat ‘steal from’, podvádˇet ‘cheat’, ukˇrižovat ‘crucify’; n=1: podfouknout ‘cheat’, probodnout ‘stab’, usmíˇrit ‘placate’, napovídat ‘dictate, say’, potˇešit ‘comfort’ Verb list 16: Verbs used in reflexive permissive function in the sample and their frequency with da(va)t in the full corpus.
200 da(va)t+inf in Czech List 16 presents the results on a per-verb basis. It shows that many of the verbs found in the reflexive permissive construction in the sample were also found in this construction in the rest of the corpus. Crucially, five verbs were found only in the sample. Since the proportion of hapax legomena types can be used as a measure of productivity (see Baayen 2002), the occurrence of hapax legomena verbs in the rather small sample already provides some evidence for productivity; the expansion of scope to the full corpus makes this claim stronger. The random sample included only about 4% of the full corpus; using a maximum likelihood estimation the number of hapax legomena verbs in the full corpus is estimated to lie at around 25 times as high, that is, at around 100 as a lower bound. These findings clearly advocate against a lexical restriction of productivity of reflexive permissive dát se. The distribution of forms of da(va)t with these complement verbs in the full corpus (not shown here) also confirmed that imperfective matrix aspect is extremely seldom, namely, used in 10 of the 463 attestions considered. This is markedly different from what we find in Polish and Russian. This seems to be linked to a special aspectual status of da(va)t, which we will turn to now. Verbal aspect of dá(va)t In the standard dictionaries, dát in reflexive permissive function is listed as biaspectual, just as in the modal passive. It is unclear on what evidence the assessment of a biaspectual status of permissive dát se is based. While in the modal passive, the imperfective dávat is ungrammatical, in the reflexive permissive it is attested, if very sporadically, e.g. in (325): (325)
Nikdy nepˇredstíral vševˇedoucnost a rád seRFX:ACC dávalIPF:PS T :3S G i pouˇcovatINF odby kteréhokoliGEN ze spolupracovník˚u. ‘He never feigned omniscience and happily let2 himself 1 be taught3 by4 any5 of the co-workers.’
Moreover, in some contexts, only dávat, not dát, is possible; cf. the next example with an durative temporal adverbial: (326)
Po 10 let seRFX dávalIPF:PS T :3S G /*dálPF:PS T :3S G podvádˇetINF svou ženou. 10 years he let (ipf/*pf)2 himself 1 be cheated3 by his wife.
i Maˇratka,
Hlaváˇcková: Josef Pelnáˇr. 2000
Causative
201
n=22: vymluvit ‘talk out of’; n=12: vzít ‘take’; n=5: namluvit ‘talk into’; n=4: rozmluvit ‘dissuade’; n=2: poruˇcit ‘command’; n=1: vymluvit ‘dissuade’, vnucovat ‘impose’, porouˇcet ‘command’, napovídat ‘dictate’, odejmout ‘take away’ Verb list 17: Verbs found with dative reflexive permissive in the full corpus. Furthermore, the analytical future, normally formed only from imperfective verbs (see 3:3.6.4.1 above), was in the full corpus found with dát only in modal passive function, not in reflexive permissive function. Accordingly, informants consistently did not accept examples involving the analytical future of dát in permissive function, such as the following: (327)
(*BudeAUX:3S G se dátINF / dá3S G se) nˇekým pˇresvˇedˇcitINF . ‘He will let (*analytical future/non-past) himself be persuaded by someone.’
I conclude that imperfective dávat is infrequent in the reflexive permissive, but not obsolete; it has functions that dát does not fulfill. While the low frequency of dávat in this construction is plausible under the assumption that dát is biaspectual, I cannot give any clear qualitative evidence supporting such a view. Furthermore, it should be noted that nech(áv)at is similar in that its imperfective form seems to be very rarely used, at least in the configurations investigated below. Interestingly, no dictionary states that nechat se is biaspectual. I tentatively conclude that dát se is not biaspectual in permissive function, and leave the question for further research.
4.2.3.2.
Dative and other reflexives
Dative reflexive are less often used, and most of the combinations of permissive da(va)t+inf and si attested in the sample should be considered lexicalizations. Two clearly compositional types were found, involving the verbs poruˇcit/porouˇcet ‘command’ as well as napovídat ‘dictate, prompt’, e.g. (328)
i Myslíš, že siRFX:DAT dám1S G odby nˇekohoGEN porouˇcetINF ? ‘You think, that I will take commands (lit.: let2 myself 2 be commanded5 by3 somebody4 )?’
i Capek, ˇ
Karel & Josef, Adam stvoˇritel, 1982
202 da(va)t+inf in Czech That dative reflexive permissives are generally less frequent than accusative reflexives is familiar from Polish, where additional searches were necessary, too. Searches in the full corpus for da(va)t+inf with si preceding or following da(va)t yielded foremost factitive instances, which will be discussed below. A list of verbs used with dá(va)t in permissive function in the full corpus is given in list 17. It was arrived at by way of manual inspection of potential candidates, involving heuristic searches using translational equivalents of Polish verbs; see below the comparison with nech(áv)at for query strings. Examples, with and without type II causee, include the following: (329)
[. . . ] ale vítˇezství siRFX:DAT pardubiˇctí hokejistéNOM:PL i nedaliNEG:PS T :3PL vzítINF . ‘... but the Pardubice hockey players2 didn’t let3 victory be taken4 away from them1 (i.e., did not relinquish victory)’
(330)
Nem˚užeme siRFX:DAT dátINF vnucovatINF podmínky odby nˇeii jakého... utopistyGEN ! ‘We cannot let2 conditions be imposed3 on us1 by4 some... utopian5 !’
The frequency of dative reflexive permissives even in the full corpus is very low. However, semantically they seem to be compositional and comparable to accusative reflexives. Some verbs are found only once in the corpus, suggesting productivity, if low functional importance. The majority of uses in the corpus concerned lexicalized types, to which I turn now.
Lexicalized types Four more dative reflexive constructions with varying degrees of idiomaticity were present. In the corpus sample, they were counted as phraseological types involving záležet ‘depend’ (n=15), rˇíci ‘say’ (n=10), líbit ‘appeal to’ (n=2) and ujít ‘get away’ (n=1). Dát si záležet ‘take seriously’ and dát si líbit ‘suffer’ are phraseological calques from German sich an etwas gelegen sein lassen and sich etwas gefallen lassen; their meaning cannot be derived compositionally (see also Toops (1992) for a discussion of such calques). Dát si rˇíci ‘listen to advice’ more clearly involves a notion of permissivity, as in i Mladá
fronta DNES, 22. 11. 2004
ii Capek, ˇ
Karel, Bílá nemoc, 1994
Causative
(331)
203
i Amálka siRFX:DAT dalaPS T :3S G rˇ íciINF a z˚ustala u muže [. . . ]. ‘Amálka listened to what she was told (lit.: let2 herself 1 be told3 ) and stayed with her husband [. . . ]’
Dát si ujít ‘let pass by, not take advantage of’ is also clearly a phraseologic unit parallel to German sich etwas entgehen lassen. The causee is typically inanimate and given in accusative, analogous to case assignment with nech(áv)at, where this combination is much more frequent (see below): (332)
Tohle zvládnu, povídám, - takovou pˇríležitostACC siRFX:DAT ii nedámNEG:1S G ujítINF . ‘I’ll handle that, I say - I won’t let3 such an opportunity1 pass4 by me1 ’
As we can see, with the majority of these lexicalizations some shade of permissive causation may be construed, and a causee may be introduced in an instrumental or prepositional by-phrase. Moreover, as we will see below, the same combinations principally also occur with nech(áv)at, which I take to be evidence of some limited compositionality of these lexicalized types.
Other cases of the reflexive Other forms of the reflexive pronoun would most prominently include the instrumental case. While (333) was readily accepted, informants differ in the acceptabiliy of by-phrases. (333)
Petr dalPS T :3S G sebouRFX:INS manipulovatINF (??odby otceGEN /?? otcemINS ). ‘Petr let himself be manipulated (?by his father).’
Clearly, nech(áv)at is more appropriate here and a type II causee is felicitous: (334)
NenechteNEG:I MP:2PL sebouRFX:INS nikýmINS manipulovatINF . ‘Don’t let1 yourself 2 be manipulated4 by anyone3 .’
Not a single case of reflexive use with an instrumental pronoun with any verb ˇ was found in all of the written CNK; this option is clearly marginal, but in view of informant judgments must be considered not completely excluded. At the same time, such uses are well attested with nech(áv)at52 .
i Cep, ˇ
Jan, Polní tráva, 1999
ii Vanˇ ek-Úvalský,
Bohuslav, Zabrisky, 2001
204 da(va)t+inf in Czech Pronoun variability: clitic and tonic As a rule, the clitic accusative pronoun cannot be replaced by the tonic pronoun. Informants did not generally accept (335) ?Sebe jsem nedal pˇresvˇedˇcit, RFX:TONIC:ACC AUX:1SG let:NEG:PST:SG persuade-INF Pavla ano . . . Pavel-ACC yes . . . ‘Me, I didn’t let myself be persuaded, but Pavel I did ...’ This test, however, is not conclusive since it is linked to the impossibility of non-reflexive permissives that are implicitly referred to in such contrastive contexts; above, #ale Pavla jsem dal pˇresvˇedˇcit is inferred, but this has factitive reading (‘he had Pavel persuaded by somebody’). However, the following attestation of the long form in a translation from Russian was accepted: (336)
Nato si Selivanovskij odvedl Oskolupova a Jakonova k sobˇe a pohrozil jim, že je oba dostane za mˇríže, ale sebeRFX:ACC že nedáNEG:3S G degradovatINF na poruˇcíka nˇekde za polárním i kruhem. ‘Then Selivanovskij took Oskolupov and Jokonov aside and threatened them that he would get them both into prison, but that he himself, he would not let himself be demoted to a lieutenant doing service somewhere north of the polar circle.’ ˇ No other examples could be found in any of the CNK corpora; note that no instance with the tonic dative pronoun sobˇe was found. I conclude that at least the accusative tonic pronoun is possible provided there is enough contextual support. Such a finding is well explained by an anaphoric function of the reflexive pronoun in this construction.
4.2.3.3.
Comparison to nech(áv)at
Since nech(áv)at is used in analogous syntax and meaning, clues for assessing the productivity of dá(va)t can be won from a comparison of the two formants. We have isolated three attested types so far: productive accusative reflexive, less clearly productive dative reflexive, and lexicalized types with a dative reflexive and some element of permissive meanings. The use of these three types is compared to nech(áv)at below. i Solženicyn,
Alexandr Isajeviˇc, V kruhu prvním I, 1992
Causative
accusative reflexive infin. dá(va)t vylouˇcit ‘exclude’ 89 pˇresvˇedˇcit ‘persuade’ 51 odbýt ‘dismiss’ 45 zlákat ‘tempt’ 44 chytit ‘catch’ 40 zabít ‘kill’ 30 vyvést ‘bring out’ 23 zastrašit ‘scare’ 16 vyrušit ‘disturb’ 16
nech(áv)at 30 62 20 98 37 66 56 31 11
dative reflexive infin. dá(va)t vymluvit ‘talk out of’ 22 vzít ‘take’ 12 porouˇcet ‘command’ 6 namluvit ‘talk into’ 5 rozmluvit ‘dissuade’ 4 poruˇcit ‘command’ 2 odejmout ‘take away’ 2 vnucovat ‘impose’ 1 napovídat ‘dictate’ 1
nech(áv)at 10 52 3 8 5 0 2 0 1
205
Table 33. Use of complement verbs with dá(va)t and nech(áv)at in compositional accusative and dative permissive reflexive constructions in the non-translated part of syn2005.
Table 33 shows the results of full corpus queries for both nech(áv)at and dá(va)t with those infinitives that were found most often in the accusative reflexive construction and cited in verb list 16 and those for the productive dative reflexive permissive given in list 17 above; translations see there. Note that the figures for dát are biased, as they also contain modal passive uses. However, none of these verbs, except, to some extent, vylouˇcit and odbýt, are very often used in the modal passive, as cursory inspection of the results showed. With dative reflexives, the numbers are reliable since
206 da(va)t+inf in Czech infin. líbit ujít
dá(va)t 44 30
nech(áv)at 192 467
infin. záležet ˇríci
dá(va)t 204 120
nech(áv)at 8 8
Table 34. Use of complement verbs with dá(va)t and nech(áv)at in phraseological formations involving si in the non-translated part of syn2005.
the queries53 were restricted to include si before or after dá(va)t and most examples were individually inspected. While instances of nech(áv)at in the accusative reflexive type are generally more numerous than those of dá(va)t, we can say that frequencies are of the same magnitude: there is no sharp contrast consistently applying across complement verbs. A similar finding applies to dative reflexive types: nech(áv)at is not used much more frequently, as one would expect if this type had become unproductive for dát. There are exceptions with markedly higher frequency, such as nechat si vzít and dát si mluvit, and a full picture would have to take a more careful analysis of nech(áv)at and the full set of complement verbs with which it appears in this function into account. However, the result of these searches suggests that in combination with these complement verbs, dá(va)t is neither the main nor a completely marginal formant. Rather, in reflexive permissive use, dá(va)t seems to be comparable to nech(áv)at in terms of functional load. A different picture obtains with lexicalized types. Table 34 shows the outcome of an analogous investigation of the four verbs found to belong to this class (see above page 202). Dát si záležet and dát si rˇici are more frequent with da(va)t, while nechat si líbit and ujít are more frequent than their counterparts. This confirms that these are lexicalized types, the formation of which does not primarily depend on meaning components expressed by matrix and complement predicates, but on specific lexemes. A further finding sets dá(va)t si zaležet apart from compositional uses: the relationship of dát to dávat is here about two to one. I see this as evidence of a low degree of compositionality with causative dá(va)t; we will see later that with some other other minor types, a similarily high proportion of dávat is observed.
Causative
matrix asp. form dávat pres. dát inf. imp. past non-past rel.:
abs 1 8 2 56 12 79
polarity pos. neg. 1 0 8 0 2 0 50 6 11 1 72 7 91% 9%
infinitive pf ipf 0 1 7 1 2 0 45 5 10 2 64 9 88% 12%
207
causee zero type II 1 0 8 0 1 1 56 0 11 1 77 2 97% 3%
Table 35. Interpersonal factitive: occurrence in the annotated sample according to form of the matrix verb; negation; complement aspect (w/0 6 biaspectual infinitives); type and occurrence of causee.
4.2.3.4.
Conclusions: permissive causation
Only with reflexive pronoun, but, in general, with any reflexive pronoun, is permissive semantics of da(va)t+inf in Czech possible. It is to different degrees productive: most clearly and functionally most importantly, with the accusative reflexive; to a lesser degree, with the dative reflexive, where lexicalized types are more frequent. The non-reflexive permissive, conversely, is not productive anymore. This finding is evidence showing that this particular morphosyntactic configuration has grammaticalized to live, so to speak, its own life after death of its parent construction, the non-reflexive permissive, from which it is derived by semantic as well as syntactic composition and which only survives in traces today.
4.2.4.
Factitive
I now turn to factitive causation. Czech is the only language in the sample where factitive is the dominant use of causative da(va)t. The most important type is curative causation, a subtype within the domain of interpersonal causation. I will start with this and then turn to other, more peripheral types. Table 35 gives an overview of all interpersonal factitive forms as met in the corpus. Concerning aspect, the picture is similar to what was seen in the
208 da(va)t+inf in Czech reflexive permissive: perfective matrix and complement aspect dominates. There is a clear difference concerning polarity and causee expression: posititive instances constitute 91% and the causee is mentioned in only 3% of all cases.
4.2.4.1.
Curative and other interpersonal factitive
Curative causation was introduced as a type in the introduction (page 18); its expression is a peripheral function of da(va)t in Polish and a central function in Czech. As outlined in the introduction, curative causation prototypically denotes causation on an object by way of a typically unexpressed intermediate animate agent, the causee. Russian and Polish, in contrast to German or English, tend not to mark such causation at all. Czech seems to be inbetween: while such contextual causatives are possible in some, notably reflexive, cases (Toops 1992: 45), the parallel corpus study clearly shows that in translations of contextual causatives into Czech, dá(va)t is introduced as an explicit marker of curative causation (see 5.1.2 from page 238). Furthermore, this also seem to bear some connection to tense and aspect functions; in (337-a), a perfective past tense example, curative causation must be marked, even if the explicit location is a place where services are typically offered by professionals. In (337-b), involving an imperfective present, however, both readings seem to be possible: (337)
a.
b.
Petr opravil auto v opravnˇe. Petr-NOM repair-PF:PST:3SG car-ACC at garage-LOC ‘Peter repaired the car at the garage’ not ‘Peter had the car repaired at the garage.’ Petr opravuje auto v opravnˇe Petr-NOM repair-IPF:3SG car-ACC at garage-LOC ‘Peter is repairing the car at the garage’ or ‘Peter is having the car repaired at the garage.’
Generally, more research on this topic is in order. We can be certain, however, that the use of contextual causatives, that is, the non-marking of curative causation, is restricted in Czech, plausibly due to contact influence from German. Usually, nech(áv)at or dá(va)t are used to express this type of causation: (338)
Petr dalPS T :3S G /nechalPS T :3S G auto opravitINF v opravnˇe. ‘Peter had (dát or nechat) the car repaired at the garage.’
Causative
209
• non-reflexive n=3: postavit ‘put’; n=2: odvézt ‘take away’, provést ‘carry out’, spravit ‘repair’, zazdít ‘seal off’; 40 hapax legomena • accusative reflexive n=2: korunovat ‘crown’, rozvést ‘divorce’; n=1: obˇrezat ‘circumcise’, odbarvit ‘bleach’, vsadit ‘put in’, zastupovat ‘substitute’ • dative reflexive n=2: ušít ‘sew’, vytisknout ‘print’, zavolat ‘summon’; n=1: potvrdit ‘confirm’, pˇredložit ‘propose’, pˇrehrát ‘play’, pˇrelakovat ‘revarnish’, provést ‘carry out’, razit ‘coin’, svázat ‘bind up’, usmažit ‘fry’, vymalovat ‘paint’, vzít ‘take’, zastˇrihnout ‘curtail’ Verb list 18: Complement verbs with curative dá(va)t, by reflexive element. List 18 gives an overview of the complement infinitive of dá(va)t in curative function in the corpus sample. Note that the spread of verbs with the curative seems to be flatter than with the reflexive permissive: more verbs are used only once, pointing to higher productivity, that is, a wider range of verbs it combines with. This might seem also plausible from a different point of view: arguably, there is simply a wider range of verbs denoting an action that may be commissioned than denoting something that one may succumb to. As in Polish, type II curatives imply the truth value of their complement, cf. (339); the causing and caused event may not be individually temporally specified, e.g. (339)
DalPS T :3S G hoACC prohlédnoutINF soudními lékaˇriINS , *ale lékaˇri to neudˇelali. Intended: ‘He had him examined by experienced doctors, but the doctors didn’t do it.’
(340) *Vˇcera dalPS T :3S G hoACC zítra prohlédnoutINF soudními lékaˇri. Intended: ‘Yesterday he had him tomorrow examined by experienced doctors’ This is evidence of event integration; one single event is expressed. As a rule, the verbs in this class are transitive, as in the next examples: (341)
Aby se její miláˇcci nezatoulali, dalaPS T :3S G zámecký parkACC i oplotitINF . ‘So that her darlings would no run astray, she had the castle’s park surrounded with a fence.’
i Sobková,
Helena, Kateˇrina Zaháˇnská, 1995
210 da(va)t+inf in Czech (342)
. . . Konstantin [. . . ] v Konstantinopoli dalPS T :3S G antickým i bohum ˚ DAT postavitINF chrámyACC [. . . ]. ‘Constantine [. . . ] had1 temples4 erected3 for the gods2 of antiquity in Constantinople [. . . ].’
The dative phrase in the last example can under no circumstances be understood as the causee; it necessarily denotes the beneficiary. This cannot be changed by the substitution of antickým boh˚um ‘ancient gods’ by pracovník˚um ‘workers’ in order to introduce a bias for its interpretation as a causee. Rather, the dative phrase is a free beneficiary dative; such datives are used frequently, often in the form of reflexive dative clitics. (343)
[. . . ] rozbˇehl se do nejbližší prodejny gramodesek, aby siRFX:DAT tam ii dal3S G pˇrehrátINF , co by právˇe mˇeli na skladˇe. ‘. . . he made for the nearest music shop to listen to (lit.: to have2 himself 1 played3 to) what they happened to have there.’
Expression of the causee is much less frequent than with the reflexive permissive. Only two cases were found in the corpus sample; one is given below: (344)
[. . . ] dejteI MP:2PL siRFX:DAT potvrditINF zamˇestnavatelemINS , u iii nˇehož jste v pracovním pomˇeru, toto: ‘have (lit.: for yourself) this confirmed by the employer you are working for, like this:’
Both attestations were with instrumental causees. In general, the causee may also be expressed by a prepositional causee formed with od. The next example from the full corpus is analogous to the last example and represents a colloquial register (with vod instead of od): (345)
DalPS T :3S G siRFX:DAT to potvrditINF vodby advokátaGEN . ‘He had it (for himself) confirmed by a lawyer.’
iv
The factors governing the choice of form of the causee, that is, as instrumental or prepositional phrase, will not be discussed here. Evidently, the factors are akin to the factors involved in the form of the passive by-phrase; this, in turn, is not well understood and at least partly attributable to register or style, as instrumental case is seen to be more typical of spisovná than obecná cˇ eština. i Ceška, ˇ Josef, Zánik antického svˇeta, 2000 iii Efeméra
- r˚uzné 2
iv Toole,
ii Linhartová, Vˇ era, Prostor k rozlišení, 1992
John, Spolˇcení hlupc˚u, 1985
Causative
infin. zavolat ‘call’ ušít ‘sew’ provést ‘carry out’ vytisknout ‘print’ spravit ‘repair’ zazdít ‘seal off’
dá(va)t 43 31 26 17 16 4
211
nech(áv)at 24 45 43 47 8 8
Table 36. Curative causation: Use of complement verbs with dá(va)t and nech(áv)at in the non-translated part of syn2005.
Note that like with curative constructions in Russian or Polish, locative expressions are also used to signal the causee. In (346), the causee is introduced with the unambiguosly locative preposition u ‘at, by’. (346)
Jste-li v pracovním pomˇeru [. . . ] a budete žádat náhradu ušlého výdˇelku za dobu úˇcasti u soudu, dejteI MP:2PL si potvrditINF uat svého zamˇestnavateleGEN [..] pr˚umˇerný hrubý hodinový výdˇelek.i ‘If you are working [. . . ] and if you want to claim compensation for loss of income during the time of your presence at court, have [. . . ] your approximate hourly income confirmed by your employer.’
While the curative construction is usually formed with transitive verbs, in some cases intransitive verbs are used, as in the next example. Type II causees are not possible with this construction then; the transitivity restriction applies (see above section 1.3.5 starting page 21). Rather, presumably under the influence of nech(áv)at, accusative causees are attested: (347)
Generál hrabˇe Kinský [. . . ] dalPS T :3S G na její poˇcest nastoupitINF ii hradní strážACC a zahrátINF vojenský pochod. ‘General Count Kinský had1 the castle guard3 assemble2 in her honour and had a military march played4 .’
This seems to be a very restricted type. At least one other intransitive verb, zatroubit ‘to sound the horn’, also stemming from a military context, may be used in the factitive construction.
i Efeméra
- r˚uzné 1
ii Sobková,
Helena, Kateˇrina Zaháˇnská, 1995
212 da(va)t+inf in Czech Comparison to nech(áv)at Table 36 gives numbers from a full-corpus investigation of complement verbs that were frequent in curative function in the sample and that were not used in the modal passive; furthermore, the queries were restricted to exclude se before and after dá(va)t in order to minimize the bias for the modal passive. Again, we see that nech(áv)at and dá(va)t are used with frequencies generally in the same order of magnitude, with nech(áv)at on average used more often. I arrive at a similar conclusion as with reflexive permissives: in curative function, dá(va)t is productive, and functionally neither marginal nor central.
Delimitation of ‘dative causee’ constructions As shown above, the causee in curative constructions formed with dá(va)t cannot be expressed in a dative phrase. Nevertheless, examples of da(va)t+inf with a dative phrase with a similar meaning are attested, if very infrequently. These cases seem to be lexically restricted; I analyze them, like Grepl & Karlík (1983) and Toops (1992), as a construction of literal give in extended transfer meaning; see also the the discussion regarding Polish on page 146 above. One such example was present in the corpus sample: (348)
Pˇredstavme si pro zajímavost, jak by tˇreba automechanici v Buenos Aires pˇrijali svého pražského kolegu a zda by muDAT dali3PL alei spoˇn zametatINF dv˚ur. ‘Let’s imagine for the sake of curiosity that for example car mechanics in Buonos Aires receive their Prague colleague and that they gave/let/had him at least sweep the courtyard.’
What exactly is the status of such a phrase? Superficially, this looks like a causative much like the curative discussed above. Grepl & Karlík (1983: 62), discussing causatives with nech(áv)at and dá(va)t from a diathesis perspective, give the following (constructed) example: (349)
PetrNOM dalPS T :3S G PavloviDAT vypoˇcítatINF složitý pˇríkladACC . ‘Peter had/give/let Pavel solve a complex (mathematical) problem’
They relate this example syntagmatically, to (350) and both examples, paradigmatically, to (351-a,b): i Komárek,
Stanislav, Opšlstisova nadace, 2002
Causative
213
(350)
Petr dal3S G PavloviDAT (zaas úkolACC ), aby (Pavel) vypoˇcítal složitý pˇríklad. ‘Peter gave Pavel (as a task) that (Pavel) solve a complicated problem’.
(351)
Pavel dostal (od Petra) uloženo, ‘Pavel got (by Peter) assigned’ a. aby vypoˇcítal3S G složitý pˇríklad. ‘that he (Peter) solve a complex problem’. b. vypoˇcítatINF složitý pˇríklad. ‘to solve a complex problem’.
This is to say that in this construction, dát is used to denote the initiating (causing) event, the act of commanding, parallel to verbs such as poruˇcit ‘command’, uložit ‘assign’, roz-/prikázat, naˇrídit ‘order’ [p.62]. Such a construction with a dative causee is thus different from the type II curative in several respects. Consider the next example with type II curative. It expresses the notion of ‘having something done’ as a single event; the caused event is implied to have taken place: (352)
Petr dalPS T :3S G dˇetiACC ostˇrihatINF odby holiˇceGEN , ??ale tenhle nemˇel czas a dˇeti neostˇrihal. Intended: ‘Petr had the children’s hair cut by a barber (lit.: had the children sheared by the barber), but he (=the barber) didn’t have time and didn’t cut the children’s hair.’
If od holiˇce is changed to a dative phrase, the interpretation of dal is transfer (the children are given to the barber) with secondary predication (so that he will cut their hair); consequently, the truth value of the infinitive is not implied: (353)
Petr dalPS T :3S G holiˇciDAT / holiˇcoviDAT dˇeti ostˇrihatINF , ale tenhle nemˇel czas a dˇeti neostˇrihal. ‘Peter gave the children to the barber to cut their hair, but he didn’t have time and didn’t cut their hair.’
Such an interpretation of transfer is not always available (as in, e.g., dal jsem se oholit ‘I had myself shaved’); therefore, type II causee in curative constructions may not generally be modified to dative phrases. Finally, note that with the dative construction, but not with the type II construction, the causing and the caused event may be temporally distinct: (354)
Petr vˇcera dalPS T :3S G holiˇcoviDAT dˇetiACC ostˇrihatINF zítra. ‘Peter yesterday gave the barber the children to cut their hair tomorrow.’
214 da(va)t+inf in Czech (355) *Petr vˇcera dalPS T :3S G dˇetiACC ostˇrihatINF odby holiˇceGEN zítra. Intended: ‘Peter yesterday had the children’s hair cut by the barber tomorrow.’ This shows that the dative causee construction expresses two events, rather than one integrated event as in the type II construction. I conclude that this marginal type is not a subtype of causative da(va)t proper, but relates to giving in an extended (metaphoric) transfer sense. Such constructions are relevant to our discussion in as much as they may have been a starting point for the grammaticalization of factitive da(va)t. I will return to this question in the diachronic part and in the conclusions. Ambiguous permissive/factitive As in Polish, in some cases, especially pertaining to reflexive constructions, permissive and factitive causation cannot be reliably distinguished; this was annotated separately. In the next examples, having implies letting: (356)
Služtiˇcka se náramnˇe divila a taky se ptala, co je ta Mnii chovohradišt’ská a dalaPS T :3S G siRFX:DAT to vysvˇetlitINF [. . . ] ‘The maid was greatly surprised and also asked about Mnichovohradišt’ská and had1 it explained3 to herself 2 (or let1 it be explained3 to herself 2 ).’
(357)
Žurnalistka Nelly Blyová se na pˇrelomu století dala zavˇrít do vˇezení, ii aby mohla v "New York World" vylíˇcit vˇezeˇnské podmínky. ‘At the turn of the century, the journalist Nelly Blyová had2 herself 1 put3 into jail (or letherself2 herself 1 be put3 into jail) to be able to describe the conditions there.’
The following verbs were categorized as ambiguous: • accusative reflexive: (n=2) fotografovat ‘take picture’ ; najmout ‘engage’, operovat ‘operate’, svézt ‘drive’ , uctívat ‘adore’, vyfotografovat ‘take picture’, zavˇrít ‘lock up’ • dative reflexive: sponzorovat ‘sponsor’, vysvˇetlit ‘explain’ Note that this is an analytic type dependent on specifics of this analysis; it does not, as, e.g., permissive reflexive or curative, represent an intensional i Stiborová, Vˇ era, Zapomeˇn, ˇreko, téci, 2000 munikace, 1995
ii Kunczik,
Michael, Základy masové ko-
Causative
215
type that plays some role in the denotational potential of da(va)t. Rather, in these cases there is an overlap in extensional meaning and syntactic restrictions (no dative causee, reflexive marker); a decision between other types cannot with certainty be made and the assessment is to some extent arbitrary. In terms of aspect and negation, note that all these cases are positive and only one imperfective matrix form was met.
Dát na sebe cˇ ekat To conclude the overview of factitive causative functions of da(va)t in the interpersonal domain, one last type has to be mentioned that does not relate to curative use: dát na sebe cˇ ekat ‘be long in coming/make people wait for oneself’. As a whole it was the single most frequent factitive combination in the sample and should be considered at least partly lexicalized; its use is exceptional in that it does not denote curative causation, involves an intransitive verb, and expresses a situation of dynamic factitive causation in which the causer, by non-action, coerces the causee into waiting. 5 examples of this type were found in the corpus, 3 of which involved an inanimate causer (cf. the discussion of cognate Polish da´c na siebie czeka´c on page 147 above). (358)
Poruˇcík dlouho nedalNEG:PS T :3S G na f or sebeRFX:T ONIC:ACC i cˇ ekatINF se šalomounským rozˇrešením trudné otázky. ‘The lieutenant did not make1 (us) wait4 for2 himself 3 for a long time (i.e. was not long in coming up) with a Salomonian solution to that difficult problem.’
(359)
Reakce devizových trh˚u na f or sebeRFX:T ONIC:ACC ii nedalaNEG:PS T :3S G dlouho cˇ ekatINF . ‘The reaction of the currency markets was not long in coming (lit.: did not make3 (people) wait4 for1 it4 for a long time)’
This is a frequent type also in the full corpus. It is often used in negative polarity, and almost always with perfective matrix verb. The dative causee is impossible: (360) *Saša dalPS T :3S G námDAT dlouho cˇ ekatINF . ‘Saša made us wait for him.’ It patterns with the curative in aspect usage (dávat is seldom used); the analogous construction with nech(áv)at na sebe cˇ ekat is even more frequent54 . It i Hašek, Jaroslav, Osudy dobrého vojáka Švejka za svˇ etové války 2000
ii Lidové
noviny, 8. 9.
216 da(va)t+inf in Czech should be noted that this collocation, also in the use with inanimate subject, is also prominently found with German lassen.
4.2.4.2.
Manipulative and impersonal
So far, types of factitive constructions relating to the interpersonal domain have been discussed. Since dá(va)t is functionally most important as a curative, this domain is the most important one. I now turn to manipulative and impersonal factitive causation, both of which are functionally marginal. Only one example of factitive manipulative causation was attested in the corpus: (361)
[. . . ] nedovedu dátINF zaniknoutINF nˇecˇ emuDAT v sobˇe, co jim i brání, aby ke mnˇe promluvili. ‘. . . I cannot let/make1 something3 inside of me dissolve2 that hinders them from speaking out to me.’
This complement verb occurred only once in the full non-translated corpus. Note that it is used with dative causee here, typical, it seems, for minor, perhaps residual, types. Verbs with this stem are also used in the impersonal domain, suggesting this case is due to a lexicalized pattern of minor productivity. Impersonal causation was comparatively frequent with 9 cases, e.g. (362)
Tahle informaˇcní chudoba je však pˇrece jen k nˇecˇ emu dobrá, nebot’ alespoˇn dalaPS T :3S G padnoutINF naivní pˇredstavˇeDAT , že ii hominizace byla [. . . ] spoˇrádaným postupem [. . . ] ‘In any case, this scarcity of information is good for something, since it at least made fall (i.e., finished with) the naive idea that hominization was . . . an orderly process [. . . ].’
Examples involved the verbs vzniknout ‘come into being’(5) vyniknout ‘eclipse’ (2), intervenovat ‘intervene’ (1) and padnout ‘fall’ (1). While the complement verbs padnout and intervenovat were not often found with a full corpus search (three / one occurrence), vzniknout and vyniknout were frequent also in the full corpus55 . They may be seen to produce formations of some degree of idiomaticity and, according to informant work, may not generally be replaced with nech(áv)at. They almost always appear i Linhartová,
Vˇera, Prostor k rozlišení, 1992
ii Reflex,
cˇ . 16/2000
Causative
dát dávat
vyniknout 44 40
vzniknout 184 44
nechat nchávat
vyniknout 10 7
217
vzniknout 4 -
Table 37. Dá(va)t and nech(áv)at vzniknout, vyniknout in the full corpus.
with causer and causee both inanimate, the latter expressed in a dative phrase: (363)
Voda mu strhla peˇclivˇe upravený obal a dalaPS T :3S G vyniknoutINF i podstatˇeDAT . The water swept away his carefully prepared outer layer and let/made1 the essence3 appear2 .
With both verbs, the proportion of imperfective matrix dávat is very high, as table 37 shows; at the same time, the imperfective partner of vzniknout, vznikat, is used only three times in the corpus, atypical for the other causative types. Furthermore, this combination is much less frequent with nech(áv)at. This suggests that this is a lexicalized type not compositionally derived from a productive factitive da(va)t in the impersonal domain. I conclude that both impersonal and manipulative causation with dá(va)t are marginal and associated with fixed rather than productive combinations. The dative causee is retained and no type II causee is possible; all verbs discussed are intransitive.
4.2.4.3.
Conclusions
The analysis of da(va)t+inf in factitive function has shown that it is specialized to the interpersonal domain and is productive in the function of curative causation. In the other domains, factitive function is marginal and mostly concerns lexicalized combinations. The uses in the interpersonal domain, including dát cˇ ekat, stand out from the other uses in frequency, syntax and aspect usage: only type II causees are possible and it is used almost exclusively with dát (rather than dávat); nech(áv)at is used comparably here. The other types are used with dative causee and to a larger proportion with dávat; these combinations do not find a parallel with nech(áv)at.
i Šmíd,
Zdenˇek, Proˇc bychom se netopili, 1987
218 da(va)t+inf in Czech • animate causer: n=13: vˇedˇet ‘know’, n=5: slyšet ‘hear’, n=4: znát ‘know’, n=2: pocítit ‘feel’ • inanimate causer: n=5: tušit ‘notice’, n=4: nahlédnout ‘look’, n=3: zapomenout ‘forget’, n=1: vzpomenout ‘recall’ Verb list 19: Verbs in cognitive causation, by animacy of causer. 4.2.5.
Cognitive causation
As in Polish and Russian, several perception or cognitive verbs combine with dá(va)t to denote differing shades of causativity; they are lexicalized to different degrees. As in Polish or Russian, in some cases, the causer may be inanimate without a strong change in semantics; markedly different from both languagees is that with some verbs, inanimate causees are more frequent than animate causees. The verbs found in this function in the corpus sample are given in verb list 19. Table 38 gives figures concerning absolute frequency, frequency in terms of the aspect of dá(va)t and a comparison to nech(áv)at; these numbers are based on full corpus searches. As we see, in almost all cases, the combination with nech(áv)at is much less frequent and the proportion of imperfective matrix dávat is comparatively high. Dá(va)t vˇedˇet ‘let know’ is the most frequent complement verb. It’s syntax is comparable to Polish da´c zna´c and Russian dat’ znat’ both also ‘let know’, but the Czech counterpart is much less frequent. This is confirmed in the parallel corpus (see below chapter 5): of 22 cases of Polish da´c zna´c, only 4 are translated with Czech dát vˇedˇet, but 13 with the Russian cognate. Dát vˇedˇet is used 5 times in the parallel corpus, vs. 21 cases of Russian dat’ znat’ and 22 cases of Polish da´c zna´c. Like its Russian and Polish counterparts dát vˇedˇet denotes factitive causation. It may be paraphrased with ‘inform’, as in (364) with dative causee/experiencer, or with ‘make oneself noticeable’, with prepositional reflexive phrase and without causee, as in (365): (364)
Kdy bude pohˇreb, tedy v kterou hodinu, dám1S G vámDAT vˇedˇetINF , i i když se nás netýká. ‘When the funeral takes place, at what hour, I’ll let you know, even if it does not concern us.’
i Mertl,
Vˇeroslav, Hˇrbitov sn˚u, 2001
Causative
vˇedˇet ‘know’ znát ‘know’ slyšet ‘hear’ pocítit ‘feel’ tušit ‘notice’ nahlédnout ‘look’ zapomenout ‘forget’
dát 290 110 75 36 28 30 101
dávat 63 47 3 14 192 46 46
dá(va)t 353 157 78 50 220 76 147
219
nech(áv)at 1 11 763 1 2 24 15
Table 38. Use of cognitive and perception verbs with dá(va)t and comparison to nech(áv)at in the full corpus. Frequent light verb constructions were filtered out; se disallowed between matrix and complement verb nahlednout, to diminish the number of modal passive usage. Note that the presence of modal passive usage in these figures biases against imperfective matrix aspect.
(365)
Návštˇeva dalaPS T :3S G oabout sobˇeRFX:LOC vˇedˇetINF tichým zai škrabáním na dveˇre. ‘The visitor made itself noticeable (made about itself know) by a soft scraping at the door.’
Dá(va)t znát in the corpus sample denoted permissive causation. Recall that this is also possible with da´c zna´c in Polish. (366)
NedávámNEG:1S G to znátINF , ale mám deprese, tenhle život mˇe niˇcíii ‘I am not letting it show, but I am depressed, this life is destroying me.’
We therefore see an interesting split in comparison to Polish, where da´c zna´c may denote both ‘to inform’ and ‘to let notice’; in Czech, dát vˇedˇet denotes the former and dát znát the latter meaning. Dá(va)t pocítit ‘feel’, analogous to Russian dat’ poˇcuvstvovat’ and Polish da´c odczu´c denotes factitive causation, as in (367)
Richard mi dalPS T :3S G poprvé pocítitINF , jaké je být krásnou a miloiii vanou ženou. ‘Richard was the first to make me feel what it’s like to be a beautiful and beloved wife.’
i Ikarie - povídky, roˇ cník 1999 2001
ii Xantypa, cˇ . 12/2003
iii Zonová, Anna: Cervené ˇ botiˇcky.
220 da(va)t+inf in Czech It is also used with inanimate causer; in the full corpus, two uses with the imperfective partner cítit were found, cf. (368)
Zelená v jarním rozvíjení pˇrírody dává3S G cítitINF plnost života i [. . . ]. ‘The developing spring green of nature lets (people) feel the fullness of life [. . . ]’.
Dá(va)t tušit ‘notice’, nahlédnout ‘look’ and vzpomenout ‘recall’, zapomenout ‘forget’ involve various shades of factitive and permissive meaning. They are used primarily with inanimate causer and animate, usually unexpressed, dative causee: (369)
[. . . ] poˇcin˚u, které dají3PL milovníkum ˚ DAT cˇ eské hudby ii vzpomenoutINF na staré cˇ asy. ‘... steps, that let/make lovers of Czech music recall the old times (i.e., remind them)’
(370)
Zvuky výbuch˚u prý nedávajíNEG:3PL obyvatelum ˚ DAT na válku iii zapomenoutINF . ‘The sound of the explosions allegedly does not let the inhabitants forget about the war.’
Dá(va)t se slyšet ‘let oneself be heard’ is different from all the other verbs in this group in two respects. It is much nearer to other causative uses of da(va)t in regard to the low use of imperfective matrix aspect; moreover, it is much more frequent with nech(áv)at than with dá(va)t. It denotes factitive causation: the causer relates something with the aim of making himself be heard. (371)
Ministr obrany Šaul Mofaz seRFX:ACC dalPS T :3S G slyšetINF , že plán iv bude realizován [. . . ]. ‘The Minister of Defence Šaul Mofaz let2 himself 1 be heard3 that the plan will be realized [. . . ].’
Dá(va)t se slyšet also differs from the much more frequent Polish da´c si˛e słysze´c, which denotes ‘to be audible’ and cannot take a complement clause, as in the example above. This type thus does not have an analogous counterpart in Russian and Polish. In contrast to the two, it is used to translate speech act verbs in the parallel corpus (2 occurrences), e.g.: (372)
“No koneˇcnˇe”, dal se slyšet. “Abou’ time”, he said.
i Jebavá, iv Právo,
Jana: Úvod do arteterapie. 1997 3. 5. 2004
(Rowling, Harry Potter) ii Právo,
31. 12. 2001
iii Respekt,
cˇ . 38/2000
Causative
221
In view of the differences to most other complement verbs, dá(va)t se slyšet should be viewed as clearly distinct from the other verbs.
Conclusions On the whole, the cognitive type is more heterogeneous than in the other languages and would deserve a more thorough examination. Both permissive and factitive causation with these verbs are attested. It seems that many instances involve residual types that partly lexicalized when the denotative potential of the dative control construction of dá(va)t diminished. In none of these types is type II syntax possible; with all of these verbs except dát se slyšet, use of the imperfective matrix verb was exceptionally high in comparison to both factitive and permissive usage. Both points I interpret to indicate that da(va)t+inf with these verbs is only loosely compositionally related to causative da(va)t+inf. 4.2.6.
Conclusions: causative dá(va)t
In this section on the causative functions of da(va)t+inf in Czech, two productive causative functions have been isolated: dá(va)t is used as a curative factitive and as a reflexive permissive formant. Both functions involve the construal of single events and permit only of type II causees. Dative causees are ungrammatical and in the vast majority of cases, the causee is omitted altogether. Dát is used much more frequently than dávat. In these functions, dá(va)t may be replaced by nech(áv)at, which, however, expresses a wider set of causative notions. Note that da(va)t is also parallel in syntax to nech(áv)at in the type II construction. Such a construction with type II causee is not observed with any other verb in Czech; in this sense, these two formants share a small paradigm of two causative formants. However, unlike da(va)t, nech(áv)at is also used as a type I causative with accusative causee. Accusative causees are observed with dá(va)t only sporadically; this might be interpreted as an effect of analogical change.A type I causative construction with a dative causee is, unlike in the other languages, no longer productive. As a type I causative, dá(va)t is restricted to some minor types that need to be considered lexicalized or of low productivity; this involves non-permissive
222 da(va)t+inf in Czech causation, factitive causation in both the manipulative and the impersonal domain and constructions with cognitive and perception verbs; the latter are heterogeneous and express both factitive and permissive causation. These types differ from productive types not only in the form of the causee, but also in the comparatively high frequency of imperfective dávat and in that nech(áv)at is not used with these verbs with comparable frequency.
4.3. 4.3.1.
Modal passive Overview
Like the cognate construction in Polish, the modal passive formed with dát se in Czech denotes dynamic causation: something is possible or impossible by virtue of circumstances, rather than permission or obligation; as in Polish, its meaning does not extend to the epistemic domain. Other than that, we will see that it differs from the Polish construction in several respects: First of all, it is more frequent; secondly, it is more synsemantic, allowing a larger class of verbs; thirdly, it is grammatically restricted in matrix aspect, but more variable in complement aspect; lastly, it does not exhibit a comparable tendency to be used under negation. I start with an overview of syntax, semantics and use of the modal passive in the corpus, providing a comparison to Polish in the process. The basic distinctions have already been introduced in relation to Polish in section 3.6.1, pages 153ff., and will be treated in less detail here. Note again that nech(áv)at does not form a modal passive and will therefore not play a role here. Dát se, like its Polish cognate, implies an unexpressed human agent. As in Polish, this agent cannot be introduced by means of a dative phrase and unlike Polish, no prepositional or instrumental phrases with this function were found. No variation of the reflexive element se is possible; it may not be replaced by sebe or with any other case form of the reflexive. Unlike Polish dawa´c, the Czech imperfective dávat cannot form the modal passive construction. In contrast to Polish, dát se in modal passive use is listed in the standard dictionaries as biaspectual. It is obvious that it may fulfill imperfective functions since it forms the analytic future which is grammatically restricted to imperfective verbs, as attested in (373):
Modal passive
(373)
223
Do Oslavice seRFX:ACC od úterka budeAUX:3S G dátINF dojetINF jen i do vesnice. ‘From Tuesday on, driving in the direction of Oslavice will only be possible (lit.: let2 itself drive) up to the village.’
In contrast to Polish, where such use is not accepted by all speakers, in Czech, this is standard, prescribed usage. What further evidence justifies treating dát in these constructions as biaspectual, not imperfective, is another matter which I will not further investigate; this would require showing that it is also used in contexts that only permit perfective aspect. The following categorization of modal passive constructions is very similar in spirit to the one for Polish (see above from page 155) but involves an important difference: in Polish, a non-agreeing construction with transitive verbs is very prominent; in Czech, this construction is completely marginal. While in the data collection phase this and potentially ambiguous types that could be construed to be either agreeing or non-agreeing were annotated, in the final analysis, this distinction was dropped; see below for a discussion. The resulting typology is determined only by the valency of the complement verb and is given below: • agreeing: transitive verb, patient as subject (374)
ii Lucern a Záhˇreb seRFX:ACC asi nedajíNEG:3PL srovnávatINF . ‘Lucern and Zagreb probably can’t be compared (lit. do not let2 themselves1 be compared3 )’
• intransitive verb: no or non-accusative object (375)
A nedáNEG:3S G seRFX:ACC tu žítINF ani choditINF , rˇíkají o souˇcasném stavu centra mˇesta Brˇnané. (no object)iii And one can’t (lit.: it does not let itself) live or walk around here, they say about the current state of the Brno city center.
(376)
30000 dˇetí dennˇe umírá na onemocnˇení jimžDAT seRFX:ACC dá3S G pˇredejítINF . (dative object)iv 30000 children die each day from diseased which1 can (lit. lets3 itself 2 ) be prevented4
i Mladá fronta DNES, 7. 8. 2004 ii Lidové noviny, 8. 8. 2000 22. 11. 2004 iv Koukolík, František, Homo sapiens stupidus, 2003
iii Mladá
fronta DNES,
224 da(va)t+inf in Czech
type transitive intransitive sent. compl. all types
abs 405 150 99 654
prop. 62% 23% 15% 100%
negated abs. rel. 119 29% 36 24% 13 13% 168 26%
imperf. inf. abs rel 131 32% 92 61% 28 28% 252 39%
past tense abs rel 116 29% 63 42% 30 30% 210 32%
Table 39. Types of modal passive in the corpus sample. The first column gives the number of occurrences in absolute and relative terms. The next two subdivisions provide data on the absolute and relative frequency of negation, infinitive aspect and matrix tense in relation to each type and, in the last row, in relation to all types together.
(377)
Naon samotuACC seRFX:ACC dá3S G zvyknoutINF , ale je to hrozné. (prepositional object)i One can (lit. it lets4 itself 3 ) get accustomed5 to1 loneliness2 , but it’s terrible.
• sentential complement (378)
Kromˇe toho seRFX:ACC dá3S G pˇredpokládatINF , žeCOMP s nar˚ustajícím poˇctem cˇ lenských stát˚u EU budou povolovány výjimky ve ii stále menším rozsahu. Besides, it can (lit. lets2 itself 1 ) be supposed3 that4 with the rising number of EU member states exceptions will be granted on an ever smaller scale.
An overview of the proportion of these types in the corpus sample as well as figures concerning negation, aspect and tense are given in table 39. Note that no uses as participle were found and one case with infinitive dát was not taken into account. Table 39 shows that like its Polish cognate, the Czech modal passive is used foremost in the present. In contrast to Polish, the Czech construction combines more freely with infinitives of both aspects; recall that in Polish, the proportion of imperfective complements was 15% (see table 29, p. 29); in Czech, it is 39%. This puts the proportion found in Czech in the same i Blesk,
8. 8. 2002
ii Žemliˇ cka,
Zdenˇek / Lukš˚u, Vladimír, Dopravní politika, 1999
Modal passive
225
general area as what we find with a central impersonal modal expressing possibility such as lze ‘one can’ where the proportion of imperfective infinitives amounts to 32% (according to a search in the full corpus restricted to infinitives directly following lze or nelze; the difference between the two distributions is statistically significant). As in Polish, the proportion of imperfective infinitives is highest with intransitive verbs. Unlike the cognate construction in Polish, the modal passive in Czech does not exhibit a special tendency to be negated: only about 26% of all cases were negative. Again, this is comparable to the distribution of lze and its negative variant nelze in the full corpus, where 11 034 of 45 793 instances, that is, about 24% are negated (same distribution at p=.34). An important difference, however, is that (ne)lze is much more frequent than modal passive dát se which on the basis of the sample is estimated to be used around 17 500 times in the full corpus. In relation to negation and aspect use, the Czech modal passive is thus more like a core modal than the Polish modal passive is. Below, I will show that it is more grammaticalized in other respects, too.
4.3.2.
Agreeing and non-agreeing construction
The most obvious structural difference to Polish concerns the use of the nonagreeing modal passive with transitive verbs. This is the most frequent type in Polish, but marginal in Czech. Like in Polish, the modal passive construction types divide into two groups according to their structure: on the one hand, there is an agreeing construction formed with transitive infinitives. Here, the accusative argument of the infinitive is expressed as a nominative subject and agrees with dát. On the other hand, there is a group of subjectless constructions where no subject is present and where dát has default third person singular neuter marking. This concerns foremost intransitive infinitives, including those with nonaccusative or sentential complement which lack an accusative argument to begin with. In Polish, there are two options for constructions with transitive infinitives: the agreeing and the non-agreeing construction; in the latter case, the accusative object is retained rather than expressed as nominative. The contrast between the two options encodes a functional difference: the agreeing modal passive expresses possibility grounded in characteristics of the patient
226 da(va)t+inf in Czech (expressed as subject), while the non-agreeing modal passive, together with all other subjectless types, does not restrict the source of modality in such a way. There are two distinct constructions in Polish, I concluded above: an older agreeing modal passive and a newer, more grammaticalized subjectless modal passive construction extending to all verb classes, including transitive verbs whose patient is coded as direct object. In Czech, the situation is very different. Non-agreeing constructions with transitive verbs are very rare. While informant work does point to a general admissability of this construction and Medová (2006) explicitly analyzes it as possible, it is much rarer in the corpus than the agreeing construction; perhaps, it is restricted to a spoken register. The corpus sample analized here included a single example (not included in the statistics above): (379)
Mluví se i o dalších vˇecech. O zásobování speak-3SG RFX:ACC and about other things. About stocking-up jídlem a jak by se dalo vyplout na food-INS and how IRR RFX:ACC let-PST:3SG:N go-out-INF to moˇre a lovit ryby. sea-ACC and catch-INF fish-ACC:PL. ‘One talks about other things. About stocking up on food and how it would be possible (lit. how it would let itself) to go out to the sea and i catch fish.’
This example involves the coordination of two infinitives: intransitive vyplout ‘go out (by ship)’ and transitive lovit ‘catch’ with the patient ryby ‘fish’ expressed as accusative object. The deletion of the common predicate dát se is only possible if the non-agreeing construction is also used with the transitive verb, since otherwise the two uses of dát would not match in respect to morphology, cf. (379)’ modified to include a agreeing passive: (379)’ [. . . ] jak by se dalo vyplout na moˇre ... how IRR RFX:ACC let-PST:3SG:N go-out-INF to sea-ACC a jak by se dali lovit ryby. and how IRR RFX:ACC let-PST:3PL catch-INF fish-ACC:PL. Such coordination of intransitive and transitive infinitive seems to be the most frequent reason for the use of the non-agreeing construction with transitive verbs. A full corpus search56 yielded eight more relevant examples; five were of this type. i Rasmussenová,
Nelly, Dˇeti severských šaman˚u, 1995
Modal passive
227
However, three of the additional examples did not involve coordination. Like the only example in Medová (2006), one of these involved a light verb construction that can be considered the equivalent of an intransitive verb: (380)
Uznávám, že v noci z 20. na 21. srpna 1968 seRFX:ACC nedaloNEG:PS T :3S G klástINF odporACC vojensky, ale myslel i jsem si, že jsme mohli svou k˚uži prodat dráže. I accept that on the night of the 20 to 21 of August 1968 it was not possible to (lit. did not let2 itself 1 ) put up3 military resistance4 , but I did think that we could have sold our skins for a more expensive price.
The other two examples, finally, cannot be explained in any such way, and are testimony to the fact that this construction does have some, at least in written usage, very limited currency: (381)
Kdyby seRFX:ACC daloPS T :3S G:N poslatINF kusACC Kunštátu za ii Tebou místo líˇcení. If only it were possible to (lit. let2 itself 1 ) send3 a piece4 of Kunštát for you instead of a description.
(382)
Zadržení ˇridiˇcského pr˚ukazu dosud bylo pˇredbˇežným opatˇrením, které se využívalo v pˇrípadech, kdy seRFX:ACC daloPS T :3S G:N oˇcekávatINF zákazACC ˇrízení, vyslovený soudem nebo správním iii orgánem. ‘The retention of a driver’s licence has been a preliminary measure, which was used in cases where one could expected (lit. let2 itself 1 expect3 ) a driving ban4 imposed by a court or administrative body.’
For the purposes of this study, non-agreeing use of dát se with transitive infinitives is considered marginal. Further research into the use of this option promises to be interesting, however, since it might shed light on general principles at work with such cases of micro-variation in Czech and beyond.
Conclusions In Czech, the non-agreeing modal passive is possible, but not of functional importance and overall infrequent. In contrast to Polish there is no semantic i Mladá fronta DNES, 8. 8. 1998 DNES, 27. 12. 2000
ii Halas,
František, Dopisy, 2001
iii Mladá
fronta
228 da(va)t+inf in Czech contrast corresponding to the two coding options given with transitive verbs. I conclude that, differently than in Polish, in Czech the modal passive functionally involves a single construction; no divergence, as in Polish, has occurred. This single construction bears close resemblance to other diathesis constructions such as those formed with the passive participle and auxiliary or the reflexive marker – the patient of a transitive infinitive is expressed as subject; if the verb is intransitive, the construction remains without subject. In contrast to many passive constructions57 , no functional difference arises from this difference in form.
4.3.3.
Further characteristics
Recall that in Polish, only the more grammaticalized subjectless construction, but not the agreeing construction allows negation of the infinitive. This restriction is not found in Czech. While only one attestation of affirmative dát with negated infinitive was present in the full corpus (383), double negation as in (384) is quite usual with both agreeing and subjectless types: (383)
Neumˇel si pˇredstavit, že by seRFX:ACC tak jemný hlasNOM dalPS T :3S G i neposlechnoutNEG:INF . ‘I couldn’t imagine that it was possible not to obey such a gentle voice (lit. such a gentle voice2 would let3 itself 1 not obey4 ).’
(384)
ii Taková vˇec seRFX:ACC nedáNEG:3S G neplánovatNEG:INF . Such a think cannot not be planned (lit. does not let2 itself 1 not plan3 ).
No instance of first or second person dát in modal passive usage was found in the full corpus. This is remarkable, as it is suggests a loss of the inflectional category of person. However, it seems this loss is a usage phenomenon related to the exceptionality of animate patients, rather than an absolute restriction, as the following constructed example is admissable: (385)
NedášNEG:2S G seRFX:ACC pˇrehlédnoutINF . ‘You cannot be overlooked (lit. don’t let yourself be overlooked)’
Evidently, low use of first or second person predicates is due to the fact that in general, animate subjects are dispreferred because they give rise to a reflexive permissive reading. i Cherryhová,
Carolyn Janice, Kukaˇccˇ í vejce, 1995
ii Blesk,
20. 5. 2003
Modal passive
229
Three cases in the corpus sample involved an animate subject referent without the possibility of a reflexive permissive reading of the whole construction. In these cases, this reading is excluded by virtue of the semantics of the resulting situation (being described or watched in the following examples just cannot be influenced by the subject referent), cf. (386)
Tak by seRFX:ACC dalPS T :3S G asi nejlépe popsatINF brýlatý cˇ lovíˇcek i. jménem Herold, který pracuje jako programátor [. . . ] ‘This is how one could best describe (lit. how he would let2 himself 1 best describe3 ) the small spectacled man named Herold, who works as a programmer [. . . ].’
(387)
ii Kde by se dali lehce hlídat a neotravovali svou d˚uležitostí. ‘Where they would be easy to watch (lit. let2 themselves1 be easily watched3 ) and wouldn’t be annoying in their importance.’
I conclude thus that there is no grammatical restriction concerning animate patients in these constructions. However, like in the Polish agreeing construction, animate patients are dispreferred because of the resulting ambiguity of the construction. Note that the Polish non-agreeing modal passive is less restricted since no such ambiguity is given for structural reasons. List 20 provides an overview of all complement verbs. It shows that modal passive dát se is productive with a wide spread of verbs attested. Two verbs are especially frequent: rˇíci ‘say’ and dˇelat ‘do’. The latter is very frequently used in the rethorical question co se dá dˇelat ‘what can be done’. Dá se rˇíci, along with some other verbs such as pˇredpokládat, is very frequently used as a parenthetical, cf. (388)
Aˇc byl urostlý, daloPS T :3S G by seRFX:ACC rˇ íctINF sportovní typ, svým dlouhým plnovousem spíše pˇripomínal poustevníka z hisiii torického obrázku. ‘Although he was tall, one can say (lit. it lets itself be said) of athletic type, with his long beard he looked rather like a hermit from a historical picture.’
In contrast to Polish, the Czech modal passive combines not only with agentive, but also with unaccusative verbs, such as umˇrít, zemˇrít or padnout, e.g. (389)
Na žlutou zimnici seRFX:ACC dá3S G umˇrítINF . From yellow fever one can die (lit. it lets2 itself 1 die3 ).
i Reflex,
ii neotravovali svou d˚ cˇ . 13/2000 uležitostí noviny, 7. 8. 2002
iv Hospodáˇrské
iii Zeman,
iv
Jiˇrí, Afrika, 2001
230 da(va)t+inf in Czech Complement verbs by type Transitive verbs: n=38: dˇelat ‘do’; n=7: rˇíci ‘say’; n=6: oˇcekávat ‘expect’; n=4: vydržet ‘endure’, vysvˇetlit ‘clarify’; n=3: cˇ ekat ‘wait’, natoˇcit ‘picture’, nazvat ‘call’, poˇrídit ‘purchase’, postavit ‘construct’, použít ‘use’, pˇrirovnat ‘compare’, stihnout ‘catch’, udˇelat ‘do’, vyjádˇrit ‘express’, zjistit ‘determine’, zvládnout ‘cope with’; n=2: dosáhnout ‘achieve’, hrát ‘play’, interpretovat ‘interpret’, nahradit ‘compensate’, najít ‘find’, napravit ‘repair’, odhadnout ‘estimate’, oznaˇcit ‘indicate’, pˇrevést ‘transfer’, provozovat ‘pursue’, rozdˇelit ‘split up’, urˇcit ‘establish’, vyˇcíst ‘read off’, vyˇrešit figure ‘out’, využít ‘use’, zajistit ‘ensure’, získat ‘gain’; n=1: 107 verbs Intransitive verbs: n=13: žít ‘live’; n=11: ˇríci ‘say’; n=10: mluvit ‘speak’; n=5: uvažovat ‘consider’; n=4: dosáhnout ‘achieve’, jít ‘go’; n=3: jet ‘drive’, pokraˇcovat ‘go on’, vˇeˇrit ‘believe’; verbs n=2: dívat ‘watch’, dostat ‘get’, dýchat ‘breathe’, jezdit ‘drive’, jíst ‘eat’, koukat ‘sea’, pˇrespat ‘sleep out’, psát ‘write’, usuzovat ‘assume’, vydržet ‘endure’, vyjít ‘come forth’; n=1: 72 verbs With sentential complement: n=50: rˇíci ‘say’; n=9: oˇcekávat ‘expect’, pˇredpokládat ‘assume’; n=3: poznat ‘learn’, tušit ‘think’; n=2: namítnout ‘object’, odhadnout ‘estimate’, spoˇcítat ‘estimate’; n=1: 19 verbs Verb list 20: Complement verbs according to types of modal passive construction.
In this sense, the modal passive construction is less restricted in Czech than in Polish in terms of the verbs it may take and consequently in its meaning. I analyze this to be a symptom of a greater degree of grammaticalization of the Czech formant in the sense of semantic attrition.
4.3.4.
Conclusions: modal passive
Since a non-agreeing construction with transitive verbs is marginal in Czech, in contrast to Polish, the Czech modal passive does not fall into two functionally differentiated groups of constructions; rather, the construction type
Residual types
231
is agreeing with transitive and non-agreeing with other verbs without a functional differentiation. Overall, the construction is functionally comparable to the more grammaticalized subjectless construction in Polish in respect to semantics and the possibility of negation of the complement infinitive. Altogether, dát se is more grammaticalized and more comparable to central modals than its Polish cognate. Like core modals in Czech, it does not have an aspectual partner and seems do display similar behavior in relation to polarity or complement aspect. Moreover, the Czech modal passive may take unaccusative verbs and is less restricted semantically than its Polish cognate. Like the Polish formant, however, it does not express epistemic modality, as core modals, according to Hansen (2001: 89f.), generally do, and should be therefore considered a modal of medium grammaticalization.
4.4.
Residual types
In this section I cover two more types annotated in the corpus, labeled give plus DIR and secondary predicate / consumptive verb during annotation. Give plus DIR is present in the following example: (390)
Dáme okapat a namáˇcíme v sirupu. Give-1PL drain-INF and soak-1PL in syrup-LOC. ‘We put it to drain and then soak it in syrup.’
i
Constructions like these are specific to Czech. Superficially, they are strongly reminiscent of Russian dat’ vysoxnut’, otstojat’sja ‘let sth. dry/settle’ etc., that is, permissive causation in the manipulative domain. However, they do not take a dative causee, but an accusative patient of give. It may take a directional adverbial: (391)
Povrch pokapeme rozpuštˇeným máslem a dáme1PL asi na hodinu ii péctINF doPREP troubyGEN . ‘We sprinkle the top with melted butter and give1 it into3 the oven4 to bake2 for about an hour.’
As Medová (2006) points out, da(va)t here is not biaspectual. The construction should be considered an instantiation of da(va)t with a secondary predicate, cf. i Miltner,
Vladimír, Indická kuchynˇe, 2001
ii Marhold,
Jiˇrí, Krkonošská kuchaˇrka, 2002
232 da(va)t+inf in Czech (392)
Tu polijeme vejci a znovu dáme1PL doPREP troubyGEN , dokud vejce i nezhoustnou. Then we pour eggs over it and put it into the oven again, until the eggs thicken.
This type is not present in other languages in the sample and is evidence of the extension of give without human recipient to mean put which, however, lies outside the scope of this study. The other group consists of constructions with secondary predication and consumptive verbs: (393)
ii
Dává3S G miDAT jístINF - jsem jeho pes. ‘He gives me to eat - I am his dog.’
This latter type is familiar from Russian and will not be discussed any further here.
4.5.
Summary and conclusions
Czech da(va)t+inf is grammaticalized in causative and modal passive function. As a causative, it is unequivocally productive in curative function and as a reflexive permissive, both type II causative constructions. Here, it is used in comparable function and syntax with nech(áv)at. Aside from that, it may express some causative types of minor productivity in the dative causee construction; clearly, it is lexically restricted here and relatively infrequent. Besides constructions with cognitive and perception verbs, these types relate to both the manipulative and the impersonal domain; both factitive and permissive causation is expressed. Here, no type II causees are possible. From a comparative point of view, the most remarkable characteristics of causative da(va)t+inf are thus the development of a productive curative, the loss of the non-reflexive permissive as well as the development of and restriction to type II syntax in all productive types. More important than the causative construction is, in terms of functional load, the use of da(va)t+inf in the modal passive. Here, dát se is a fairly central modal, clearly more grammaticalized than its Polish cognate.
i Novotná,
Jana & Dora, Španˇelská kuchynˇe, 1999
ii Ikarie
- povídky, roˇcník 1999
Chapter 5 Czech, Polish and Russian in parallel
5.1.
Introduction
Additionally to the above corpus studies, the use of da(va)t+inf in Czech, Polish and Russian was investigated in the Slavic parallel corpus ParaSol58 . The examination of parallel translations serves three functions: • it was used as a heuristic device in the initial stages of investigation in order to find relevant functional types that coincide or differ between the three languages. This hypothesis-building process is not described in detail here. • it provides a constrained enlargement of scope to competing formants of da(va)t; the set of translational equivalents of a da(va)t+inf in one language so to speak opens a window into the onomasiologic field of formants in the other language. • it serves to provide additional evidence concerning differences in frequency found in the monolingual corpora. This has already been referred to above; e.g. the higher frequency of non-reflexive permissives in Russian in comparison to Polish found in monolingual corpora is validated by data from the parallel corpus. Most of the results immediately relevant to the synchronic comparison have already been cited above. In this section, only a very brief overview of the conducted analysis will be given. ParSol, a Parallel Corpus of Slavic and Other Languages is a parallel corpus containing original and translated texts in 14 Slavic languages developed at the Slavic departments of the Universities of Bern, Switzerland and Regensburg, Germany. The corpus is available through a web interface; see von Waldenfels (2006, 2011) and the corpus web sites at parasol.unibe.ch and www-korpus.uni-r.de/ParaSol for more information. Since texts in ParaSol are lemmatized and morphologically annotated, the same queries as in the synchronic corpus studies could be employed. At the time of the present study the corpus was in a pilot stage and not well balanced. Generalizations based on these data therefore have to be made with caution. Factors such as the combination of source and target languages,
234 Czech, Polish and Russian in parallel texts Mixail Bulgakov: Master i Margarita Stanisław Lem: Astronauci, Fiasko, Głos pana, Pami˛etnik znaleziony w wannie, Pokój na Ziemi, Solaris Joanne K. Rowling: Harry Potter and the Sorcerer’s Stone Arkadij i Boris Strugackie: Piknik na oboˇcine all
Czech 132 072 517 870
Polish 167 464 573 025
Russian 145 213 519 000
94 649
100 460
90 823
57 311
71 973
61 799
801 902
912 922
816 835
Table 40. Texts in the Polish-Russian-Czech subcorpus with size in tokens.
texts Mixail Bulgakov: Master i Margarita Stanisław Lem: Astronauci, Fiasko, Katar, Kongres futurologiczny, Pami˛etnik znaleziony w wannie, Pokój na Ziemi, Powrót z gwiazd, Solaris, Wizja lokalna Milorad Pavi´c: Hazarski reˇcnik Joanne K. Rowling: Harry Potter and the Sorcerer’s Stone Arkadij i Boris Strugackie:Piknik na oboˇcine, Gadkie lebedi Patrick Süskind: Das Parfüm all
Polish 167 464 826 493
Russian 145 213 745 547
105 118 100 460
105 836 90 823
162 257
138 406
78 609 1 440 401
80 226 1 306 051
Table 41. Texts in the Polish-Russian subcorpus with size in tokens.
influence of the textual basis, contextual parameters and a more detailed analysis of semantics in context would need to be taken into account in order to reach more definite conclusions. For this study, two subcorpora were used: a Russian-Polish and a smaller Russian-Polish-Czech subcorpus. Their textual composition is given in tables 40 and 41 (for bibliographic references, see the corpus webpage).
Introduction
5.1.1.
235
The permissive
In Russian and Polish both permissive and reflexive permissive are productive constructions. However, as we have seem above, non-reflexive constructions are the main type in Russian, while in Polish, the reflexive constructions are more grammaticalized. In the Polish texts, of the 32 non-reflexive permissives with da(va)t+inf, most corresponded to the cognate construction in Russian (15 instances). The next most frequent type involved the verb pozvolit’ ‘allow’ with 5 cases. The remaining 12 cases were spread across different types, the next most frequent equivalent being free translations. In Russian texts, non-reflexive permissive da(va)t was more frequent. Of 57 non-reflexive cases, only 1659 , that is, 28% corresponded to the Polish cognate. Another 11 (19%) corresponded to pozwoli´c ‘allow’: (394)
(395)
RU Ja ne dalPS T :S G emuDAT skazat’INF bol’še ni slova. PL [. . . ] ale nie dałemPS T :1S G muDAT doj´sc´ INF do słowa. ‘I didn’t let him say another word.’
i
PL Obstapiono ˛ go i nie pozwalanoIPS przej´sc´ INF . ii RU Ego obstupili i ne davaliPS T :PL projtiINF . ‘They blocked his path and didn’t allow him to pass (RU) / let him pass (PL).’
The next largest group included free translations and other not further categorized cases; the following presents such an example which illustrates the lack of manipulative permissive in Polish. Translations with niech form another important group that has already been discussed above in section 2.2.6.1 from page 85. (396)
RU K e˙ tomu on dobavil [. . . ] cibetina, [. . . ] dalPS T :S G nastojat’sjaINF i profil’troval vo vtoruju butylku. PL Dodał [. . . ] cybetu, [. . . ] pozostawiłPS T :3S G do wytrawieniaV BN:GEN i odfiltrował do innej butelki. DE Dazu gab er [. . . ] Zibet, [. . . ] ließPS T :3S G digerierenINF und iii und filtrierte ab in eine zweite Flasche. ‘To this he added [. . . ] civet, [. . . ] let it digest (RU/GE) / put it somewhere to digest (PL), and filtered it into a second bottle.’
i Stanisław Lem, Pami˛ etnik znaleziony w wannie. lebedi iii Patrick Süskind, Das Parfüm
ii Arkadij
i Boris Strugackie, Gadkie
236 Czech, Polish and Russian in parallel The parallel corpus thus clearly reflects the asymmetry in grammaticalization of non-reflexive permissive da(va)t in Russian and Polish: in Russian, the formant is more frequent and involves a more diverse set of translations into Polish than the other way around. Reflexive permissives, conversely, were more frequent in Polish than in Russian. Here, the most frequent equivalent was Russian pozvolit’ ‘allow’ with 24 of 73 Polish instances. The Russian cognate construction involving dat’ sebja was only used in 17 cases. The third most frequent type from Polish to Russian involved different configurations of arguments and participants, roughly captured as a change in diathesis (11 cases), as in (397): (397)
RU Nekotorye soglašajutsja3PL , osobenno ljudi s sem’jami. PL [. . . ] niektórzy daja˛3PL si˛eRFX:ACC namówi´cINF , zwłaszcza i je˙zeli maja˛ rodziny. ‘Some agree (RU) / let themselves be talked into it (PL), especially if they have family.’
In this case, the function of the reflexive permissive to express an agentive patient is mirrored in the use of a middle verb. To conclude, note that almost all reflexive permissives in Russian (16 of 19) corresponded to the Polish cognate construction, and two involved pozwoli´c. All reflexive permissive cases were found in translations, rather than in Russian original texts. We see that in both languages, pozwoli´c/pozvolit’ ‘allow’ is an important verb expressing a subset of functions of da(va)t. Somewhat simplifying matters, we can say that non-reflexive pozwoli´c and da´c express to similar degrees in Polish what is expressed by non-reflexive dat’ in Russian alone; conversely, Polish da´c si˛e is often translated as pozvolit’ sebja into Russian. Generally, the parallel corpus confirms the conclusions based on the basis of the monolingual corpora: the non-reflexive permissive function of da(va)t+inf is more general in Russian and encompasses functions expressed by heterogeneous means in Polish, including the analytical imperative formant niech and other more explicit means as in example 2. Polish da´c si˛e, in turn, is more general than its Russian cognate, which is reflected in a similarily diverse set of equivalents that prominently involves middle verbs.
i Arkadij
i Boris Strugackie, Piknik na oboˇcine
Introduction
237
Czech equivalents of Polish and Russian permissive da(va)t+inf The pattern of use of the Czech formant in the (smaller) trilingual corpus is very different, as expected in view of its much more marginal status found in the monolingual corpus. The parallel corpus contained only 16 attestations of permissive dá(va)t. Only 9 were clearly productive reflexive permissives, 2 of which did not correspond to Polish or Czech da(va)t+inf. The rest included one case of dávat spát ‘let sleep’ (see (316)) and 6 cases of intermediate degrees of idiomaticity such as dát si rˇíct ‘listen to advice’ (lit. ‘let things be said to oneself’), nedát se odbýt ‘lit. not let oneself be dismissed’, which did not have counterparts in the Russian and Polish versions. Alltogether there were 72 cases of either Polish or Russian containing either reflexive or non-reflexive permissive da(va)t+inf in the trilingual corpus; only 7 of these corresponded to a cognate construction in Czech:, as e.g., (398)
PL Ja jednak, postanowiwszy twardo, z˙ e wi˛ecej nie dam1S G i si˛eRFX:ACC omami´cINF , szybko wyrzuciłem: CZ Já jsem se však pevnˇe rozhodl, že seRFX:ACC víckrát nedámNEG:1S G omámitINF , a rychle jsem vyrazil: ‘But having made the decision that I would not let myself be deluded anymore, I quickly said:’
The most frequent translation of permissive da(va)t in Polish and Russian was nech(áv)at with 18 cases altogether (399), followed by free translations in general, especially using modal verbs (9 cases) as in (400): (399)
(400)
PL Wypu´sciłPS T :3S G mnieACC pierwszego z windy. CZ NechalPS T :3S G mˇeACC vystoupitINF z výtahu prvního. RU On dalPS T :S G mneDAT pervomu vyjtiINF iz lifta. ‘He let me exit the lift first.’ a.
b.
i Stanisław
ii
Ale nie dam1S G podej´sc´ INF nikomuDAT do rakiety w zwykłym iii skafandrze — odezwał si˛e Gosse. ‘But I will not let1 anybody3 approach2 the rocket in a simple space suit - said Gosse.’ Ale v obyˇcejném skafandru k raketˇe nikdoNOM nesmí3S G ," prohlásil Gosse. ‘But nobody is allowed to (approach) the rocket in a simple space suit, said Gosse’
Lem, Pami˛etnik znaleziony w wannie. znaleziony w wannie. iii Stanisław Lem, Fiasko
ii Stanisław
Lem, Pami˛etnik
238 Czech, Polish and Russian in parallel Dovolit ‘allow’, unlike its cognates in Polish and Russian, was used only 5 times to translate da(va)t; in all cases it was used with a complement clause, rather than an infinitive. The parallel corpus very clearly shows the marginal status of permissive dá(va)t as well as the role of nech(áv)at as a prominent exponent of permissive causation that competes with and at least partially outcompetes da(va)t.
5.1.2.
The factitive
Things are very different when we turn to the factitive. This function is nonexistent in Russian, marginal in Polish and central in Czech. A single unambiguously factitive instance was found in Polish, translated not by dá(va)t, but by nech(áv)at, the second causative auxiliary, into Czech: (401)
PL Bogaci starcy dawaliPS T :3PL si˛eRFX:ACC pochowa´cINF w płyni nym azocie z nadzieja,˛ z˙ e b˛eda˛ kiedy´s wskrzeszeni. CZ Bohatí starci seRFX:ACC nechávaliPS T :3PL pochovatINF v kapalném dusíku v nadˇeji, že budou nˇekdy oživeni. ‘Old rich people had themselves buried in liquid nitrogen in the hope that they would be resurrected in the future.’
With respect to the factitive function of da(va)t in Czech, the parallel corpus data provides interesting additional insights: it was often added in translation in order to make non-explicit causation in the other languages explicit. Dá(va)t in curative function appears 16 times in the Czech part of the corpus, never as a translation of da(va)t+inf in other languages. In 14 cases, a contextual causative was used in Polish, Russian, or both. Cf. the next two examples with contextual causatives in Polish and Russian: (402)
RU Ja prizvalPS T :S G tebja, cˇ toby ty pokazal mne nož, kotoryj byl u ii tebja. CZ DalPS T :S G jsemAUX:1S G tˇe zavolatINF , abys mi ukázal sv˚uj n˚už. ‘I called you (RU) / had you called (CZ) in order to ask you to show me your knife.’
i Stanisław
Lem, Fiasko
ii Mixail
Bulgakov, Master i Margarita
Introduction
(403)
239
PL [. . . ] obelisk, który dziad mój, tak˙ze astronom, wzniósłPS T :3S G i ku czci Einsteina. CZ [. . . ] obelisk, který dalPS T :3S G postavitINF na poˇcest Einsteina m˚uj dˇed, rovnˇež astronom. ‘[. . . ] an obelisk, which my grandfather, also an astronomer, erected (PL) / had erected (by someone; CZ) in honour of Einstein’
The second most frequent type of equivalence with 4 of 16 involve pleosemantic verbs in Russian and Polish, in particular speech act verbs such as prikazyvat’ ‘order’, prosit’ ‘ask’ and the more general kaza´c ‘command, have sth. done’ in Polish, as in (404): (404)
PL Jaki´s madry ˛ in˙zynier kazałPS T :3S G pomalowa´cINF kadłub i ii skrzydła na jasnobrazowy ˛ kolor [. . . ]. CZ Nˇejaký moudrý inženýr dalPS T :3S G natˇrítINF trup i kˇrídla svˇetlehnˇedou barvou. RU Kakoj-to umnik-inžener velelPS T :3S G okrasit’INF korpus i kryl’ja samoleta v svetlo-koriˇcnevyj cvet [. . . ]. ‘Some clever engineer had the body and the wings painted in a light brown color.’
It is significant that only two of those contextual causatives involved reflexive; in much of the literature, perhaps going back to Issaˇcenko, the contextual expression of causation is linked to the reflexive (see also Geniušien˙e 1987), as in on postrigsja ‘he had his hair cut’. These data clearly show that contextual causatives in Russian or Polish do not necessarily or prototypically include the reflexive. The data point to a reduced potential of contextual causatives in Czech, since translators evidently feel the need to introduce an explicit analytical causative where Russian and Polish do not mark causation.
5.1.3.
The modal passive
The modal passive was investigated in the trilingual corpus. As may be expected, the Czech formant was much more frequent with 255 instances than its Polish cognate with 123 instances. The degree of overlap was surprisingly low: only 34 cases, that is, about 28% of the Polish and about 13% of the i Stanisław
Lem, Astronauci
ii Stanisław
Lem, Astronauci
240 Czech, Polish and Russian in parallel Czech examples have the cognate modal passive as an equivalent in the other language. An analysis of the formants used in translation shows that Polish dá´c si˛e corresponded to Czech modal constructions in 92 of 123 cases, amounting to about 75%. 60 In a similar way, Czech dát se corresponds to Polish modal constructions in 180 of 255, i.e., 71%.61 The range of translational equivalents in the modal domain of both languages is quite wide. The translational equivalents in Russian suggest that the two constructions differ in their centrality. In the case of Polish da´c si˛e, the Russian version included modal constructions in 82 of 123 cases, equivalent to 67%; most prominent were the central modals možno/nel’zja ‘it’s possible/impossible’ with 23%. Czech dát se corresponded to modal constructions in Russian in 182 cases (=71%). This concerned the central modals možno/nel’zja to a larger degree, namely in 92 of 255 cases (37%). It seems that these data reflect that the Czech modal passive is more central than its Polish cognate, not only because of its higher frequency, but also in view of the high proportion of central modals as translational equivalents in Russian. A full analysis of translational equivalents will not be attempted here, as it would need a more balanced and larger corpus as well as a more detailed discussion. However, we see that these formants in both languages relate to modal formants to about two thirds of the examined cases in respect to one another as well as to Russian. This pattern of modal equivalents does not extend to modal passive constructions with inattentive perception verbs in Polish. such as dostrzec ‘see’, wy-/odczu´c ‘feel’ and słysze´c ‘hear’. Only daˇc si˛e słysze´c will be briefly discussed here, as only this lexical verb was found in sufficient number in the Polish part of the corpus. This type has already been mentioned in section (370) on page 220 above and its diachronic development will be discussed in section 6.5.3 from page 272 below. We have seen that this type does not have a strict counterpart in Czech, where perception verbs are not especially frequent as complements of da(va)t; the parallel corpus data additionally shows that it is a distinct type in relation to its translation into both Czech and Russian. Da´c si˛e słysze´c was the most frequent lexical verb in Polish with 18 occurrences (followed by zrobi´c with n=8). It was not translated by analytical modal constructions; rather, in 9 out of 18 occurrences the corresponding Czech text contained the reflexive verb ozvat se ’be heard, resound’ and in
Introduction
241
5 cases the lexicalized být slyšet ‘be audible, lit. be to hear’. In Russian, da´c słyšet also had equivalents not conforming to the general pattern: the Russian version of these 18 occurrences included 12 cases of the middle verb poslyšat’sja ‘to be audible’, (405)
a. b. c.
Z gromadki dałPS T :3S G si˛eRFX:ACC słysze´cINF radosny pomruk. Z hlouˇcku seRFX:ACC ozvalPS T :3S G radostný šum. PoslyšalsjaPS T :3S G radostnyj šepot. (Stanisław Lem, Astronauci) ‘A happy murmur was audible from the group.’
The only firm conclusion at this point is the confirmation of the finding that, first, the Czech formant is more frequent and, tentatively, its more central status is seen in the translation with more central modals in Russian as well as Polish; second, that modal passives with inattentive perception verbs, which are used in Polish, but not Czech, form a special group among modal passives also in respect to their translation into the other two languages.
5.1.4.
Conclusions
This short overview of the parallel corpus study mainly confirms the findings concerning different functional load of da(va)t+inf in the compared language; the corpus was highly useful in the initial, hypothesis building phase of this study. Moreover, it offered a window into the onomasiological fields that are formed by the equivalents of da(va)t+inf in the languages under comparison. In Polish and Russian, this concerns most of all pozwoli´c and pozvolit’ which compete with da(va)t mainly in those domains where it is less grammaticalized: the reflexive permissive in Russian and the non-reflexive permissive in Polish. For Czech, the results emphasize the role of factitive causation marking da(va)t+inf fulfills in this language and not paralleled by formants in Russian and Polish that either resort to implicit marking in contextual causatives or use unambiguous pleosemantic factitive lexemes. The data for the modal passive is quite complex; the field of equivalents in all three languages covers a broad range of modal formants that cannot be covered in detatil here. However, it is conclusive to see that da´c si˛e słysze´c, and combinations with other perception verbs not explicitly treated here, differ from other verbs inasmuch as they are typically translated with lexicalized equivalents.
Chapter 6 da(va)t+inf in Old Church Slavonic and earlier stages of Polish and Czech
6.1.
Overview
In this diachronic chapter, I concentrate on specific questions involving those characteristics that differ between Russian, Polish and Czech. Issues that are relevant here are • Since when, and with what characteristics, is causative meaning with da(va)t attested in Slavic? When, and under which circumstances does factitive and permissive causation arise and when does it disappear? • When does the reflexive permissive grammaticalize in Czech and Polish? • What can the data tell us about the loss of the dative (Czech), and the evolution of the type II causee (in Polish and Czech)? • When, and in which form, do the modal passive constructions evolve and how do they grammaticalize? The section is organized in three sections. First, I establish which functions need to be considered inherited, proceeding from the state in Old Church Slavonic. I then provide a short overview of the hypothesized development in Russian; this is not analyzed in depth. Next, I describe two corpus-based investigations into the changes in the syntax and semantics of da(va)t+inf in Czech and Polish and try to establish their chronology. Finally I summarize.
6.2. da(va)t+inf in Old Curch Slavonic In order to establish the functions of da(va)t+inf in OCS, I examined dictionaries covering the earliest period as well as some of their published sources. The permissive construction formed from da(va)t and an infinitive or a complement clause is found already in the earliest texts. According to the Slovník Jazyka staroslovˇenského and Cejtlin et al. (1994), OCS da(ja)ti
244 da(va)t+inf in OCS and earlier stages of Polish and Czech ‘give’ in this construction translates the following Greek verbs (translations following Liddell & Scott 1940): • ἀφιέναι ‘to send off, discharge’, with infinitive: ‘suffer, permit’ • ἐᾶν with infinitive ‘let, suffer, allow, permit’; negated ‘not to suffer, forbid, prevent’ • συγχωρεῖν ‘come together, meet’; with infinitive: ‘allow, concede, grant’ • some other verbs such as παραχωρεῖν, ἐθέλειν All these verbs denote various shades of permissive causation. In the translation of Luke 4:41 shown under (406), dati translates ἐᾶν; in (407) from the Codex Suprasliensis dati translates συγχωρεῖν: (406)
a. b. c. d.
(407)
a. b.
i zaprˇeštaj˛e ne daˇešeI MPF:3S G ımъDAT g(lago)latıINF . eˇ ko vˇedˇeaxo˛ x(rist)a samogo so˛šta. (Luke 4:41 Mar; dadˇeaše Zogr) καὶ ἐπιτιμῶν οὐκ εἴαI MPF:3S G αὐτὰACC λαλεῖνINF , ὅτι ᾔδεισαν τὸν Χριστὸν αὐτὸν εἶναι. (Koine) et increpans non sinebatPS T :3S G eaACC loquiINF quia sciebant ipsum esse Christum. (Vulgata) But he rebuked them and would not allow them to speak, because they knew he was the Christ. (New Int. Version) [. . . ] i ne dajo˛štouPRS :PTCP:DAT približitiINF s˛e imi· white white white white white (Supr 184, 11) [. . . ] καὶ μὴ συγχωροῦνταPRS :PTCP:ACC ἐγγίσαιINF αὐτοῖς. ‘[. . . ] not let them go near the beasts.’
Both translations are from before the 11th century. No factitive use is referenced in the dictionaries; no instance of a reflexive permissive is cited. Slovník Jazyka staroslovˇenského notes that dati in this function is used predominantly negated; recall that this is still the case in Russian today. Examples of permissive dati involving animate causer and causee are numerous and translated from several verbs, which allows us to conclude that we are dealing with an established construction. In one of the citations found in the Codex Suprasliensis, the causee is inanimate. The narrative goes that an angel rescues the two martyrs Paul and Julian from the fire they were thrown into by removing it:
da(va)t+inf in Old Curch Slavonic
(408)
a. b.
245
ne dastъAOR:3S G ognjuDAT sъto˛žitiINF ima (Supr 5, 25; 67) και ὀυκ εἴασενAOR:3S G τὸ πῦρACC [. . . ] παρενοχλῆσαιINF αὐτοῖς. ‘he did not let the fire harm them’
This might be an indication of an advanced stage of semantic bleaching, where the causee of dati is not necessarily animate, which it must have been in the very earliest stages of the development from ‘give’ to ‘let’. However, this evidence is hardly conclusive since we are dealing with very close translations that do not necessarily reflect actual usage. Examination of the occurrences of the Greek source verbs in the 11th century Codex Marianus (Jagi´c 1883) and Codex Suprasliensis (Zaimov & Kapaldo 1982) as well as the 14th century translation of the Exposition of John of Damascus (Weiher 1987) showed that these verbs were translated not only by dati, but also by other verbs including ostaviti ‘leave’, cf. (409) where ostavljajuštou translates ἐῶντος. (409)
i ne ostavljajuštouPRS :PTCP:DAT podatiINF nam svˇet. white white white white white (John of Damascus, 175 b5) ‘. . . and (the body of the moon) doesn’t let (the sun) give us light . . . ’
Ostaviti, as well as pustiti ‘let off, discharge’ are more central permissive causative in Bosnian/Croatian/Serbian than their cognates are in the North Slavic languages today. A derivative of pustiti was, however, found only once in the translations of the above Greek verbs in the texts surveyed. While a thorough treatment of the onomasiological field of permissive causation in OCS remains outside the scope of this study, I conclude from these findings that dati was, among other verbs, a regular member of the class of verbs expressing such causation. In view of the importance of translated texts in the Slavic traditions, we will now turn to a small excursus on permissive use of the Greek (and Latin) verbs as possible influences on the use of da(va)t. Permissive give in Latin and Greek In two cases, OCS dati translates Greek διδόναι ‘give’ in permissive reading: ˇ (410) a. Eko ne ostaviši dšj˛e moєj˛e vъ adˇe: ni dası2S G prˇepodobъnomouDAT svoєmou vıdˇetıINF ıstъlˇenьiˇe. white white white white white (Psalt. Sin., Ps. 15,10)
246 da(va)t+inf in OCS and earlier stages of Polish and Czech b. c.
VΟτι οὐκ ἐγκαταλείψεις τὴν ψυχήν μου εἰς ᾅδην, οὐδὲ δώσειςFUT :2S G τὸν ὅσιόνACC σου ἰδεῖνINF διαφθοράν. . . . , because you will not abandon me to the realm of the dead, nor will you let your faithful one see decay. (New Int. Version)
Here, the Greek source reflects an analogous Hebrew source with permissive give62 ; for the next example, this is not obvious: (411)
a. b.
voini tˇelo davъšePS T :PTCP:NOM:PL oukrastiINF . (Supr 441,6) οἱ στρατιῶται τὸ σῶμα δόντεςPS T :PTCP:NOM:PL κλαπῆναιINF ‘[. . . ] the soldiers, having let the body be stolen, [. . . ]’
However, according to the standard dictionaries covering the classics, in both Latin and Greek, ‘give’ is not normally used with proper permissive reading like in the above examples. Bauer & Aland (1988: 388) cite (410) above from the Psalter above merely as an example involving an a.c.i. construction for a subtype of the sense ‘grant’. This latter sense is normally used with a finite complement clause and present already in the pre-Christian period in prayers. It is reflected in many other European languages as well and can be accounted for by a conceptualization where what man does is in fact given to him by divine authority, cf. Mark 10:37: (412)
a.
b. c.
daždьI MP:2S G namaDAT daCOMP edinъ o desno˛jo˛ tebe · i edinъ o lˇevo˛jo˛ tebe · s˛edevˇe1DU kъ slavˇe tvoeı white white white white white (Zogr Mar As Sav Ostr) δὸς2S G ἡμῖνDAT ἵναCOMP εἷς σου ἐκ δεξιῶν καὶ εἷς ἐξ ἀριστερῶν καθίσωμενAOR:COND:1PL ἐν τῇ δόξῃ σου. (Koine) daI MP:2S G nobisDAT utCOMP unus ad dexteram tuam et alius ad sinistram tuam sedeamusCOND:1PL in gloria tua. (Vulgata) ‘Let one of us sit at your right and the other at your left in your glory.’ (New Int. Version)
In Latin, I have not found any examples of permissive usage; merely a sense ‘grant’ in poetic usage is listed under dare in Lewis & Short (1879)63 . The following example is given there: (413)
a.
DaI MP:2S G mihiDAT perpetua, genitor carissime, dixit, virginitate fruiINF : deditPRF:3S G hoc pater ante Dianae. ‘My dearest father let me live a virgin always, for remember Jove did grant it to Diana at her birth.’ (Ovid, Daphne)
A further use cited in Lewis & Short (1879) involves modal meaning:
da(va)t+inf in Old Curch Slavonic
247
Datur, it is permitted, allowed, granted; with subj. clause: quaesitis diu terris, ubi sistere detur, Ov. M. 1, 307: interim tamen recedere sensim datur, Quint. 11, 3, 127 : ex quo intellegi datur, etc., Lact. 5, 20, 11. (Lewis & Short 1879)
Judging by the quotations and a simple Google search, the third person passive datur with perception and cognitive verbs such as intellegi datur, audire datur is found abundantly in post-classical Latin texts. An example is given below: (414)
[. . . ] ex quo intellegi datur, quanta illa fuerint quae studio propter horrorem repressa sunt [. . . ] Orosius: Historiarum Adversum Paganos (5. century A.D.)
We are dealing with a dynamic modality: intellegi datur means ’it can be understood’, not ’it is allowed to understand’. The influence of Latin for Polish structures such as da´c si˛e widzie´c, da´c zna´c, da´c do wiedzenia can thus not be discarded. According to my sighting of OCS dictionaries, this usage is, in any case, not attested in the earliest Slavonic texts.
Conclusions A survey of the use of da(va)t in OCS has shown that permissive use is found in the earliest texts and, as regards permissive meaning proper, cannot be considered a calque. In neither Latin nor Greek did the use of ‘give’ regularly extend into permissive or factitive causation proper, so an influence of the classic literature is not plausible beyond specific subtypes (see below). Dati serves to translate several Greek verbs, which in turn have also other equivalents in OCS; this shows that it must have been a regular member of the onomasiological field of permissive causation. Since the permissive use of da(va)t+inf is furthermore attested in all Slavic languages, but not, e.g., in Baltic, it is a safe assumption that it is inherited at least from Proto-Slavic and is perhaps a Slavic innovation. From this survey I also conclude that the tendency to be expressed under negation in the modern languages should be considered a conservative trait rather than an innovation, since it seems to have been the case already in OCS; secondly, that the same is plausible for the function of da(va)t to denote manipulative causation with inanimate causees, as found in early Czech, Polish and modern Russian but lost in modern Czech and Polish. I take factitive causation to be a later innovation since it is not found in OCS and does not extend to Russian.
248 da(va)t+inf in OCS and earlier stages of Polish and Czech Only in the abstract sense of ‘grant’ does da(va)t+inf correspond to Greek διδόναι or Latin dare. It is instructive to see that in both Latin and Greek, some meaning shift towards a causative is thus observed. The usage of ‘give’ in the sense of ‘grant’, which can be observed also in English, German or French, may well be influenced by translated biblical texts; at the same time, this extension seems to be very elementary and may well have evolved independently. No modal meaning of reflexive or passive dati, nor causation with cognitive verbs is attested in OCS. It is, however, seen in post-classical Latin, which might have played a role in these developments in the West Slavic languages. Such usage is bound to have exerted some influence on Polish and Czech in later developments.
6.3.
A side glance at da(va)t+inf in Russian
Russian seems to have changed little from the state attested in OCS, the notable exception being the rise of the various extended functions of the imperative. Two small corpus studies were conducted: Nestor’s chronicle, originally compiled in Kiev 1113, in the edition of O.V.Toporov64 as well as the 18th century part of the RNC were systematically annotated. Besides that, other texts of various ages were searched for uses of the imperative and reflexive permissive. These results will not be presented here in detail, most importantly, because they do not provide any new insights and at the same time must be considered incomplete. Six examples of permissive da(va)t were found in Nestor’s Chronicle, none of which was reflexive and all of which were negated. In the 18th century corpus, 83 permissive uses were found of which 40 were negated, 11 contained imperatives and only 6 were reflexive; no factitive instances were found. No clues of significant change in terms of those characteristics of da(va)t that set Russian apart from Czech and Polish were thus found. The remaining issue would therefore be the development of the constructions involving the imperatives daj(te) and davaj(te); here, evidence is hard to come by. Searches in the Novgorod Birchbark parchments, as well as in early texts that are known to be influenced by the spoken register, such as Aavakum’s autobiography, did not yield any uses of the grammaticalized imperative. This function certainly first developed in the spoken languages, and
A side glance at da(va)t+inf in Russian
249
therefore it is plausible that textual evidence is of a time when it had already grammaticalized to a large degree (but see Barentsen’s (2003) work on the development of the hortative in the 18th to 20th century). It should be noted, however, that, generally speaking, the hortative construction is more plausibly analyzed as a reinforcement of the hortative function than as deriving from the permissive. To see that, note that there are no cases of extensional overlap between the permissive and the hortative: in contrast to the first person imperative, where causer and causee may be straightforwardly mapped to adressee and performer of the action, the difference between permissive and hortative as in davajte mne pet’ ‘let me sing’ and davajte pet’ ‘let’s sing’ hinges on the set of referents of the second person: inclusive and exclusive reading do not permit an overlap. Rather, it seems plausible that davaj(te) had already acquired an autonomous incitative reading with only loose connection to its syntactic environment prior to the emergence of the hortative construction. According to such a scenario, it would then have combined with the second person imperative or indicative in imperative function (Spoem(te)! Budem(te) pet’! ‘Let’s sing’) in the function of a reinforcing marker that later developed to predominately signal the meaning it initially supported. This is known as reinforcement (Lehmann 20022 : 20); the best-known example for such developments is Jespersen’s cycle of negation, e.g. in the development of French pas ‘step’. Such an hypothesis of development presupposes the autonomous function coming chronologically before the hortative function. There is little chance of falsifying this hypothesis, however, since convincing textual evidence would need to reflect spoken language. Some further issues would need to be investigated; the synchronic state, in any case, strongly suggests that davaj(te) pet’ ‘let’s sing’ has developed from davaj(te) budem pet’ ‘let’s we’ll sing’ with auxiliary and perhaps even davaj(te) budem(te) pet’ with imperative auxiliary, rather than that it is in any way connected to davaj(te) nam pet’ ‘let us sing’. The specifics of the transition of give! to an incitative marker and ultimately to the hortative is thus left for further research; none of these issues will be taken up in this study again.
250 da(va)t+inf in OCS and earlier stages of Polish and Czech 6.4. da(va)t+inf in Czech ˇ This section relies on the diachronic part of the CNK as of autumn 2007. The make-up of the corpus is given in table 42. It is composed of roughly equal parts by texts from the 14th to the 16th century and texts from the 18th and 19th centuries. All potential cases of dát with infinitive following or preceding were extracted. Since the corpus is not lemmatized and since spelling conventions in such texts need to be considered unstable, this was accomplished by querying for all words beginning with (ne)da, (ne)dá or (ne)dej preceded or followed within 4 tokens by a word ending in -c, -ci, -t or -ti. The results of these queries were input into a database and the lists of candidate words that corresponded to this query for forms of dát and infinitives were manually checked for plausible forms. This eliminated 1300 false positives. The remaining 997 potentially relevant hits were each inspected and categorized with the same tags as the synchronic data, but with some additional, heuristic tags. Tables 43 and 44 provide a detailed overview of causative types as found in the corpus. Table 45 summarizes this data irrespective of differences pertaining to animacy or form of the causee and adds the number of modal passive cases found. While the corpus is obviously very small and merits only tentative conclusions, it shows that the modal passive is initially used only much less frequent than the causative and becomes gradually more frequent; this fits well to the fact that Gebauers Staroˇceský slovník contains examples only of the causative type, not of modal passive usages. The following discussion first deals with the causative types, then focus on the emergence of type II causees before turning to the modal passive.
6.4.1.
Permissive and factitive
Permissive The non-reflexive permissive is, as may be expected, present already in the earliest texts. The tendencies are familiar from the synchronic analysis: most non-reflexive examples are from the interpersonal domain, they are often negated and usually contain a dative causee, as in (415):
da(va)t+inf in Czech cent. 14th. (34%)
15th. (4%) 16th. (14%)
18th. (20%)
19th. (28%)
all
date 14th c., sec. half 14th c., end 1389-1401 ca. 1400 mid 15th 1491-92 1532 1552 1558 1580 1580 1581 1585 1705 1732 1736 1738 1768 1775-1820 1792 1803 1828 1832-33 1833 1833 1834 1834-35 1834-35 1835 1835 1835 1836 1869 1890
title Pasionál muzejní (Muz III D 44) (R) Pˇreklad prorok˚u Izaiáše ˇ ci besední (Budyšínský rkp. 20 56) Reˇ Pulkavova Kronika král˚u cˇ eských Hvˇezdáˇrství krále Jana (R) ˇ Cesta z Cech do Jeruzaléma a Egypta (KapPraž O 35) (R) O krvi pouštˇení žilami Krátkej spis o morové nemoci Knížka o štˇepování rozkošných zahrad. Nové praktiky dvˇe Služba kˇrtu svatého Snáˇr Obrácení pohan˚uv v Jáponˇe Knihy hospodáˇrské hospodáˇrství polního Harfa nová na hoˇre Sion znˇející Jerusalem nova Desatero pˇripíjení mládenecké Historie pobožná a velmi pˇríkladná Píseˇn nová aneb ˇremeslu mlynáˇrskému... ˇ ˇ Ceské amazonky aneb dˇevˇcí boj v Cechách pod zprávou rekynˇe Vlasty. Dobrá rada v potˇrebˇe Luˇcba cˇ ili chemie zkusná Klášter sázavský (R) Návrat (R) Pout’ krkonošská (R) Rozbroj svˇet˚u Kˇrivoklát (R) Obrazy ze života mého Cikáni (R) Deník na cestˇe do Itálie (R) Deník z roku 1835 (R) Valdice (R) ˇ Ceský študent Filosofská historie
251 size 15 9661 74 230 56 381 70 227 28 328 17 274 4 034 12 862 10 335 10 811 1 581 61 211 40 754 143 886 24 262 7 103 1 002 11 632 376 20 167
32 556 67 653 1 220 1 089 2 743 853 12 448 4 261 28 411 4 372 2 970 1 791 100 375 30 337 1 047 196
ˇ Table 42. Texts in the diachronic part of CNK as of autumn 2007, with size and number of examples extracted per text. See http://ucnk.ff.cuni.cz/diakorp-texty.html for more details.
252 da(va)t+inf in OCS and earlier stages of Polish and Czech
century 14th (n=51)
type non-refl. perm.
refl. perm.
15th (n=15)
16th (n=79)
ambiguous non-refl. perm. refl. perm. factitive
ambiguous non-refl. perm. refl. perm.
factitive
ambiguous
reflexive – – – – acc. acc. acc. dative dative other other other other – – dative – acc. acc. acc. – – – acc. acc. acc. acc. other – acc. dative dative other – – acc. other
causee? none dative dative dative none dative dative none dative none none dative none none dative none none none dative none none dative dative none dative dative type II none none none none (loc) dative none none none none
inanim. causer?
inanim. causee?
yes yes
yes
yes yes
yes
yes yes
yes yes
yes
n= 3 11 5 1 9 6 6 2 1 1 2 2 2 3 5 1 2 1 1 2 1 1 3 3 2 1 1 4 29 19 4 2 1 1 1 5 1
subgroup
20
29 2 8 1
4 2
5
11
55
8
Table 43. Czech: causative types in 14th to 16th century texts. The data are grouped into four major types: permissive, reflexive permissives, factitive and ambiguous and contain detailed information on presence, form and animacy of the causee as well as animacy of the causer. Other as a reflexive type includes instrumental, prepositional phrase and reflexive possessive pronouns.
da(va)t+inf in Czech
century 18th (n=63)
type non-refl. perm. refl. perm.
factitive
ambiguous
19th (n=74)
non-refl. perm. refl. perm.
factitive
ambiguous
reflexive – – – acc. acc. acc. dative other – – dative other – acc. acc. – – – acc. dative other – – – acc. acc. dative dative – – – – acc. dative
causee? none dative dative none dative type II none none none type II none none dative none none none dative dative none none none none none type II none type II none none none dative dative acc. none none
inanim. causer?
inanim. causee?
yes
yes
yes
yes
yes yes
yes yes yes
n= 1 8 3 5 2 1 5 1 26 4 2 1 1 2 1 1 2 2 9 4 2 35 2 1 4 1 3 1 1 1 1 1 1 1
253
subgroup
Table 44. Czech: causative types in 18th and 19th century texts. See table 43.
12
14
33
4
6
15
47
6
254 da(va)t+inf in OCS and earlier stages of Polish and Czech century tokens query results non-refl. permissive refl. permissive factitive ambiguous all causative modal passive
14 360 499 890 20 29 0 2 51 -
15 45 602 140 8 1 4 2 15 3
16 141 588 311 5 11 55 8 79 2
18 208 428 466 12 14 33 4 63 33
19 291 079 460 6 15 47 6 74 64
Table 45. Overview of text size, query results, causative types and modal passive per century.
[. . . ] poslal vojsko k Tepléj jemu vstˇrieci, aby jemuDAT vtrhnútiINF i nedaliNEG:PS T :3PL . ‘[. . . ] he sent the army to Teplá towards him, so that that they [=the army] would not let2 him1 invade3 ’
(415)
Both manipulative and impersonal causation are likewise attested in the 14th century: (416)
[. . . ] proˇc by jie druhého poklínanie nezbavil, toˇcíš proˇc by byl jejie svatému tˇeluDAT v zemi hnítiINF dalPS T :3S G . (14th c., sec. half)i ‘[. . . ] why would he not forgive her the second curse, so why would he let3 her holy body1 rot2 in the earth.’
(417)
[. . . ]alé mé svˇedomie nedáNEG:3S G miDAT tomu vˇerˇ itiINF . white white white white white (14th c., sec. half)ii ‘[. . . ] but my conscience does not let me believe that’
Already at this stage the reflexive permissive is very frequent, with comparable frequency to the non-permissive. Both animate and inanimate dative causees are attested: Nade vším svˇetem vítˇezˇe, i dáš2S G sˇeRFX:ACC jednomu popuDAT pˇreˇ mociINF . (14th c., sec. half)iii ‘You triumph over the whole world, but you let1 yourself 2 be overcome4 by a priest3 .’
(418)
i ca.
1400/ Pulkavova Kronika král˚u cˇ eských muzejní
iii Pasionál
i Pasionál
muzejní
ii Pasionál
muzejní
da(va)t+inf in Czech
255
(419)
Dˇréve sem pravil, že svú volí padnúti v hˇriech móžem a svým netbáním dáme1PL sˇeRFX:ACC uvázatiINF hˇriechuDAT ; (1389-1401)iv ‘Earlier I said that we can fall into sin by our own will, and by our negligence we let ourselves be bound by sin’
(420)
Co jest tˇem pak ˇréci, ješto sˇeRFX tak dadie3PL hnˇevuDAT oslepitiINF , že [. . . ] (1389-1401)i ‘But what should one tell those who let2 themselves1 be so blinded4 by anger3 , that [. . . ]’
(421)
[. . . ] i budeš múdrým tak slúti, když mladým vichrómDAT v uši sobˇeRFX:DAT nedášNEG:2S G dútiINF . (ca. 1400)ii ‘[. . . ] you will be known as a wise man, if you do not let3 young storms1 (= young people65 ) blow4 (lit. yourself 2 ) into your ears’
Recall that looser selectional restrictions concerning non-agentive causees in reflexive permissives were taken to be indicative of a higher degree of grammaticalization of this structure in Polish in comparison to Russian. The two main characteristics of this difference – higher frequency of the reflexive permissive and neutralization of the causee – seem to obtain already at this early point. The overview of these types given in table 45 as well as the more detailed figures in table 43 and 44 show that reflexive permissives are very frequent from the earliest text beginning on, with the exception in the 15th century where absolute numbers are very low due to the small textual basis; this is in contrast to early East Slavic texts, where the small sample adduced above contained mostly non-reflexive permissives. Concerning the qualitative restrictions in Russian discussed on 136f., we find attestations with nonagentive causees, such as hˇriech ‘anger’ in (419) above that are impossible in Russian already in the oldest Czech texts. We will see that the diachronic picture in Polish is similar in this respect.
iv Tomáš ii Pulkavova
ˇ ci besední Štítný ze Štítného: Reˇ Kronika král˚u cˇ eských
i Tomáš
ˇ ci besední Štítný ze Štítného: Reˇ
256 da(va)t+inf in OCS and earlier stages of Polish and Czech Factitive The oldest clearly factitive example in the corpus is the following. (422)
O té bránˇe slyšal sem, že by ji dala3S G jedna židovka ustavˇetiINF na svój náklad, a to jedne proto, aby pamˇet jejie zuostala po nie, neb tu brány na tom miestˇe potˇrebie nic nebylo jest [. . . ]. (1491-92)i ‘About this gate I have heard that one jewess had it built at her own expense, only so that her memory should stay behind her, since a gate at this place was not of any use [. . . ].’
It represents curative meaning and is constructed without causee. The profile of the curative type - usually causeeless, often with a reflexive – as it represents the most frequent causative construction formed with dá(va)t today is the same already in these earliest attestations. It is the most frequent use of causative dá(va)t in the corpus from the 16th century on, that is, already from the Humanist period of Middle Czech; of 55 uses in the 16th century, only 3 involve an explicit causee. This use is distributed across 5 of 7 texts in the 16th century and we can therefore assume that is not specific to a certain author or a genre. Let us turn to the question of dative causees with curative causation. The first such example is found in the same source as the first unambiguous causeeless factitive cited above from 1491. It is necessary to cite more context to justify a factitive interpretation: (423)
Potom se toho dovˇedˇel sám pán jeruzalimský i obeslal nás, abychom pˇred ním stáli, neb se nˇekterým židóm nás zslitovalo, neb rozumˇeli, že se nám kˇrivda stala, i šli s námi pˇred pána. Pán seRFX:ACC námDAT dal3S G pˇred sebú zpravitiINF a kázal nám o té všie vˇeci povˇedˇeti, i pravili sme jemu zkrze tlumaˇce, kterak jest nám ten zlý cˇ lovˇek uˇcinil neprávie. (1491-92)ii ‘Afterwards the lord of Jerusalem himself heard about this and summoned us, so that we would stand before him, because some Jews had found pity on us, because they understood that we had been wronged, and they went with us to the lord. The lord had3 us2 justify4 ourselves1 before him and commanded us to tell him about the whole thing, and we did that through a translator, what injustice that evil person did to us.’
i Martin Kabátník: Cesta z Cech ˇ do Jeruzaléma a Egypta ˇ Cech do Jeruzaléma a Egypta
ii Martin
Kabátník: Cesta z
da(va)t+inf in Czech
257
The context makes it clear that the narrator is summoned and questioned; however, doubt remains and a permissive interpretation cannot be excluded. In a similar vein, the other putative example with dative causee found in the corpus, from the 16th century, may be either interpreted as giving in the donatory sense or as factitive curative: (424)
Avšak oni pˇredce poˇctem se neumenšili, ale vždycky jich vícejí pˇribejvalo, dávajíceCV B kˇrtítiINF 66 své synyACC jednomu starému kˇrest’anuDAT [. . . ]. (1585)i ‘But they didn’t become less in number, rather more of them kept coming all the time, giving1 their sons3 to one old christian4 to baptize2 . . . ’ or, alternatively: ‘. . . having1 one old christian4 baptize2 their sons3 [. . . ]’
However, the causative interpretation seems much less likely here than the donatory interpretation. No later examples of dative causees with factitive dá(va)t were found in the corpus. In the synchronic part, we have seen that dative causees are rare today and that the factitive status of constructions formed with a dative causee is questionable (p. 212f. above). The diachronic data shows the same kind of ambiguity and limited frequency also in earlier stages. If with permissives the dative causee construction is clearly the starting point in a development leading to the current state where dative causees are obsolete, the same cannot be said of the factitive construction. Rather, this data suggests that the factitive use of dá(va)t became frequent without causee, that is, in essentially the same configuration as it is found today. This is an important finding that we will take up in the reconstruction of these developments below.
6.4.2.
The development of type II causees
Type II causees, that is, causees expressed as argument-adjunct in a form resembling the passive by-phrase, are today the only option in both the reflexive permissive and the factitive construction; dative causees are impossible in both constructions. The data show that this was rather different in the past: dative causees are found in abundance with both reflexive and non-reflexive permissives from early on, while they were basically never well attested with factitive causatives. We are thus dealing with converging developments in two initially different constructions. i Hostounský
Baltazar: Obrácení pohan˚uv v Jáponˇe
258 da(va)t+inf in OCS and earlier stages of Polish and Czech Type II and dative causees with reflexive permissives Generally, data on causee coding is rather sparse with reflexiv permissives. Overall causee mentioning drops from around half in the 14th to about a third in the 18th century to zero out of 15 cases in the 19th century. Causee mentioning in the earliest texts is in the dative, like with the non-reflexive permissive. Type II causees are attested from the 16th century on, cf. the following example with an inanimate causee: (425)
A protož, milí lidé, [. . . ] dejmeI MP:1PL seRFX:ACC aspoˇn jako kˇrest’ané dobrotouINS božskou pohnoutiINF , kteráž nás k sobˇe volá, ˇrka: (1580)i ‘And therefore, good people, [. . . ] let’s let1 ourselves2 at least as christians be moved4 by the goodness3 of God, which calls out to us, saying:’
Since the relative pronoun kteráž agrees with dobrotou božskou ‘goodness of god’, it is clear that it is, technically speaking, god’s goodness, not god himself who moves and calls out; in other words, the instrumental case of dobrotou božskou does not denote an instrument, but the causee / agent of the caused situation. However, less complex examples would clearly be desirable. No more data is available for the 16th century; in the 18th century, we find a single example in a song bock by the baroque exile writer Jan Liberda translated from German; judging by a possible source the reflexive permissive se zavát daly ‘let themselves be blown’ loosely translates a passive construction die getrieben ‘which were blown’ here: (426)
O, jak jsou blahoslavené duše s Kristem sjednocené, jenž odPREP lásky vˇetˇríˇckaGEN tak mocnˇe seRFX:ACC zavátINF dalyPS T :3PL , až též cele tam z˚ustaly, kdež jich magnet byt sv˚uj má. (1732)ii German: O wie seelig sind die Seelen, die mit Jesu sich vermählen! Die sein sanfter Liebeswind so gewaltiglich getrieben, daß sie ganz iii daselbst geblieben, wo sich ihr Magnet befind. ‘Oh, how blessed are the souls that are united with Christ, which let5 themselves3 be so mightily blown4 by1 the wind2 of love that they also came to stay the place where their magnet was to be.’
No further example was found in the 19th century. i Georg
ii Jan Liberda: Harfa nová na hoˇre Sion znˇ Ursinus: Nové praktiky dvˇe ející Psalter-Spiel Oder Sammlung, von Alten und Neuen auserlesenen Geistlichen Gesängen, Büdingen, fifth edition 1775, p. 672 (Google Books) iii Davidisches
da(va)t+inf in Czech
259
When, then, were dative causees with reflexive permissives replaced by type II causees? As the tables show, the latest attestations of the dative causee in the corpus are from the 18th century: (427)
Vˇernost pravá neustává, s hˇríchem stále bojuje, vˇetruDAT seRFX:ACC toˇcitINF nedává3S G , válˇcí až triumfuje. (1732)i ‘True faith does not tire, it fights with sin incessantly, it does not let4 itself 2 be swayed3 by the wind1 , it fights until it triumphs.’
A tentative conclusion in view of the sparseness of data is that overall causee mention seems to have gradually diminished. While type II causees with reflexive permissives have been in sporadic use since the 16th century, certainly initially only as a secondary option rather like in Polish today, dative causees went out of use no earlier than in the 18th century, probably later.
The emergence of type II coding with factitives More data is found concerning type II causees with factitive dá(va)t. Before discussing these, locative phrases in causee-less factitives that are attested from the 16th century, such as in the next example, should be mentioned. (428)
Po vyˇcištˇení zlého povˇetˇrí má se nˇecˇ ím vonným zakouˇriti, bud’to bílým kadidlem, myrrou, jalovcem a neb tímto prachem, který podlé toho latinského spisu m˚užeš sobˇeRFX:DAT dátiINF vPREP apatéceLOC pˇripravitiINF . (1552)ii ‘After the cleaning out of the bad air one needs to fumigate (the room) with something odorous, either with white incense, myrrh, juniper or with that powder that you can have2 (yourself 1 ) prepared5 in3 a pharmacy4 according to a Latin recipe’
Such cases are relevant to the question of type II causees since they provide a means of specifying the institution where the service referred to is provided; on page 211 I have cited such an example with modern Czech dá(va)t. Such locative phrases are typical for curatives and contextual causatives; in Russian and Polish, they are the preferred way to express the agent with contextual causatives (e.g. Russian podpisat’ cˇ to-to u kogo-to ‘have something signed by somebody’, lit. ‘at somebody’; see Weiss 1999). The example thus shows an alternative option to specify the causer typical for this type. i Jan Liberda: Harfa nová na hoˇre Sion znˇ ející nemoci
ii Jan
Voˇcehovský: Krátkej spis o morové
260 da(va)t+inf in OCS and earlier stages of Polish and Czech However, use of this option remained sporadic; rather, passive-like typ II causees become grammaticalized. The earliest attestations of such causees are found in the 18th century part of the corpus. The following example involves a type II causee with the preposition skrze lit. ‘through’, a regular passive by-phrase formant at the time. Three attestations with this preposition were found in the corpus, e.g.: (429)
Opatrní hospodáˇri [. . . ] zrna [. . . ] oddˇelují a nad to jistou cˇ ástku skrzePREP cˇ eled’ACC dávají3PL po zrnách pˇrebíratiINF , aspoˇn v té míˇre, aby se na budoucí cˇ asy dobrým semenem zaopatˇrili. (1705)i ‘Prudent farmers put aside some crop [. . . ] and have3 a certain part of that crop sorted4 grain for grain by1 the menials2 , at least so much that they will have enough good seed for the future.’
Two cases involved podle ‘by, according to’ in arguably the same function and one case involved instrumental case; cf. the examples below. No attestation of a causee introduced by od ‘from’ was found. ˇ (430) [. . . ] poslal svého ještˇe vˇerného k knˇežnˇe Castavˇ e a dalPS T :3S G jí podléPREP nˇehoGEN toto oznámitiINF : (1792)ii ˇ ‘. . . he sent his still faithful servant to the dutchess Castava and had1 the following announced4 to her by3 him4 :’ (431)
[. . . ] týmiž labutˇemiINS do Delf˚u seRFX:ACC dovéztiINF dalPS T :3S G . white white white white white (1869)iii ‘[. . . ] he had4 himself 2 carried3 to Delphi by these swans1 .’
Generally, the general pattern of mostly unmentioned causees and rare type II causees with factitives found in the synchronic analysis was already present in earlier stages. A difference is that there was more variation in the coding of the type II causees than there is today. While the individual characteristics of these coding options clearly merit more research, causative and passive byphrase coding changed in parallel over time. The nature of this connection is one of the issues to be considered in the concluding section. This will be discussed together with analogous findings concerning Polish.
i Kryštof Fišer: Knihy hospodáˇrské hospodáˇrství polního ˇ zonky aneb dˇevˇcí boj v Cechách pod zprávou rekynˇe Vlasty
ˇ Šedivý: Ceské amaštudent
ii Prokop iii Ceský ˇ
da(va)t+inf in Czech
type agreeing ambiguous intransitive sentential complement all
15th
16th
3
2
18th 16 17
3
2
33
261
19th 34 25 2 3 64
Table 46. Modal passive attestations in the diachronic Corpus of Czech by century.
6.4.3.
The modal passive
Table 45 shows that modal passives (not attested in, e.g., Gebauer’s Staroˇceský slovník) occur already in old Czech. Table 46 gives an overview. In most of these cases, including the earliest occurrences, the formant in question is dát; only one modal passive use of dávat was found in the 18th and the 19th centuries, respectively. No examples were found from the 14th century. All examples from the 15th century pertain to cognitive verbs (dá se znát, poznát and vˇedˇet) and it is unclear whether they should be counted as regular, modal passives; they may also be interpreted as causative constructions with a cognitive verb and inanimate subject: (432)
Pˇrirozenie cˇ lovˇecˇ ie nepromˇení sˇe, leˇc nemóž anebo jiný nedostatek co pˇreruší. A to sˇeRFX:ACC cˇ asto dá3S G vˇedˇetiINF . (mid 15th)i ‘Human nature does not change, if sickness or a different failing causes a problem. And this can be (lit. lets2 itself 1 ) often seen3 or this often makes itself noticeable’
(433)
A to sˇeRFX:ACC dá3S G na každém dietˇeti poznatiINF , pod kterú sˇe zvˇerocestí poˇcne neb urodí, té tváˇrnost pˇrijme [. . . ]. (mid 15th)ii ‘And that can (lit. lets2 itself 1 ) be recognized3 in every child, under which zodiac it begins or is born, that is the nature it will take on [. . . ].’ or ‘It shows (lit. makes2 itself 1 known3 ) on a child . . . ’
Note that the only examples of a modal passive-like construction which Jungmann˚uv slownjk from 1835 cites involve the verb znati in a similar use. It is from the 16th century that we find the first unequivocal examples of modal passives in the corpus: i Hvˇ ezdáˇrství
krále Jana
ii Hvˇ ezdáˇrství
krále Jana
262 da(va)t+inf in OCS and earlier stages of Polish and Czech (434)
Item vezmi galbanum a na šátek roztažené pˇrilož, jestli tuze pˇrilne a nedáNEG:3S G seRFX:ACC snadnˇe odtrhnoutiINF , jest zhojitedlný; pakli toho nic není, zlé znamení. (1552)i ‘So take galbanum and put it on a spread out cloth, if it stubbornly sticks there and cannot be easily ripped off (lit: does not let itself be ripped off), it is spoiled; it is good for nothing, a bad omen.’
This example is one of two in the 16th century and typical for the use of the modal passive also in the 18th century, where this construction is much more frequently attested. Examples from this time seem to deal with properties of concrete objects and involve transitive verbs with a concrete patient; dát se here denotes modality grounded in some property of this patient, promoted to subject position, much as in the Polish personal modal passive. Correspondingly, up to the first half of the 19th century, the construction is especially frequent in texts dealing with properties of objects (and plants): 29 cases were found in a book on agriculture from 1705, among them the only use of imperfective dávat se: (435)
Vytáhneš-li pak pˇred cˇ asem, nedáNEG:3S G seRFX:ACC pˇrístiINF a nesnadno seRFX:ACC trusky její dajíNEG:3PL vytˇrástiINF a oddˇelitiINF od jádra a niti budou plné plev, k tkání nepohodlné, a bude z ni velmi ostré plátno. (1705)ii ‘If you pull it out before its time, it cannot (lit: does not let1 itself 2 ) be woven3 and the spills cannot (lit.: do not let5 themselves4 ) be shaken6 off easily and separated7 from the meat and the threads will be full of husk [. . . ]’
(436)
Borovice, smrk, modˇrin [. . . ]jsou témˇeˇr stejní stromové 160. Milují hory. Dávají3PL seRFX:ACC tence štípatiINF a hodí se truhláˇru˚ m k všelikému [. . . ] táflování[.] (1705)iii ‘Pine-tree, spruce, larch . . . are practically the same trees. 160. They like mountains. They can (lit: let themselves) be cut thinly and are used by carpenters for all kinds of [. . . ] paneling.’
However, some doubt concerning this profile remains. While in the corpus only examples of the transitive type were found, still strongly reminiscent of reflexive permissives, the (isolated?) use of an intransitive infinitive with modal passive ‘se dát’ is already attested67 in a translation from Latin from the beginning of the 16 century: i Jan
ii Fišer, Knihy hospodáˇrské Voˇcehovský: Krátkej spis o morové nemoci iii hospodáˇrství polního Fišer, Knihy hospodáˇrské hospodáˇrství polního
da(va)t+inf in Czech
(437)
263
Ale pakli ty uteˇceš, ját’ také uteku. [. . . ] Neviem, dá3S G li seRFX:ACC ujítiINF . (1516)i ‘But if you run away, I’ll also run away. [. . . ] I don’t know, whether it will be possible to (lit. let itself) run away.’
This may be evidence of early and occasional use of the construction with intransitive verbs. Judging by our corpus, the profile of almost exclusively transitive verbs changes only in the the course of the 19th century. On the one hand, a book on chemistry from 1828 – naturally dealing with properties of many substances – contained 31 cases very much like the examples from the 18th century: (438)
Tažnost a kujnost dle rozdílné teploty rozdílná: v obyˇcejné jest velmi kˇrehký [. . . ]; opatrnˇe maˇckán m˚uže se na plechy dost tenké vypra-viti; v teplotˇe 100◦ C a trochu vyšší snadno seRFX:ACC dá3S G kovatiINF a skrze válce provalovatiINF [. . . ]. (1828)ii ‘Tractability and forgeability is different at different temperatures: generally it is very fragile [. . . ]; carefully pressed it can be spread out on metal sheets quite thinly; at a temperature of 100◦ C and slightly higher it can (lit: lets2 itself 1 ) be easily forged3 and passed4 through a roller [. . . ].’
At the end of the 19th century, however, 21 uses of a more general kind are ˇ contained in Ceský študent from 1869, a large text covering a general domain: (439)
Vlˇckovi je tam ale daleko, myslí, že by seRFX:ACC to také ve Vrbicích odbýtiINF daloPS T :3S G [. . . ]. (1869)iii ‘For Vlˇcek this is too far away, he thinks that it could also (lets3 itself 1 ) take2 place in Vrbice [. . . ].
All corpus examples up to 1835 contain transitive verbs and a patient promoted to sentence subject. The first example involving a sentential complement is from 1835 and the first intransitive complement is from 1869: (440)
Vyslouženec Bárta navštˇevoval, jako posud, každého veˇcera židovnu, a býval-li vesel, když oni truchlili, dá3S G seRFX:ACC snadno pomyslitiINF , žeCOMP nebyl truchliv, když oni než jindy byli veselejší. (1835)iv ‘The veteran Bárta, as before, went to the jewish pub every evening, and if he was cheerful while they were sad, it can (lit: lets itself) be
i Mikuláš ii Presl
Konáˇc z Hodíštkova, Enea Siluia Poethy o Sstiestij y diwný y vžyteˇcný Sen ˇ J.S.: Luˇcba cˇ ili chemie zkusná iii Ceský študent iv Karel Hynek Mácha: Cikáni
264 da(va)t+inf in OCS and earlier stages of Polish and Czech easily thought (imagined) that he wouldn’t have been sad, had they been happier than at other times.’ (441)
[. . . ] bez hajného nedaloNEG:PS T :3S G by seRFX:ACC ani na žádný pokrok myslitiINF ; proto spokojila se krvolaˇcná spravedlnost kozákovská prozatím jen jednou obˇetí. (1869)i ‘[. . . ] without the gamekeeper no thought of any progress was possible (lit: it did not let itself to think of any progress); therefore, for the time being, the bloodthirsty justice of Kozákov was satisfied with only one victim.’
On the basis of this restricted data set the following chronological picture emerges: Modal passives are used no later than the 16th century onwards. At the beginning, they are rare. In the 18th and the 19th century, modal passives generally become more frequent; as far as can be judged on the basis of this restricted corpus, they in the beginning mostly denote some characteristic of a concrete patient that provides the basis for something that can be done with it. By the second half of the 19th century, the semantic and syntactic class of verbs has broadened and unambiguosly subjectless constructions have appeared. However, it still does not approach the relative dominance over causative uses in terms of frequency that we have seen from the synchronic analysis. We will see that a similar picture is valid for Polish.
Summary The investigation of dá(va)t in the diachronic corpus has shown that symptoms of a comparatively high functional load and degree of grammaticalization of the reflexive permissive may date to the Old Czech period. Already at this stage, inanimate causees are possible with this type. Both reflexive and non-reflexive permissive are attested at comparative frequencies in the oldest part of the corpus; over the centuries, there is a decrease of the non-reflexive type. However, numbers for the permissives are rather low and therefore not robust. The profiles of non-reflexive and reflexive permissives support the hypothesis that they differ in frequency of expression of the causee, and explicit causees become increasingly infrequent with reflexive permissives up to the 19th century. The youngest dative causee used in a reflexive permissive construction is from the 18th century; the first examples containing type i Ceský ˇ
študent
da(va)t+inf in Polish
265
II causees were found already in the 16th century, but overall, they are only sporadically found in the orpus. The curative use of dát is attested since the end of the 15th century. Already in the 16th century, this function is more frequently encountered than the reflexive and non-reflexive permissive use taken together, resembling the profile of these constructions in Czech today. In an overwhelming majority of cases, no causee is expressed. No unequivocal examples of curative dative causee construction could be found, suggesting that the curative does not derive from a comparatively frequent dative causee construction, as is the case with the permissive; rather, it seems to have gained currency already in a causee-less configuration. Type II causatives were found only in the 18th and the 19th century; however, they may have well been present earlier, since they are comparatively infrequent. Their form is not stable and reflects parallel changes in the expression of the passive by-phrase. Clearly productive modal passive uses appear in the 16th century in personal constructions with transitive verbs. They grammaticalize to express more general modality and accommodate non-transitive verbs no later than by the second half of the 19th century; at least sporadically, this was possible also earlier, but the low number of examples prohibits any clear conclusions based on the Czech corpus alone.
6.5. da(va)t+inf in Polish At the moment of writing, no diachronic corpus spanning the history of Polish was publicly available; evidence concerning the development of da(va)t+inf was therefore collected from a variety of sources. Rich information was found in dictionaries, namely the Słownik staropolski, covering the earliest period, and Słownik polszczyzny XVI wieku, concerned with the 16th century. The latter can be considered a condensed corpus of its own, since it contains extensive citations and statistics on all occurrences of its entries in a large corpus of canonical texts. However, the semantic and syntactic criteria underlying these statistics differ from the one employed in this study; for instance, we are able to see which verb was attested how frequently, but we may not see how often it was used with a reflexive pronoun, with causee, or which type of causee. For this reason, Rozmy´slania przemyskie from around 1450 was taken as a further source for the earliest period. It was downloaded in modernized
266 da(va)t+inf in OCS and earlier stages of Polish and Czech cent. 15th. (2%) 16th. (16%)
date ca. 1450
text Rozmy´slania przemyskie
1566
Łukasz Górnicki: Dworzanin polski
1568
˙ Mikołaj Rej: Zywot człowieka poczciwego Andrzej Frycz Modrzewski: O obyczajach. O prawach. Fragmenty. Przekład Cypriana Bazylika Jan III Sobieski: Listy do Marysie´nki [wybór] J˛edrzej Kitowicz: Opis obyczajów za panowania Augusta III Ignacy Krasicki: Mikołaja Do´swiadczy´nskiego przypadki Maria Wirtemberska: Malwina, czyli domy´slno´sc´ serca Maurycy Mochnacki: O literaturze polskiej w wieku dziewi˛etnastym Józef Kope´c: Dziennik podró˙zy Juliusz Słowacki: Anhelli Henryk Rzewuski: Pamiatki ˛ J. Pana Seweryna Soplicy, cze´snika parnawskiego Józef Bohdan Dzieko´nski: S˛edziwoj ˙ Narcyza Zmichowska: Poganka Jan Potocki: R˛ekopis znaleziony w Saragossie (transl. by Edmunt Chojecki)
1577
17th. (10%) 18th. (18%)
1665 1743-1804 1778
19th. (55%)
1812 1830 1837 1838 1839-1841
1845 1846 1847
all
size 23 771 118 714 43 716 22 107
120 144 166 886 43 514 62 180 48 456 33 694 11 298 122 365
70 443 57 474 242 685
1 187 447
Table 47. Texts in the diachronic corpus compiled for Polish. Rozmy´slanie przemyskie according to Twardzik & Keller (1998); all other texts see for more details.
da(va)t+inf in Polish
267
othography from an XML-coded text collection published by the Instytut Je¸zyka Polskiego PAN in Cracow68 . This text reflects the edition published by Twardzik & Keller (1998); the paper edition provides extensive commentaries and Latin sources. To cover later periods, a variety of further prose texts covering the 16th to 19th centuries was downloaded from the wirtualna biblioteka literatury polskiej collected by Gda´nsk University69 . Rozmy´slania przemyskie and these texts were input into a database and categorized, employing the methods and annotation schemes reported above. The resulting corpus is heavily biased toward the 19th century; a list of texts is given in table 47. Texts in the selfcompiled corpus are referenced by year, author and abbreviated title; examples from one of the two dictionaries give the year and, if applicable, author and the abbreviation used there. For the late 19th and early 20th century, the older prose part of the IPI PAN sample corpus was used. This subcorpus includes classical prose from a variety of authors, published between 1880 and 1925 and has an overall size of around 3 million tokens; recall that it was excluded from synchronic analysis. This corpus has the advantage that it may be straightforwardly compared to the newer prose contained in the sample corpus that also amounts to about 3 million tokens. Textual evidence is thus heavily biased toward the beginning (well covered by the dictionaries) and the end (the older prose part of the IPI PAN sample corpus). Tables 48 and 49 give the figures for the occurrence of causative da(va)t. It should be mentioned that a construction composed of daj and a finite purposive clause is used in the very first Polish sentence that has come down to us (Klemensiewicz 1999: 197f.): (442)
[Sine, ut ego etiam molam.] Hoc est in polonico: DayI MP , utCMPL ia pobrusa1S G , a ti poziwai. ‘This is in Polish: Let1 me, I shall grind3 , and you take a rest.’
However, in order to constrain the analysis and keep with the general delimitation of the subject to infinitive complements, such cases were not taken into account here or in the Czech section. Investigating the link of this and the infinitive construction is left for further research.
268 da(va)t+inf in OCS and earlier stages of Polish and Czech
century 15th (n=29)
type non-refl. perm. refl.perm. factitive
16th (n=43)
ambig. non-refl. perm. refl.perm.
factitive
17th (n=16)
ambig. non-refl. perm. refl.perm.
factitive 18th (n=19)
non-refl. perm. refl.perm.
factitive
reflexive – – acc. acc. – other – – – – acc. acc. acc. dative dative other other – dative acc. – – acc. acc. acc. dative – dative – – acc. acc. acc. acc. –
causee? none dative none dative none none dative none dative dative none dative type II none dative none dative none none none dative dative none dative dative none none none dative dative none dative dative type II none
inanimate.. causer causee
yes yes yes
yes
yes
yes
yes yes
n= 3 7 11 1 2 1 4 1 3 1 10 10 1 2 1 1 1 9 1 2 2 1 5 2 1 1 3 1 1 3 9 1 3 1 1
subgroup 10 12 3 4
5
26 10 2 3
9 4 4
14 1
Table 48. Polish: Causative types in 15th to 18th century texts. The data are grouped into permissive, reflexive permissive, factitive and ambiguous and then categorized in terms of presence, form and animacy of causee/causer.
da(va)t+inf in Polish
century 19th (n=85)
type non-refl. perm.
refl. perm.
IPI PAN corpus (1880 to 1925) (n=217)
fact. ambig. non-refl. perm.
refl. perm.
reflexive – – – – – acc. acc. acc. acc. acc. dative other other – acc. – – – – – acc. acc. acc. dative other other
ambig. acc. acc. dative
causee? none none dative dative dative none dative dative type II type II none none dative none none none none dative dative dative none dative dative none none dative dative none dative none
inanimate.. causer causee yes yes yes
yes yes
yes yes yes
yes
yes
n= 2 2 9 2 3 38 6 6 3 3 6 2 1 1 1 20 8 54 7 1 66 22 16 15 2 1 1 2 1 1
269
subgroup
18
65 1 1
90
122
5
Table 49. Polish: causative types in 19th century texts and in the IPI PAN older prose subcorpus. See table 48; other includes instrumental, prepositional and reflexive possessive pronouns.
270 da(va)t+inf in OCS and earlier stages of Polish and Czech 6.5.1.
Permissive and reflexive permissive
The non-reflexive and reflexive permissive is – as may be expected – present in the corpus from the very beginning; Rozmy´slania przemyskie contains 22 permissive uses, 12 of which are reflexives, usually without causer. From the earliest texts in the corpus on, reflexive permissives are more frequent than non-reflexive permissives.
Manipulative causation Manipulative, non-reflexive causation is today felicitous in Russian and ungrammatical in Polish. While no uses with inanimate causee were referenced in the Słownik Staropolski or found in Rozmy´slania przemyskie, that is, the sources of the 15th century, both corpus and the Słownik XVI wieku provide ample evidence for such use in the sixteenth century, e.g. (443)
A nie dajI MP:2S G czasowiDAT nigdy darmo upłyna´ ˛cINF ; widzisz, z˙ e´c to szkodliwa utrata [. . . ] ‘And never let the time pass away in vain; you see, it it a harmful loss [. . . ]’ (Rej)
(444)
á rospu´sc´ iwszy miód z oliwa/ ˛ dayI MP:2S G dobrze wezwrze´cINF white white white white white (1564)i ‘and the dissolved honey with olives/ let boil well’
However, neither in any later texts, nor in any of the 19th and early 20th century dictionaries such as Słownik Linde, Słownik Wile´nski or Słownik Warszawski is such use attested. Slownik j˛ezyka Adama Mickiewicza contains a single example of a slightly different type of manipulative causation in a translation from English70 While more data would be necessary to ascertain the chronology, I conclude that this usage must have existed in earlier stages of Polish, but must have declined already before the 19th century, since not a single example was found in a comparatively large amount of data concerning this century.
i Marcin Siennik: Lekarstwa do´swiadzone, ktore zebrał uczony lekarz Pana Jana Pileckiego
da(va)t+inf in Polish
271
Evolution of causee coding Dative causees of reflexive permissives, animate and inanimate, are attested from the 16th century on in the corpus as well as in many citations in the Słownik XVI wieku. The corpus, moreover, gives an example of a type II causee already in the 16th century. It is introduced by od, which was also an option with passives in this period71 : (445)
[. . . ]aby´s te˙z wiedział powinno´sc´ swoj˛e, a nie dałPS T :3S G si˛eRFX:ACC unosi´cINF w powinno´sci swojej odPREP plotekGEN rozlicznie wymy´slonych s´wiata tego [. . . ]. (1568)i ‘[. . . ] so that you know your duty, and do not let yourself be carried away (governed) in your sinfulness by frequently invented rumours of this world [. . . ].’
Type II causees are especially frequent in the 19th century part of the corpus, more frequent than would be expected on the basis of the synchronic study; however, 5 of these examples were from a single text (Rzewuski), so the conclusion that it was generally more productive cannot be drawn from this finding. Of the 6 cases with type II causee, three appeared in a phrase introduced by przez, one with od and one in instrumental case: (446)
[. . . ] czy˙z to jest przest˛epstwem ofiar˛e z siebie czyni´c dla wdzi˛eczno´sci, da´cINF si˛eRFX:ACC najtkliwsza˛ lito´scia˛INS i głosemINS obowiazku ˛ zwyci˛ez˙ y´cINF ? (1812)ii ‘[. . . ] so is it a crime to make a victim of oneself for thankfulness, let1 oneself 2 be defeated5 by tender compassion3 and the voice4 of responsibility?’
I conclude that the possibility of type II causees with reflexive permissives was given already in the 16th century, and that its expression has changed along with the expression of the passive by-phrases; this is what we also found for factitives in Czech.
6.5.2.
Factitive causation
Factitive causation is already well attested in Rozmy´slania przemyskie, among others in a citation from Mathew (6,2): i Mikołaj Rej: Zywot ˙ człowieka poczciwego domy´slno´sc´ serca
ii Maria
Wirtembergska: Malwina, czyli
272 da(va)t+inf in OCS and earlier stages of Polish and Czech (447)
Tegodla kiedy czynisz jamu˙zn˛e, nie dajI MP:2S G przed soba˛ tr˛ebi´cINF , jako czynia˛ licemiernicy w synagogach a na ulicach, aby je ludzie chwalili. (ca. 1450)i Cum ergo facis eleemosynam, noliNEG:I MP tuba canereINF ante te, sicut hypocritæ faciunt in synagogis, et in vicis, ut honorificentur ab hominibus. (Vulgata) Therefore when thou dost an alms-deed, sound not a trumpet before thee, as the hypocrites do in the synagogues and in the streets [. . . ]. (New intern. version).
The Latin version contains a contextual causative72 translated into Polish with factitive dát. The construction is also frequently found in texts in the following centuries, especially in Jan Sobieski’s letters from the 17th century, a genre presumably quite close to spoken language: (448)
Z drogietu tego dałemPS T :1S G sobieRFX:DAT urobi´cINF kontusz, bom nie miał w czym chodzi´c [. . . ]. ‘From the droguet I had myself a kontusz made, because I didn’t have anything to wear [. . . ].’
No unequivocal dative factitive was found. The same use is still possible in Polish today, but, as we have seen in the synchronic analysis, it is very infrequent. In comparison to the permissive the factitive is still quite prominent in the 16th and 17th century texts, but much less so in the texts from the 18th and the 19th centuries. Its frequency relative to the permissive decreases over the centuries; note that the only 19th century example involves phraseologized da´c czeka´c ‘make wait’ and that no example was attested in the older prose IPI PAN sub corpus. This reflects the difference between Czech and Polish that we have found in the synchronic part. The decline of factitive da´c that we see in the data was explicitly commented on by the lexicographer Linde at the beginning of the 19th century, noting that "[. . . ] da´c instead of kaza´c [‘have done, command’] is more seldom used nowadays; the Germans therefore translate their bitten lassen as da´c prosi´c against [ the rules of] the Polish language"73 (Linde 1854-1860: 402; my translation).
6.5.3.
The modal passive
No example of the modal passive is attested in the corpus before the 16th century. The Słownik Staropolski contains one citation that may perhaps be i Rozmy´slania
przemyskie
da(va)t+inf in Polish
type non-refl. perm. reflexive perm. factitive ambiguous all causative słysze´c widzie´c modal passive
a) compiled corpus 15th 16th 17th 23 771 162 430 120 144 10 6 3 12 25 9 3 10 4 4 2 4 29 43 20 1 2 -
18th 210 400 4 14 1 0 19 3 19 4
273
19th 648 595 18 65 1 1 85 31 10 35
b) IPI PAN sample corpus older p. modern p. type ca. 3 mln. ca. 3 mln. non-refl. perm. 90 56 reflexive perm. 122 232 factitive 0 1 ambiguous 5 9 all causative 217 298 słysze´c 72 18 widzie´c 1 0 modal passive 47 456 Table 50. Causative and modal passive types in (a) the self-compiled corpus of 1.2 mln. tokens spanning the 15th to the first half of the 19th century, and (b) the two subcorpora from the IPI PAN sample corpus, each with about 3 million tokens and containing artistic prose from 1880-1925 and 19982003, respectively. Figures for the modal passive are given excluding combinations wiht widzie´c ‘see’ and słysze´c ‘hear’ which are counted separately.
interpreted as modal passive: (449)
Ad id est commune proverbium: mowa szya˛RFX:ACC da3S G movyczINF , a chleb yesczINF (1449) ‘There is a common proverb concerning this: language can (lit. lets2 itself 1 ) be spoken3 , and bread eaten4 .’
274 da(va)t+inf in OCS and earlier stages of Polish and Czech Without a broader context, however, this proverb remains unclear. Note that the Słownik j˛ezyka Jana Chryzostoma Paska (1636–1701) does not contain a single example of a modal passive. In the Słownik XVI wieku, however, several unequivocal examples are cited, as in (450), where the modal passive is used like the personal modal passive today in that it predicates something about the properties of the subject referent: (450)
póki ro˙zczká młoda/ dáie3S G si˛eRFX:ACC náchyli´cINF y náwie´sc´ INF gd´zie potrzebá/ á vrosła˛ trudno c´ iagn ˛ a´ ˛c. (1595)i ‘As long as the twig is young/ it can (lets itself) be bent and guided where necessary / and the grown-up (twig) is hard to grow.’
While the construction is thus attested from the 16. century on, the earliest and most frequent use in the corpus up to the 19th century is with cognitive verbs, in particular widzie´c ‘see’ and słysze´c ‘hear’. The figures for these constructions are given, among others, in table 50. Recall that this construction is also a point of difference between Czech and Polish today, as the synchronic and parallel data have shown: Czech dát se slyšet lit. ‘let itself be heard’ regularly means ’to utter’ and is not frequent; the Polish cognate may be translated with ‘be audible’ and is a frequent type, not translated by explicit modal means into Russian and Czech (section 5.1.3 above). These finding are corroborated by this diachronic evidence showing that da(va)t with these verbs were collocations even before the modal passive became fully productive; note that judging by the diachronic corpus analyzed in this study, this was never the case for the Czech equivalent. (451)
Je´sli W´c bardziej kochasz, ni˙zeli ja, da3S G si˛eRFX:ACC to widzie´cINF , kto i w tej tak delikatnej i lubej materii g˛estszych i wi˛ekszych za˙zyje ponctualites. ‘If you love more, than I do, it is apparent (lets itself be seen), who in this both delicate and fine matter deserves denser and greater ponctualites.’ (1665)ii
(452)
Marszałek starej laski z posłami, przeciw którym nie dałaPS T :3S G si˛eRFX:ACC słysze´cINF z˙ adna protestacja, był s˛edzia˛ zarzutów. white white white white white (1743-1804)iii ‘The Marszałek of the old sejm with those delegates, against which no protest could be (lit. let itself be heard), was to be the judge regarding these accusations.’
i Piotr Skarga (SkarKaz) ii Jan Sobieski: Listy do Marysie´ nki obyczajów za panowania Augusta III
iii J˛ edrzej
Kitowicz: Opis
da(va)t+inf in Polish
type personal ambiguous impersonal intransitive sentential compl.
comp. corpus 19th cent. 648 595 17 17 1 35
275
IPI PAN sample corpus older prose modern prose ca. 3 mln. ca. 3 mln. 33 87 12 103 2 181 66 19 47 456
Table 51. Modal passive construction types in the 19th century part of the compiled corpus, in the older and modern prose parts of the IPI PAN sample corpus.
Da´c si˛e widzie´c first appears in the 16th century and is by far the most frequent use of the modal passive in the 17th and the 18th centuries74 . In the first half of the 19th century, it is surpassed by da´c si˛e słysze´c; taken together, these two verbs are more frequent than the sum of all other uses. As far as can be seen from the corpus data, the modal passive proper gains frequency only in the 19th century. Da´c si˛e słysze´c is still more frequent than all other modal passive uses together in the late 19th / early 20th century part of the IPI PAN corpus. This picture changes at some time in the 20th century. The modern prose part of the IPI PAN corpus exhibits a marked difference to the older prose corpus: the frequency of the productive modal passive is now clearly much higher than any other use; da´c si˛e słysze´c is comparatively frequent, but clearly outranked by the modal passive as a whole, which is used with a variety of verbs. A breakup into construction types of the modal passive is given in table 51. As can be seen, the increase in frequency is linked to the use of subjectless modal passives, be they constructed with transitive verbs or with intransitive verbs. In the 19th/early 20th-century corpora, in contrast, most examples either contain an overt subject or are ambiguous in regard to the presence of a subject; only three of 82 are subjectless. The first subjectless uses appear in the 19th century. Neither Słownik Linde, Słownik Wile´nski, Słownik Warszawski nor Doroszewski cite such usage; in fact, only Linde and Słownik Wile´nski cite any modal passive use aside from those with cognitive verbs. Linde’s citation suggests that it must have
276 da(va)t+inf in OCS and earlier stages of Polish and Czech been in use in spoken language: “O trunkach i potrawach; da si˛e pi´c, da si˛e je´sc´ = niezłe, albo te˙z: bardzo przednie (p.406) [About beverages and food: lets itself drink, eat = not bad, or also: very good (my transl.)]”. The Slownik j˛ezyka Adama Mickiewicza cites one intransitive subjectless example75 from a letter written in 1847. The first unequivocal subjectless example with a transitive object in the corpus is from the same year; here, the object is marked genitive by virtue of a general rule concerning objects in negated clauses: (453)
By´c mo˙ze chciała przez to powiedzie´c, z˙ e mojej metodyGEN nie dałobyIRR:3S G si˛eRFX:ACC zastosowa´cINF do liczb pierwszych, poniewa˙z dziela˛ si˛e one tylko przez jeden. (1847)i ‘Perhaps she wanted to say that one cannot (lit. it does not let2 itself 3 ) employ4 my method1 to the first figures, because they can be divided only by one.’
It seems that, like in Czech, the expansion of subjectless uses of the modal passive was first a gradual and slow process; they are found only very sporadically in the older prose part of IPI PAN. The frequency of the modal passive then increases quite dramatically by an order of magitude in the newer prose part of IPI PAN as well as in the relevant corpora in Czech. This change is accompanied by an expansion of subjectless uses. Apparently, this is to be attributed to a rapid change in both semantic and syntactic respects that happen in close temporal vicinity in what may be expected to be a s-shaped curve of linguistic change. An important clue of which of the two aspects - semantic or syntactic changed first is found in the following example from turn of the 19th to the 20th century: (454)
[. . . ] przepowiadaja˛ [. . . ] upadek miasta lub co najmniej ruin˛e wielkiego domu, któraREL:NOM tylko nadzwyczajnymi ofiarami da3S G si˛eRFX:ACC odwróci´cINF . (1896)ii ‘[. . . ] they predict [. . . ] the fall of the city or at least the ruin of the great house that can (lit. lets itself) only be averted with the help of extraordinary sacrifices.’
This example would be more felicitously expressed in an non-agreeing impersonal construction today: i Jan
Potocki: R˛ekopis znaleziony w Saragossie
ii Sienkiwicz:
Quo vadis
Conclusions
(454)’ [. . . ,] któr˛eREL:ACC tylko si˛eRFX:ACC odwróci´cINF .
nadzwyczajnymi
ofiarami
277 da3S G
This suggests that the semantics of the personal construction first generalized to a more general modal, and that only subsequently two distinct constructions arose. This question will be returned to in the concluding discussion.
6.6.
Conclusions
I will now give a chronological overview of the developments delineated above. We have seen that permissive causation with dative causee is, hardly surprisingly, the oldest among the constructions involving da(va)t+inf. Initially, it included also the manipulative domain, that is, independent processes involving inanimate causees that are under indirect control of the causer. While it is initially found in all three languages, it is subsequently lost in Polish and Czech, but retained in Russian. A tendency towards negative polarity is a characteristic trait of the permissive in Russian that seems to be already present in Old Church Slavonic. From this perspective, Russian is the most conservative and seems to have changed little; the notable exception is the rise of the various extended functions of the imperative, not found in Czech or Polish. Grammaticalization of the reflexive permissive construction in Czech and Polish is evident already in the earliest texts stemming from the 14th and the 15th centuries, respectively; it can be shown to have been used with a frequency comparable to the non-permissive and to have offered, unlike Russian today, the possibility of being used with inanimate, non-agentive causees. From the 15th century on, factitive use is attested in both Czech and Polish. It is an intriguing question which cannot be resolved here whether the absence of factitive uses in Czech in the 14th century reflects a coincidence or whether this function actually developed in this time. In any case, Polish and Czech exhibit a close resemblence: da(va)t expresses curative causation without explicit mention of the causee. Today this is a productive profile in Czech and to a much lesser degree in Polish; no development from a different configuration is evident. While this function was never dominant in Polish, it was initially more frequent and declined towards the 18th century; it has been a seldom used
278 da(va)t+inf in OCS and earlier stages of Polish and Czech possibility since. In Czech, conversely, it has been the most important factitive function since late Middle Czech. In relation to type II causatives with permissives and factitives, the record is uneven in these two languages, possibly due to the low number of examples. Type II causatives with reflexive permissives are attested in both languages from the 16th century on; in both languages, they were thus used alongside dative causees for at least 200 years, since the last Czech dative causee is attested in the 18th century. Note that in view of the low numbers of examples in Czech, dative causees may well have been used much longer. Today, they are ungrammatical in Czech while in Polish, dative causees dominate and type II causees are only marginally possible. Type II causatives with factitive da(va)t are not attested in Polish; this is hardly conclusive, since altogether only 20 cases of factitive causation were found. In Czech, the first type II factitives are from the 18th century; this type is felicitous to this day. The data is thus compatible with the hypothesis that type II causatives developed only in those contexts where they are present today, that is, with reflexive permissives and curatives. Both types in Czech and Polish share the characteristic that the causee is regularly backgrounded. In both cases, the form of the causee changes with the form of the passive by-phrase. We thus see a functional and formal parallel to the passive. This will be discussed in the conclusions at some length. Modal passives arise later in both languages, first predominantly in the agreeing variant with subject and transitive verb. In both languages, subjectless uses of the modal passive are attested from the 19th century on; they first appear only sporadically and then seem to increase in frequency quite abruptly, ultimately developing into the most frequently used function of da(va)t+inf in the second half of 20th century in both languages.
Chapter 7 Conclusions and directions for further research
7.1.
Overview
Da(va)t+inf in Russian, Polish and Czech presents an intriguing network of constructions that are synchronically and diachronically interrelated within and across these languages. Three approaches have been employed to investigate the grammaticalization of these constructions: a synchronic approach aiming to assess the current state; a diachronic approach aiming to reconstruct and assess the evolution of this state; a comparative approach supplementing both. It is the contrastive (including diachronic-contrastive) perspective that has helped us understand what is idiosyncratic and what follows a regular, expected pattern. This especially applies to the various sub-constructions that we have seen emerge as relevant in the analysis: both the imperative of the permissive construction in Russian (daj podumat’ ‘let me think’) and the reflexive permissive in Polish and Czech (dát se zastrašit od nˇekoho ‘let oneself be intimidated by someone’) would not seem remarkable if its closely related neighbors would not be so different in comparison. Similar things can be said about the expression of manipulative causation with da(va)t+inf in Russian (dat’ kraske vysoxnut’ ‘let paint dry’), which has become infelicitous with its Western neighbors or the non-agreeing modal passive construction in Polish (da si˛e czyta´c ksia˙ ˛zk˛e one can read the book) that add another piece to the puzzle of non-agreeing structures in this and other Slavic languages. Many properties of these constructions could only be touched upon here and in many cases it would be worthwhile to look at other evidence to test the analysis forwarded in this study. In the present chapter, I provide an overview of the results of this investigation from a wider perspective. In doing so, I discuss some explanations for the developments found, including possible contact influence of German as well as similarity with other constructions. Inevitably, these explanations raise new questions that, however, are left for further research.
280 Conclusions and directions for further research 7.2. 7.2.1.
Developments of da(va)t+inf General remarks
Generally, the development of da(va)t+inf does not primarily concern the grammaticalization of one of the elements (least of all there is morphological evidence of change). Rather, we see gradual syntactic change proceeding from specific morphosyntactic configurations that become less and less compositional as they chrystalize into new constructions of their own right. We have seen that this change is gradual and configurations with certain semantic characteristics and morphosyntactic regularities in use become fixed in the sense that their categorial characteristics can be traced to statistical characteristics of a former state. In Polish and Czech, but not in Russian, the erstwhile compositional formation of the reflexiv permissive has further grammaticalized to a construction of its own. This new construction is, like the compositional formation, characterized by frequent causee omission; however, it additionally allows a type II, passive-like by-phrase to introduce the causee. Yet further grammaticalization of this reflexive permissive construction has led to the formation of the modal passive. This construction closely follows the causee-less form of the reflexive permissive in surface syntax and generalizes this tendency in a categorial restriction: the causee, or erstwhile agent, can no longer be introduced. Like the permissive and reflexive permissive, this construction is used foremost with negation and with perfective complement aspect, albeit not as strongly as its predecessors. As we proceed towards the more grammaticalized types, it seems aspectual variation of da(va)t becomes less common, even if this is difficult to assess across languages. In Polish, we can diagnose a certain loss of the aspectual opposition of da(va)t in the modal passive, while in the most grammaticalized construction, the modal passive in Czech, no variation in matrix aspect is possible and only dát may be used. Here the statistical tendencies concerning negation and perfective aspect of the lexical infinitive are no longer noted: dát se is in this respect quite similar to other core modals of possibility. We have seen that morphological attrition stricto sensu has not played a role in the processes concerned: in all cases, no univerbation on the phonological level is found, da(va)t continues to behave as a single word. However, we have seen a reduction in form, specifically the (not unequivocal) loss of
Developments of da(va)t+inf
281
person and (only in Polish) number, as well as aspect. The assessment of the loss of aspect depends on whether one views the aspectual partners as forms of one verb, or as two verbs that stand in a derivative relationship. In the first case, aspect loss is analyzed as a loss of inflectional categories, in the second, as a reduction of the paradigm of verbs that enter this construction. In both cases, aspect loss is a clear symptom of grammaticalization: since the vast majority of verbs have aspect, being biaspectual is a symptom of small-class syntax, diagnosed either as loss of integrity (reduced potential to inflect) or loss of paradigmatic variability prima facie. The domain of the structural scope of da(va)t+inf in the various constructions is a difficult question that is not discussed in complete detail above. We have seen that initially, the permissive expresses a complex event, albeit composed of two events; da(va)t makes a clear semantic and syntactic contribution here, with several subtypes of meaning present in different configurations of participant and verb classes. At the end point of this development, in the modal passive, meaning becomes more uniform and less dependent on verb classes and participant properties; the construction is now unequivocally uniclausal and expresses a single event. Selectional restrictions do not depend any more on da(va)t as much as on its infinitive. In the course of these changes, two syntactic developments are especially remarkable: the development of type II causatives as well as the development of an impersonal or subjectless modal construction, where the patient of the lexical verb is expressed as object, strongly reminiscent of the impersonal construction involving si˛e in Polish. These will be given interpretations below. In the remainder of this section, I present a sketch of the development of da(va)t+inf in a step-by-step fashion, taking up these two developments in more detail. I proceed chronologically: I first discuss the emergence and status of the permissive, the reflexive permissive and the factitive and then turn to the development of type II causatives; finally, I conclude with an overview of the development of modal passives.
7.2.2.
From give to permissive
The initial grammaticalization of da(va)t+inf from giving to letting took place before the earliest written documents. This development involved a change in both semantics and syntax: the initial structure is a ditransitive construction with a nominative agent, a dative recipient and an accusative
282 Conclusions and directions for further research patient, as in (455-a). The resulting structure involves a nominative causer, a dative causee and an infinitival predicate, optionally with arguments of its own; an example is given in (455-b). (455)
a.
b.
On dal emu [NPacc knigu]. He gave.PST.SG him.DAT book.ACC ‘He gave him a book.’ On dal emu [VPin f otdoxnut’ He gave.PST.SG him.DAT.SG relax.INF /ˇcto-to skazat’]. something.ACC.SG say.INF ‘He gave let him relax / say something.’
Causer and causee thus find direct counterparts in the initial construction, while the infinitival clause takes the place of the patient. I see two possible accounts for the emergence of this structure. The development may have started out from an extension of giving in metaphorical usage. In this scenario, the event that is granted is framed as something that is given; such a metaphor would have started out as creative usage and would have become conventionalized. This development is clearly plausible with cognitive verbs, as in davat’ ponjat’, znat’ ‘let/give understand, know’. Such a metaphor is well attested in uses of give in the meaning ‘to grant’ with a subordinate finite clause. A second path would involve metonymy, or the conventionalization of implicature, captured in a model of different context types indicative of stages in grammaticalization. (456)
a. b.
c.
On dalPS T :S G emuDAT kniguACC . (transfer) ‘He gave him the book.’ On dalPS T :S G emuDAT kniguACC poˇcitat’INF . (transfer/perm.) ‘He gave him the book to read.’ / ‘He let1 him2 read4 the book3 .’ On dalPS T :S G DATINF dogovorit’INF . (permissive) ‘He let him speak.’
(456-a) represents the initial stage; (456-b) represents an instance of give where a secondary predicate sharing a patient with give is introduced. In such a construction, the implicature may be invited that the receiver wanted to read the book, and that by giving it to the receiver, the agent enables the recipient to read it. Then, both permissive and transfer readings are possible. This implicature may have become conventionalized; finally, the expansion into
Developments of da(va)t+inf
283
environments such as (456-c), representing the stage of conventionalization where no object is transferred, would have become possible. No empirical data is available for a decision between these models. In view of the fact that the meaning shift from give to let is very widespread in the languages of the world (Heine & Kuteva 2002), the latter model, relying on specific functions of the infinitive as both a purposive marker and infinitive, might seem less plausible. However, purposive and infinitive functions are closely related (Haspelmath 1989). Leino (2005: 10-23) explicitly posits such a development on the grounds of morphosyntactic changes in the corresponding construction in Finnish. In Finnish, causative syntax and syntax of give with purposive infinitive do not overlap today, unlike Russian; however, in documented former stages of the languages these two were identical, and the erstwhile purposive marker has evolved to be the infinitive marker used only in the causative construction. In general, these models do not exclude each other: both invited implicature and metaphor may have both been forces driving the change. The resulting permissive expresses a single, internally complex event. The grammaticalization of this construction in the languages considered relates foremost to its semantics; syntactically, this verb largely patterns with other object control verbs. Some symptoms of a tight bond of the infinitival clause and da(va)t are notable: the impossibility of pronominalization of the infinitival clause, difficulty of independently negating the infinitive and the use of negative polarity items in the complement clause under matrix negation. The status of these findings is, however, not certain: other verbs, such as pozvolit’ ‘allow’ in Russian, seem to share the same properties, suggesting that these properties follow from the semantics of permissive causation. Alternatively, it may be that verbs such as pozvolit’ should also be seen as grammaticalized. We see variation in the grammaticalization of da(va)t as a permissive most clearly in relation to verb classes and participant properties: in this sense, Russian da(va)t is more grammaticalized than its Polish cognate, since it is also used with inanimate causee, that is, in the domain of manipulative causation. Polish as well as Czech initially also allowed such use of permissive da(va)t+inf, as evidence from the diachronic corpora shows. This as well as the loss of the non-reflexive, but not the reflexive permissive in Czech, may be interpreted either as a decrease in generality of the permissive marker – i.e., a case of degrammaticalization – or the more expected loss of a specific sense in a network of senses. This depends on how one views the relationship of the different usage patterns.
284 Conclusions and directions for further research 7.2.3.
From permissive to reflexive permissive
Initially and in contemporary Russian, the reflexive permissive is a transparent combination of the reflexive and the permissive construction. The diachronic study in chapter 6 has shown that it had gained special status in Czech and Polish already quite early, possibly before the beginning of documented history, while it didn’t grammaticalize any further in Russian. The parameters that set this construction in the West Slavic languages apart from its Russian cognate are • greater frequency • greater combinatorial freedom as regards the restraints on the causee, which are structural rather than semantic in the West Slavic languages (Polish dałPS T :3S G si˛eRFX:ACC przekona´cACC argumentowiDAT vs. Russian *dalPS T :S G sebjaRFX:ACC ubedit’INF argumentuDAT , both intended ‘let oneself be persuaded by an argument’) • greater combinatorial freedom regarding complement verbs / semantic bleaching of the function expressed (abstract responsibility in dałPS T :3S G si˛eRFX:ACC wyprzedzi´cINF vs. conscious, deliberate action of the causer in dalPS T :S G sebjaRFX:ACC operedit’INF , both lit. ‘let oneself be overtaken’) • gain of passive-like type II causee coding (discussed in detail below) The path of development seems clear: the initially compositional combination of reflexive and permissive becomes more and more entrenched and gradually forks off from the permissive proper, forming a separate construction. A special status of reflexive permissive constructions is not, as far as I know, described in the literature; they are mentioned in the context of the development of passive constructions from causatives (Haspelmath 1990; Knott 1995). It is a contribution of this study that this construction would need to be represented in a separate node in a complex semantic map or network of causative constructions. Reflexive permissives may be said to exhibit important similarities to passive diatheses. As in passive constructions, the patient is expressed as the sentence subject, while the agent is either not expressed at all or given as an oblique object. As in the passive, and unlike the other causative constructions, this does not increase valency: no argument is added. The resulting construction involves an agentive patient, that is, a patient that is construed to have at
Developments of da(va)t+inf
285
least some theoretical control and responsibility for the action that is affecting him. Initially, this is a compositional construction type that is first differentiated from the non-reflexive permissive only in usage, primarily in terms of the class of possible verbs and a regular omission of the causee. In the West Slavic languages, this construction type gradually undergoes grammaticalization, leading to an increase in functional load. Eventually, this configuration needs to be considered a construction in its own right, as restrictions concerning the causee and its expression in non-reflexive and reflexive permissives develop differently. Ultimately, in Czech, the non-reflexive and reflexive permissive become unrelated, as the former goes out of use and the latter is preserved. Such a development is well explained from a usage-based perspective on the emergence of linguistic structure that accords both holistic and compositional status to individual configurations. According to such a perspective, a certain compositional combination, in this case, the reflexive permissive, becomes entrenched as a whole through frequent usage and may develop differentiating traits of its own in the process: behavioral properties lead to coding properties. In the case of the reflexive permissive, such a trait concerns foremost the emergence of type II causees in Czech and Polish which remain impossible with non-reflexive permissives. The status of these phrases will be taken up below from a broader point of view.
Influence of German Among the possible factors driving the grammaticalization of the reflexive permissive as a construction of its own, language contact with German, where the causative construction formed with lassen is highly grammaticalized with many similarities to Polish and Czech da(va)t+inf, is a very plausible candidate. Unfortunately, the existing research into the development of lassen is not sufficient for a definite assessment of this influence. In his monograph devoted to the comparison of modern lassen and OHG lâzen based on late 12th and early 13th century texts, Ide (1996) does not mention a special tendency of lâzen to be used in reflexive constructions; but since it is the specific comparative and corpus-based perspective employed in the present study that makes such a tendency remarkable, this omission cannot be seen to mean that the reflexive construction was infrequent; the ex-
286 Conclusions and directions for further research amples cited in Ide (1996) suggest the opposite. Note that Ide did not find type II causee in his corpus even though by-phrases were current in the passive at this point (p. 79). According to Bondzio (1959), type II causees with reflexive lâzen become frequent in the 15th century (note that he does not differentiate permissive from factitive meaning). Influence of German on Polish and Czech regarding the use of type II causees is therefore generally a chronologically plausible factor, while its influence in relation to the grammaticalization of reflexive permissives per se is less clear. This similarly applies also to factitives, to which we now turn.
7.2.4.
Factitive: from give or from the permissive?
Factitive use of da(va)t+inf is fully productive only in Czech. Here, it is frequent, forms with nechat a small class of causative formants and admits of a type II causee, as only a restricted number of constructions do. In Polish, this function is much less grammaticalized, since the construction is infrequent and generally does not admit of type II causees. In both languages, da(va)t expresses the curative subtype of factitive causation, not general factitive causation, cf. (457): (457)
Dal jsem pokoj vymalovat malíˇrem. Have-PST:SG AUX:1SG room-ACC paint-INF painter-INS ‘I had the room painted (by a painter).’ (Czech)
It is an intriguing question whether the cross-linguistically wide-spread grammaticalization of give to a factitive causative is evidence of an independent path of grammaticalization of give or of further grammaticalization of the permissive. This question has been addressed by Lord et al. (2002: 232) on the basis of a sample of unrelated African and South East Asian languages. Based on the distribution of the grammaticalized functions of give in these languages they hypothesize the following grammaticalization path: Lexical verb → Permissive → Factitive This is all the more convincing as the transition from permissive to factitive is well attested regardless of the original lexical source of the formant, specifically in the Germanic languages (including English, where this function has receded, see Fischer et al. 2000: 234). Ide (1996: 191f) forwards a cautious hypothesis concerning this change involving the conventionalization of invited implicatures in dialog, a use characteristic of OHG lâzen in his corpus.
Developments of da(va)t+inf
287
Elaborating on this idea, we could expect this change to involve understatement: just as you may go now is used to convey a polite request (instead of you must go now), let somebody do something can in the appropriate context be understood as have somebody do something. However, such an account does not seem probable for the Slavic formant, since there are general arguments against a scenario in which the factitive function of da(va)t+inf developed via the permissive. Crucially, such an expansion of permissive meaning would not explain why factitive da(va)t has always been, as far as we can see, attested in the form and meaning with which it is productive now, that is, without causee and only with curative meaning. If the factitive developed from the permissive by conventionalized implicature in the way outlined above, we could expect to see a gradual expansion of the meaning of da(va)t formed with dative causee, proceeding from permissive to underspecified causative to encompass both factitive and permissive meaning. No evidence is found in the corpus supporting such an account of the developments. Other grammaticalized causative auxiliaries that express both factitive and permissive causation, such as German lassen or Czech nech(áv)at, are underspecified rather than ambiguous; this is clearly not the case for Czech dá(va)t which is used in specific disjoint permissive and factitive functions rather than as an underspecified causative. In fact, very few ambiguous examples are found in the corpus. Generally, they do not concern switch contexts, i.e., situations where permissive usage invites a factitive implicature; rather, these examples are ambiguous because the researcher cannot with certainty reconstruct the situation. The grammaticalization of factitive da(va)t may have also involved metaphorical extension where a proposition is framed as the patient of give, similar, but not relying on, such an account for the development of permissive da(va)t as given above. The crucial point of evidence here is that also in contemporary language, give may denote such a metaphorically extended meaning with dative recipients, albeit without further grammaticalization. Such an example was quoted above on page 212: (348)
Pˇredstavme si pro zajímavost, jak by tˇreba automechanici v Buenos Aires pˇrijali svého pražského kolegu a zda by muDAT dali3PL alei spoˇn zametatINF dv˚ur.
i Komárek,
Stanislav, Opšlstisova nadace, 2002
288 Conclusions and directions for further research ‘Let’s imagine for the sake of curiosity that for example car mechanics in Buonos Aires received their colleague from Prague and that they gave/let/had him at least sweep the courtyard.’ However, this configuration never grammaticalized any further in any of the languages considered and is not found frequently it any of the corpora. While the curative construction is not derived from this specific configuration, it may be seen to rely on the same conceptual basis — namely metonymic use of ‘give’ to denote the giving of a task. This conceptual basis, I assume, provided the semantic basis for the grammaticalization of a construction without dative phrase. This would have involved contexts where ‘giving’ in the sense of ‘giving a task’ is used with an omitted, contextually known or generic, causee / recipient of the task; this then arguably became conventionalized as a construction in its own right, where da(va)t is further bleached to denote not only the speech act, but the complex process of having something done, that is, causing and thereby effecting. Note that the development of factitive meaning from a causeeless outset makes another hypothesis implausible, namely the scenario involving an infinitive that was originally a secondary predicate, outlined above for the permissive. Since purpose-denoting infinitives are unspecified regarding the modality of the proposition (if someone gives someone else ‘a book to read’, it might be because the recipient is supposed to read it or because he wants to read it), the use of ‘give’ with purpose-denoting infinitives may also yield factitive meaning. Again, however, it seems this would have led to an ambiguous or underspecified factitive-permissive causative with a dative causee, rather than to a causeeless construction.
Influence of German Language contact is, again, an obvious factor in the development of factitive da(va)t. As Toops (1992: 40) notes, da(va)t has developed factitive meaning only in that subset of the Slavic languages known to have been strongly influenced by German and that have developed a modal passive construction, namely in all West Slavic and the West South Slavic languages, strongly suggesting an influence based on lassen. However, the more fine-grained analysis of causative functions in the present study has shown that the development of factitive da(va)t concerns, by and large, only curative causation; more general factitive causation, e.g.
Developments of da(va)t+inf
289
causation aiming at controlling the causee’s action (‘make somebody do something’), rather than the accomplishment of some deed (‘have something done’), is not and has never been, as far as we can see, expressed by da(va)t in Czech or Polish. The Western European languages possess highly grammaticalized causative formants, used to denote a wide range of causative functions. These formants — cognates of French laisser and faire, English let, have, make, German lassen — are used in type I syntax to express more general factitive causation (see the examples for Spanish on page 21ff.) Da(va)t has never been used this way in the languages considered76 . Considering the influence of German lassen on da(va)t, then, we see that the latter did not develop to simply mirror the denotative potential of German factitive lassen; rather, language contact seems to have facilitated the extension of da(va)t based on the meaning potential provided by its lexical source, ‘give’. This observation — namely that factitive dá(va)t has never, as far as one can tell on the evidence compiled for this study, grammaticalized as a noncurative or underspecified factitive with a dative type I causee — is interesting as it has consequences for the assessment of the competition of nech(áv)at and dá(va)t that has played such a role in Czech prescriptive writing. Nech(á)vat did not simply replace likewise ambiguous permissive-factitive dá(va)t– something that would seem to be remarkably unmotivated, as Toops (1992) points out. Rather, with nech(áv)at, Czech developed a formant modeled on German lassen more closely not only in terms of lexical source, but, crucially, also in terms of syntax and wide range of functions in accordance with the pattern of type I and type II causatives observed in other Standard Average European languages. The details of this development remain to be empirically investigated.
7.2.5.
The emergence of by-phrases: causatives as diathesis constructions
Da(va)t+inf in causative function instantiates several abstract constructions, including what may be called control constructions, infinitive complement constructions, dative object constructions, etc. The exact modelling of these constructions differs from framework to framework, and of course they are not seen as atomic by many theories; however, they are subject to analysis in most. No exhaustive modelling of the syntax of da(va)t+inf has been attempted in this thesis in the sense that only some parallels and differences to
290 Conclusions and directions for further research other verbs taking part in these constructions, as well as regularities of the use of da(va)t in these constructions have been addressed. An important issue in the above discussion and description of da(va)t+inf have been type II causative constructions; it emerged that it is an peculiar property of da(va)t+inf that they may govern type II causees in their most grammaticalized functions. In this section I discuss the status of these phrases and their connection to other constructions, such as the passive; henceforth, I will refer to them as by-phrases. I first show that by-phrases are themselves grammaticalized constructions that are used with a set of diathesis constructions; I then elaborate on the interpretation of the causative constructions in question as diathesis constructions. From a construction-based point of view, the by-phrase must be considered a construction of its own; this is implicitly assumed in many approaches that ascribe it some sort of identity across governing construction types. Its identity lies (a) in its restriction to appear only in a restrained set of environments, (b) in a certain morphosyntactic form, and (c) in its function. Below, the grammaticalization of by-phrases is approached from all three of these perspectives. I give only a general outline, since a full account of by-phrases with deverbal nouns and passives would go beyond the scope of this study. Environments: The syntagms that take by-phrases in Czech and Polish involve passives (458-a), deverbal nouns (458-b) and permissive (458-c) and factitive (458-d) causatives: (458)
a.
PL CZ
b.
PL CZ
Piotr został oszukany przez brata. Peter AUX.PST.3SG cheat.PPP.SG by brother.ACC Petr byl oklamán bratrem / od Peter AUX.PST.3SG cheat.PPP.SG brother.INS / by bratra. brother.ACC ‘Peter was cheated by his brother.’ Oszukanie Piotra przez Pavla. cheating.VBN Peter.GEN by brother.ACC Oklamání Petra Pavlem / *od cheating.VBN Peter.GEN brother.INS / by Pavla. brother.ACC ‘The cheating of Peter by Paul.’
Developments of da(va)t+inf
c.
PL
CZ
d.
CZ
291
Piotr dał si˛e oszuka´c przez Peter let.PST.3SG RFX.ACC cheat.INF by brata. brother.ACC Petr se dal oklamat Pavlem / od Peter RFX.ACC let.PST.3SG cheat.INF brother.INS / by Pavla. brother.ACC ‘Peter let himself be cheated by Paul’ Petr dal Jana oklamat Pavlem / od Peter had.PST.3SG Jan.ACC cheat.INF brother.INS / by Pavla. brother.ACC ‘Peter had Jan cheated by Paul.’
Form: By-phrases are expressed by limited sets of language specific formants. In Polish, this primarily concerns PPs introduced by przez; additionally, instrumental case by-phrases are possible, arguably depending on animacy of the referent (Siewierska 1988: 251ff., Bartnicka et al. 2004: 359). In earlier Polish, this set also included the preposition od; all three options are reflected in use with both causatives and passives. In Czech, two options are given today: the instrumental case and a PP introduced by od. The instrumental seems to be dispreferred in spoken language; od is reported to have been more frequent in the past. A further possibility in older Czech were PPs introduced by skrze; on the basis of corpus evidence, this was found to apply to both passive and causative. The corpus studies have shown that the form of the by-phrase in causative and passive is synchronically linked; their form changed over time, but it changed in both surroundings. The situation with deverbal nouns is more complicated: while Polish przez and the Czech instrumental is used in all three environments, Czech od is not used with deverbal nouns. The highly interesting task of taking a diachronic perspective on this distribution cannot be undertaken here. Function: Functionally, by-phrases denote and agent or, more generally, the participant expressed as subject in unmarked diathesis. This function is grammatical rather than lexical: the by-phrase does not express a certain semantic content, but a structural relation, i.e., the referent that is the ‘logical subject’ of the base verb. Some semantic content may be retained in traces,
292 Conclusions and directions for further research such as, for example, concerning the difference between przez and the instrumental in Polish. This is a characteristic of grammaticalization: the size of paradigms decreases with grammaticalization, while the semantic content of their members is rendered more and more abstract. Instead of a wide range of semantically differentiated formants that are used, e.g., in locative adverbials, with by-phrases, only one or two formants are given to choose from. The more integrated the paradigm, the less semantic content is retained; cf. English, where by is the only member of the class, and competing von and durch in German (see Wiemer 2004: 305-313 for an account of the grammaticalization of the by-phrase with passives in Slavic). Use of the by-phrase is generally controlled by grammatical rules. In the passive the by-phrase is the only option to introduce the subject argument of the lexical verb. With deverbal nouns, the by-phrase may be used only if both agent and patient of the lexical verb are present; then, it is the obligatory form. In relation to causatives, the situation in the two languages is different from one another. While in Czech, no other coding option is given, in Polish, type II causees with reflexive permissives are only a secondary option never obligatory and not accepted by all users. This will be discussed below. I now turn to the question how we can generalize over the environments governing by-phrases. Expression of the subject in a by-phrase is related to peripheral status in use in all three cases, i.e., the subject in these constructions usually remains non-expressed. Crosslinguistically, the passive is used predominately without agent; many languages do not provide such a possibility (see, e.g. Siewierska (1984: 35ff.)); an important function of the passive is precisely to defocus the agent. Furthermore, it seems plausible that this is also true of deverbal nouns accompanied by the objects of their base verbs; presumably, the introduction of the agent is rather exceptional in such cases. Finally, that the causee in reflexive permissive and curatives is seldom used has been shown for Russian, Czech and Polish above. I assume that this is generally the case; some data from other languages have been adduced in the relevant section above concerning the curative, but, to my knowledge, this has not been investigated in detail, least of all for the reflexive permissive which does not figure prominently in the linguistic literature. We thus see that by-phrases correlate with regular omission of the base verb subject. All three environments may be viewed as diatheses: this is trivially true of the passive and a plausible feature of deverbal nouns that represent a transposition involving change of argument coding.
Developments of da(va)t+inf
293
The causative as a valency-decreasing diathesis construction Causatives is sometimes seen as a valency-increasing diathesis (e.g. Kulikov 2001: 889): besides adding some semantic content to the exposition of the event, causatives generally add a participant to the situation. This is the case for the permissive, cf. Piotr oszukał Pawła ‘Peter cheated Paul’ vs. Jan dał Piotrowi oszuka´c Pawła ‘Jan let Peter cheat Paul’. Here, Jan has been added as additional participant. As a rule, all participants are expressed: to say Jan dał Pawła oszuka´c ‘Jan let (someone) cheat Paul’ in a permissive reading is unusual (see pages 2.2.6.2ff. above). In combination with a reflexive, however, the number of participants is not changed in comparison to the unmarked case, cf. Polish Piotr oszukał Pawła ‘Peter cheated Paul’ and Paweł dał si˛e oszuka´c Piotrowi ‘Paul let himself be cheated by Peter’; this combination is valency-preserving. Moreover, the causee is usually not mentioned, yielding a configuration where one participant is not expressed and the number of arguments is decreased in relation to the unmarked diathesis: Paweł dał si˛e oszuka´c ‘Peter let himself be cheated’ is, in all the languages covered, the most frequent structure. This is prima facie a usage phenomenon; but I will argue shortly that the tendency to omit the causee in usage is linked to the emergence of passive-like by-phrases, that is, a phenomenon of grammatical encoding, rather than usage. Like the permissive, the curative adds the causer as an additional participant that is expressed as subject. Unlike in the permissive, but very much like in the reflexive permissive, the causee is usually not expressed. In relation to Czech Petr oklamal Pavla ‘Peter cheated Paul’, the causative Jan dal Petra oklamat ‘Jan had Peter cheated’ thus involves the same number of arguments, but the participants are different: the initial agent, the causee, is replaced by the causer. Semantically, the curative construction abstracts from the specific agent to the commissioner of the action and is therefore in respect to the base clause both valency-increasing (it adds a causer) and valency-decreasing (it demotes the causee, the immediate agent). It is in this form — without causee — that the curative construction seems to have developed in the languages covered in this study. In Czech, the causee may be introduced as a by-phrase, but this is a rarely-used possibility; in this sense, it is a secondary option. While the causee may be left unexpressed in all three constructions — non-reflexive permissive, reflexive permissive and curative — only in the case of the non-reflexive permissive, non-expression of the causee is exceptional.
294 Conclusions and directions for further research In the reflexive permissive, it is a regular phenomenon; in the curative, omission of the causee is even more frequent and to a certain degree part of the definition of the construction. I would like to propose that it is not by coincidence that the by-phrase is possible only in the latter two, but that the regularity of omission makes the use of the by-phrase possible. Specifically, I propose that on the grounds of their common characteristics, those types where omission of the causee is regular enter a paradigm of constructions with other diathesis constructions where the participant expressed as the subject in unmarked diathesis (i.e. an agent) is also omitted. Because of this, as with those other diathesis constructions, use of the byphrase becomes a possibility to reintroduce the usually omitted causee. Put differently, with further grammaticalization and frequency of use, omission of the causee with reflexive permissives comes to be not merely a usage phenomenon, but an inherent characteristic of this construction: causee omission becomes primary, as with the other diathesis constructions. By analogy to the other diathesis constructions, then, the ‘demoted’ agent may be reintroduced in an optional by-phrase. I assume that the common tendency of curatives and reflexive permissives not to mention the causee essentially follows from their function and therefore has cross-linguistic relevance. This is supported by the fact that some languages, such as Latin or English, use passive morphology with curatives and reflexive permissives, as in I had it repaired by a mechanic (with the bare participle) and I let myself be persuaded. Rather than taking this as evidence for the plausibility of deriving type II causatives from an underlying passive, I consider this fact to be evidence of the common functional characteristic of perpherality of the causee in particular causative subtypes. In these language, this is coded compositionally with the help of passive constructions in the verbal complement. Note that this predicts that the usage profile of these construction coincides with what we have found in Slavic - namely, that in such languagess, curative and reflexive causation is typically expressed with a passive, rather than active complement.
The by-phrase in Polish: explaining marginality While in Czech, a passive-like by-phrase is the only way to express the causee in a reflexive permissive construction, in Polish, such a possibility is given only marginally (see section 3.2.4, p. 134ff.). This can be explained by as-
Developments of da(va)t+inf
295
suming that the reflexive permissive in Polish has, so to speak, two identities. It is, at the same time, • a compositional construction involving the permissive and the reflexive • a diathesis construction in its own right, related to other diathesis constructions Because the non-reflexive permissive in Polish has not, like in Czech, gone out of use, the reflexive permissive can be in principal seen to be compositionally derived from a combination of the permissive and the reflexive. While I have shown that the resulting permissive reflexive is more than the sum of these parts and has grammaticalized as a construction of its own (section 3.2.4, p. 129ff.), the non-reflexive permissive still provides a model for the use of dative coded type I causees. Thus, based on the first identity, that is, an analysis of the reflexive permissive as essentially compositionally derived from the non-reflexive permissive, the form of the causee should be dative. However, the Polish reflexive permissive is likewise a diathesis construction with a passive-like function of expressing the patient as subject and backgrounding the agent. This second identity provides the basis for the use of passive-like by-phrases. Note that the by-phrase is possible only with the two most frequent subtypes of the reflexive permissive in Polish, with dative and accusative reflexives. It may not be used with the much less frequent instrumental or prepositional reflexives. Note, furthermore, that with dative reflexives, causee omission is much more regular than with accusative reflexives; at the same time, in the seldom occurring case when the causee is introduced, type II causees are relatively more frequent than in the accusative reflexive. There is thus a clear correlation of (1) causee omission and type II form of the causee and (2) overall frequency across subtypes of the reflexive permissive. I conclude from this that the regularity of causee omission makes the second identity, as a diathesis construction in its own right, more prominent. This double nature of the Polish reflexive permissive reflects a basic property of language: the same structures may be accessed from different perspectives, as ready-made chunks or as freely formed compositional formations (see Haspelmath 1999; Lehmann 2002; Tomasello 2003). No decision between the two possible analyses must be made; rather, both are relevant and provide the basis for dative and by-phrase causees, respectively. Compo-
296 Conclusions and directions for further research sitionality in this sense is very similar to analogy: instead of saying that the permissive reflexive is compositionally derived from, we may also say it is coded by analogy to the non-reflexive permissive. In Czech, the non-reflexive permissive does not exert any influence, because this function has gone out of use. Productive da(va)t+inf is restricted to those causative types that involve causee suppression and where its identity as a diathesis construction is foregrounded; consequently, only type II causees are possible. Note that in Czech, analogy works rather towards use of type II causees, since they are also used with causative nech(áv)at.
Conclusions I regard by-phrases in the languages considered to be abstract, synsemantic, small-paradigm constructions that are used to reintroduce demoted subjects in diathesis constructions; they should be viewed as grammaticalized, arising from more semantic coding of nominal phrases in several environments and converging to a small paradigm of desemanticized coding options used over a constrained range of environments. Their use with causatives is a symptom of their own grammaticalization as well as of the grammaticalization of causatives as ‘subject-demoting’ diathesis constructions. I suggest that two factors come together in the development of these phrases to be used with certain types of causatives in the languages investigated: first, the causative function concerned is grammaticalized (including high frequency), and this function involves regular demotion of the erstwhile subject of the lexical verb. It is under these conditions that by-phrases became possible in Polish and Czech. With the reflexive permissive in Russian, by-phrases did not become possible because this construction did not further grammaticalize; it is essentially still a composition of the reflexive and the permissive without a separate structural identity. In Polish, by-phrases may not be used with curatives by a similar token: the construction is overall little grammaticalized and marginal.
7.2.6.
From reflexive permissive to agreeing modal passive
The modal passive, as in Czech to se nedá udˇelat ‘that cannot be done’, has grammaticalized both in Polish and in Czech. On pages 222ff. above, I have
Developments of da(va)t+inf
297
shown that the Czech construction is more grammaticalized than its Polish counterpart in relation to several parameters, making it more alike central modals. This notwithstanding, the development of the modal passive has proceeded along the same lines in both languages. I distinguish two stages: an initial stage, where the reflexive permissive gives rise to structurally similar, agreeing modal passive, and a subsequent stage in which also subjectless modal passives evolve. It is the second stage that sees a marked loosening of selectional restrictions and a sharp increase in frequency in both languages. It is this stage where the two languages take slightly different paths of development. Similar to what was said concerning the emergence of type II syntax and factitive semantics with da(va)t+inf, the role of language contact with German is very probably a contributing factor to the development of this construction. The modal passive only developed in those Slavic languages that were strongly influenced by German, i.e. in Polish, Czech, Upper and Lower Sorbian, Slovak, Slovene as well as in Bosnian/Croatian/Serbian, but not in any of the East or South East Slavic languages. However, since details of the development and of the German modal passive formed with sich lassen are not known, this issue cannot be investigated in detail here. The change from permissive reflexive (a) to agreeing modal passive (b) is schematically represented in the following Polish examples: (459)
a.
b.
Piotr daje si˛e (komu´s) oszuka´c. → Piotr lets RFX sbdy-DAT cheat-INF ‘Peter lets himself be cheated (by somebody)’ Ksia˙ ˛zka daje si˛e czyta´c. Book-NOM lets RFX read-INF. ‘The book lends itself to reading, reads well.’
This change involves an expansion of the class of possible subject referents towards inanimate, non-agentive referents and the semantic reduction of permissive causation to the modality of possibility (see section 1.4, p. 27ff.). Structurally, the participant which corresponds to the former causee / agent of the reflexive permissive construction is understood to be of either specific or generic reference and can no longer be expressed. The transition from permissive to modal involves the transition from a dynamic to a static, delocalized situation. Such delocalization may be observed with perfective reflexive permissives with future meaning; cf. (460) from Russian, where a modal passive interpretation is not available:
298 Conclusions and directions for further research (460)
On ne dast3S G sebjaRFX:ACC obmanut’INF . ‘He won’t let himself be cheated.’
(Russian)
This is, first of all, a statement about a future non-event: He will not let himself be deceived. It refers to a potential event that is conceptualized as an eventuality in discourse. Indirectly, it it is also a statement about a characteristic of the subject referent: He is somebody of a kind that will not let himself be deceived. This is not strictly implied by the statement: we may say both (461-a) and (461-b): (461)
On ne dast sebja obmanut’, . . . (Russian) ‘He won’t let himself be cheated, . . . ’ a. . . . xotja on takoj cˇ elovek, kotorogo legko obmanut’. ‘although he is such a person that is easy to cheat.’ b. . . . potomu cˇ to on takoj cˇ elovek, kotorogo nel’zja obmanut’. ‘because he is a person that is impossible to cheat.’
That the statement expresses something about the characteristics of the potential patient / causer is therefore an implicature. Such implicatures may have conventionalized, as conventionalization of implicatures is a frequent process in language change and an important mechanism in grammaticalization. The foregrounding of this implicature is facilitated by a backgrounding of the agentive component, plausible with inanimate causers; it is, for example, also evident in the use of inanimate causer in impersonal causation, cf. (103) citet on page 80 above: (103)
Demokratija, koneˇcno, daet3S G otdel’nomuDAT :S G cˇ elovekuDAT :S G i razvernut’sjaINF . ‘Democracy, of course, lets (enables) the individual person (to) develop.’
The following Russian example, repeated from page 99 above, contains metaphoric use of an inanimate causer in a reflexive permissive and illustrates that in creative usage, a reading close to the modal passive, also with a generic agent, may arise in a language where this structure has not grammaticalized: (144)
A tol’ko strannaja vešˇc’ sˇcast’e. Srazu nikogda sebjaRFX:ACC ne ii dast3S G razliˇcit’INF . ‘But happiness is a strange thing. It never lets2 itself 1 be recognized3
i Aleksandr
Genis. Dovlatov i okrestnosti (1998) iz tumana (1997)
ii Marina
Višneveckaja. Vyšel mesjac
Developments of da(va)t+inf
299
immediately.’ This is clearly creative usage and such examples are extremely infrequent; however, such usage is a possible starting point for the development of modal passives. In this example, sˇcast’e is personified, but not an agent; semantically, it is very like the modal passive by virtue of the impossibility of an agentive reading of the ‘causer’. Rather, the unspecified causee remains an agent. Cf. an English example that is testimony to the same extension: (462)
She was tired a lot, as if she was living not one life but two: her own, and another, shadowy life that hovered around her and would not let itself be realized, the life of what would have happened if Lucy had i not stepped sideways and disappeared from time.
The emergence of modal passive constructions out of reflexive permissives has thus probably been influenced by a process of invited implicature as well as, more generally, semantic bleaching concomitant with a loosening of selectional restrictions concerning the causer. While the details of this development remain speculation, generally, this change may be accounted for by standard mechanisms of language change.
Persistence of an actional component In the grammaticalization of the reflexive permissive, the causer’s agentive role becomes weakened: in dał si˛e oszuka´c ‘he let himself be cheated’, the causer is also the patient. This component is further weakened with the development of the modal passive, used with inanimate patients. A shade of agentivity is retained, however, since modality in the less grammaticalized, agreeing modal passive, as it is still used in Polish, is based on this participant (section 3.6.1, p. 153ff.). Actional elements are likewise retained. Let us have a look at uses of the reflexive permissives in past tense, as in Russian (463)
On dalPS T :3S G sebjaRFX:ACC obmanut’INF . ‘He let himself be cheated.’
The statement asserts that there was a causee who succeeded in deceiving the causer despite some capacity of the causer to somehow obstruct this; the causer (sentence subject) thus carries some responsibility. The (unmentioned) i Margaret
Atwood, Death by landscape
300 Conclusions and directions for further research causee intended an action that was not guaranteed to be successful from the outset and than succeeded. That he did succeed is asserted. Such an assertion of success, that is, aspired action, asserted possibility and realization of this action combined is often met in the context of modals. Hansen (2001: 174f.), among others, does not consider Russian smoˇc’ a modal since it denotes both possibility and realization; often, central modals may also denote success, arguably as an implicature rather than by intensional meaning. This is the case for Polish da´c si˛e, as in (464)
Otwór wygladał ˛ na lisia˛ nor˛e, lecz pokruszony łupek łatwo dałPS T :3S G si˛eRFX:ACC rozkopa´cINF [. . . ] ‘The hole looked like a fox’s den, but the crushed slate was easy to (lit. let itself easily) dig up.’
In this case, the modal passive implies successful completion of the event, unlike the use of the modal predicative łatwo: (465)
a.
b.
Łupek łatwo dałPS T :3S G si˛eRFX:ACC rozkopa´cINF , *ale nie rozkopali´smy go. ‘The slate was easy (let itself easily) to dig up, *but we didn’t dig it up.’ Łupek łatwoADV byłoAUX:PS T :3S G rozkopa´cINF , ale nie rozkopali´smy go. ‘The slate was2 easy1 to dig3 up, but we didn’t dig it up.’
A full analysis of this complex would require more research on when such interpretations arise with more central modals. I would like to make the point here that such a meaning of success may be accounted for as a phenomenon within the domain of modals, but also be interpreted as persistence, that is, a meaning component ‘inherited’ from the reflexive permissive. In this sense, the fact that the permissive asserts the completion of the complement event allows for the expectation that the resultant modal also carries the dynamic component of realization. Note that persistence of an actional component in modal passives may also explain why complement infinitives in Polish, but not in Czech are mostly perfective. These observations concern the first stage of the development of the modal passive constructions. At this stage only agreeing constructions are found, and the construction is generally not nearly as frequent as today. At this stage, non-agreeing constructions appear only sporadically; they seem to become more frequent towards the end of this stage. The second stage, commencing somewhere between the turn of the 19th to the twentieth century and the mid-
Developments of da(va)t+inf
301
twentieth century in both Czech and Polish, is characterized both by a marked increase in frequency of this construction and the emergence of subjectless modal passives. We will turn to this stage now.
7.2.7.
From agreeing to subjectless modal passives
In its earlier, agreeing variant, the modal passive is restricted to transitive verbs, following the model of the reflexive permissive; it is fairly infrequent. Extrapolating from contemporary Polish data, the modal passive in this first version probably shared several more characteristics with the reflexive permissive: a marginal possibility of using non-accusative reflexive pronouns, no independent negation of the infinitive, and the very marginal possibility of introducing an agent in a przez-phrase is in evidence (see section 3.6.2 on page 162ff.). All in all, the delimitation between the agreeing modal passive and the reflexive permissive is not absolute77 ; most importantly, the limitation to transitive verbs is evidence that si˛e in this construction retains traits of its anaphoric function. The second stage of development of the modal passive involves a sharp increase in frequency, an expansion of the class of infinitives to include also intransitive verbs and more general modal semantics. The most conspicuous change is the development of a new, frequent subtype, the non-agreeing modal passive, as under (b) in the following schematic illustration using Polish data: (466)
a.
b.
Ksia˙ ˛zka da si˛e czyta´c → Book-NOM lets RFX read-INF ‘The book (NOM) can be read’ Ksia˙ ˛zk˛e da si˛e czyta´c. / Da si˛e spa´c. Book-ACC lets RFX read-INF / Lets RFX sleep-INF ‘The book (ACC) can be read. / It’s possible to sleep.’
This subjectless structure results trivially if the infinitive does not govern an accusative complement that could be promoted to subject position. While in Czech, this structure is almost completely restricted to intransitive verbs, in Polish, it is also generalized to transitive verbs, as shown in (466-b). Since the reflexive morpheme does not have an antecedent in the subjectless construction, the construction loses its link with the reflexive permissive: si˛e can only be interpreted as a lexicalized member of the construction, rather than a syntactically conditioned reflexive pronoun. This new syntactic type is a relatively recent phenomenon. While agree-
302 Conclusions and directions for further research ing modal passives are found as early as the 16th century, subjectless modal passives in both languages are initially found only sporadically and start to appear more frequently in the 19th century. At some point around the turn of the 20th century, however, modal passives in general become much more frequent and the relative proportion of subjectless modal passives increases dramatically. In Polish, the subjectless construction with transitive verbs becomes the most frequent type. Concommitant with the expansion of this type there is a change in semantics. While the earlier, solely agreeing modal passive in Czech and Polish was, as far as we can see, restricted in meaning, later modal passives denote modality of possibility in a more general way. In Polish, there is a split: with transitive infinitives, the agreeing modal passive type retains this restriction, while the subjectless construction has more general semantics. We thus see a second step of complex grammaticalization of the modal passive. The class of possible infinitives shifts to include intransitive verbs, while the reflexive pronoun is reanalyzed as an abstract element co-signalling the construction. This reanalysis goes hand in hand — in a dialectic process — with semantic bleaching, that is, loss of the semantic characteristics linking the original, personal modal passive to reflexive permissives. This process seems to have reached a decisive stage some time around the turn of the 19th / 20th century, resulting in a remarkable overall increase in frequency of the construction. While we cannot say for sure, it is plausible that this process also involved the reduction in aspectual variation of da(va)t that we see today. In Polish, da´c today clearly also fulfills functions of an imperfective verb in the modal passive (section 3.6.4 from p. 171), while in Czech, the formerly perfective verb dát is the only remaining option. he Czech construction is overall more grammaticalized and more like central modals than its Polish counterpart in relation to several parameters (section 4.3, p. 222ff.), including • higher frequency • use with a wider range of lexical verbs, including unaccusatives • no polarity sensitivity, no special preference to complement aspect • no variation in matrix aspect
Directions for further research
303
This notwithstanding, the development of the modal passive has proceeded along the same lines in both languages. However, Polish and Czech diverge in that Polish develops a split construction with an older, less grammaticalized agreeing structure and a newer, more grammaticalized subjectless structure, respectively. Czech develops a uniform construction, and avoids the subjectless construction with transitive verbs, giving precedence to an agreeing structure with the patient framed as subject with transitive verbs and to a subjectless structure with intransitive verbs. Syntactically, the non-agreeing modal passive in Polish is very similar to two other subjectless constructions, namely an agent-suppressing diathesis construction with the reflexive morpheme (czyta3S G si˛eRFX ksia˙ ˛zk˛eACC ‘people read the book’) as well as a similar construction with a former passive participle, the so-called bezosobnik. Both share the characteristic that the undergoer of the transitive verb is not expressed as a subject, i.e., that the object is not promoted; similar impersonal reflexive constructions exist in Ukrainian, Slovene, and western dialects of Bosnian/Croatian/Serbian, but not in Czech, where such subjectless constructions are used only if the verb is intransitive. In this sense, the development of the modal passive in Polish represents a further case of what Weiss (1984, 1983) has analyzed to be a general preference for non-agreeing syntactic patterns in this language.
7.3.
Directions for further research
Apart from many loose ends and speculative analysis in the present book that merit more attention, I would like to pick out a few issues where the present study may serve as a point of departure. Generally, many details of the diachronic developments point beyond this particular construction and remain to be understood. Why did the modal passive expand so rapidly when it did sometime in the first half of the 20th century, and what was the role of the German model? Are we dealing with a general move towards structures long present in the spoken language, or did extensive translation and cultural influence of German have a stronger impact? A corpus based approach could well pinpoint the factor different text types played in this development. The above analysis of non-agreeing modal passive in Polish suggests that the development of the impersonal reflexive construction (czyta3S G si˛eRFX:ACC ksia˙ ˛zk˛eACC ‘one reads the book’) in Polish may have been anal-
304 Conclusions and directions for further research ogous and may have involved the generalization of a construction with intransitive verbs to transitive verbs, rather than a reanalysis of ambiguously marked subjects, as Pisarkowa (1984) suggests. Using a more comprehensive diachronic corpus of Polish, balanced in genre, text size and time spans, the development of this coding option with the modal passive and other subjectless constructions could be put into a common perspective, more generally investigating the emergence of non-agreeing diathesis constructions in Polish and other Slavic languages. As concerns causative constructions in general, it is clear that a more comprehensive, cross-linguistic, corpus based and typologically minded approach could make the factors in the development of this category and its recurrent sub-types clearer. Specifically, we could ask whether certain patterns of use of analytical causative constructions, such as the overt marking of curative causation, constitute a Standard Average European (Haspelmath 2001) feature and whether the Czech development constitutes a move of this language towards SAE. Generally, the grammaticalization of da(va)t+inf in the Slavic languages has doubtlessly been influenced by German, but many of the details remain obscure. An important project for further research would be to take a much more complex and comprehensive perspective on the influence of German on Slavic (and possibly vice versa) and look at the developments in the area of modal, passive and other grammatical constructions in Slavic and German against a contact linguistic and socio-linguistic background. Here, there is much to be learned about patterns of contact influence that promises to be relevant also beyond the Slavic-German realm.
Notes
1. Verbs that denote accidental causation like Rs. ronjat’are cases in point. 2. ‘Of any one subject, one thing must be either asserted or denied.’ Aristoteles, Metaphysics 1011b24, cited from Horn (Fall 2006). 3. Note that all these statements contain presuppositions, not discussed here: To state that Walt Whitman is alive or not is only informative under the presupposition, stable under negation, that there is a referent for the nominal phrase Walt Whitman. 4. But they do exist, as in Russian ne mog ne, or German er konnte nicht umhin, das zu tun; see de Haan (2002) for Russian. 5. “Der Prozentsatz negierter Phrases [. . . ] ist zum grossen Teil direkt proportional dem Prozentsatz permissiver Phrasen”, Nedjalkov (1976: 50). 6. Note that throughout this study, I use the term object control in a non-technical sense; I do not delimit it against ECM verbs, the decisive characterization conveyed being obligatory co-reference of an object and the infinitive’s understood subject. 7. Unfortunately, no exact information on the size of the RNC were available at that time, so nothing can be said about the absolute frequency of this construction in the corpus. 8. Two queries, one searching for da(va)t with and one without preceding ne originally yielded a total of 17,808 hits, of which 4,451 were negated. The results were then filtered to exclude • the lexicalized passive participle dannyj ‘given’ (1,286 instances) • nouns between da(va)t and the infinitive that form frequent light verb constructions where the infinitive is an argument of the noun (3,600 instances), as in e.g., dat’ pravo ‘give the right’: (467) Kto dalPS T :S G ljudjamDAT pravoACC sudit’INF o vešˇcax, kotoryx oni ne ponimajut? [Ne mešajte nam verit’! // "Argumenty i fakty / Moskva", 2001.06.06] ‘Who gave people the right to judge about things they do not understand?’ The following nouns were excluded: svobodu ‘freedom’, nadeždu ‘hope’, nakazanie ‘punishment’, šans ‘chance’, slovo ‘word’, znak ‘sign’, povod ‘occasion’, soglasie ‘consent’, zadanie ‘task’, osnovanie ‘basis’, vozmožnost’ ‘possibility’, ukaz ‘command’, pravo ‘right’, sovet ‘advice’, vremja ‘time’, kljatvu ‘oath’, komandu ‘command’, trud ‘work’ • the parenthetical (ne) daj bog ‘god forbid’ (108 instances) • full stops, question marks, colons, semicolons and exclamation marks or less than 2 commas between da(va)t and the infinitive in order to exclude sentence boundaries (2,211 instances) During annotation work, the rather unusual usage of da(va)t in the writings of Anatolij Blum, the metropolit of Sourozh in Britain, proved especially vexing. da(va)t is very frequent in his writings and offers usages that differ from the patterns established in the rest of the corpus. After initially pursuing the hypothesis that this was evidence of a certain style or register, I discovered that he spent his whole life abroad, living in Britain
306 Notes
9.
10.
11.
12.
13.
14.
15.
and France, and even spoke Russian with a slight accent. I decided that these patterns are due to intervention, and chose to exclude his writings from the corpus. (589 cases). For example, the aspect-tense distribuperf. imperf. tion of the telic verb otkryt’/otkryvat’ ‘to 1,859 (20%) open’ shows a similar tendency. The ta- non-past 529 (6%) 6,110 (66%) 740 (8%) ble to the right displays the results of a past search of the RNC (post-1945 subcor6,639 (72%) 2,599 (28%) pus, 99.5 mio tokens) for perfective otkryl/otkroet vs. imperfective otkryval/otkryvaet. The contrast in distribution is large (Cramer’s V =.65) and highly significant (χ2 = 3933, p<.001). “Kogda oborot daj/te+infinitiv upotrebljaetsja bez mestoimennogo aktanta, to otsutsvujušˇcij aktant - e˙ to praktiˇceskij vsegda 1. l. ed. cˇ ., v iskljuˇcitel’nyx sluˇcajax – 1 l. mn. cˇ . (Xrakovskij & Volodin 1986: 130)” ˇ “Castica davaj/te (i v men’šej stepeni daj/te) kak budto by vydvigaetsja na rol’ standartnogo markera imperativa dlja vsej paradigmy liˇcnyx form. (Xrakovskij & Volodin 1986: 130)” In an earlier work, Podlesskaya (2005b: 95) adduces word order restrictions to show that the finite construction with daj has grammaticalized, as only dajteI MP:2PL proidu1S G ! but not ?proidu1S G dajteI MP:2PL ! is possible for ‘please let me pass’ (lit. ‘give I’ll pass’), while both are possible with the permissive: dajI MP:2S G projtiINF / projtiINF dajI MP:2S G ‘let me pass’. However, this argument should be approached with caution, since it seems to be just as true for other formants such as razreši ’allow’, cf. ?projdu1S G razrešiI MP:2S G ! vs. razrešiI MP:2S G projdu1S G ! ‘allow me to pass’. I regard this restriction in word order to be a characteristic of constructions involving a juxtaposition of two finite clauses. In the query, dumat’ ‘think’ and govorit’ ‘speak’ were excluded as lexical verbs, as they resulted in many false hits such as davaj, dumaju ‘Davaj, I thought’. Generally, high numbers are not reliable since interpunctuation between the imperative and lexical verb could not be excluded; it would have been necessary to check and categorize each example for relevance. In the case of an intervening pronoun (daj ja podumaju, ‘let me think’) numbers are more reliable. In general, low numbers are reliable, since it was feasible to examine the whole set. “. . . ] v konstrukcii s infinitivom reˇc’ idet o nastojašˇcej pros’be, pri e˙ tom govorjašˇcij igraet dovol’no passivnuju rol’. V liˇcnoj konstrukcii rol’ govorjašˇcego javljaetsja, naoborot, kak raz oˇcen’ aktivnoj, cˇ to prepjatstvuet upotrebleniju e˙ toj konstrukcii pri javnyx nekontroliruemyx dejstvijax (Barentsen 2006: 27)” In order to ascertain this, one would need to systematically compare the onomasiological fields concerning this field in several languages; this is beyond the scope of the present study. A further interesting finding is that in reflexive imperatives, the reflexive pronoun coreferent to the causer is most often replaced by the personal pronoun, as in: (468) DajI MP:2S G xot’ potrogat’INF tebjayou:ACC ... - govorit Ekaterina [. . . ] Zapovednik (telemelodrama) (1991)) ‘Let me at least touch you...’
(Bitov.
The prescribed reflexive pronoun was used only in one of 7 relevant cases. 16. Causatives have given rise to passive constructions in several languages. It is a possible
Notes
307
course of events that in these cases, the agentivity of the causer/patient, possibly within a reflexive construction, was even further deemphasized in the course of language change, yielding the canonical passive. See Knott (1995); von der Gabelentz (1861); Haspelmath (1990). 17. Recall RNC results were filtered to produce a post- and pre-1945 corpus. Hitherto, only texts from after 1945 were considered in the analysis 18. These examples are from the later enlarged RNC and were not present in the original sample; therefore, they do not show up in the above list. 19. Topoli´nska (1993: 237) gives a third structure of da(va)t with subjunctive complement clause, as in (469) Jurek nie daje, z˙ eby kto´s mu co´s Jurek NEG give.3SG, COMP.IRR somebody.NOM him.DAT something.ACC powiedział na ten temat said.PST.3SG to that topic.ACC ‘Jurek doesn’t let / suffer that somebody tells him anything about that.’ She states this to be a colloquialism. My informants generally do not approve of these cases, and Szlifersztejnowa (1968: 66) explicitly states that this construction is not possible, noting “that not even in the dialects the infinitive is replaced by a subordinate clause involving a complementizer or by a deverbal noun, as it is done after kaza´c ’command, have’, poleci´c ’recommend’, dopu´sci´c ‘afford, let’ etc. ” (my translation) “Doda´c trzeba, z˙ e bezokolicznik nie bywa po nim zast˛epowany zdaniem podrz˛ednym spójnikowym czy rzeczownikiem odsłownym nawet w gwarach, jak to si˛e zdarza po kaza´c, poleci´c, dopu´sci´c itp.”
20. 21. 22. 23.
24.
25. 26.
In the full IPI PAN corpus, not a single instance was found: the query string [base="daw?a?a´c"] [tag!=".*interp.*"] {,1} "," "˙zeby" yielded 54 hits, none of which were of the relevant structure. http://korpus.pl Recall that of the initial 18,000 Russian examples, alone 3,500 potential light verb constructions were filtered out. See http://nkjp.pl; this corpus appeared after the main body of this study was completed. There is a morphological difference: in Polish, as in all West- and Southslavic languages, there is an opposition of clitic and non-clitic reflexive pronouns, while in Russian, the clitic form has developed into a postfix that obligatorily attaches to the verb after the desinence. “Jak w ka˙zdej konstrukcji b˛edacej ˛ w u˙zyciu od wieków, i w tej wytworzyły si˛e pewne odcienie, zale˙znie od typu rzeczowników podmiotu i dopełnie´n dalszych (Wilczewska 1966: 155)” “[. . . ] cało´sc´ połaczenia ˛ oznacza uleganie podmiotu wpływem czynników subiektywnych lub od niego niezale˙znych [. . . ] (Wilczewska 1966: 155)” Wilczewska (1966) cites the following example of an causee in instrumental case: (470) Trzeba przepracowa´c uczciwie sze´sc´ lat, [. . . ] nie da´cINF si˛eRFX:ACC zmócINF ci˛ez˙ ka˛INS praca˛INS .
308 Notes ‘One needs to work honestly for six year [. . . ] not let oneself be overwhelmed by hard work.’ (Fic. Wspom. 123, cited in Wilczewska (1966: 156), my translation) In (470), ci˛ez˙ ka˛ praca˛INS ’hard work’ is indeed the causee; it cannot designate the instrument, since one cannot say *zmógłem go czym´s ‘I overwhelmed him with hard work’. However, neither Szlifersztejnowa nor Bartnicka mention the possibility of introducing instrumental causees. It is difficult to find further examples with instrumental phrases in the corpus that unambiguously relate to causees, not to instruments of an unexpressed agent. The only candidates found among the examples extracted from the IPI PAN corpus were of the following type: (471) NieNEG dajcieI MP:2PL si˛eRFX:ACC bałamuci´cINF kłamstwamiINS prasy i radia. Don’t let yourself be confused by the lies of the press and the radio. What or who confuses? Are the ‘lies of the press and radio’ the agent or are they instruments of the metonymically present ‘liars’? A decision between causer and instrument is not possible here and in all other cases found in the corpus. Furthermore, substitution of datives by instrumentals in cases where an instrument reading is not possible is generally not felicitous, cf. for example: (472) DałemPS T :1S G si˛eRFX:ACC ponie´sc´ INF nerwomDAT / *nerwamiINS . ‘I let myself be carried away by nerves (DAT/*INS ’ I therefore conclude that instrumental causees are not generally possible in Polish today. 27. Note that with this verb, the argument may also be given in instrumental case, as this is the default case for means or effectors: (473) Straw dałPS T :3S G si˛eRFX:ACC przekona´cINF argumentemINS , z˙ e odwołanie poi jedynku naraziłoby na du˙ze straty [. . . ]. Straw was persuaded (lit.: let himself be persuaded by the argument) that a cancellation of the duel could lead to heavy losses.
28.
29.
30. 31.
In this case, however, the instrumental phrase is simply licensed by the semantics of the verb. The queries were: [base="daw?a?´c"][pos!=interp]{,5}[tag=".*inf.*"][pos!=interp]{,5} [orth="przez"] and [base="daw?a?´c"][pos!=interp]{,3}[orth="przez"][pos!=interp]{,3} [tag=".*inf.*"]. This list includes manipulowa´c ‘manipulate’, a verb that takes the instrumental in active voice, but appears with an accusative reflexive in this list. This is a behavior parallel to what is seen in the passive, where the instrumental object may be promoted to subject: przetarg był manipulowany ‘the tender was manipulated’. Query string: [base="daw?a?´c"][pos!=interp]{,3}[orth="sobie"][pos!=interp]{,5} [tag=".*inf.*"] The following example is a quotation from the 17th century: (474) – Jak zapisał Marcin Goli´nski – musieli´smy da´c nowe d˛ebowe szubienice troiste. (...) Dali d˛eby s´cia´ ˛c, pal sosnowy, na koniec ostrze dali´smy odrobi´c. These two instances concern a situation where the authorities have wood cut and a pole prepared for an execution. i Dziennik
Polski, 2000-01-29, Wracaj, skad ˛ przybyłe´s
Notes
309
32. “[C]zynno´sc´ po˙zadana˛ czy niepo˙zadana˛ (Szlifersztejnowa 1970: 78, my translation)”. 33. Query string: [base="daw?a?ć"][tag!="interp"]{,5}"do" "my´slenia|zrozumienia" 34. The relationship of impersonal reflexive si˛e and the non-agreeing modal passive involving da si˛e is not one of compositionality, as it might appear. To see this, consider that the impersonal si˛e construction takes a lexical verb as argument and demotes the agent: (475) Mówi si˛e ⇔ludzie mówia.˛ Say.3SG RFX.ACC people.NOM.PL say.3PL ‘One says ⇔people say’ Consequently, the combination of permissive da(va)t with impersonal si˛e yields generic letting: (476) Da mu si˛e to zrobi´c ⇔ludzie daja˛ mu let.SG he.DAT RFX.ACC that.ACC do.INF ⇔people.NOM.PL let.3PL he.ACC to zrobi´c that do.INF ‘One lets him do that - people let him do that’ The following corpus example from Dziennik Polski contains this combination: here, a generic agents lets somebody do something: (477) [. . . ] jak kto´s nagrał 98 audycji, to when somebody.NOM record.PST.3SG 98 auditions then RFX.ACC si˛e mu daje dojecha´c do setki. he.DAT let.3SG go.INF to 100 ‘[. . . ] when somebody already recorded 98 auditions, then one lets / people let him finish the hundred’ 35. Not all these criteria apply to nic ‘nothing’: nic in direct object function does not necessarily change to genitive in negative sentences (nicNOM /niczegoGEN nie zrobiono ‘nothing (nom./gen.) was done’), and can trigger neuter agreement (nicNOM:M tam nie byłoPS T :N ‘nothing was there’); see Mendoza (2004: 347). 36. “Konstrukcje te, na trwałe zwiazane ˛ z modalno´scia˛ aletyczna,˛ nie moga˛ wyra˙za´c [. . . ]ani modalno´sci deontycznej, ani epistemicznej[.] Rytel (1982)” 37. No analysis of the distribution of the construction types per genre was attempted. However, the parliament protocols do not represent an exception to the overall predominance of the non-agreeing construction. In the sample corpus (analyzed completely), 29 agreeing, and 183 non-agreeing constructions were found in texts from this genre. 38. There is a systematic bias for positive sentences to be ambiguous, since the genitive marking of the direct object under negation is in praxis an important diagnostic for an assignment of cases with transitive verbs to either the agreeing or non-agreeing type. However, even if we assume that all ambiguous instances were in fact undiagnosed agreeing modal passives and take them as one class, the proportion of negative instances is at only 183 of 546, that is, 33%. Non-agreeing modal passives involve overt negation to about 79%; again, if we add the ambiguous to the non-agreeing cases, this still results in a rate of 65%. 39. Estimated on the basis of the full corpus with infinitive immediately following mo˙zna (n=21,103) and nie mo˙zna (n=3,303). 40. To exclude the influence of individual lexemes, we restrict the calculation to infinitives
310 Notes that were only used once with this type in the corpus:
agreeing non-agreeing ambiguous intransitive
all 111 216 112 64
dawa´c 91 (82%) 6 (3%) 16 (14%) 4 (6%)
imperf. inf. 24 (22%) 34 (16%) 12 (11%) 19 (30%)
negated 48 (43%) 171 (79%) 30 (27%) 48 (75%)
41. Variation in person is only theoretically possible; first and second person usage almost always involves a change of meaning to the reflexive permissive. 42. E.g., in Russian there is the apparent aspectual pair moˇc’ and smoˇc’; however, as Holvoet 1989: 73f. puts it, smoˇc’ must be considered either the future variant of moˇc’ (*budet moˇc’ is not possible), or a different lexeme (with an additional element of achievement). See also the relevant section on the Russian modal system in Hansen (2001) as well as Barentsen (2002). 43. On the internet, however, examples can be found, if not very numerous. 44. According to Jagodzi´nski, the supposedly not acceptable analytical future needs to be used: Nie martw si˛e, kiedy´s b˛edziesz potrafił to zrobi´c; this is supported by my own informant work. Jagodzi´nski’s evidence is not conclusive in all details. However, in any case he shows that the position of the WSPP that the analytical future of potrafi´c is not acceptable is clearly not in accordance with current usage. (O polszczy´znie i poprawno´sci j˛ezykowej, http://grzegorj.w.interia.pl/lingwpl/mn.html, accessed on Oct. 4, 2007). 45. Search conducted with google, 4-10-2007. 46. A heterogeneous set of frequent irrelevant word forms such as przeszło, podział and dorosły were excluded. The query term: b˛edzie [tag!=interp &tag!="praet:.*imperf.*"&orth!="mo˙zna"] {,2} [tag="praet:.*:perf" & tag!=".*imperf.*" & orth!="przeszło|podział|zaciek.*|doros.*|dojrz.*"] Note that this query could not find all occurrences of this sort; also, I did not search for analytical future formed with the infinitive as the large number of infinitives governed by other verbs makes such a search impractical. In order to control for bias involved in this search, I used a different query for future forms such as b˛edzie/b˛edziemy da´c with the query "b˛ed.*" [orth!="mo˙zna"&pos!="interp"]{,3} [orth="da´c|dał[oay]|dali"]. Of 158 instances, only one proved relevant. It involved reflexive permissive da´c and will be discussed below. 47. All examples from the full IPI PAN query involving the analytical future of da´c si˛e are cited here for reference. Please refer to the IPI PAN corpus for more details. (478) Przez wadowicki rynek niedługo nie b˛edzie si˛e dało spokojnie przej´sc´ . (479) Poza tym zamierzamy wystapi´ ˛ c do Powiatowego Urz˛edu Pracy oraz do Gminnego O´srodka Pomocy Społecznej, mo˙ze tak˙ze z tej strony b˛edzie si˛e dało pomóc rodzinie Pauliny. (480) - No, przyprowadził, ale przecie wiadomo, co jakby ino młodzi mogli, to by chcieli sami mieszka´c i jak si˛e b˛edzie dało, to zara z chałupy uciekna˛ na swoje. (481) Z takim ju˙z zupełnie nie b˛edzie si˛e dało dogada´c. (482) Na pewno nie b˛edzie si˛e go dało tak po prostu odbudowa´c.
Notes
311
(483) Je´sli dzi´s zostanie podj˛eta decyzja o sprzeda˙zy cz˛es´ci hali, ju˙z jej nie b˛edzie dało si˛e odwróci´c. (484) Wielu strat (podobnie jak po powodzi w 1997 r.) nie b˛edzie si˛e dało prawdopodobnie zrekompensowa´c ju˙z nigdy. (485) Wcia˙ ˛z niektórzy inwestorzy licza,˛ z˙ e co´s b˛edzie si˛e dało zarobi´c w trakcie zamiany walut uzyskanych ze sprzeda˙zy TP SA. (486) Wszyscy z tych łak ˛ korzystamy, ka˙zdy chce si˛e przej´sc´ na spacer, wyprowadzi´c psa, ale jak tak dalej pójdzie to niedługo nie b˛edzie si˛e dało t˛edy w ogóle przej´sc´ - mówi mieszkanka osiedla. (487) Za darmo nie b˛edzie si˛e natomiast dało pozyska´c nowych mieszka´n. (488) Szkoleniowiec Boche´nskiego ostrzegał wówczas, z˙ e nie zawsze i nie ze wszystkim dru˙zynami b˛edzie si˛e tak dało gra´c. (489) - Pó´zniej tych zada´n nie b˛edzie si˛e dało fizycznie wykonywa´c, bo poczatek ˛ sierpnia jest tradycyjnym okresem pielgrzymkowym. (490) Mo˙ze mur - wi˛ec nawet na spacer nie b˛edzie si˛e dało przej´sc´ . (491) Jego dom znajduje si˛e na zboczu góry, która si˛e osuwa i nie wiadomo czy b˛edzie si˛e dało w nim mieszka´c. (492) - Prawdopodobnie nie b˛edzie si˛e tego dało zrobi´c z dnia na dzie´n, cho´cby z tego powodu, z˙ e w okolicach pasa˙zu handlowego trzeba wybudowa´c dodatkowa˛ zatok˛e autobusowa˛ w kierunku K˛et. (493) Teraz trzeba zrobi´c wszystko, z˙ eby inwestycj˛e doprowadzi´c do takiego punktu, z którego nie b˛edzie si˛e dało wycofa´c. 48. The null hypothesis is that mistakes happen on a random basis and that therefore mistakes involving da´c si˛e should be roughly proportional to its frequency of use. This was not investigated in detail, but note that the least one would then expect is that false b˛edzi˛e dał in its literal sense ‘will give’ should be attested on a comparable scale or more frequent. This is not the case; not one such mistake was found in the corpus. 49. In the course of these searches, frequent light verb constructions were filtered out by disallowing any of the following strings between da(va)t and complement verb: možnost, šanci, moc, odvahu, oprávnˇení, sílu, volnost, d˚uvod, nadˇeji, právo, motivaci, pˇríležitost, pˇríkaz, svobodu, pˇredpoklad; za cíl. 50. Four searches for matrix verb followed by infinitive within three verbs were conducted: negated and non-negated, with dá(va)t and nech(áv)at. Dá(va)t was found 231 times, of which 192 contained overt negation; of 184 cases with nech(áv)at, only 33 cases were negated. Cases with nech(áv)at often appeared with animate, agentive causer/permitter. 51. I estimate these to constitute fewer than 20% on the basis of manual inspection of 20 instances, which contained 3 modal passive uses. Note that the query yields all relevant constructions, plus false positives; it thus involves high recall and low precision. 52. Note that manipulovat ‘manipulate’ may be combined with accusative reflexive; this is parallel to such a construction with nech(áv)at. The reflexive pronoun is taken to refer to the sentence subject: (494) Petr dal/nechalPS T :3S G seRFX:ACC /sebouRFX:INS manipulovatINF . ‘Peter let (nechat/dát) himself (acc/instr.) be manipulated.’
312 Notes
53.
54. 55. 56.
All four combinations of matrix verb and reflexive pronoun are felicitous, but only with nech(áv)at were both combinations found: 8 cases with sebou, 19 with se. I take this to be a sign of high transitivity that plays a role both in the passive and the reflexive permissive; note that manipulovat may form the passive participle, too. Query templates with MTR = dá(va)t or nech(áv)at, VRBS = set of infinitives): [lemma="MTR"] [tag!="ˆ.:.*"]0,2 "si" [tag!="ˆ.:.*"]0,4 [tag="Vf.*"&lemma="VRBS"] and "si" [tag!="ˆ.:.*"]0,2 [lemma="MTR"] [tag!="ˆ.:.*"]0,4 [tag="Vf.*"&lemma="VRBS"] E.g., the query +"na""sebe" [lemma="dáv?a?t|nechá?v?at"] []0,3 "ˇcekat"+ in the full corpus yields 109 for nech(áv)at and 59 results for dá(va)t. [lemma="nechá?v?at|dáv?a?t"][tag!="^.:.*"]{0,5}"v[zy]niknout|v[zy]nikat" In view of the dearth of examples, a number of heuristic corpus searches in several maˇ jor CNK corpora was conducted for se dalo with an infinitive directly followed by an accusative phrase. Examples such as the following are not, as they seem to be, examples of the non-agreeing construction: (495) Za cˇ tvrt miliardy by se dalo udˇelat spoustu For quarter billion-GEN IRR RFX:ACC let-PST:3SG:N do-INF a-lot potˇrebných vˇecí. necessary-GEN:PL things-GEN:PL. ‘For the quarter of a billion a lot of necessry things could be done (lit. a lot of necessary things would let itself do)’ (Syn2000, mf990512)
57. 58.
59. 60.
Here, spoustu ‘a lot’ is a lexicalized quantifier, triggering third person singular neuter agreement, rather than a feminine noun in accusative case, as it appears to be in view of the desinence -u; the same is the case of trochu ‘a bit’, polohu ‘a half’. I am indebted to Markus Giger, Basel, for pointing this out to me. E.g., according to Frajzyngier (1982), passives of intransitive verbs generally differ from identically formed passives of transitive verbs in implying a human agent. ParaSol stands for Parallel Corpus of Slavic and Other Languages. At the time of analysis a first version of this corpus was used, then under the name Regensburg Parallel Corpus of Slavic languages. Note that in one case a reflexive permissive in Polish corresponded to a non-reflexive in Russian; this example is thus found in different groups in these two languages. In particular, this included • • • • •
dát se: 34/123 (28%) šlo, nešlo: 15/123 (12%); nelze, lze: 13/123 (11%); je možno, možné: 10/123 (8%); adjectives or adverbs with modal meaning components such as popsatelny (n=7), the modal construction je slyšet (n=6) as well as moci (n=4), together amounting to 12% of all cases.
61. This included: • mo˙zna: 80/255= 31% • da´c si˛e: 34/255= 13% • adjectives, adverbs and predicative elements with a modal component, e.g.
Notes
• • • •
313
forms of trudno, łatwo, wygodnie, etc.; wida´c: 24 /255=10% by´c do pomy´slenia, przewidzenia, stracenia, etc.: 10/255= 4% móc: 10/255= 4% modal uses of the infinitive: 7/255= 5% others: 15/255=8%
62. I am indebted to Yannis Kakridis, Bern, for pointing this out to me. 63. It is interesting to see that Lexicon Mediae et Infimae Latinitatis Polonorum, the dictionary of medieval Latin in Poland, cites factitive use of dare, evidently reflecting an influence of Polish on Latin. 64. Available from the website of the IRL RAN at http://pushkinskijdom.ru 65. This metaphorical phrase is opposed to the older people in the sentence before. 66. Note that in general, dát kˇrtíti ‘to have somebody baptized’ was annotated as ambiguous permissive/factitive in most other examples. 67. I am indebted to František Martínek for pointing this attestation out to me and for extensive comments on the examples in this section. 68. Korpus tekstów staropolskich do roku 1500, available at http://www.ijp-pan. krakow.pl/pl/publikacje-elektroniczne/korpus-tekstow-staropolskich (13.12.2011). 69. Marek Adamiec (ed.): Wirtualna biblioteka literatury polskiej, available at http:// univ.gda.pl/~literat/ (13.12.2011) 70. I da warkoczom płyna´ ˛c przez ramiona translates When left to roll its folds below from Byron’s Giaur. 71. Cf.: A jeden pan Lupa Podlodowski, któremu sie jako Włoszy sami, tak rzeczy od nich wynalezione nie podobaja,˛ miałby wielka˛ pogod˛e strofowa´c to wszytko . . . (Mikołaj Rej). 72. Note Martin Luther’s translation here where lassen also does not have an equivalent in the Latin text: Wenn du Almosen gibst, sollst du nicht lassen vor dir posaunen, wie die Heuchler tun in den Schulen und auf den Gassen, auf daß sie von den Leuten gepriesen werden. I have not examined the Greek original. 73. “[. . . ] teraz rzadziej si˛e u˙zywa da´c zamiast kaza´c; wbrew za´s przeciw Polszczy´znie Niemcy swoje bitten lassen tłumacza˛ da´c prosi´c” (Linde 1854-1860: 402). 74. Note that there are other uses of da´c (si˛e) widzie´c in the sense of to show (oneself) that were annotated as causation with cognitive verbs and will not be discussed here 75. A single subjectless example was found in the Slownik j˛ezyka Adama Mickiewicza: (496) . . . stanowisko [. . . ] z którego da si˛e działa´c skutecznie w sprawie polskiej, najwi˛ekszej na s´wiecie. (1847, L3 153). ‘A position [. . . ] from which one can (lit. it lets2 itself 1 ) act effectively regarding the Polish question, the most important on earth.’ 76. Note that in Upper Sorbian da(va)t seems to follow the pattern of lassen more closely, being used also in type I syntax with accusative causee. 77. Accordingly, the Inny Słownik J˛ezyka Polskiego dictionary groups the personal modal passive with the reflexive permissive and has a separate entry only for the subjectless use of da´c si˛e.
Bibliography
Alsina, Alex. 1992. On the argument structure of causatives. Linguistic Inquiry, 23(4), 517–555. Łazi´nski, Marek. 1996. Bezokolicznik czasownika dokonanego jako człon wymagany w zdaniu. Poradnik J˛ezykowy, 21–29. Baayen, R. Harald. 2002. Word Frequency Distributions. Text, Speech and Language Technology, no. 18. Dordrecht et al.: Kluwer. Babby, Leonard H. 1983. The relation between causative and voice: Russian vs. Turkish. Wiener slawistischer Almanach, 11, 61–88. Barentsen, Adrian. 2002. O russkix glagolax smoˇc’ i sumet’. Pages 7–29 of: Ivanickij, V.V. (ed), Osnovnye problemy russkoj aspektologii. SanktPeterburg: Nauka. Barentsen, Adrian. 2003a. O nekotoryx izmenenijax v upotreblenii vyraženij prizyva k sovmestnomu dejstviju v russkix tekstax poslednix dvux stoletij. Pages 13–40 of: Honselaar, Wim, van der Tak, Hans, de Haard, Eric, Barentsen, Adrie, & Popovski, Josif (eds), Time flies. A Festschrift for William R. Veder. Amsterdam: Pegasus. Barentsen, Adrian. 2003b. O pobuditel’nyx konstrukcijax s ispolnitelem 1-go lica. Amsterdam, New York: Rodopi. Barentsen, Adrian. 2006. K voprosu o vidovoj oppozicii v konstrukcijax tipa daj pomogu - davaj pomogu. Pages 37–66 of: Lehmann, Volkmar (ed), Glagol’nyj vid i leksikografija. Semantika i struktura slavjanskogo vida IV. Slavolinguistica, no. 7. München: Otto Sagner. Bartnicka, Barbara. 1982. Funkcje semantyczno-składniowe bezokolicznika we współczesnej polszczy´znie. Wrocław et al.: ZNiO. Bartnicka, Barbara, Hansen, Björn, Klemm, Wojtek, Lehmann, Volkmar, & Satkiewicz, Halina (eds). 2004. Grammatik des Polnischen. Slavolinguistica, no. 5. München: Otto Sagner.
316 Bibliography Bauer, Alter, & Aland, Kurt und Barbara. 1988. Griechischdeutsches Wörterbuch zu den Schriften des Neuen Testaments und der frühchristlichen Literatur. 6., rev. edn. Berlin, New York: Walter de Gruyter. Bogusławski, Andrzej. 2000. Die slawische Korrelation: Aspekte – Negation. Pages 125–138 of: Katny, ˛ Andrzej (ed), Aspektualität in germanischen und slawischen Sprachen. Pozna´n: Wydawnictwo UAN. Bondzio, Wilhelm. 1959. Die Herausbildung der präpositionalen Konstruktion beim Verbum "lassen". Wissenschaftliche Zeitschrift der HumboldtUniversität Berlin. Gesellschafts- und sprachwissenschaftliche Reihe, IX(3), 205–219. Bordelois, Ivonne. 1988. Causatives: from lexicon to syntax. Natural Language and Linguistic Theory, 6, 57–93. Brus. 1894. Brus jazyka cˇ eského. Praha. Bulygina, Tatjana V., & Shmelev, Aleksej D. 1999. On the semantics of some Russian causative constructions. Aspect, control, and Types of causation. Pages 98–113 of: Abraham, Werner, & Kulikov, Leonid (eds), Tenseaspect, transitivity and causativity: essays in honor of Vladimir Nedjalkov. Studies in language: Companion series, vol. 50. Amsterdam etc.: John Benjamins. Bybee, Joan. 2003. Mechanisms of change in grammaticalization: The role of frequency. Pages 602–623 of: Joseph & Janda (2003). Bybee, Joan L., Pagliuca, William, & Perkins, Revere D. 1994. The Evolution of grammar: tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Campbell, Lyle. 2000. What’s wrong with grammaticalization? Language Sciences, 23(2/3), 113–161. Cejtlin, Ralja M., Veˇcerka, Radoslav, & Bláhová, Emilija. 1994. Staroslavjanskij slovar’ po rukopisjam 10–11 vv. Moskva: Russkij jazyk. Chappell, Hilary, & Peyraube, Alain. 2006. The Analytic Causatives of Early Modern Southern Min in Diachronic Perspective. Pages 973–1012 of: Ho,
Bibliography
317
D.-A., Cheung, H.S., Pan, W., & WuK, F. (eds), Linguistic studies in Chinese and neighboring languages. Taiwan: Academia Sinica. Cole, Peter. 1983. The grammatical role of the causee in universal grammar. International Journal of American Linguistics, 49, 115–133. Comrie, B. 1985. Causative verb formation and other verb-deriving morphology. Pages 309–348 of: Shopen, T. (ed), Language typology and syntactic description, vol.3, Grammatical categories and the lexicon. Cambridge: Cambridge University Press. Comrie, Bernard. 1976. The syntax of causative constructions: crosslinguistic similarities and divergences. Pages 261–312 of: Shibatani (1976a). Croft, William A. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. da Silva, Augusto Soares. 2007. Verbs of letting: Some cognitive and historical aspects. Pages 171–200 of: Delbecque, Nicole, & Cornille, Bert (eds), From Action and Motion to Transitivity and Causality. Berlin, New York: Mouton de Gruyter. de Haan, Ferdinand. 1997. The Interaction of Modality and Negation: A Typological Study. New York: Garland Press. de Haan, Ferdinand. 2002. Strong Modality and Negation in Russian. In: Reppen, R., Fitzmaurice, S., & Biber, D. (eds), Using Corpora to explain Linguistic Variation. Amsterdam: John Benjamins. de Haan, Ferdinand. 2005. Modality in Slavic and semantic maps. Pages 3–24 of: Hansen, Björn, & Karlík, Petr (eds), Modality in Slavonic Languages. New perspectives. Amsterdam: Sagner. Döhmann, Karl. 1974. Die sprachliche Darstellung der Modalfunktoren. Pages 57–91 of: Menne, Albert, & Frey, Gerhard (eds), Logik und Sprache. Bern, München: Francke. Dickey, Stephen M. 2000. Parameters of Slavic aspect. a cognitive approach. Stanford: CSLI Publications.
318 Bibliography Dixon, R.M.W. 2000. A typology of causative constructions: form, syntax and meaning. Pages 30–83 of: Dixon, Robert M. W., & Aikhenvald, Alexandra Y. (eds), Changing valency: case studies in transitivity. Cambridge: Cambridge University Press. Eckardt, Regine. 2006. Meaning change in grammaticalization. An enquiry into semantic reanalysis. Oxford: Oxford University Press. Evert, Stefan, & Lüdeling, Anke. 2001. Measuring morphological productivity: Is automatic preprocessing sufficient? Pages 167–175 of: Rayson, P., Wilson, A., McEnery, T., Hardie, A., & Khoja, S. (eds), Proceedings of the Corpus Linguistics 2001 Conference. Lancaster: UCREL. Fielder, Grace. 1984. The Aspect of the Russian Infinitive Complement. Ph.D. thesis, University of California at Los Angeles. Fillmore, Charles J., Kay, Paul, & O’Connor, Mary C. 1988. Regularity and idiomaticity in grammatical constructions: The case of let alone. Language, 64(3), 501–538. Fischer, Olga, Kemenade, Ans Van, Koopman, Willem, & der Wurff, Wim Van. 2000. Pathways of change. Grammaticalization in English. Studies in language: Companion series, no. 53. Amsterdam: Benjamins. Folli, Raffaella, & Harley, Heidi. 2007. Causation, obligation and argument structure: On the nature of little v. Linguistic Inquiry, 38.2, 197–238. Frajzyngier, Zygmuni. 1982. Indefinite agent, passive and impersonal passive: A functional study. Lingua, 58, 267–290. Frajzyngier, Zygmunt. 1999. Domains of point of view and coreferentiality: system interaction approach to the study of reflexives. Pages 125–153 of: Frajzyngier, Zygmunt, & Curl, Traci S. (eds), Reflexives. Forms and Functions. Typological Studies in Language, no. 40. Amsterdam: John Benjamins. Gehrmann, Maria. 1981. Semantischer Inhalt und syntaktische Verbindungen der kausativen Verben im Deutschen und Polnischen. Zeitschrift für Slavistik, 26(1), 26–29. Gehrmann, Maria. 1983. Zu den polnischen Entsprechungen modalpassivischer Konstruktionen des Deutschen. Zeitschrift für Slavistik, 28(1), 9–17.
Bibliography
319
Geniušien˙e, Emma. 1987. The Typology of Reflexives. Empirical Approaches to Language Typology, no. 2. Berlin, New York, Amsterdam: Mouton de Gruyter. Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Grepl, Miroslav, & Karlík, Petr. 1983. Gramatické prostˇredky hierarchizace sémantické struktury vˇety. Brno: Univerzita J.E. Purkinˇe. Guasti, Maria-Teresa. 1996. Semantic restrictions in Romance Causatives and the Incorporation Approach. Linguistic Inquiry, 27, 294–312. Gunkel, Lutz. 1998. Causatives in German. Pages 65–79 of: Meurers, Detmar, & Kiss, Tibor (eds), Proceedings of the ESSLLI-98 Workshop on Current Topics in Constraint-Based Theories of Germanic Syntax. Saarbrücken. Hansen, Björn. 2001. Das slavische Modalauxiliar. Semantik und Grammatikalisierung im Russischen, Polnischen, Serbischen/Kroatischen und Altkirchenslavischen. Slavolinguistica, vol. 2. München: Sagner. Hansen, Björn. 2004. The boundaries of grammaticalization. The case of modals in Russian, Polish and Serbian/Croatian. Pages 245–271 of: Wiemer et al. (2004). Hansen, Björn, & de Haan, Ferdinand. 2009. Concluding chapter: Modals Constructions in the languages of Europe. Pages 511–560 of: Hansen, Björn, & de Haan, Ferdinand (eds), Modals in the Languages of Europe. A Reference Work. Mouton de Gruyter. Hashimoto, K. 1989. Zu den Kausativkonstruktionen im Deutschen und Japanischen. Eine kontrastive Untersuchung am Beispiel der lassen- und sase-Konstruktionen. Arbeitspapiere der Universität Bern, no. 26. Haspelmath, Martin. 1989. From purposive to infinitive — a universal path of grammaticization. Folia Linguistica Historica, 10, 287–310. Haspelmath, Martin. 1990. The grammaticization of passive morphology. Studies in Language, 14(1), 25–72.
320 Bibliography Haspelmath, Martin. 1998. Does grammaticalization need reanalysis? Studies in Language, 22(2), 49–85. Haspelmath, Martin. 1999. Optimality and diachronic adaptation. Zeitschrift für Sprachwissenschaft, 18(2), 180–205. Haspelmath, Martin. 2001. The European linguistic area: standard average European. In: Haspelmath et al. (2001), 1492–1510. Haspelmath, Martin. 2003. The geometry of grammatical meaning: semantic maps and cross-linguistic comparison. Pages 211–42 of: Tomasello, M. (ed), The new psychology of language: Cognitive and functional approaches to language structure. Vol. 2. Mahwah, NJ: Laurence Erlbaum Associates. Haspelmath, Martin, König, Ekkehard, Oesterreicher, Wulf, & Raible, Wolfgang (eds). 2001. Language typology and language universals. Sprachtypologie und sprachliche Universalien. La typologie des langues et les universaux linguistiques. An international handbook / Ein internationales Handbuch / Manuel international. Vol. 2. Berlin, New York: Walter de Gruyter. Heine, Bernd. 1993. Auxiliaries. Cognitive forces and grammaticalization. New York, Oxford: Oxford University Press. Heine, Bernd. 2003. Grammaticalization. Pages 575–601 of: Joseph & Janda (2003). Heine, Bernd, & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge University Press. Heine, Bernd, Claudi, Ulrike, & Hünnemeyer, Friederike. 1991. Grammaticalization: A conceptual framework. Chicago: University of Chicago Press. Hoffmann, Sebastian. 2004. Are low-frequency complex prepositions grammaticalized? On the limits of corpus data — and the importance of intuition. Page 171–210 of: Lindquist, Hans, & Mair, Christian (eds), Corpus Approaches to Grammaticalization in English. John Benjamins. Holvoet, Axel. 1989. Aspekt a modalno´sc´ w jezyku polskim. na tle ogólnosłowia´nskim. Prace slawistyczne, vol. 77. Wrocław et al.: Ossolineum.
Bibliography
321
Hopper, Paul J., & Thompson, Sandra A. 1980. Transitivity in grammar and discourse. Language, 56, 251–300. Hopper, Paul J., & Traugott, Elizabeth C. 1993. Grammaticalization. Cambridge: Cambridge University Press. Horn, Laurence R. 1989. A natural history of negation. Chicago: University of Chicago Press. Horn, Laurence R. Fall 2006. Contradiction. In: Zalta, Edward N. (ed), The Stanford Encyclopedia of Philosophy. Ide, Manshu. 1996. Lassen und lâzen: eine diachrone Typologie des kausativen Satzbaus. Würzburg: Königshausen und Neumann. Jagi´c, Vatroslav. 1883. Quattuor evangeliorum versionis palaeoslovenicae codex Marianus glagoliticus. (Repr. Graz 1960). Berolini, Sanktpeterburg. Joseph, Brian D. 2000. Is there such a thing as "grammaticalization?". Language Sciences, 23(2-3), 163–186. Joseph, Brians, & Janda, Richard (eds). 2003. The Handbook of Historical Linguistics. Blackwell Handbooks in Linguistics. Malden: Blackwell. Kemmer, Suzanne. 1993. The middle voice. Typological studies in language, no. 23. Amsterdam, Philadelphia: John Benjamins. Kemmer, Suzanne, & Verhagen, Arie. 1994. The grammar of causatives and the conceptual structure of events. Cognitive Linguistics, 5(2), 115–156. Klemensiewicz, Zenon. 1999. Historia j˛ezyka polskiego. 7th edn. Warszawa: Wydawnictwo naukowe PWN. Knott, Judith. 1995. The causative-passive correlation. Pages 53–59 of: CBennett, David, Bynon, Theodora, & Hewitt, B. George (eds), Subject, voice and ergativity. Selected essays. London: School of Oriental and African Studies. Koo, Myung-Chul. 1997. Kausativ und Passiv im Deutschen. Frankfurt a.M.: Peter Lang. Koschmieder, Erwin. 1987. Aspektologie des Polnischen. Selecta Slavica, vol. 11. Neuried: Hieronymus-Verl.
322 Bibliography Kozinsky, Isaac, & Polinsky, Maria. 1993. Causee and patient in the causative of transitive: coding conflict or doubling of grammatical relations? Pages 177–240 of: Comrie, Bernard, & Polinsky, Maria (eds), Causatives and Transitivity. John Benjamins. Kudli´nska-St˛epie´n, Halina. 1995. Ob aktualizacii vidovogo znaˇcenija sinkretiˇceskix form dvuvidovyx glagolov (na materiale russkogo i pol’skogo jazykov). Pages 375–387 of: Jelitte, Herbert, & Troškina, Tat’jana (eds), Innerslavischer und deutsch-slavischer Sprachvergleich. Beiträge zur Slavistik, vol. 27. Frankfurt a. M. et al.: Peter Lang. Kulikov, Leonid I. 2001. Causatives. Pages 886–898 of: Haspelmath et al. (2001). Kural, Murat. 1996. Verb Incorporation and Elementary Predicates. Ph.D. thesis, University of California at Los Angeles. Kural, Murat. 2000. Modality in Causatives. In: Juge, Matthew L., & Moxley, Jeri L. (eds), Proceedings of the 23rd Meeting of the Berkeley Linguistics Society, 1997. Berkeley Linguistics Society. Lehmann, Christian. 1988. Towards a typology of clause linkage. Pages 181– 225 of: Haiman, John, & Thompson, Sandra A. (eds), Clause combining in discourse and grammar. Proceedings of a conference at Rensselaerville Institute, Albany, N.Y. Typological Studies in Language, no. 18. Amsterdam: Benjamins. Lehmann, Christian. 1991. Predicate classes and PARTICIPATION. Pages 183–239 of: Seiler, Hansjakob, & Premper, Waldfried (eds), Partizipation: das sprachliche Erfassen von Sachverhalten. Tübingen: Günter Narr. Lehmann, Christian. 2002. New reflections on grammaticalization and lexicalization. Pages 1–18 of: Wischer & Diewald (2002). Lehmann, Christian. 20022 . Thoughts on grammaticalization. First published 1982, rev. edn. Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt, no. 9. Lehmann, Christian. 2005. Theory and method in grammaticalization. Zeitschrift für Germanistische Linguistik, 32(5), 152–187.
Bibliography
323
Lehmann, Volkmar. 1999. Der Aspekt. Pages 214–242 of: Jachnow, Helmut (ed), Handbuch der sprachwissenschaftlichen Russistik und ihrer Grenzdisziplinen. Harassowitz. Leino, Jaakko. 2005. Frames, profiles and constructions: Two collaborating CGs meet the Finnish Permissive Construction. Pages 87–120 of: Östman, Jan-Ola, & Fried, Mirjam (eds), Construction Grammars. Cognitive grounding and theoretical extensions. Amsterdam: John Benjamins. Lewis, Charlton T., & Short, Charles. 1879. A Latin Dictionary. Founded on Andrews’ edition of Freund’s Latin dictionary. Revised, enlarged, and in great part rewritten by Charlton T. Lewis, Ph.D. and. Charles Short, LL.D. Oxford: Clarendon Press. Liddell, Henry George, & Scott, Robert. 1940. A Greek-English Lexicon. Revised and augmented throughout by Sir Henry Stuart Jones with the assistance of Roderick McKenzie. Oxford: Clarendon Press. Linde, Samuel Bogumil. 1854-1860. Słownik j˛ezyka Polskiego. 2nd edn. Löbner, Sebastian. 1990. Wahr neben falsch. Duale Operatoren als die Quantoren natürlicher Sprache. Linguistische Arbeiten, no. 244. Tübingen: Niemeyer. Loewenthal, Judith. 2003. Meaning and use of causeeless causative constructions with laten in Dutch. Pages 97–129 of: Verhagen, Arie, & van de Weijer, Jeroen (eds), Usage-Based Approaches to Dutch. Lexicon, grammar, discourse. Utrecht: LOT. Lord, Carol, Yap, Foong Ha, & Iwasaki, Shoichi. 2002. Grammaticalization of ‘give’: African and Asian perspectives. Pages 217–235 of: Wischer & Diewald (2002). Lötzsch, Ronald. 1972. Zum Ausdruck der Kausativität in den slawischen Sprachen. Zeitschrift für Slawistik, 17(3), 360–70. Majsak, Timur A. 2005. Tipologija grammatikalizacii konstrukcij s glagolami dviženija i pozicii. Moskva: Jazyki slavjanskich kul’tur. Markowski, Andrzej (ed). 2000. Nowy słownik poprawnej polszczyzny PWN. 2nd edn. Warszawa: PWN.
324 Bibliography Medová, Lucie. 2006 (April). Complex structures with se. Unpublished manuscript. Mendoza, Imke. 2004. Nominaldetermination im Polnischen. Die primären Ausdrucksmittel. Habilitationsschrift, München. ´ Smiech, Witold. 1971. Funkcje aspektów czasownikowych we współczesnym j˛ezyku ogólnopolskim. Łód´z: LTN. Nedjalkov, Vladimir P. 1971. Kauzativnye konstrukcii v nemeckom jazyke – Analytiˇceskij kauzativ. Leningrad: Nauka. Nedjalkov, Vladimir P. 1976. Kausativkonstruktionen. Studien zur deutschen Grammatik, no. 4. Tübingen: TBL. Nedjalkov, Vladimir P., & Nikitina, T. N. 1965. O priznakax analitiˇcnosti i služebnosti (na materiale kauzativnyx konstrukcij). Pages 170–193 of: Žirmunskij, V. M., & Sunik, O. P. (eds), Analitiˇceskie konstrukcii v jazykax razliˇcnych tipov. Leningrad: Nauka. Nedjalkov, Vladimir P., & Sil’nickij, Georgij G. 1969. Tipologija kauzativnyx konstrukcij. Tipologija morphologiˇceskogo i leksiˇceskogo kauzativov. Pages 5–50 of: Xolodoviˇc, A.A. (ed), Tipologija kauzativnyx konstrukcij. Morfologiˇceskij Kauzativ. Leningrad: Nauka. Newman, John. 1996. Give. A cognitive linguistic study. Berlin, New York: Mouton de Gruyter. Paduˇceva, Elena V. 1994. Tipy kauzativnych otnošenij v semantiˇceskoj strukture odnoj leksemy. Russian Linguistics, 18(1), 1–16. Palmer, Frank R. 20012 . Mood and modality (2nd edition). Cambridge Textbooks in Linguistics. New York: Cambridge University Press. Panevová, Jarmila. 2003. O jednom typu kauzativní konstrukce v cˇ eštinˇe. Pages 379–385 of: Bany´s, W., Bednarczuk, L., & Pola´nski, K. (eds), Etudes Linguistique Romano-Slaves offertes à Stanisław Karolak. Cracovie: Edukacja. Pennanen, Esko V. 1986. On the so-called curative verbs in Finnish. FinnischUgrische Forschungen, 47(2-3), 163–182.
Bibliography
325
Pereltsvaig, Asya. 2006. Negative Polarity Items in Russian and the “Bagel Problem”. Pages 153–178 of: Przepiorkowski, Adam, & Brown, Sue (eds), Negation in Slavic. Bloomington: Slavica. Perlin, Jacek. 2003. Czas tera´zniejszy w funkcji przyszłego w polszczy´znie: szczegółowe warunki semantycznej neutralizacji czasowo-aspektowej. Poradnik J˛ezykowy, 19–29. Pisarkowa, Krystyna. 1984. Historia składni j˛ezyka polskiego. Prace Instytuta J˛ezyka Polskiego, vol. 52. Wrocław et al.: Ossolineum. Piwowar, Jolanta. 2002. Próba klasyfikacji polskich czasowników dwuaspektowych i dwuaspektowych homonimów czasownikowych. Prace J˛ezykoznawcze (Olsztyn), 4, 47–58. Podlesskaya, Vera I. 2004. Glagol’nye kategorii vo vzaimodejstvii: russkie permissivnye konstrukcii so vspomagatel’nym glagolom dat’. Pages 412– 428 of: Volodin, A. P. (ed), Tipologiˇceskie obosnovanija v grammatike. K 70-letiju V. S. Xrakovskogo. Moskva: Znak. Podlesskaya, Vera I. 2005a. ‘Give´-verbs as permissive auxiliaries in Russian. Sprachtypologie und Universalienforschung, 58(1), 124–138. Podlesskaya, Vera I. 2005b. Russkie glagoly dat’/davat’: ot prjamyx upotreblenij k grammatikalizovannym. Voprosy Jazykoznanja, 49(2), 89–103. Podlesskaya, Vera I. 2006. Auxiliation of ‘give’ verbs in Russian. Discourse evidence for Grammaticalization. Page 271–298 of: Tsunoda, Tasaku, & Kageyama, Taro (eds), Voice and Grammatical Relations. In Honor of Masayoshi Shibatani. Typological Studies in Language, vol. 65. John Benjamins. Rappaport, Gilbert C. 1986. On anaphora binding in Russian. Natural language and linguistic theory, 4, 97–120. Reis, Marga. 1976. Reflexivierung in deutschen AcI-Konstruktionen. Ein transformationsgrammatisches Dilemma. Papiere zur Linguistik, 9, 5–82. Rozina, Raisa I. 2000. Ustranenie pregrady. Semantika i aspektual’noe povedenie gruppy russkich glagolov. Pages 67–77 of: Artjunova, N.D., & Levontina, I.B. (eds), Logiˇceskij analiz jazyka. Jazyki prostranstv. Moskva: Jazyki russkoj kul’tury.
326 Bibliography Rytel, Danuta. 1982. Leksykalne s´rodki wyra˙zania modalno´sci w j˛ezyku czeskim i polskim. Monografie slawistyczne, vol. 47. Wrocław et al.: Ossolineum. Saksena, Anuradha. 1982. Topics in the Analysis of Causatives with an Account of Hindi Paradigms. University of California Press. ´ Saloni, Zygmunt, & Swidzi´ nski, Marek. 2004. j˛ezyka polskiego. 4., rev. edn. Warsaw: PWN.
Składnia współczesnego
Shibatani, Masayoshi (ed). 1976a. The grammar of causative constructions. Syntax and Semantics, no. 6. New York: Academic Press. Shibatani, Masayoshi. 1976b. The grammar of causative constructions: a conspectus. Pages 1–40 of: Shibatani (1976a). Shibatani, Masayoshi, & Pardeshi, Prashant. 2002. The causative continuum. Pages 85–126 of: Shibatani, Masayoshi (ed), The grammar of causation and interpersonal manipulation. Typological Studies in Language, no. 48. Amsterdam, Philadelphia: John Benjamins. Siewierska, Anna. 1984. The passive. A comparative linguistic analysis. London: Coom Helm. Siewierska, Anna. 1988. The passive in Slavic. Pages 243–289 of: Shibatani, Masayoshi (ed), Passive and Voice. Typological Studies in Language, no. 16. Amsterdam, Philadelphia: John Benjamins. Skoumalová, Hana. 2003. Multiverb expressions in Czech. In: Beermann, Dorothee, & Hellan, Lars (eds), Proceedings of the workshop on MultiVerb constructions, Trondheim Summer School 2003. Suchsland, Peter. 1987. Zur Syntax und Semantik von lassen. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung, 40(5), 652–667. Szlifersztejnowa, Salomea. 1968. Bierne czasowniki zaimkowe (reflexiva) w j˛ezyku polskim. Wrocław: Wydawnictwo Polskiej Akademii Nauk. Szlifersztejnowa, Salomea. 1970. O polskim causativum „analitycznym”. Prace Filologiczne, 20, 72–79.
Bibliography
327
Talmy, Leonard. 1988. Force Dynamics in Language and Cognition. Cognitive Science, 12(1), 49–100. Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol. 1. MIT press. Testelec, Jakov G. 2001. Vvedenie v obšˇcij sintaksis. Moskva: RGGU. Tomasello, Michael. 2003. Constructing a Language. A Usage-Based Theory of Language Acquisition. Harvard University Press. Toops, Gary H. 1985. Grammatival Causativity in Slavic. Ph.D. thesis, Yale. Toops, Gary H. 1987. Russian Contextual Causatives. Slavic and East European Journal, 31(4), 594–611. Toops, Gary H. 1989. The Syntax and Semantics of Lithuanian Curative Constructions. Pages 249–282 of: Aronson, Howard I. (ed), The Non-Slavic Languages of the USSR. Linguistic Studies. Chicago: Chicago Linguistic Society. Toops, Gary H. 1991. Russian da(va)t’ + Infinitive: ‘To Give’, ‘To Let’, or ‘To Have’? Russian Language Journal, 45(4), 39–53. Toops, Gary H. 1992. Causativity in Czech: The verbs dá(va)t and nech(áv)at. Canadian Slavonic Papers, 34(1-2), 39–56. Toops, Gary H. 1993. Causative Constructions in the Upper Sorbian Literary Language. Die Welt der Slaven, 102–136. Topoli´nska, Zuzanna. 1993. The verb da´c/dawa´c in Polish periphrastic constructions. Pages 229–249 of: Hentschel, Gerd, & Laskowski, Roman (eds), Studies in Polish morphology and syntax. Twardzik, Wacław, & Keller, Felix (eds). 1998, 2000, 2004. Rozmy´slanie przemyskie. Transliteration, Transkription, lateinische Vorlagen, deutsche Übersetzung / Transliteracja, transkrypcja, podstawa lacinska, niemiecki przekład. Monumenta Linguae Slavicae, vol. XL, XLII, XLIII. Freiburg i. Br.: Weiher. van der Auwera, Johan. 1985. Language and logic. A speculative and condition-theoretic study. Pragmatics and Beyond Companion Series, no. 3. Amsterdam & Philadelphia: John Benjamins.
328 Bibliography van der Auwera, Johan. 1996. Modality: the three-layered scalar square. Journal of Semantics, 13, 181–195. van der Auwera, Johan. 2000. On the typology of negative modals. Pages 23–48 of: Hoeksema, Jack, Rullmann, Hotze, Sanchez-Velencia, Victor, & van der Wouden, Ton (eds), Perspectives on negation and polarity items. Linguistik aktuell, vol. 40. Amsterdam: Benjamins. Van Valin, Robert D., & Wilkins, David P. 1996. The case for “Effector”: Case roles, agents and agency revisited. Pages 289–322 of: Shibatani, Masayoshi, & Thompson, Sandra A. (eds), Grammatical Constructions: Their Form and Meaning. New York: Oxford University Press. von der Gabelentz, Hans. 1861. Über das Passivum. Eine sprachvergleichende Abhandlung. Abhandlungen der Königlich-Sächsischen Gesellschaft der Wissenschaften, 450–546. von Waldenfels, Ruprecht. 2003. Die mit lassen gebildete deutsche analytische Kausativkonstruktion und ihre Äquivalente im Russischen. M.Phil. thesis, Humboldt Universität zu Berlin. von Waldenfels, Ruprecht. 2004. The grammaticalization of Finnish antaa and Russian davat’ ‘to give’ as causative auxiliaries. M.Phil. thesis, Freie Universität zu Berlin. von Waldenfels, Ruprecht. 2006. Compiling a parallel corpus of Slavic languages. Text strategies, tools and the question of lemmatization in alignment. Pages 123–138 of: Brehmer, Bernhard, Ždanova, Vladislava, & Zimny, Rafał (eds), Beiträge der Europäischen Slavistischen Linguistik (POLYSLAV) 9. von Waldenfels, Ruprecht. 2011. Recent Developments in ParaSol: Breadth for Depth and XSLT based web concordancing with CWB. Pages 156–162 of: Majchráková, Daniela, & Garabík, Radovan (eds), Natural Language Processing, Multilinguality. Proceedings of Slovko 2011, Modra, Slovakia, 20–21 October 2011. Bratislava: Tribun EU. von Waldenfels, Ruprecht. forthc.. Finnish antaa and Russian davat’ ‘to give’ as causatives: a contrastive analysis. In: Leino, Jaakko, & von Waldenfels, Ruprecht (eds), Analytical causatives: from makin to laskma. Munich: Lincom Europa.
Bibliography
329
von Wright, Georg Henrik. 1963. Norm and action. A logical enquiry. London et al.: Routledge & Paul. Wachowicz, Krystyna. 1976. Some universal properties of morphological causatives. Working papers on language universals (Stanford), 20, 59–106. Weiher, Eckhard (ed). 1987. Die Dogmatik des Johannes von Damaskus in der kirchenslavischen Übersetzung des 14. Jahrhunderts. Monumenta linguae slavicae dialecti veteris, vol. LX. Freiburg i. Br.: Weiher. Weiss, Daniel. 1983. Zur typologischen Stellung des Polnischen (ein Verˇ gleich mit dem Cechischen und Russischen). Pages 229–261 of: Brang, P., Nivat, G., & Zett, R. (eds), Schweizerische Beiträge zum IX. Internationalen Slavistenkongreß in Kiev, September 1983. Slavica Helvetica, no. 22. Bern, Frankfurt: Peter Lang. Weiss, Daniel. 1984. Kongruenz vs. Kongruenzlosigkeit: zur typologischen Entwicklung des Polnischen. Zeitschrift für Slavische Philologie, 44(1), 144–192. Weiss, Daniel. 1999. Ob odnom predloge, sdelavšem blestjašˇcuju kar’eru (vopros o vozmožnom agentivnom znaˇcenii modeli "u + imjarod"). Pages 173–186 of: Tipologija i teorija jazyka. Ot opisanija k ob"jasneniü (sbornik v cˇ est’ A.E.Kibrika). Wiemer, Björn. 2001. Aspect choice in non-declarative and modalized utterances as extensions from assertive domains – lexical semantics, scopes, and categorial distinctions in Russian and Polish. Pages 195–221 of: Bartels, H., Störmer, N., & Walusiak, E. (eds), Untersuchungen zur Morphologie und Syntax im Slavischen : Beiträge zur dritten Tagung der Europäischen Slavistischen Linguistik (POLYSLAV) Toru´n / Thorn 1999. Oldenburg: BIS. Wiemer, Björn. 2004. The evolution of passives as grammatical constructions in Northern Slavic and Baltic languages. Pages 271–331 of: Wiemer et al. (2004). Wiemer, Björn, Bisang, Walter, & Himmelmann, Nikolaus P. (eds). 2004. What makes grammaticalization? A look from its fringes and its components. Berlin, New York: Mouton de Gruyter.
330 Bibliography Wierzbicka, Anna. 1998. The semantics of English causative constructions in a universal-typological perspective. Pages 113–153 of: Tomasello, M. (ed), The new psychology of language: Cognitive and functional approaches to language structure. Mahwah, NJ: Laurence Erlbaum Associates. Wierzbicka, Anna. 2002. English causative constructions in an ethnosyntactic perspective: focusing on let. Pages 162–203 of: Enfield, N. J. (ed), Ethnosyntax. Explorations in grammar and culture. Oxford: Oxford University Press. Wilczewska, Krystyna. 1966. szczy´znie. Toru´n: TNT.
Czasowniki zwrotne we współczesnej pol-
Wischer, Ilse, & Diewald, Gabriele (eds). 2002. New Reflections on Grammaticalization. Typological Studies in Language, no. 49. Amsterdam, Philadelphia: John Benjamins. Wolff, Phillip. 2003. Direct causation in the linguistic coding and individuation of events. Cognition, 88, 1–48. Xrakovskij, Viktor S. 2001. Hortative constructions. Pages 1028–1038 of: Haspelmath et al. (2001). Xrakovskij, Viktor S., & Volodin, Aleksandr P. 1986. Semantika i tipologija imperativa. Russkij imperativ. Leningrad: Nauka. Zaimov, Jordan D., & Kapaldo, Mario (eds). 1982, 1983. Supras˘alski ili Retkov sbornik. Suprasl’skaja rukopis’. Sofija: Izdat. na B˘algarskata Akad. na Naukite.
Index accidental causation, 76 ambiguous causatives, see factitive causatives, ambiguous to permissive aspect and tense, 306 corpus attestations of analytical future of da´c , 310 loss of, 174, 183, 302, 310 of lexical infinitive, 57, 74, 107, 122–123, 151, 168, 224, 280 of da(va)t, 53, 57, 74, 98, 122–124, 168, 171–183, 200, 222, 223, 280 with modals, 174 auxiliarization, 46, 47, 101 bezosobnik, see subjectless constructions by-phrases, see type II causees causative domains, 19–20 cognitive, 103, 105, 150–152, 218–221, 275 impersonal, 128, 216, 254 manipulative, 127–128, 216, 245, 254, 270, 279, 283 causatives and diathesis, 212, 228, 236, 290, 292–295 and modals, 27–31, 297 modal domains, 29–31 and passive, 20, 132, 290, 291, 294, 307
and polarity, 65–71 contextual, 145, 208, 238, 239, 272, 313, see curative causatives definition, 14 causee accusative, 195, 203 animacy, 127–128, 198, 245 in reflexive permissives, 134–137 as locative phrase, 146, 208, 210, 259 omission, 89, 95–97, 100, 124, 141, 143, 145, 197, 210, 256, 287, 292 type I, 21–27, 146, 189, 193– 195, 198, 210, 212, 216– 217, 221, 250, 254, 256, 282, 287, 295, 313 type II, 21–27, 138–143, 164, 189, 196, 197, 210, 221, 257–260, 271, 286, 289– 296, 307 type II and passive, 24–26, 294 complement verb aspect, see aspect of lexical infinitive unaccusative, 127, 161, 229 curative causatives, 144–146, 148, 208–212, 238, 256, 272, 287, 289, 293, 294, 304, 308, 313 definition, 18–19
332 Index hybrid constructions, 49, see secondary predicates
translation, 148, 208, 272 da(va)t with finite complement, 42, 86–89, 114, 267, 307 degrammaticalization, 283 enablement, 60 enablement and non-intervention, 16, 21, 29, 40, 64, 75, 85, 126, 129 event integration, 96, 145, 146, 149, 199, 209, 281, 283 factitive causatives, 104, 114, 144–147, 216–218, 271– 272, 286, 289 ambiguous to permissive, 15, 147–149, 214, 287–289, 313 earliest attestations, 256 translation, 238–239 type II coding, 259 German, see language contact with German grammaticalization definition, 9 frequency, 10 introduction, 5–13 of by-phrases, 289–296 of factitive, 286 of modal passive, 296–303 of permissive, 281–283, 289 of reflexive permissive, 284– 286 hortative, see imperative da(va)t as hortative
of
imperative first person singular vs. permissive, 91 non-use of reflexive pronoun, 306 of permissive, 31, 54, 85–93, 121, 248, 249, 279 and causee omission, 89 translation, 90 of da(va)t as hortative, 2, 118–119, 248, 249 impersonal constructions, see subjectless constructions indefinite-personal constructions, see subjectless constructions infinitive, see complement verb language contact with German, 202, 208, 285, 288–289, 297, 304 lexical verb, see complement verb modal passive and passive, 228 and reflexive permissive, 162, 228, 297 aspect, see aspect comparison to other modals, 161, 166, 174, 178, 179, 224–225, 231, 239–241, 300, 302, 312 delimitation to subjectless reflexive construction, 309, see also subjectless constructions
Index
earliest attestations, 261, 272–275 expression of agent, 164, 222 negation of infinitive, 166 polarity, 168, 224, 225, 228 reflexive pronoun in, 165, 222 semantics, 157–162, 262 translation, 239–241, 312 types agreeing, 153, 225–227, 261, 274, 296–301 bias in annotation, 309 comparison, 157–171, 184, 225–227, 310 definition and annotation, 153–157, 223–224 non-agreeing, 156, 225– 227, 276, 279, 301 subjectless, 225–227, 263, 275, 301–303 modals and causatives, see causatives and modals modal deverbal adjective, 169 polarity, 305 polarity and frequency, 38 modes of causation, 15 nech(áv)at, 188, 204–206, 212, 216–218, 220, 221, 237, 238, 289, 296 negation and causatives, 32–40, 47 and modals, 32–40 non-intervention, 59, 66 passive and causatives, see causatives
333
and passive, reflexive permissives and passive of causative, 105 of give, 48, 109, 150 permissive causatives, 270–271 ambiguous to factitive, see factitive causatives comparison, 127, 128 definition, 16–18 non-reflexive in Czech, 193– 196 polarity, 54, 58, 71, 77–78, 81–85, 121, 127, 128, 247 translation, 65, 235–238 reflexive
permissives, 93–100, 129–134, 270–271, 279, 284, 293, 294 and passive, 284, 285 comparison across languages, 120, 129–132, 136, 137 earliest attestations, 255 polarity, 98, 122, 198 translation, 235–238 type II coding, 258 with dative reflexive, 132, 140, 201–202, 206, 295 with other reflexive, 203 with tonic reflexive, 204 reinforcement, 249 secondary predicates, 49–52, 110, 118, 231, 282, 288 Standard Average European, 289 subjectless constructions, 154, 157, 281, 303, 304, 309
334 Index unaccusative verbs, see complement verb word order, 88, 102