Constituent Order in Classical Latin Prose (Studies in Language Companion Series)

Constituent Order in Classical Latin Prose

Studies in Language Companion Series (SLCS) This series has been established as a companion series to the periodical Studies in Language.

Editors Werner Abraham University of Vienna

Elly van Gelderen

Arizona State University

Editorial Board Bernard Comrie

Max Planck Institute, Leipzig and University of California, Santa Barbara

William Croft

University of New Mexico

Östen Dahl

University of Stockholm

Gerrit J. Dimmendaal University of Cologne

Ekkehard König

Free University of Berlin

Christian Lehmann University of Erfurt

Brian MacWhinney

Carnegie-Mellon University

Marianne Mithun

University of California, Santa Barbara

Heiko Narrog

Tohuku University

Johanna L. Wood

University of Aarhus

Volume 117 Constituent Order in Classical Latin Prose by Olga Spevak

Constituent Order in Classical Latin Prose Olga Spevak University of Toulouse

John Benjamins Publishing Company Amsterdamâ•›/â•›Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Spevak, Olga. Constituent order in classical Latin prose / Olga Spevak. p. cm. (Studies in Language Companion Series, issn 0165-7763 ; v. 117) Includes bibliographical references and index. 1. Latin language--Word order. 2. Latin language--Semantics. I. Title. PA2293.S58â•…â•… 2010 475--dc22

2009048325

isbn 978 90 272 0584 1 (Hb ; alk. paper) isbn 978 90 272 8851 6 (Eb)

© 2010 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents Preface Abbreviations

Introduction 1. The traditional approachâ•… 2 2. The typological approachâ•… 2 3. The generative approachâ•… 3 4. Pragmatic approachesâ•… 4 4.1 Firbas’s communicative perspective (Panhuis 1982)â•… 4 4.2 The perspective of Functional Grammarâ•… 6 4.2.1 Pragmatic functions in Functional Grammarâ•… 6 4.2.1.1 Topicâ•… 6 4.2.1.2 Focusâ•… 7 4.2.2 General principles and hierarchiesâ•… 8 4.2.2.1 Formal hierarchiesâ•… 8 4.2.2.2 Dominance hierarchiesâ•… 9 4.2.2.3 Familiarity hierarchiesâ•… 9 5. Other aspects of Latin constituent orderâ•… 10 5.1 Colon subdivisionâ•… 10 5.2 Stylistic ornamentationâ•… 11 6. Aim and methodology of this studyâ•… 11

xiii xv

1

chapter 1

Placement constraints and liberties in Latin constituent order 1. First sentence- and clause-positionâ•… 13 2. Second sentence-position and enclitics enim ‘indeed’, autem ‘but’ and vero ‘however’â•… 16 3. Enclitics -que ‘and’, -ve ‘or’ and -ne â•… 17 4. Adpositionsâ•… 19 5. Focusing particlesâ•… 20 6. Principles of domain integrity and head proximityâ•… 22 7. Hyperbaton or discontinuity of noun phrasesâ•… 23

13



Constituent order in Classical Latin prose

chapter 2

Pragmatic functions 1. My approach to Latin constituent orderâ•… 27 1.1 Introductionâ•… 27 1.2 Problems in the analysis of Latin constituent orderâ•… 28 1.3 The position of constituents in the sentenceâ•… 28 1.3.1 Topics and sentence-initial positionâ•… 28 1.3.2 Focus positionâ•… 29 1.4 Situational and contextual dependencyâ•… 32 1.5 Question testsâ•… 35 1.6 Principles of analysisâ•… 36 2. Focusâ•… 39 2.1 Focus related to newnessâ•… 39 2.2 Focus-first strategyâ•… 41 2.3 Presentative and what-happens sentencesâ•… 41 2.4 Complex informationâ•… 44 2.5 Contrastâ•… 45 2.5.1 Contrastive Topic and Focusâ•… 45 2.5.2 Verum Focusâ•… 46 2.6 Emphasisâ•… 47 2.7 Particlesâ•… 49 2.7.1 The function of focusing particlesâ•… 49 2.7.1.1 Et, etiam and quoque ‘also’ â•… 51 2.7.1.2 Quidem â•… 52 2.7.2 The function of connective particlesâ•… 53 2.8 Cleft constructionsâ•… 55 3. Topicâ•… 56 3.1 Future Topicâ•… 56 3.1.1 Introduction of new discourse participantsâ•… 56 3.1.2 Reintroduction of discourse participantsâ•… 60 3.2 Referential chainsâ•… 61 3.3 Discourse Topicâ•… 64 3.4 Sub-Topicâ•… 64 3.5 Sentence Topicâ•… 65 3.6 Discourse Topics Iugurtha and Caesar â•… 66 3.6.1 Dataâ•… 67 3.6.2 Sentence Topic and Discourse Topicâ•… 68 3.6.3 Setting constituent and Discourse Topicâ•… 68 3.6.4 Discourse Topic in a late positionâ•… 72 4. The placement of pronounsâ•… 73 4.1 Pronominal anaphoraâ•… 73 4.2 Topicalisationâ•… 75

27



Table of contents 

Resumption of Focusâ•… 77 4.2.1.1 Initial is and hic â•… 77 4.2.1.2 Is versus hic â•… 79 4.2.1.3 Is and hic in an internal positionâ•… 80 4.2.1.4 Special casesâ•… 81 4.2.2 Resumption of larger contentâ•… 82 4.2.2.1 Is and hicâ•… 82 4.2.2.2 Summarising function of hicâ•… 83 4.2.3 Selection of a constituentâ•… 85 4.2.4 The adverbs ibi, eo, huc, etc.â•… 85 4.3 The connecting relative qui â•… 87 4.4 Ille â•… 89 4.5 Conclusionsâ•… 91 4.6 Personal pronounsâ•… 92 4.6.1 Personal pronouns in the nominativeâ•… 92 4.6.1.1 Contrast and emphasisâ•… 92 4.6.1.2 Subjectivityâ•… 93 4.6.1.3 The placement of egoâ•… 94 4.6.2 Personal pronouns in an oblique caseâ•… 94 Ellipsisâ•… 96 5.1 Ellipsis of the verbâ•… 96 5.2 Non-expression of the first argumentâ•… 98 5.3 Ellipsis of the second argumentâ•… 99 5.3.1 Zero anaphora in complex sentencesâ•… 100 5.3.2 Zero anaphora overlapping a sentenceâ•… 102 5.3.3 Zero, or pronominal anaphora? The case of coordinationâ•… 103 5.4 Conclusionsâ•… 106 Themeâ•… 107 6.1 Definition. Problems of identificationâ•… 107 6.2 Typology of Theme constituentsâ•… 107 6.3 Theme versus Topic constituentsâ•… 108 6.4 Theme versus prepositional phrases with de â•… 109 6.5 Subordinate clausesâ•… 111 6.6 Conclusionsâ•… 111 Tailâ•… 111 7.1 Tail constituentsâ•… 111 7.2 Tail-head linkingâ•… 114 4.2.1

5.

6.

7.

 Constituent order in Classical Latin prose

chapter 3

Declarative sentences 115 1. Bivalent transitive verbsâ•… 115 1.1 Verbs of actionâ•… 118 1.1.1 Dataâ•… 118 1.1.2 The pattern first argument > second argument > verbâ•… 119 1.1.3 The pattern first argument > verb > second argumentâ•… 122 1.1.4 The pattern second argument > first argument > verbâ•… 124 1.1.5 The pattern verb > second argument > first argumentâ•… 125 1.1.6 Conclusionsâ•… 125 1.2 Verbo-nominal constructionsâ•… 125 1.2.1 Terga verto ‘to flee’ â•… 127 1.2.2 Castra pono ‘to encamp’, castra moveo ‘to move camp’ and iter facio ‘to march’ â•… 128 1.2.3 Verba facio ‘to speak’ â•… 129 1.2.4 Mentionem facio ‘to mention’ and finem facio ‘to end’ â•… 129 1.2.5 Operam do ‘to do one’s best’ â•… 130 1.2.6 Gratias ago ‘to thank’ â•… 130 1.2.7 Conclusionsâ•… 131 2. Trivalent verbsâ•… 131 2.1 The verb mitto ‘to send’â•… 132 2.1.1 Dataâ•… 132 2.1.2 Analysisâ•… 134 2.1.3 Legatos ad Caesarem mittunt ‘they send a delegation to Caesar’ â•… 136 2.1.4 Ad te litteras misi ‘I sent you a letter’ â•… 138 2.2 The verb duco ‘to lead’ â•… 141 2.3 The verb do ‘to give’ â•… 142 3. Passive sentencesâ•… 145 3.1 Introductionâ•… 145 3.2 Dataâ•… 146 3.3 Contextually independent second argumentsâ•… 146 3.4 Contextually dependent second argumentsâ•… 148 3.5 Analytic forms of the passive voiceâ•… 149 3.5.1 Dataâ•… 150 3.5.2 Analysisâ•… 151 3.6 Analytic forms of deponent verbsâ•… 155 3.7 Conclusionsâ•… 156 4. Verbs of thinking and sayingâ•… 157 4.1 Verbs of thinkingâ•… 157 4.2 Verbs of communicationâ•… 159 4.2.1 Verbs of saying + AcI clauseâ•… 159



Table of contents 

5.

6.

7.

4.2.2 Verbs of saying and direct and indirect speech or questionsâ•… 162 4.2.3 The verb inquam as a parenthetical expressionâ•… 163 Bivalent intransitive verbsâ•… 164 5.1 Dataâ•… 165 5.2 Placement of directional complementsâ•… 166 5.3 Complex informationâ•… 169 5.4 The reflexive verb se recipere ‘to retire’â•… 170 5.5 Conclusionsâ•… 171 Monovalent verbsâ•… 172 6.1 Introductionâ•… 172 6.2 Dataâ•… 174 6.3 The accurro ‘to run’ type: dynamic controllable processesâ•… 175 6.4 The nascor ‘to be born’ type: dynamic uncontrollable processesâ•… 177 6.5 The permaneo ‘to remain’ type: permanent or temporary statesâ•… 178 6.6 Conclusionsâ•… 179 The verb sum ‘to be’â•… 180 7.1 The predicative adjectiveâ•… 181 7.2 Predicative nouns (or noun phrases)â•… 184 7.3 Existential sum â•… 187 7.3.1 Existential sentencesâ•… 189 7.3.2 Locative sentencesâ•… 190

chapter 4 Interrogative sentences 1. Introductionâ•… 195 2. Word-questionsâ•… 196 3. Sentence-questionsâ•… 198 3.1 Sentence-questions with an interrogative particleâ•… 199 3.1.1 The particle -ne â•… 199 3.1.2 The particles num and nonne â•… 200 3.1.3 The particle an â•… 201 3.2 Sentence-questions without interrogative particlesâ•… 202 4. Disjunctive questionsâ•… 203 5. Conclusionsâ•… 204 chapter 5 Imperative sentences 1. Introductionâ•… 205 2. A typology of imperative sentencesâ•… 207 3. Periphrastic imperativesâ•… 208 4. Imperatives of verbs of saying, thinking, knowing, and perceivingâ•… 210 4.1 Metadirective imperativesâ•… 210

195

205



Constituent order in Classical Latin prose

4.2 Communicative expressionsâ•… 211 5. Verbs of actionâ•… 213 6. Intransitive verbsâ•… 216 7. Trivalent verbsâ•… 216 8. Future imperativesâ•… 218 9. Imperatives and oblique personal pronounsâ•… 219 10. Conclusionsâ•… 221 chapter 6 Noun phrases 223 1. Introductionâ•… 223 2. Adjective modifiersâ•… 224 2.1 Placement of adjectivesâ•… 224 2.2 Distinctive function of adjectivesâ•… 226 2.3 Fixed formulasâ•… 228 2.4 Juxtaposed attributive adjectivesâ•… 229 2.4.1 Hierarchical principle of adjective orderingâ•… 229 2.4.2 Dataâ•… 233 2.4.3 Hierarchical and non-hierarchical ordering of adjectivesâ•… 234 2.4.4 Other orderingsâ•… 236 2.5 Coordination of adjectivesâ•… 237 2.5.1 Attributive adjectivesâ•… 237 2.5.2 Predicative adjectivesâ•… 238 3. Demonstrative determinersâ•… 239 4. Indefinite modifiersâ•… 242 5. Quantifiersâ•… 244 5.1 Non-numerical quantifiersâ•… 244 5.2 Numerical quantifiersâ•… 246 5.3 Temporal expressionsâ•… 248 6. Possessive modifiersâ•… 250 7. Juxtaposed determinersâ•… 254 7.1 General observationsâ•… 254 7.2 Quantifiers and indefinite determinersâ•… 257 7.3 The anaphor is and the demonstratives hic and ille â•… 259 7.4 Possessive modifiersâ•… 260 7.5 Multiple determinersâ•… 263 7.6 Multiple determiners and adjectivesâ•… 263 8. Genitive complementsâ•… 265 8.1 Simple instancesâ•… 267 8.2 Complex instancesâ•… 268 8.3 Hyperbaton within the domain of the noun phraseâ•… 272



Table of contents 

Hyperbaton produced by alien elementsâ•… 274 9.1 Frequency of hyperbaton of noun phrasesâ•… 275 9.2 Hyperbaton produced by alien elementsâ•… 275 10. Conclusionsâ•… 280

9.

Conclusion

283

References

287

Index locorum

299

Index rerum

303

appendix Three commented texts 1. Sallust, The Conspiracy of Catiline (55.1–6)â•… 305 2. Cicero, AgainstVerres (4.86)â•… 310 3. Cicero, A letter to Atticus (11.10.1)â•… 315

305

Preface The research on which this book is based started in 2004. I had the opportunity to present part of the findings to various audiences. I hope they will find that their reactions have contributed to the final version I am presenting here. I wish to express my thanks to Prof. Michèle Fruyt (Paris – Sorbonne) who supervised my habilitation (2006), in which some of the findings of this book were first presented, and who supported me throughout my academic career in France. My special thanks go to Prof. Harm Pinkster (Amsterdam). His work on Latin syntax and Latin constituent order was my major inspiration. He attentively read through the whole manuscript, provided me with useful comments and prevented me from several errors. For the book version, I am also indebted to Prof. Roger Wright (Liverpool) for his invaluable help with reading the manuscript and for his pertinent remarks. I am grateful to the editors of SLCS for having accepted this book for publication and to the reviewers for their comments. Finally, I thank Jeremy Brightbill for his critical reading of the final version of the book. All remaining errors are mine. Dr. Dirk Panhuis’s pioneering work on Latin word order from a pragmatic point of view, based on insights developed in the Prague school of linguistics, has been a continuous source of inspiration. It is a pleasure for me to see my book published by Benjamins as well. Although I do not always agree with his analysis, I appreciate very much the work he has done on this complicated matter. It is not easy to examine the constituent order of a language only known from literary tradition: we ignore its prosody; we do not practice it in everyday life. We can only try to “read” something from preserved texts: how information is organized, what strategies are used in order to present it, in what ways Latin constituent order is distinctive in contrast with modern European languages. Furthermore, everyone who examines Latin constituent order makes conscious or unconscious comparisons with his native language and “projects”, so to speak, his own intuitions and ideas on it. I am aware of this difficulty. Although as a Czech native speaker I was tempted several times to note paralells between Latin and Czech, I sought to concentrate on Latin as much as I could.

Abbreviations

An. Ø AcI A1 A2 A3 AD L V X>Y

anaphora zero anaphora accusative plus infinitive clause first argument (subject) second argument (direct object) third argument (indirect object) directional complement locative complement verb X is followed by Y

Capitals indicate contrastive or emphatic element. * marks unattested expressions. Abbreviations of ancient authors and works follow the Oxford Latin Dictionary (Oxford, Clarendon Press, 1968)

Introduction Latin constituent order is variable1 in the sense that the position of a constituent in its sentence or phrase is not determined by the grammatical structure of that sentence (for example, whether it is subject in its sentence), nor is it indicative for an addressee (a hearer or reader) of its grammatical function. In Latin constituent order is rather determined by and indicative of (i) the role of a constituent within the discourse to which its sentence belongs (for example, whether the constituent is present in some way in the preceding discourse), (ii) the speaker or writer’s estimation of what the addressee knows and expects, and (iii) how important the speaker and hearer consider a constituent within the overall communication. For these three, partially overlapping, aspects I will use the term “pragmatic value”. In other words, in Latin a constituent can change its place without changing its syntactic function;2 however, this does not mean that the information conveyed is the same. Much has been written about Latin constituent order. In this introductory chapter, I will briefly sketch the main approaches and the problems involved. I will first deal with different approaches to Latin constituent order: the traditional (Section 1), typological (Section 2), generative (Section 3), and pragmatic (Section 4) approaches. As pragmatic approaches are the most important ones for my description, I will pay special attention to Firbas’s communicative perspective as it was applied to Latin by Panhuis (Section 4.1) and to a general presentation of the perspectives of Functional Grammar (4.2). In Section 5, I briefly mention other aspects of Latin constituent order. The aim of my study and information about the methodology and corpora used are presented in Section 6.

1.

For the term “free” constituent order, see below, Section 1 and Chapter 2, Section 1.

2. It is worth pointing out that already Weil (1844:€25) rightly observed that there is a – pragmatic – difference between the following sentences: Romulus Romam condidit. ‘Romulus founded Rome.’ Hanc urbem condidit Romulus. ‘This town was founded by Romulus.’ Condidit Romam Romulus. ‘Romulus founded Rome.’ Weil wrote: “These three sentences inform us about different things because their elements, remaining the same, are distributed in a different way.” See Chapter 2, Section 1.5 on question tests; indeed, these sentences convey information answering three different implicit questions: what did R. found?; who founded this town?; and what happened?



Constituent order in Classical Latin prose

1. The traditional approach The traditional approach to Latin constituent order arose from a comparison of Latin with other languages: modern European (Romance, Germanic and Slavic languages) and ancient ones (Sanskrit and Ancient Greek). From this point of view, constituent order in Latin was labelled “free”, in contrast with the “fixed” order in French and German, for example. This distinction was introduced by Weil (1844:€54) and elaborated by Meillet (1903). The traditional approach focuses on several properties that are observed in other languages, especially the position of the finite verb and the placement of enclitics. In traditionally oriented works, the verb is considered the most important element of a sentence, and therefore, its position has drawn much attention. As Classical Latin authors, in particular Caesar and Cicero, frequently put the verb at the end of a sentence,3 this position is described as “normal”; the other empirically observed positions, initial and internal, are described as “deviations” (cf. Schneider 1912). Kühner and Stegmann’s Latin grammar (1914, II: 590 ff.) shares the same point of view and talks about “traditional” and “occasional” order. The same distinction between final, initial and internal verb position is found in Marouzeau’s monograph (1938). He considers patterns with a final verb as “unmarked”; on the other hand, “marked” initial and internal patterns have to be explained with the help of a set of factors responsible for these “deviant” placements of the verb.

2. The typological approach Ideas about “normal” or “unmarked” order, which have a long tradition in descriptions of constituent order,4 led to a typological classification of languages by Greenberg (1966).5 According to him, languages are supposed to exhibit one basic, unmarked order of the S(ubject), the O(bject), and the V(erb); consequently, other orderings are viewed as deviations. Although it is generally stated that from a typological point of view, Latin exhibits a basic SOV order, the question is more complicated. An SOV language is also claimed to use postpositions rather than prepositions, and to place adjectives as well as genitive modifiers before their head nouns, which does not hold true for Latin. Discussion of Greenberg’s criteria showed that Latin is perhaps not only 3. This statement is based not only on empirical observation but also on a passage from Quintilian (Inst. 9.4.26) who advises his readers to put the verb in the sentence-final position, because the ‘force’ of the sentence is in the verb. For a recent analysis of this passage, see de Jonge (2008:€321 ff.). 4. They mainly reflect a distinction between the subjective and the objective order, made first by psychologically oriented studies on word order and also used by Mathesius, a founder of the Prague School. 5.

For typological approach to Latin, see Bauer (2009:€241–271).



Introduction

an SOV language, but also SVO.6 This divided scholars into two groups: some of them hold that Archaic and Classical Latin has the SOV order as the basic one, which underwent a typological shift to SVO, the dominant order in the Romance languages – at least in some of them, especially in French. Others argue that the SVO order is already present in Archaic Latin and that SVO coexisted along with SOV (Adams 1976a) throughout the language’s history. The typological approach to Latin constituent order suffers from several difficulties that were discussed in detail by Pinkster (1991; cf. 1995:€ 218). The first is that Latin constituent order largely depends on pragmatic rules and has no syntactic implications. Consequently, it is not very helpful to describe it in syntactic terms such as subject, object, verb because pragmatic values cannot be translated into the formulas such as SOV, SVO, or OVS. On the other hand, without claiming a one-to-one correspondence, we can say that a syntactic order is the outcome of certain pragmatic conditions (cf. Pinkster 1992:€ 522). The second difficulty is that descriptions using the syntactic functions of subject, object, and verb are centred on clauses containing transitive verbs; other types of verbs (monovalent and trivalent) are usually neglected. The third problem concerns the considerable variety of patterns occurring not only in different Latin authors, but also in the works of one author (Pinkster 1992:€521 f. and 1991:€72).7 For all these reasons, I will not discuss the question of the “basic order”. Indeed, how is one to decide what order is the basic one when we have to deal with a large range of pragmatic arrangements? In sum, Greenberg’s typology poses more problems than it solves and therefore, it is no longer adopted as a sustainable framework (Sörés 2004).

3. The generative approach The approach to word order within the framework of Generative Grammar resembles in a sense the typological one: it is based on the hypothesis that there is one basic, unmarked word order and a set of transformational rules producing marked patterns. Latin word order is supposed to exhibit one neutral, basic SOV order; other orderings represent deviations. In order to explain these “deviations”, Generative Grammar works with the concept of movement rules, scrambling, as well as pragmatic features (Topic, Focus, Contrast, Emphasis, Heaviness).8 Recent contributions9 within this framework to Latin constituent order are Devine and Stephens (2006), Polo (2004), and Salvi (2004). As I have discussed several problems posed by these contributions 6. See Panhuis (1984a)€for details. He himself considers Latin as an ambivalent language (p. 154). 7. See also Panhuis (1981) on non-final placement of the verb in descriptive passages in Caesar, and Spevak (2005a). 8. The concepts of Topic and Focus are different in Functional Grammar, see below, Section 4.2. 9. The transformational approach was first applied to Latin by Ostafin (1986).





Constituent order in Classical Latin prose

elsewhere (Spevak 2006c and 2007c), I limit myself here to saying that Generative Grammar is a well-constructed theory in which nowadays pragmatics and discourse play a certain role. However, due to its continuing insistence on a basic (underlying) word order it does not really help a Latinist to better understand Latin constituent order, especially because of the great number of “deviations” that we must admit in that case. It also poses problems with deciding what counts as empirical evidences, for there is much variation in constituent ordering not only between Latin authors but also between works of individual authors.

4. Pragmatic approaches 4.1

Firbas’s communicative perspective (Panhuis 1982)

Panhuis’s study (1982) was the first attempt to provide a systematic description of Latin constituent order from a pragmatic point of view. As a general framework, he uses the theory of Functional Sentence Perspective, elaborated by linguists of the Prague School (mainly Mathesius, Daneš, and Firbas).10 In the theory of Functional Sentence Perspective, the distribution of elements in a sentence is viewed as a result of both the communicative aims of the speaker or writer and the function of that sentence in its context, textual as well as situational. The analysis works with the distinction between a “theme” and a “rheme”.11 The theme is defined as the less informative element whereas the rheme is the most informative one. Since communication represents a dynamic process where information is conveyed from a speaker or writer to an addressee, it is supposed to proceed in a forward direction. Elements of a sentence contribute in different degrees to the progression of the message: they have different degrees of “communicative dynamism”. What governs the distribution of elements in a sentence (in a normal, non-contrastive or non-emphatic context) is the principle of increasing communicative dynamism: the element bearing the lowest degree of communicative dynamism comes first, the element with the highest degree – or we can say the most informative one – comes last. Firbas’s theory was elaborated on the basis of constituent order phenomena in Czech that he compared with English and German. Whereas the constituent order is predominantly governed by the Functional Sentence Perspective

10. Firbas (1992) concerns a synthesis of his previous works. 11. It is worth noticing that the distinction between “theme” and “rheme” was already suggested by O. Behaghel (1932:€4): “the element that is less important (or already known to the listener) is placed before the€more important element” (Behaghel’s Second Law). Firbas’s (1992) concept of the “theme proper” partially corresponds to Dik’s (1997, I-II) concept of Topic. However, Theme in Functional Grammar covers another pragmatic function (see below, Chapter 2, Section 6). Firbas’s “rheme proper” corresponds to Focus in Functional Grammar.



Introduction

(i. e. by increasing communicative dynamism) in Czech, English and German are less sensitive to this principle. For Latin, Panhuis (1982:€54 f., cf. 1981:€300) claims three main arrangements: (i) a normal, non-emotive arrangement {theme > rheme > verb}, and two “emotive” arrangements (ii) {rheme > theme > verb}, and (iii) {verb (rheme) > theme}. In other words, a Latin sentence is supposed to obey the principle of communicative dynamism just as in Czech. The only difference is that the Latin verb takes the final position regardless of the degree of its communicative dynamism in arrangements (i) and (ii). In order to show the normal, non-emotive arrangement, I borrow an example from Panhuis (1982:€44 and 55) that illustrates the principle of increasing dynamism. In a passage from Plautus, a young man Phaedromus complains to his slave about his situation because a pimp (leno) is tormenting him. When the slave asks why (Quid est?), Phaedromus replies:12 (1)

Alias me poscit pro illa triginta minas, sometimes me-acc he asks for her thirty-acc minae-acc alias talentum magnum sometimes talent-acc great-acc ‘Sometimes he asks me thirty minae for her, sometimes a great talent.’(Pl. Curc. 62)

This sentence can be divided into two parts: a thematic and a rhematic one. The understood subject (leno ‘pimp’) belongs to the thematic part, as well as the setting element alias ‘sometimes’ and the personal pronoun me referring to the speaker. In the rhematic part, we have the verb poscit ‘(he) asks’ (that may be labelled as “transition” in Firbas’s terminology), followed by its non-obligatory complement pro illa ‘for her’ that is rhematic but less than triginta minas ‘thirty minae’ that is the most informative constituent of the sentence. This sentence can schematically be represented as follows: Thematic part (leno) theme proper >

alias theme >

Rhematic part me theme >

poscit rheme >

pro illa rheme >

triginta minas rheme proper

12. Examples quoted in this book are accompanied by interlinear word-to-word glosses; their aim is to provide information about constituent ordering, and not a full morphological analysis (a complete morphological analysis of examples is available on Perseus, collection Classics, www.perseus.tufts.edu). Case marking is provided for nouns and noun phrases; prepositional phrases are signalled by prepositions. In some cases, when not interrupted by other elements, constituents are treated as a whole and bear only one case marking, when it is evident what belongs together. Only sentences or clauses that are analysed are glossed; preceding or following context is provided between brackets [ ]. Abreviations used for glossing follow the Leipzig glossing rules (www.eva.mpg.de/lingua/pdf/LGR09_02_23.pdf).





Constituent order in Classical Latin prose

Panhuis’s aim to analyse sentences in their context and to determine what is new information deserves appreciation. On the other hand, the principle of communicative dynamism raises several other questions, such as how to decide about the degree of communicative dynamism of individual constituents. This difficulty concerns not only the arrangement of thematic constituents (cf. Panhuis 1982:€55 f.) but also rhematic ones. Which criteria allow us to say that one constituent is more thematic or more rhematic than another one? The absence of native Latin speakers renders this task difficult. The principle of increasing communicative dynamism governs the ordering of constituents in Czech, as Firbas, among others, has convincingly shown, and in normal conditions (in the absence of contrast), the final sentence-position receives the most informative element, the rheme proper. However, does the same principle always work in Latin? In Panhuis’s analysis, a number of examples can be regarded as conforming to the principle going from the less informative to the most informative constituent. In other cases, in particular in sentences with a final verb, the most informative constituent does not necessarily stand before the verb.13 In other words, the final or pre-verbal position is not reserved for the rheme proper constituent, and hence this position does not seem to be a reliable indicator that the constituent occupying it has rheme function.

4.2

The perspective of Functional Grammar

4.2.1 Pragmatic functions in Functional Grammar In order to describe constituent order, Functional Grammar distinguishes between two main pragmatic functions: Topic, the entity the sentence is about, and Focus, the salient element of the sentence (Dik 1997, I: 312 f.). These pragmatic functions affect the formal side of the sentence: special positions are reserved for Topic and Focus constituents and they can be marked by specific means.

4.2.1.1 Topic The pragmatic function of Topic, “what is being talked about”, can be assigned to entities (persons, objects, localities, etc.).14 A discourse may contain several Topics: one can be more central than others, the importance of Topics can change and new Topics can be introduced. According to their status in the discourse, Functional Grammar distinguishes several types of Topics: Discourse Topic, Future (New)15 Topic, Given Topic, Resumed Topic and Sub-Topic.16 13. See Chapter 2, Section 1.3.2. p. 29 ff. 14. Several scholars argue that also verbs can be considered as Topics – H. Dik (1995:€207) and Bolkestein (1998a: 197). 15. I will follow the suggestion made by Bolkestein (2000) and call it “Future Topic”. Indeed, in spite of its name, Future Topic has focal character. It functions as Topic not in the presentative sentence but afterwards (Toth 1994:€178, and Wehr 1984:€7). 16. In my study, I will not use all these types, but only some of them.



Introduction

Discourse Topic (Dik 1997, I: 313–326) is the entity about which a discourse (story, paragraph, chapter, etc.) gives information. In order to function as Discourse Topic, the entity has to be introduced and presented. This procedure, necessary for ensuring a coherent development of the discourse, is labelled the “introduction of Future Topic” (Dik 1997, I: 315). Once introduced, the new entity is treated as a Given Topic and can be maintained in the text for a shorter or a longer time. Nevertheless, it is necessary to keep the entity present in the mind with the help of topical or anaphoric chains. For example:

(2) Yesterday, I got a phone call from the tax inspector. He/The man/The joker wanted me to come to his office, and he/Ø gave me the impression that I was in for some trouble.17

The tax inspector is a new entity that is introduced into the discourse (Future Topic). It is maintained as Given Topic with the help of a personal pronoun (he), a noun that specifies the category it belongs to (the man), a qualifying noun (the joker), or zero anaphora (Ø). These resumptive means differ as to their force: a noun is a stronger form of anaphora than the anaphoric pronoun, zero anaphora is the weakest means (cf. Givón 1983:€17). After the intervention of another Topic, the speaker can return to the previous one, by referring to it as the tax inspector. It is also possible to deduce a Topic by association with another word. For example, on the basis of shared knowledge that music is played at parties, the music is not a new element but a Sub-Topic, which explains the presence of the definite article the:

(3) John gave a party last week, but the music was awful.

4.2.1.2 Focus Focus is the salient, or most informative, element of a sentence (Dik 1997, I: 326 ff.). The pragmatic function of Focus can be assigned to entities as well as to actions. Focus adds new information or replaces information that we already have. It need not convey completely new information but may be in contrast with another element, either explicit or implicit. This is exemplified in (4a) with Focus represented by new information and in (4b) with Focus that contrasts with another, implicit element. The sentence in (4b) is a reaction to (4a). Saliency due to explicit contrast is illustrated in (4c), presenting two judgments concerning John and Bill; nice and boring function as explicit contrastive Focus constituents. (4) a. I’ve just bought a Peugeot.18 b. Did you buy a Peugeot??? c. John and Bill came to see me. John was nice, but Bill was rather boring.

17. Here and below, I quote Dik’s examples (1997, I: 314 and 318). 18. In this section, I quote Dik’s examples (1997, I: 326 and 328).





Constituent order in Classical Latin prose

Languages have various means of signalling the saliency of an element. These include prosodic prominence, one or several special positions reserved for this type of elements, focusing particles, and special syntactic constructions, such as cleft constructions (cf. also Miller 2006:€128). 4.2.2 General principles and hierarchies Functional Grammar, mainly based on works by Dik (1997, I-II),19 views constituent order as an interplay of a diverse range of syntactic, semantic, and pragmatic factors. The factors involved can be subcategorised and described as linearization rules or linearization hierarchies. The remarks that follow are grounded in observations of Siewierska (1988:€29), who discusses in detail three main groups of hierarchies: formal, dominance, and familiarity hierarchies.

4.2.2.1 Formal hierarchies The formal hierarchies are formulated as the Language Independent Preferred Order of Constituents (LIPOC) schema (Dik 1997, I: 411). They include the rule of increasing complexity20, a principle that is responsible for the rightward placement21 of syntactically heavy constituents such as complex noun phrases, for example M. Scaurus in (5), which is expanded by a subordinate clause and an apposition. (5)

In quis fuit M. Scaurus, de quo supra among them was M. Scaurus-nom of whom above memoravimus, consularis et tum senatus princeps. we have spoken ex-consul and then of Senate leader ‘Among them was Marcus Scaurus, of whom I have already spoken, an exconsul and at the time the leader of the Senate.’ (Sal. Jug. 25.4)

The formal hierarchy also includes the rule of domain integrity 22 according to which “what belongs together should be kept together” (Dik 1997, I: 402). Constituents belonging to a domain such as noun phrase, infinitive or participial clause, etc. are not expected to leave it and go to another domain. An example of

19. The work referred to as Dik (1997, I-II) is the second, enlarged version of his previous grammar published in 1978. 20. When talking about the principle of increasing complexity, it is worth recalling Behaghel’s “Law of Increasing Terms” (Gesetz der wachsenden Glieder, Behaghel 1909 and 1932:€6); according to him, “given two phrases, when possible, the shorter precedes the longer”. 21. It is important to stress that, in contrast with the Generative Grammar framework (see Section 3), linearization principles following Dik (1997, I: 392) are template principles and not “movement” principles. 22. Cf. “First Behaghel’s Law”: “Elements that belong close together intellectually will also be placed close together” (Behaghel 1932:€4).



Introduction

contiguous syntactic units is given in (6). As we will see (p. 22), Latin may rather easily violate this rule. (6)

His rebus impulsus / equitatum omnem / prima nocte / by these things urged cavalry-acc all-acc at nightfall ad castra hostium / mittit / ad flumen Bagradam. to camp of enemies he sends to river Bagrada ‘Under the influence of these factors, he sends off all his cavalry at nightfall to the enemy camp at the River Bagradas.’ (Caes. Civ. 2.38.3)

4.2.2.2 Dominance hierarchies “Dominance” hierarchies concern priority given to two categories: person and animate entities, and semantic roles, as is represented in the following schema (from Siewierska 1988:€30): The Personal hierarchy 1st person > 2nd person > 3rd person human > higher animals > other organisms > inorganic matter > abstract entities The Semantic role hierarchy agent > patient > recipient > benefactive > instrumental > spatial > temporal In accordance with the personal hierarchy, the first person has priority over the second person; nouns denoting human beings are preferably placed before nouns denoting inanimate or abstract entities. The effect of the semantic role hierarchy on the linearization of constituents is that the agent has priority over the patient, the patient over the recipient, and so on. At the same time, the semantic role is largely connected with the eligibility of the agent as syntactic subject, and of the patient as object.

4.2.2.3 Familiarity hierarchies The hierarchies labelled as “familiarity” (Siewierska 1988:€61) represent a group of hierarchies concerning the speaker’s personal involvement, and encompasses mainly givenness, definiteness, referentiality and iconicity. The Familiarity hierarchy more familiar > less familiar given > new definite > non definite referential > non-referential According to these hierarchical principles, the entity that is more familiar to the speaker has priority; given information tends to be placed before new information, a definite entity before an indefinite one, a referential entity before a non-referential one. The familiarity hierarchies occupy the highest place on the scale of hierarchies. This is why the Topic > Focus arrangement (given > new hierarchy) does not necessarily coincide





Constituent order in Classical Latin prose

with the agent > patient preference in the semantic role hierarchy (Siewierska 1988:€65). The preference for Topic being placed first is observed in Latin (Pinkster 1995:€123) as well as in Dutch and in Polish, whereas in French and English initial non-subject constituents are unusual. The following example is taken from Caes. Civ. 2.23.5. (7)

Hunc secutus Marcius Rufus quaestor navibus XII. him-acc followed M. Rufus-nom quaestor with 12 ships Hem volgde de quaestor Marcius Rufus met twaalf schepen. Za nim podążył kwestor Marcius Rufus z 12 statkami. Le questeur Marcius Rufus l’avait poursuivi avec douze vaisseaux. ‘The quaestor Marcius Rufus followed him with twelve ships.’

Linearization of constituents can also reflect our perception of extra-linguistic facts. Following the iconicity principle, especially concerning the temporal order (Siewierska 1988:€79), actions or events are presented in their actual temporal succession, as in (8). Information about Pompey’s trip is organised according to the iconic principle: his point of departure (Luceria) is given first, then comes the action (proficiscitur), followed by the direction (Canusium, Brundisium), in their order of occurrence. (8) Pompeius... Luceria proficiscitur Canusium atque Pompey-nom from Luceria sets out to Canusium and inde Brundisium. thence to Brundisium ‘Pompey... goes from Luceria to Canusium and thence to Brundisium.’  (Caes. Civ. 1.24.1)

5. Other aspects of Latin constituent order 5.1

Colon subdivision

Describing the structure of Latin sentences, Fraenkel (19642 and 1965) used an ancient rhetorical concept and divided a simple or a complex sentence into syntactically defined units called cola. A colon may be a participial and infinitive clause, an ablative absolute, an adverbial prepositional phrase, or indeed any subject, direct object, and indirect object composed of more than one element. Habinek (1986) elaborated Fraenkel’s theory in more detail.23 Apart from the above-quoted Example (6), that illustrates domain integrity as well as colon subdivision, I give another one in (9). Slashes mark a colon boundary. 23. Habinek (1986:€ 11) proposed a distinction between rhetorical cola and rhythmical cola. Especially what he calls rhythmical cola, defined as “short, detachable, grammatical constituents of the Latin sentence”, are of interest for our subject. The colon division also inspired Adams (1994b: 109) when explaining the placement of Latin pronouns.



Introduction

(9)

Cuius adventu nuntiato / L. Plancus / qui legionibus his arrival being announced L. Plancus-nom who legions-dat praeerat / necessaria re coactus / locum capit superiorem /... commanded by necessity compelled place-acc takes higher-acc ‘On the report of his approach Lucius Plancus, who was in command of the legions, under the stress of necessity occupies the higher ground...’ (Caes. Civ. 1.40.5)

In a number of cases, Latin complex sentences can be seen as combinations of several successive cola. On the other hand, not every sentence allows a clear colon subdivision, as pointed out by Janse (1997:€111; cf. Salvi 2004:€145). Furthermore, it would make no sense to talk about cola in cases such as (10) which contain only single constituents. (10) Curio Marcium Vticam navibus praemittit. Curio-nom Marcius-acc to Utica with ships sends forward ‘Curio sends Marcius ahead to Utica with the ships.’ (Caes. Civ. 2.24.1)

5.2

Stylistic ornamentation

Latin constituent order in literary prose (as well as poetry) can reflect various stylistic figures that serve as ornamentation (see, for example, Kühner and Stegmann 1914, II: 616–622). These figures will not be described in this book. I only mention two of them: firstly, chiasmus, which is an arrangement following the pattern a b – b a; for example in (11), a are the nouns mala and voluptates, b the adjectives praesentia and praeteritae. In other words, the second noun phrase has a different order of elements. Secondly, parallelism, which copies the same ordering of constituents, for example the verb and the subject in (12). (11)

Non enim video, quo modo sedare possint mala neg indeed I see how allay can evils-acc praesentia praeteritae voluptates. present-acc past-nom pleasures-nom ‘I do not see how past pleasures can allay present evils.’

(12) Luget senatus, maeret equester ordo. mourns Senate-nom grieves equestrian order-nom ‘The Senate mourns; the equestrian order is inconsolable.’

(Cic. Tusc. 5.74)

(Cic. Mil. 20)

6. Aim and methodology of this study This book aims to provide a systematic description of constituent order in Classical Latin prose from a pragmatic point of view, in order to elucidate the pragmatic value of Latin sentences; in other words, what information is conveyed by a sentence or a clause. For doing so, I adopt the approach of Functional Grammar, combining it with several





Constituent order in Classical Latin prose

aspects worked out by the Prague School, such as contextual dependency. However, my aim is not to apply a theory to Latin constituent order but to try to understand more about it. My study integrates and discusses recent works on pragmatic aspects of Latin constituent order (Bolkestein, de Jong, Panhuis, Pinkster, among others); whereas these works are concentrate on particular questions or expressions – after all, Panhuis’s monograph is a case study just as well –, my aim is to describe Latin constituent order in a systematic way. Furthermore, the reader will find suggestions to solve several questions that are traditionally discussed in Latin grammars, such as the placement of the verb in sentence-initial position, the placement of pronouns, coordinators, and ellipsis. This book also aims to treat problems of more general linguistic interest. Indeed, the Latin language does not figure in the Eurotyp project (Siewierska 1998a) and other typologically oriented studies. Thus, attention is paid to various peculiarities of Latin in the domains of the noun phrase, focusing particles, and cleft constructions. The present study is based on several corpora. The main one is a corpus of Classical Latin prose including the works of Cicero, Caesar and Sallust. It contains different text types: philosophical treatises, speeches, correspondence, and historical narrative.24 There is a comparable amount of material – about 21,000 words – in each of the text types. This corpus has served for the chapters on interrogative and declarative sentences. For the examination of Latin noun phrases (Chapter 6), another corpus was set up.25 The chapter on imperative sentences mainly rests upon an examination of data provided by LASLA.26 Also, various case studies were made with the help of the CD-ROM Bibliotheca Teubneriana Latina (BTL); for example, with respect to the positioning of the verb mitto ‘to send’, or verbo-nominal constructions in the chapter on declarative sentences. However, the present study concentrates on the verb and its obligatory arguments; it does not examine in a systematic way optional complements and subordinate clauses. In the near future, I will elaborate these points in detail and devote some articles to them. The book is organised in the following way: Chapter 1 provides a survey of constraints and liberties in Latin constituent order. Chapter 2 presents my approach to Latin constituent order and principles of analysis, as well as the pragmatic functions Focus, Topic, Theme, Tail, and related phenomena. Chapters 3–5 discuss three main sentence types: declarative, interrogative, and imperative sentences. Chapter 6 is devoted to the order within the noun phrase. The most important results are summarised at the end of each section or chapter. 24. This corpus 1 includes: Cicero’s treatise Tusculan Disputations (books 1 and 3); speeches On His House and Philippics (1 and 4); correspondence (Letters to Atticus, 13.50 – 16); Caesar’s The Civil War (1 – 3.30), and Sallust’s The Jugurthine War. 25. The corpus 2 contains Cicero’s correspondence (Letters to Atticus, 1–4,€ 30,954 words), Caesars’s The Gallic War (books 1–5,€28,696 words), and Sallust’s The Conspiracy of Catiline and The Jugurthine War (32,360 words). 26. LASLA (Le Laboratoire d’Analyse Statistique des Langues Anciennes) is available for registered users on the Internet: http://www.cipl.ulg.ac.be/lsl.htm.

chapter 1

Placement constraints and liberties in Latin constituent order In the variable constituent order of Latin, there are a certain number of placement constraints as well as certain liberties that are not common in modern European languages such as Romance and Germanic, nor in the more flexible Slavic languages. The aim of this chapter is to catalogue at least the most important characteristics of Latin constituent order. One may note that Latin poetry is much more flexible than prose and shows even more liberties in the placement of constituents.

1. First sentence- and clause-position Because of their semantic-referential and functional properties, coordinators, connective particles, subordinators and interrogative words are usually found in initial position in Latin. The regular position of coordinators such as additive et, atque, ac ‘and’, neque, nec ‘and not’, alternative aut, vel, seu ‘or’, and adversative sed ‘but’ is at the beginning of a clause or, if they coordinate constituents, before the coordinated constituent. In Classical Latin prose, the initial position is obligatory for the connective particles nam ‘for’ and itaque ‘so’. Whereas other connective particles may appear in another position, for example igitur ‘therefore’ (see Example 1) and ergo ‘accordingly, then’, nam (2) and itaque never leave the initial position, and thus they are placed before subordinators and interrogative words. (1) Hanc tu igitur dedicationem appellas...? this-acc you therefore dedication-acc call ‘Is this, then, what you call dedication...?’ (2) Nam quis hoc non intellegit...? for who this neg understands ‘For who cannot see that...?’

(Cic. Dom. 118)

(Cic. Ver. 1.9)

The first sentence-position is reserved for the connecting relative qui, which has an anaphoric function. In the case of its co-occurrence with another candidate for the initial position, for example with a subordinator (quia), the connecting relative has priority, as illustrated in (3) where cuius picks up the preceding memoria rerum gestarum ‘the recording of past events’.



Constituent order in Classical Latin prose

(3) (memoria rerum gestarum) Cuius de virtute quia of which about virtue because multi dixere, praetereundum puto. many-nom have spoken to be passed over I consider ‘(the recording of past events) since many have spoken of its merits, I do not think I need say more about it.’ (Sal. Jug. 4.2) Subordinators are expected to appear in the first position of a subordinate clause. In Classical Latin, relative pronouns and subordinators introducing a complement clause or a final clause mainly have this introductory function and mark the beginning of the subordinate clause. But temporal (cum ‘when’, dum ‘as long as’, ubi ‘when’), causal (quia ‘because’, quoniam ‘since’) and conditional (si ‘if ’) subordinators give way to another constituent(s) in 20% to 30% percent of cases.1 They are typically preceded by anaphoric pronouns or noun phrases containing one (cf. Example 3 cuius de virtute ‘of its merits’), and by Sentence Topics (Tubero in 4).2 Focal or contrastive (palam ‘in public’) constituents are also found in sentence- or clause-initial position, as is shown in (5). (4)

Tubero cum in Africam venisset, invenit in provincia Tubero-nom when to Africa had come finds in province cum imperio Attium Varum. with command Attius Varus-acc ‘Tubero on reaching Africa finds Attius Varus in the province in military command.’ (Caes. Civ. 1.31.2)

(5)

Vbi id a Caesare negatum et, palam si conloqui when this by Caesar was refused and publicly if negotiate vellent, concessum est... they want was accorded ‘While Caesar refused this request, but allowed them, if they wished, to negotiate in public...’ (Caes. Civ. 1.84.2)

This placing of subordinators introducing adverbial clauses is not restricted to Classical literary prose.3 Similar arrangements also appear in technical prose; for example in Vitruvius’s On Architecture. In (6), the initial position is occupied by a noun phrase containing the connecting relative qui (Topic) followed by the contrastive Focus Auster ‘the South wind’ and then comes the subordinator cum. 1. 30% in Caesar (Civ. 1), 21% in Cicero (Tusc. 1). These figures are even higher than Marouzeau’s (1949:€123) estimation of 15–20% in Caesar and Cicero (without distinguishing subordinate clauses). In Sallust, other placement than initial is exceptional (see Marouzeau, ibid.). Latin grammars usually speak about traiectio, displacement of the subordinator (Kühner and Stegmann 1914, II: 614). See also Chapter 2, Section 3.6.3. p. 70. 2. See Panhuis (1982:€80 ff.), Pinkster (1995:€221), and Amacker (1998). 3. Poetry is even more flexible; a focal element standing before a subordinator may also be the verb, cf.: superat quoniam Fortuna, sequamur (Verg. A. 5.22) ‘Since Fortune is victor, let us follow.’



Chapter 1.╇ Placement constraints and liberties in Latin constituent order

(6)

In qua civitate Auster cum flat, homines aegrotant; in this city Auster-nom when blows people-nom are ill cum Corus, tussiunt. when Corus-nom cough ‘In this city when the South wind blows people fall ill; when the North-west, they cough.’ (Vitr. 1.6.1)

Question-words and interrogative particles (num, nonne) usually start questions, i. e. stand as the first word; they may be preceded by a coordinator (et, sed) or by a sentence-initial connective particle (nam, itaque). When a question-word is used in a noun phrase it can be separated from its head noun (7). In about 7–13% of cases, interrogative pronouns and particles occupy a position other than initial and are mainly preceded by a Topic constituent. Every type of Topic is found, contrastive (tu in Example 8), Sentence Topic or Resumed Topic (totum prope caelum in 9).4 However, apart from instances of ellipsis, interrogative pronouns and particles do not appear in sentence-final position.5 (7)

Qui sunt homines a Q. Metello, fratre tuo, consule in who are men-acc by Q. Metellus your brother consul in senatu palam nominati...? Senate publicly named ‘Who are the men whom the consul Quintus Metellus, your brother, publicly named in the Senate...?’ (Cic. Dom. 13)

(8) Atque ille tamen ad collegium rettulit, tu ad quem rettulisti? and he yet to College submitted you to whom submitted ‘Yet he submitted his dedication to the College; and to whom did you submit yours?’ (Cic. Dom. 132) (9)

Totum prope caelum, ne pluris persequar, nonne whole-nom nearly sky-nom neg further I pursue is not humano genere completum est? by human race filled up ‘Is not nearly the whole sky (to give no further examples) filled up with the human race?’ (Cic. Tusc. 1.28)

4. In poetry, we can find instances where the verb precedes an interrogative word, for example€(separor is contextually dependent): Separor a domina cur ego saepe mea? (Ov. Am. 2.16.41–42) ‘Why am I often separated from the mistress of my heart?’ 5.

See Chapter 4, Section 2. p. 198.





Constituent order in Classical Latin prose

2. Second sentence-position and enclitics enim ‘indeed’, autem ‘but’ and vero ‘however’ Any constituent whatsoever may occupy the second sentence-position, but in Classical Latin there are only three words for which this position is obligatory: the connective particles enim ‘indeed’, autem ‘on the other hand’, and vero ‘however’.6 In Caesar, Sallust and Cicero, these particles are excluded from the initial position in 100% of cases. Exceptionally, they are placed third (enim 6%, autem 3% and vero 0.5%) under well-defined conditions,7 when the author aims to keep together a syntactic unit, for example: difficile est enim ‘for it is hard’ (Cic. Sulla 31), quam multi enim ‘how many (orators)’ (Cic. Brut. 138), non est autem ‘but there is no’ (Cic. Div. 2.41). These placement properties allow us to consider them as enclitics, i. e. words requiring a full, autonomous word placed before them. Since they resemble words bearing their own accent, they can be regarded as “non-bound” enclitics (cf. Siewierska 2004:€26 and Spevak 2006b). It is important to define the second position occupied by these enclitics: it is the position after the first full word in a sentence, as is shown in (10). The negations non and neque, subordinators such as si, etsi, ut, cum, ubi, pronouns and the verb sum ‘to be’ in whatever function, including the auxiliary of passive perfect forms and the perfect of deponent verbs, also count as full words (see est... profectus ‘is gone’ in 11). (10)

Adventus enim L. Nasidi summa spe et arrival-nom indeed of L. Nasidius with utmost hope and voluntate civitatem conpleverat. goodwill city-acc had filled ‘For the arrival of L. Nasidius had filled the community with the utmost hope and goodwill.’ (Caes. Civ. 2.4.4)

(11) Est enim profectus in Hispaniam Dexius. be-aux but gone to Spain Dexius-nom ‘For Dexius is gone in Spain.’

(Cic. Fam. 7.23.4

One- and two-syllable prepositions do not behave as full words in Classical Latin. Thus, sentence enclitics do not follow prepositions but the first word of the prepositional phrase.8 (12) De morte enim ita sentit, ut... about death indeed so he thinks that ‘For his view of death is to think that...’

(Cic. Tusc. 5.88)

6. See Spevak (2006a, Chapter 7, Section 8) and Spevak (2006b). 7.

See Watt (1980), Adams (1994a), Spevak (2006a, Chapter 7, 8.3), and (Spevak 2006b).

8. The example quoted by Marouzeau (1953:€73), post autem eum (Cic. Tim. 46) ‘however after him’ is the only example from Classical Latin prose I know.



Chapter 1.╇ Placement constraints and liberties in Latin constituent order 

As they must come after the first full word, the enclitics enim, autem et vero split up noun phrases and prepositional phrases. In (13), the first full word is ea belonging to the prepositional phrase in ea civitate. It is noteworthy that in Latin even such closely related noun phrases as lexical units, for example res publica ‘State, republic’ (14), or proper names may be discontinuous. Sentence enclitics then stand after the first word and, in the case of proper names, between the praenomen (first name) and the nomen gentile (family name), for example after Quintus in (15).9 (13) In ea enim civitate mentio facta est. in this indeed city mention-nom was made ‘That was the state in which the matter was mentioned.’ (14) Rem vero publicam penitus amisimus. republic-but-republic-acc entirely we lost ‘But we have lost for ever our republic.’ (15) Q. enim Ligarius... Q. indeed Ligarius-nom ‘Indeed Quintus Ligarius...’

(Cic. Fam. 3.8.3)

(Cic. Off. 2.29)

(Cic. Lig. 2)

3. Enclitics -que ‘and’, -ve ‘or’ and -ne Apart from sentence enclitics placed in second sentence-position, there is another set of enclitic elements in Latin: the additive coordinator -que ‘and’, the disjunctive coordinator -ve ‘or’ and the interrogative particle -ne. Whereas enim, autem et vero resemble full words in the sense that they are phonologically independent and bear accent, -que, -ve and -ne depend phonologically on their host (Allen 1973:€25). Due to this property, they can be described as “bound” clitics (cf. Siewierska 2004:€26). The position of these enclitics depends on their function. The coordinator -que and the disjunctive -ve accompany the first constituent of a clause that is connected to another by coordination or disjunction, or the first (word of a) constituent involved in coordination or disjunction. In (16), -que hosted by eam coordinates a clause whereas -que tacked onto cellam coordinates a constituent. If the constituent is a complex phrase, -que and -ve are tacked onto the first word of the phrase, as in (17). Consequently, they split up noun and prepositional phrases, including lexical units such as patres conscripti ‘senators’ in (18) and proper names, for example Gaius Norbanus in (19).

9. There are three counter-examples, among them: De Gai Gracchi autem tribunatu (Cic. Amic. 41) ‘But about the tribuneship of C. Gracchus’, mentioned by Devine and Stephens (2006:€274).



Constituent order in Classical Latin prose

(16)

Cui (Atticae) quidem ego totam villam cellamque her-dat indeed I entire house-acc store-acc=and tradidi eamque cogitabam V Idus videre. I handed over her-acc=and I intended on 11th see ‘I handed the entire house and stores over to her (to Attica) and intend to see her on the 11th.’ (Cic. Att. 14.19.6)

(17) in hac autem peregrinatione militiave nostra in this but tour campaign=or our ‘on my late tour or military campaign’ (18)

(Cic. Att. 6.2.2)

ut aliquando vos patresque conscriptos Italiamque that sometimes you-acc fathers-acc=and conscript Italy-acc=and universam memoria mei misericordiaque all-acc memory-nom of me pity-nom=and and desiderium teneret desire-nom should hold ‘that at some time pity and regret for me should come upon you and the conscript fathers, and all Italy’ (Cic. Red. pop. 1)

(19) et in causa M’. Aquili Gaique Norbani nonnullisque aliis and in case of M’. A. of Gaius=and Norbanus several=and others ‘in the cases of Manius Aquilius, Gaius Norbanus and a number of others’  (Cic. De orat. 2.188) In the case of prepositional phrases, -que and -ve are frequently found after the first full word (20). However, unlike sentence enclitics, -que and -ve can also be attached to prepositions in Classical Latin. This occurs especially in two cases (Marouzeau 1947:€300 ff.): when the prepositional phrase contains the anaphoric pronoun is, both alone and in attributive function10 (21), and when two complex prepositional phrases are coordinated and the preposition is repeated (22).11 However, repetition of the preposition is optional. In the absence of a sufficient number of examples, at least in Cicero, there is no real support for Marouzeau’s statement (1947:€304 f. and 1953:€63) that dissyllabic prepositions behave in a different way from monosyllabic ones.12 10. In Cicero, Caesar and Sallust, however, sequences of monosyllable prepositions in or ex, -que, and the pronoun is, that is in/exque + is vs. in/ex + is + -que seem to be simple variants, with a slight preference for the case when -que joins the pronoun: in/exque + is occurs 12 times whereas in/ex + is + -que occurs 15 times. As for the preposition de, we find only deque + is (13 occurrences); de eoque and similar are avoided for understandable phonological reasons. 11. The choice between preposition + -que and preposition ... -que is difficult to establish. For example, deque occurs in 78% of cases, de ... que only in 22%, but inque appears in 29% of cases and in ... que in 71% (see BTL for Caesar, Sallust and Cicero). 12. Dissyllabic prepositions circum, sine, super, circa, apud, praeter, ultra, iuxta and inter occur rarely in combination with -que. A few examples of circumque (2 occurrences), sineque (1) and



Chapter 1.╇ Placement constraints and liberties in Latin constituent order 

(20) istum reliquit de provinciaque decessit. this-acc he abandoned from province=and went away ‘he abandoned the accused and left the province’ (21)

(Cic. Ver. 2.48)

Erat in Italia eius exercitus inque eo exercitu ipsius was in Italy his army-nom in=and that army of very tribuni plebis, inimici mei, fratrem praefecerat. plebeian tribune of my ennemy brother-acc he had put in charge ‘His army was in Italy, and in that army he had appointed to a command the brother of that very plebeian tribune, my enemy.’ (Cic. Sest. 41)

(22) de temporibus deque universa re publica... collocuti sunt. about bad times about=and whole state they discussed ‘they had discussions concerning the crisis and the state of politics generally’  (Cic. De orat. 1.26) The interrogative particle -ne is usually tacked onto the initial word of an interrogative sentence, as the initial word is typically the Focus of the question. In about 11% of its occurrences in Cicero (see BTL) is the word bearing -ne not initial, as for example in (23) where the preceding constituent is Topic. (23)

De reliquis rei publicae malis licetne dicere? Mihi about remaining of State evils is allowed=Q speak me-dat vero licet... indeed is allowed ‘Am I permitted to speak of the remaining ills of the State? I am permitted...’  (Cic. Phil. 1.14)

4. Adpositions Latin adpositions are predominantly prepositions. The placement of enclitics has shown that prepositions behave mostly as non-autonomous, proclitic words in Classical Latin prose. They do not usually constitute the first sentence-position and they may or may not serve as host to -que, -ve and -ne. As for their placement, prepositions normally come first in the prepositional phrase (cf. Example 22 above). However, in well-defined conditions, it is possible to find prepositions postposed or placed inside prepositional phrases (Marouzeau 1947:€307 f. and 1953:€64 f.). This is the rule for the connecting relative qui, for example: praeterque (2) attested in Cicero do not permit formulating more general conclusions. On the other hand, there are a few discontinuous sequences, for example with apud: apud... -que (2 occurrences). Roughly speaking, co-occurrences of dissyllabic prepositions with -que are rare in Cicero: in case of coordination (cf. Example 22), these prepositions are usually not repeated or the coordinator et is chosen instead of -que.



Constituent order in Classical Latin prose

qua de re ‘wherefore’ (Cic. Att. 1.20.1) and quam ob rem ‘for this reason’, more or less lexicalised as ‘therefore’ (Sal. Jug. 42.5). It also occurs with “simple” relatives in Classical Latin prose, especially in legal or administrative style (Marouzeau 1953:€64): (24) Hoc ipsum interdictum quo de agitur consideremus. this very injunction-acc which about is discussed we should consider ‘Let me deal with the actual injunction with which we are concerned.’  (Cic. Caec. 55) Besides these two cases, there are also examples13 with another pronoun (aliquam in rem ‘on any matter’, Cic. Inv. 2.178) or with an adjective (magno in honore ‘in high honour’, Caes. Civ. 1.77.2) that precede the preposition. According to Marouzeau (1947:€307, 320 and 1953:€64 ff.), it is not very common to find prepositions postposed or placed inside a complex prepositional phrase as in (25). In any case, “exceptions” listed in this paragraph do not invalidate the statement that prepositions are usually placed first in their domain. (25) plurimis iustissimisque de causis many very strong=and for reasons ‘for many strong reasons’

(Cic. Ver. 2.2)

There are a few postpositions in Latin, words governing a noun phrase but placed post-nominally, for example: causā, gratiā ‘on account of ’ or tenus ‘up to’.14 When the governed phrase is complex, causā and gratiā are placed not after the first word but as final words of the whole phrase, for example: rei frumentariae causā ‘for foraging purposes’ (Caes. Civ. 1.16.1).

5. Focusing particles Latin has a number of particles, for example et, etiam, quoque, vel, iam, quidem, solum or modo whose scope may cover just one constituent or an entire sentence.15 Focusing particles with a constituent in their scope have variable placement in the sense that they accompany a constituent placed anywhere in the sentence (König 1991).16 Some Latin focusing particles are prepositive, for example et, etiam, vel or solum, some are postpositive, especially quoque and quidem. Prepositive particles stand before the modified word (26), postpositives are placed after it; in the case of noun phrases, as in (27), postpositive particles regularly produce discontinuity. 13. See ThLL, s. v. ad, in and per. 14. For anteposition of these words, see Marouzeau (1947:€319). 15. It is thus necessary to distinguish focusing particles from connectives such as enim, autem or igitur. 16. In particular, see remarks concerning additive and restrictive particles (König 1991:€164 and 166 f.) in modern languages. Cf. also C. Kroon (1995:€282).



Chapter 1.╇ Placement constraints and liberties in Latin constituent order 

(26) Fateor me sectorem esse, verum et alii multi. I confess me broker-acc be-inf but also others many-nom ‘I confess that I am a broker, but so are many others.’ (Cic. S. Rosc. 94) (27) ...nosque cum multum litterae, tum non minimum us-acc=and both much literature-nom and neg least Idus quoque Martiae consolentur. Ides-nom also of March should console ‘...and allow literature to console us much, but the Ides of March in particular.’  (Cic. Att. 14.13.3) Furthermore, Latin has a discontinuous particle ne... quidem that marks the exclusion of an entity in a negative context. Also ne... quidem produces discontinuity of complex noun phrases: (28) [quis nostrum adfuit Vargunteio?] Nemo, ne hic quidem Q. Hortensius. nobody not this-nom even Q. Hortensius-nom ‘[Which of us helped Vargunteius?] No one, not even Quintus Hortensius here.’  (Cic. Sul. 6) When focusing particles modify an entire sentence, they stand at the beginning of it, as in (29). Etiam can also be placed as the second word, as is shown in (30). Such a placement is required for the postpositive particle quidem. When the first constituent is a noun phrase, it is regularly separated by quidem (31).17 (29)

Etiam homines novi, qui..., furtim et per latrocinia... even men-nom new who by stealth and through banditry ad imperia et honores nituntur. for command and office strive ‘Even new men who... now strive for command and office stealthily and through banditry.’ (Sal. Jug. 4.7)

(30) [Diserti autem Q. Varius, C. Carbo, Cn. Pomponius...] C. etiam Iulius, aedilis curulis, cotidie fere G. also Julius-nom curule aedile-nom daily almost accuratas contiones habebat. careful-acc harangues-acc delivered ‘[But speakers of real ability were Q. Varius, C. Carbo, Cn. Pomponius...] C. Iulius too as a curule aedile delivered carefully prepared harangues almost daily.’  (Cic. Brut. 305)

17. There are a few counter-examples, that maintain the proper name as a unit: M. Caeli quidem epistulam (Cic. Att. 10.9.2) ‘however the letter of Marcus Caelius’.



Constituent order in Classical Latin prose

(31) C. quidem Caesar non exspectavit vestra decreta. G. indeed Caesar-nom neg awaited your decrees-acc ‘C. Caesar indeed did not wait for your decrees.’ (Cic. Phil. 8.5)

6. Principles of domain integrity and head proximity Rijkhoff (2002:€250) argues that there is a cross-linguistic tendency for constituents to remain within their proper domain. For example, phrases (noun, prepositional) or clauses (participial, infinitive) keep their elements together. This principle, called the principle of domain integrity, is also valid in Latin.18 On the other hand, it can be “violated” – see next section on discontinuity of noun and prepositional phrases.19 At the same time, Rijkhoff (2002:€261) distinguishes another principle, head proximity, requiring that constituents occur adjacent to their head. This principle also holds true in Latin, as we will see in section on genitives (p. 265 ff.) that are often adjacent to their governing noun. On the other hand, there are instances of distantiation of genitive modifiers that occur for pragmatic reasons (see p. 274 ff.). Furthermore, there are other cases of violation of the principle of head proximity, for example complex sentences such as in (32a). The verb scribas governs an infinitive clause (horti – sint) which is not adjacent to it. This sentence is rewritten from the point of view of dependency in (32b).20 (32) a.

Horti quam in diem proscripti sint velim ad me gardens-nom which for day are advertised for sale I would like to me scribas. you write ‘I would like you to write to me what day the sale of the gardens is advertised for.’ (Cic. Att. 12.49.3)

b. Velim ad me scribas in quam diem horti proscripti sint. Such arrangements can be explained by pragmatic reasons. In fact, horti is a Topic constituent placed in initial position, and the indirect question to which it belongs, governed by scribas, conveys new information; velim ad me scribas in the given context is expected, less informative and thus it comes later. Incidentally, we note that horti exceeds its proper domain of subordinate clause (the so-called anticipatio). I add one more example of this phenomenon that is not exceptional: in (33) with an anticipated noun (signum). Note also that the governing clause is disrupted by its subordinate clause. 18. See Introduction, p. 9, Example (6) and p. 11, Example (9). 19. Violation of the principle of domain integrity in other cases, for example in subordinate clauses, requires further research. 20. This phenomenon also requires further research.



Chapter 1.╇ Placement constraints and liberties in Latin constituent order 

(33)

ita continuo signum ut demolirentur et Messanam so promptly statue-acc that they take down and to Messina deportarent imperavit they convey he ordered ‘he promptly ordered them to take the statue down and convey it to Messina’  (Cic. Ver. 4.84)

7. Hyperbaton or discontinuity of noun phrases Latin allows discontinuity, i. e. separation of elements forming a noun or a prepositional phrase.21 Whereas French or English noun phrases22 are normally not interrupted by other elements, discontinuity can be observed in other modern European languages, for example in Polish and Russian (Siewierska 1988:€170). In Latin, several types of discontinuity may be distinguished, depending on the elements that produce it (cf. Pinkster 2005b and Gettert 1999). In Sections 2, 3, and 5, we have seen that noun phrases and prepositional phrases are separated by enclitics and postpositive particles. However, this is not speaker’s or author’s choice but a constraint due to the enclitic and postpositive property of these elements. Enclitics and postpositive particles are not the only words that separate noun and prepositional phrases. Discontinuity can be produced by a constituent such as a verb, an adverb, or by several constituents. This type of hyperbaton is not obligatory but optional. It is usually regarded as an elaborated stylistic device; however, if we consider Siewierska’s (1984) observations on Polish, such a description seems to be misleading. Discontinuity of noun or prepositional phrases in Polish is not just a literary device but also occurs in everyday speech.23 Discontinuity does not represent a norm, and speakers exhibit variability in the production and the acceptance of discontinuous phrases. In other words, discontinuity is an optional device that occurs in contrastive contexts, for example (Siewierska 1984:€57 and 63): (34) Piękny Markowi kupili obraz. beautiful-acc to Mark they bought painting-acc ‘They bought Mark a beautiful painting.’

21. I will call a simple constituent (orator) a noun and a complex constituent (orator bonus) a noun phrase. By the way, I consider that the concept of verb phrase is not adequate for a description of Latin constituent order (cf. Siewierska 1988:€146). Forms such as profectus est ‘he set out’ are analytic forms of the passive voice and not verb phrases. 22. In English, determiners and possessive pronouns can be separated from their nouns under well defined conditions, but adjectives cannot (Siewierska 1988:€167). 23. See also Gasparov (1978) on hyperbaton in Russian impromptu speech.



Constituent order in Classical Latin prose

(35) Do mojego włamali się mieszkania. into my they broke refl flat ‘They broke into my flat. / My flat was broken into.’ Discontinuity tends to occur in short clauses such as (34) and (35). Discontinuous elements are placed in initial and final position, i. e. they occupy the positions of Topic and Focus. In (34), the topical adjective (piękny) is placed first and the focal noun is found in final position (obraz); this example can be paraphrased as: “they bought Mark (not a beautiful statue but) a beautiful painting”. In (35), Focus comes first (mojego): “my (and not someone else’s) flat was broken into”. There are several restrictions on discontinuity in Polish (Siewierska 1984:€62 ff.), including: discontinuous noun phrases contain typically just one modifier; the noun and the modifier can only be separated by the verb (and not, for example, in (34) by the indirect object Markowi alone). Furthermore, only one instance of discontinuity per sentence is allowed. I give this brief sketch description of hyperbaton in Polish in order to argue that there is no reason to consider it as a literary phenomenon in Latin. This type of hyperbaton is a pragmatically grounded device that serves to give prominence to an important element. It is not restricted to the literary language but is also found in text intended for oral presentation, for example in Plautus’s comedies. Furthermore, as we will see, Latin noun phrases are even more flexible than those in Polish. In Latin, hyperbaton produced by alien elements, which do not belong to the noun phrase, involves for example adverbs, pronouns, and verbs: (36)

Quibus rebus Romam nuntiatis tantus repente these things to Rome being announced such-nom suddenly terror invasit ut... panic-nom attacked that ‘When these events were announced at Rome there was immediately such panic that...’ (Caes. Civ. 1.14.1)

(37) Hae permanserunt aquae dies complures. these-nom lasted floods-nom days several-acc ‘The floods lasted for several days.’

(Caes. Civ. 1.50.1)

(38) Dvas a te accepi epistulas heri. Ex priore... two-acc from you I received letters-acc yesterday from first ‘I received two letters from you yesterday. The first...’ (Cic. Att. 14.2.1) These examples show that constituents of discontinuous noun phrases are not necessarily confined to initial and final positions, but can also appear in an internal position. The verb can stand outside the discontinuous noun phrase, as in (36). The pragmatic analysis of discontinuous noun phrases requires more detail (see p. 274 ff.) but from now on we can see that discontinuous elements are Topics (hae aquae in 37) or Focus (tantus terror in 36). It is also possible that only one of elements of the noun phrase is focal, as duas in (38) is the contrastive element.



Chapter 1.╇ Placement constraints and liberties in Latin constituent order 

The head noun usually has one modifier, but noun phrases containing more modifiers may also be discontinuous, for example the possessive, the noun and the genitive in (39): (39) ut... ad nostra me studia referam litterarum that to our refl pursuits I return of literature-gen ‘that... I return to my former literary pursuits.’

(Cic. Fam. 1.8.3)

In Latin prose, it is usual to find one hyperbaton per sentence; but two hyperbata can also occur in one sentence, especially of the type exemplified in (40), where two noun phrases (ornamentum proprium and Academiae meae) are entwined.24 (40) Est ornamentum Academiae proprium meae, quod... is ornament-nom Academia-dat appropriate my-dat because ‘It is an ornament appropriate to my ‘Academia’ for two reasons:...’(Cic. Att. 1.4.3) Discontinuity produced by alien elements occurs especially in noun phrases; it is less common in prepositional phrases. Marouzeau (1953:€68) quotes one example of it in Classical Latin prose: qua in re (in qua re) is interrupted by ego: (41)

Deinde, si qua ego in re fratri tuo rei publicae secondly if any I in matter brother your-dat of State causa restiterim... for sake I have opposed ‘Secondly if in any matter I have opposed your brother for the sake of the public good...’ (Cic. Fam. 5.2.6)

Proper names can be considered as very compact units; they are normally not disrupted by anything except enclitics and postposed particles. The example given in (42) is unique; however, it can be explained as an instance of contrast between two persons. As Devine and Stephens (2006:€275) point out, there are no examples with the intervention of a verb or a noun within proper names. (42) M. ad me Brutus, ut consueverat, cum T. Pomponio venerat. M. to me Brutus-nom as accustomed with T. Pomponius had come ‘Marcus Brutus had come to visit me, as usual, together with Titus Pomponius.’  (Cic. Brut. 10) There is another kind of hyperbaton: noun phrases or prepositional phrases can be separated by modifiers of their own such as a genitive noun (phrase) or a prepositional phrase (cf. Gettert 1999:€ 25 and passim). For example, tuas litteras in (43) is separated by its modifier, represented by a prepositional phrase. In (44), the prepositional phrase de virtutibus is disrupted by its genitive complement. Discontinuity produced by one or more complement(s) of a noun phrase is optional.

24. Poets, on the other hand, can integrate up to three hyperbata in a verse (Marouzeau 1953:€11).



Constituent order in Classical Latin prose

(43) tuasque de istis rebus litteras exspectabo your-acc=and about these matters letter-acc I will await ‘and I will look forward to letters from you about these matters’(Cic. Att. 7.17.5) (44) in epulis canere... de clarorum hominum virtutibus at banquets sing-inf about of famous men virtues ‘singing at banquets... in honour of the virtues of famous men’(Cic. Tusc. 1.3) As Siewierska and Uhlířová (1998:€135 f.) point out, in East Slavic languages a phenomenon similar to (43) can be observed. Prepositional phrases are regularly incorporated into noun phrases, especially with prenominally placed adjectives. Thus the noun and its adjective form a frame for delimiting the whole unit, as in (45). When the adjective is placed postnominally, the ordering as in (46) is chosen. Both examples are from Polish. (45) bogaty w doświadczenia człowiek rich-nom in experience man-nom ‘a human being rich in experience’ (46) człowiek bogaty w doświadczenia To sum up, there are several types of hyperbaton in Latin depending on the element that produces it: (i) discontinuity caused by enclitics and postpositive particles that is obligatory; (ii) discontinuity produced by alien element(s) that is optional, as well as (iii) discontinuity within the domain of the noun phrase caused by a genitive modifier or a prepositional phrase. Apart from the constraints on constituent ordering enumerated in this chapter, other Latin words can occupy any position whatsoever within a sentence.

chapter 2

Pragmatic functions 1. My approach to Latin constituent order 1.1

Introduction

Latin constituent order is often labelled as “free”. This term, which goes back to Weil (1844:€54),1 is somewhat unsatisfactory, because it is understood in various ways. On the one hand, unlike English, Latin constituent order is not subject to syntactic rules, and the position of a constituent can vary without changing its syntactic function (cf. Pinkster 1991).2 On the other hand, Latin constituent order strongly obeys pragmatic rules in the sense that not only the semantic value of constituents but also their position build up together the message the that speaker or author wants to communicate.3 The pragmatic approach of Dik’s Functional Grammar, which I will adopt in this book, has not yet been applied to Latin in a systematic way. However, another pragmatic approach has been, the Functional Sentence Perspective, following Firbas (Panhuis 1982). These two approaches are conceptually different but they have some features in common, in particular the consideration of the pragmatic status of constituents. Indeed, Panhuis (1982) as well as H. Dik (1995) – in her study of Ancient Greek from the perspective of Functional Grammar – deem it indispensable to examine whether a constituent already figures in the immediate context or not. The Functional Grammar approach seems to me convenient for describing Latin constituent order particularly because of the distinction it makes between two main pragmatic functions, Topic and Focus. I find it very difficult to decide on the informative contribution of each element in the sentence, as postulated by the Functional Sentence Perspective (cf. Section 1.3.2 below). However, I will integrate its concept of contextual

1. Weil established a distinction between “languages with a free construction” and “languages with a fixed construction”, based on his observations of Greek, Latin, French and German. 2. Well-known examples in English are sentences such as: Peter beats Paul and Paul beats Peter where Peter and Paul do not have the same function (Marouzeau 1953: ix). 3. It is indeed in this sense that the Prague School uses the term “free constituent order”. Free does not mean that a word can be placed anywhere in the sentence but that its position has incidence on the meaning of the sentence. Consequently, the speaker can choose which word he considers the most informative one and place it in a prominent position (cf. my remarks concerning neštěstí, indicated below, Examples (4) and (5)).



Constituent order in Classical Latin prose

dependency into the Functional Grammar approach, in order to arrive at some principles governing constituent order in Classical Latin prose.

1.2

Problems in the analysis of Latin constituent order

General studies of constituent order examine the ordering of constituents in modern languages and try to define theoretical principles underlying the observed arrangements; with modern languages one is in a position to consult speakers. For the Latinist, the perspective is different: we try to determine the pragmatic value of constituents in a given sentence, and to explain why a particular ordering is found and what it means. We also observe regularities in the organisation of Latin sentences, drawing on theoretical observations made on modern languages. Since Latin is a dead language known only from its written tradition, we take no account of its intonation and prosody, which, if they were available, could have given us a better understanding of the principles underlying sentence structure. However, it is indispensable to distinguish texts that were intended for oral presentation – such as comedies or speeches – and work, at least to a certain extent, with prosodic realisation, from texts that are purely literary. According to Firbas (1992:€ 218), literary language does not make use of prosodic means. The writer must be sure that sentences he formulates will be understood as such irrespective of their specific prosodic realisation.4 As Firbas (ibid.) points out, literary language is a code, produced as well as perceived. When decoding a text and its units, readers rely on the signals the sentences give them. These are specifically: syntax serving as a vehicle for the semantic content, the position of each constituent in the sentence, and its relationship with the immediately relevant context. Spoken language utilises prosodic means in addition to these signals. Leaving aside the first of these (syntax), I will pay attention to the position of constituents (Section 1.3), and the relationship of constituents with the pertinent context (Section 1.4).

1.3

The position of constituents in the sentence

1.3.1 Topics and sentence-initial position The sentence-initial position is a prominent one and is usually occupied by pragmatically marked constituents, mainly Topic and contrastive elements.5 Topic, the entity spoken about, usually goes in this position in Latin, as is shown by Pinkster (1995:€222).6 4. On the other hand, the author of a text intended for oral presentation must make sure that his sentences will be understood without written support. For example, when using discontinuous phrases–if he/she is conscious at all of doing so– the author assumes that his audience is able to understand them. 5.

However, I am not arguing that every sentence-initial constituent is prominent.

6. According to Knoth (2006:€ 211), Topics occupy the sentence-initial position in 40% of cases, Focus in 38%, settings in 15% and other elements in 7%.



Chapter 2.╇ Pragmatic functions 

For example, sentence (1) conveys information about Marius and this constituent functions as (Discourse) Topic: (1)

Marius postquam confecto quo intenderat Marius-nom after concluded on which concentrated negotio Cirtam redit. business to Cirta returned ‘After Marius had concluded the business on which he had been concentrating, he returned to Cirta.’ (Sal. Jug. 104.1)

1.3.2 Focus position Whereas there exists a consensus, more or less, that the entity about which information is given stands in initial position (Pinkster 1991:€ 74, 1995:€ 222, and Panhuis 1982:€57; passim on “theme proper”), the position occupied by the most informative element is more difficult to establish in Latin. Panhuis (1982:€54 f.; cf. 1981:€300) argued that the normal order of a Latin sentence observes the principle of communicative dynamism formulated by Firbas (1992:€7), i. e. that the constituent conveying the most salient information stands at the right-hand end of the sentence or, in Classical Latin prose, before the verb.7 Such a claim is not surprising, because there is a more general tendency for given information to precede new information, the so-called given > new hierarchy.8 However, if this were so in Latin, the final – or pre-verbal – position would be an indicator of saliency. In order to demonstrate this, it may be helpful briefly to present the functioning of increasing communicative dynamism in Czech, a language that has served as a point of departure for Firbas’s concept. This principle claims that the organisation of a sentence goes from the least informative element to the most informative one. See the following examples in French (2), English (3), and Czech (4, not involving a special prosodic realisation; cf. Firbas 1992:€136): (2) Hier, un accident s’↜est produit. Hier, il s’est produit un accident.

(3) Yesterday there was an accident.

(4) Včera se stalo neštěstí. yesterday refl occurred accident-nom In French, there is no notable difference in meaning between the sentences given in (2), although the sentence with a dummy subject pronoun (il) and late placement of the indefinite noun phrase is preferred in sentences containing a verb such as ‘to be’, ‘to 7.

For Pinkster’s (1995:€211) reservations on this, see also Chapter 3, Section 2.1. p.132.

8. Introduction, Section 4.2.2.3. p. 9. See also Hannay and Martínez Caro (2008) on sentencefinal Focus phenomena. They argue that focal information presented sentence-finally is the outcome of the speaker’s specific discourse strategies.



Constituent order in Classical Latin prose

appear’ (Vet 1981:€151). The same tendency, i. e. placement of an indefinite noun rightwards in the sentence is observed in English (3). In Czech, sentence (4) informs us about what happened yesterday: neštěstí (‘accident’), placed in sentence-final position, is the most informative element. It is possible to place another constituent in this position, for example the adverb včera (‘yesterday’): (5) Neštěstí se stalo včera. accident-nom refl occurred yesterday This sentence has another meaning – or another pragmatic value: it says when the accident took place. The adverb včera (‘yesterday’) does not function as a setting element but as the Focus and specifies the process expressed by the verb. The noun neštěstí in (5) (‘accident’) does not have the same relationship with the context as in (4): in sentence-final position, it is read and understood as a new element that was not mentioned before. The sentence-initial position in (4) signals that it is given information. Otherwise, an indefinite marker such as nějaké neštěstí ‘a (certain) accident’ would have to accompany it, in order to avoid the normal definite reading. Compare now an example from Caesar (6) with its translations into English and Czech (7): (6)

Accidit etiam repentinum incommodum biduo, quo happened also unforeseen-nom disaster-nom two days rel haec gesta sunt. these-nom were done

(7)

Dva dni po těchto událostech došlo ještě two days after these occurrences happened also k neočekávanému neštěstí.9 to unforeseen disaster ‘There also happened an unforeseen disaster within two days of these occurrences.’ (Caes. Civ. 1.48.1)

The Czech sentence starts with the temporal setting related to the preceding context (dva dni po těchto událostech ‘within two days of these occurrences’); the Focus, neočekávanému neštěstí10 (‘an unforeseen disaster’), stands in sentence-final position, which signals its newness. The Latin sentence starts with an initial, non-focal, verb accidit, followed by the particle etiam ‘also’. Then comes the Focus repentinum incommodum ‘an unforeseen disaster’ and finally the temporal expression placed in sentence-final position. According to the principle of increasing communicative dynamism, the temporal expression biduo, quo haec gesta sunt ‘within two days of 9. The Czech translation is taken from Gaius Iulius Caesar: Válečné paměti. O válce občanské, Prague (1972:€327). 10. The Czech sentence presents a different syntactic construction, an impersonal verb + prepositional phrase as its argument. This has no relevance to the ordering of the constituents.



Chapter 2.╇ Pragmatic functions

these occurrences’ should be the Focus of the sentence. According to considerations about “emphatic” placement of the verb in initial position,11 one would interpret the verb accidit ‘it happened really’ as the Focus. However, examination of the context permits one to reject both interpretations. Caesar informs us about what rendered the situation, already difficult, even worse. Incommodum was not mentioned before; furthermore, its attribute repentinum (‘sudden’) is, due to its lexical value, a signal of newness (Torrego 2005). The verb accidit is not prominent; it only serves to introduce the repentinum incommodum that is described in the subsequent context. The temporal expression, explicitly related to the preceding context (haec), appears, to my astonishment, as a speaker of Czech, in sentence-final position. The above quoted example is far from isolated. Since I have discussed elsewhere how the Focus position in Latin does not always match that of Czech (Spevak 2008), I give just one additional example. It concerns a recurrent formula in Sallust’s and Caesar’s historical narratives: “to send a delegation”.12 In such sentences, the Focus is represented by nuntios or legatos ‘deputies’, who in some way announce the content of the message to be communicated. In (8), the constituent nuntios follows the Topic constituent Bocchus, whereas in Czech (9), poslů takes the sentence-final position. (8)

Nam Bocchus nuntios ad eum saepe miserat: for Bocchus-nom envoys-acc to him often had sent [velle... populi Romani amicitiam…] ‘For Bocchus had been sending messengers to him frequently: [he wanted the friendship of the Roman people...]’ (Sal. Jug. 88.5)

(9)

Bocchus s ním nepřetržitě vyjednával prostřednictvím Bocchus with him continually negotiated through poslů; [vzkazoval, že si přeje mír s římany...]13 envoys

To sum up, whereas in Czech the principle of increasing communicative dynamism functions well, it is difficult to claim it for Latin in a systematic way. Consequently, the sentence-final – or pre-verbal – position is not a reliable indicator of the saliency of a constituent. This does not mean that Focus constituents cannot be found in the final and preverbal positions, but they can also be placed elsewhere in the sentence, especially after the Topic, or in the initial position. This being so, the major question is how to identify Focus constituents in Latin. The following section proposes useful criteria for recognising them.

11. See Kühner and Stegmann (1914, II: 590), among others. 12. Attested Latin patterns are analysed in detail in Chapter 3, Section 2.1.3, p. 136 ff. 13. The Czech translation is taken from Gaius Sallustius Crispus: Catillinovo spiknutí, Válka s Jugurtou, Prague (1988:€141).





Constituent order in Classical Latin prose

1.4

Situational and contextual dependency

As we are dealing with a language known only from written records, the relationship with the context is often the only means of assessing the informative value of a constituent and, consequently, the pragmatic value of a sentence.14 In this study, I will systematically exploit Firbas’s concept of contextual dependency or boundness, applied to Latin by Panhuis (1982:€11 and passim), and to English by Hannay.15 According to Firbas (1992:€22), a constituent is dependent if it is inferable from the discourse situation or is mentioned in the preceding context. Dependent on the situation (Firbas 1992:€24 f.) are all words that refer to it, in particular the discourse participants, the speaker and the addressee, as well as words referring to the extra-linguistic situation concerning them. These constituents may or may not have a pragmatic function. Compare the personal pronoun (ad) me, without a specific function in (10), occupying an internal position in the sentence, with contrastive ego and tibi in (11),€in the prominent sentence-initial position. The fact that pragmatically unmarked pronouns appear in an early, non-initial position is to be ascribed to their situationally dependent or bound status.16 (10) Lepidus ad me heri vesperi litteras misit Antio. Lepidus to me yesterday evening letter-acc sent from Antium ‘Lepidus sent me a letter yesterday evening from Antium.’ (Cic. Att. 13.47a) (11) Ego tibi irascerer? Tibi ego possem irasci? I you-dat be angry you-dat I can be angry ‘I be angry with you? Could I be angry with you?’

(Cic. Q. fr. 1.3.1)

Contextually dependent are elements retrievable from the preceding context. The formulation “preceding context” calls for further specification, because not the whole but only the pertinent context has to be considered (Firbas 1992:€23). For example, Caesar mentions a valley in Section 27.3 of the second book of The Civil War (see (12) below); the subsequent context concerns other events. When Caesar wants to return to the valley (in Section 34.1, Example 13), he has to reintroduce it because the mention of other material in the meantime has made this valley inaccessible to the reader. Furthermore, Caesar uses an explicit reminder formula ‘as explained above’ that allows the reader a better orientation to what has already been said and what it is related to.

14. However, this relationship is sometimes not evident or is difficult to determine; this problem concerns in particular Cicero’s correspondence, because we often do not know the content of the letter to which Cicero is replying. 15. Hannay (1985b) with reference to Daneš (1974). 16. Cf. Panhuis’s (1982:€56 f.) remarks on the co-occurrence of several dependent elements in a sentence. He considers that contextually given constituents are more “thematic” than situationally given ones, for example id tibi in Pl. Men. 1035.



Chapter 2.╇ Pragmatic functions 

(12) ...atque una valle non magna interiecta suas and one valley-abl neg big-abl lying between his-acc uterque copias instruit. each forces-acc draws up ‘... and they both drew up their forces separated by a single small valley.’  (Caes. Civ. 2.27.3) (13)

Erat vallis inter duas acies, ut supra demonstratum was valley-nom between two lines as above shown est, non ita magna... Hanc... was neg so big-nom this-acc (valley) ‘Between the two lines there was, as described above, a valley, not particularly deep... This valley...’ (Caes. Civ. 2.34.1)

While there is no reliable way of measuring the non-pertinence of a constituent in a given context,17 contextual dependency is readily observable. Firbas (1992:€31) distinguishes four categories of means: resumption by a noun, a pronoun, morphological devices,18 and ellipsis. Latin examples are given below in (14) – (16). (14) [Huic Sp. Albinus qui proximo anno post Bestiam cum Q. Minucio Rufo consulatum gerebat, persuadet... regnum Numidiae ab senatu petat.] Avidus consul belli gerendi movere quam senescere keen-nom consul-nom of making war move than let languish omnia malebat. all-acc preferred ‘[This was the man whom Spurius Albinus, who held the consulship with Quintus Minucius Rufus in the year after Bestia, persuaded to seek the kingdom of Numidia from the Senate...] The consul, being keen to make war, preferred to stir everything up rather than let things go stale.’ (Sal. Jug. 35.2–3) (15) (Iugurtha) At ille sese in loca saltuosa et natura but he refl to places woody and naturally

17. Cf. the concept of ‘predictability’ applied to Polish by Siewierska (1998b) following Givón (1983). The predictability that a constituent appears (Givón’s “Task urgency principle”) can be theoretically measured using referential distance, by taking into consideration the number of sentences between the examined occurrence of a word and its occurrence in the preceding context. However, Siewierska (1998b), working with a distance of twenty sentences, arrives at the conclusion that the analysis of Polish – although it brings interesting results – does not really confirm predictability as a test. Cf. also Bolkestein and van de Grift (1994:€286 f.). 18. Resumption by morphological means concerns, for example, the use of third person pronouns in languages such as English that do not admit zero anaphora: ‘(Marius)... he arrived at a hillocky area..., and there he waited’ (cf. Example 16).



Constituent order in Classical Latin prose

munita receperat. protected had retired ‘But the man had retired to an area full of wood and naturally well protected.’  (Sal. Jug. 54.3) (16) (Marius)... Ø pervenit in locum tumulosum... ibique... Ø opperitur. arrives to place hillocky there=and waits ‘(Marius)... he arrived at a hillocky area..., and there he waited.’  (Sal. Jug. 91.3) Contextual dependency is often signalled by means of anaphoric expressions. In (14), Spurius Albinus is referred to by a noun indicating his function: consul; the pronoun ille refers back to Jugurtha (15), the principal character in the text; Marius in (16) is the Discourse Topic of the paragraph continued by means of zero anaphora (Ø). Contextually dependent elements may or may not be assigned a pragmatic function (mainly Topic). In any case, we cannot say that these elements always go in the initial position; instances of later placement of a contextually given entity are not exceptional at all. Here is an example of an explicative sentence with the contextually dependent noun oratio ‘speech’ that is not selected as Topic;19 the Focus of this sentence is on the manner. (17) [Brutus noster misit ad me orationem suam habitam in contione Capitolina petivitque a me ut eamne ambitiose corrigerem ante quam ederet.] Est autem oratio scripta elegantissime sententiis, is but speech-nom written very finely in thoughts verbis, ut nihil possit ultra. in words that nothing can farther ‘[Brutus sent me the speech that he made at the Capitol meeting asking if I might be kind enough to correct it before he publishes it.] But the speech is extremely well written; both content and wording nothing could be better.’  (Cic. Att. 15.1a.2) To these basic means we can add other co-referential means and expressions inferred to be associated with an explicitly mentioned word (cf. Hannay 1985b: 61). Consider the following examples: (18) [Erat in celeritate omne positum certamen, utri prius angustias montesque occuparent...] Confecit prior iter Caesar atque... aciem instruit. completed first journey-acc Caesar-nom and line-acc drew up ‘[The whole contest turned on speed, which of the two would first seize the pass and the hills...] Caesar completed his journey first and he drew up a battle-line...’  (Caes. Civ. 1.70.1–3) 19. Oratio could be omitted.



Chapter 2.╇ Pragmatic functions 

(19)

[Caesar...duas legiones suas antecedere, reliquas subsequi iussit...] eique negotio Q. Fufium Calenum legatum praeficit. this-dat=and duty-dat Q. Fufius Calenus-acc legate puts in charge ‘[Caesar ordered his own two legions to march first, the rest to follow close...] and set his legate, Q. Fufius Calenus, in charge of this duty.’  (Caes. Civ. 1.87.4)

There is competition between Caesar and Afranius as to who will first take an advantageous position. In (18), confecit iter is deducible from the intention of seizing the hills: in order to do this, one must travel. The idea of ‘first’ (prior) is in essence repetition of prius. In (19), the verb praeficit is deducible from the extra-linguistic fact that such operations usually have a leader. The noun phrase ei negotio ‘this duty’ summarizes the content of Caesar’s orders. All these constituents are contextually dependent. By contrast, Caesar and Q. Fufium Calenum are the only non-deducible constituents that, at the same time, function as the most informative ones: that is, as Focus. It is worth pointing out that there is no contradiction between the Focus function of the constituent Caesar in (18) and the fact that he is a well-established character in The Civil War. Here, the reader is expecting the following information: Who will arrive first, Caesar, or Afranius? The result is not predictable.

1.5

Question tests

Each sentence or clause is expected to convey some information that will contribute to the development of the discourse. It may include other elements, but it must contain something new: the Focus. The most informative part of a sentence can be determined with the help of a question test.20 Indeed, each sentence can be viewed as the reply to an explicit or implicit question. Consider for example: (20) Yesterday, I went to the cinema. Depending on the concrete context, this sentence can represent a reply to at least three questions. (21) a. b. c.

Where did you go yesterday? – Yesterday, I went to the cinema. To the cinema. What did you do yesterday? – Yesterday, I went to the cinema. I went to the cinema. What happened? – Yesterday I went to the cinema.

20. See, among others, Sgall et al.€(1986:€207), and Daneš (1968:€57). For Latin, see de Jong (1989:€522 f.) and Bolkestein (2000:€199) who mention the problem without treating this question in detail.



Constituent order in Classical Latin prose

The reply may contain elements that are already known, or be reduced to the most informative part, as To the cinema in (21a) and I went to the cinema in (21b). No reduction is possible in (21c). This is a “what-happens” sentence that could occur for example in the following context: c.′ [I will tell you what happened.] Yesterday I went to the cinema. I met Paul there. He told me about... Determining the explicit and implicit questions to which the sentence under examination represents an adequate answer is important for understanding its pragmatic value. This permits the identification of the most informative part of the sentence. Focus constituents are most clearly observable in dialogues. Question-words signal what piece of information is missing and the “minimal” answer is provided in the reply; for example: (22)

Quot in partes distribuenda est omnis doctrina how many in parts to be distributed is whole theory-nom dicendi? – In tres. – Cedo quas? – Primum in ipsam vim of speaking in three please what-acc first in itself nature oratoris, deinde in orationem, tum in quaestionem. – of speaker then in speech then in topic In quo est ipsa vis? – In rebus et verbis. in what is itself nature in matter and words ‘Into how many parts ought the theory of rhetoric as a whole to be divided? – Three. – Tell me what they are. – First, the speaker’s personal resources, second the speech, and third the topic. – In what do the speaker’s personal resources consist? – In matter and in language.’ (Cic. Part. 3)

However, in literary prose texts, we are naturally often confronted with expressions that are more complex than these. On the one hand, they often contain contextually bound elements; on the other hand, a message need not be delivered piece by piece but can be given all at once.

1.6

Principles of analysis

The aim of this section is to describe the principles of analysis that I will follow in my study. In a prose text such as those that make up my corpus the objective is to determine what implicit question underlies the sentence under examination. This task is not always easy and it is necessary to use contextual clues, especially the contextual dependency of a constituent and its continuation in the subsequent context. As Firbas (1992:€42) points out, contextually independent elements are the strongest candidates for Focus function. The town of Taurois in (23) has such a contextual status: (23) [Massilienses... ex portu exeunt] et Tauroenta, quod est and to Taurois which is



Chapter 2.╇ Pragmatic functions 



castellum Massilensium, ad Nasidium perveniunt fortress-nom of Massilia to Nasidius they come ibique naves expediunt... there=and ships-acc get ready ‘[The Massiliots... come out of port] and reach Nasidius at Taurois, a fortified outpost of Massilia; there get their ships ready...’ (Caes. Civ. 2.4.5)

The Massiliots (the Discourse Topic), as well as Nasidius, are already known from the preceding context; Taurois has not yet been mentioned. This place will constitute the setting for the subsequent context and can thus be considered as the most informative element of its sentence. Furthermore, the verb perveniunt ‘they reach’ is deducible from the context: if the Massiliots set out (exeunt), we expect them to arrive somewhere; ‘where?’ is thus the underlying question that this sentence answers. Consider now the following example: (24) Eum libellum Caesari dedi. that petition-acc to Caesar I handed ‘That petition I handed to Caesar.’

(Cic. Att. 16.16a.2)

As the constituent eum libellum is contextually dependent, we can envisage that this sentence answers the question ‘to whom?’ or ‘what did I do?’ The context enables us to decide on the correct interpretation: (25) [Commotus Atticus libellum composuit. Eum mihi dedit ut darem Caesari; eram enim cenaturus apud eum illo die.] Eum libellum Caesari dedi. [Probavit causam.] (Cic. Att. 16.16a.2) ‘[Atticus was alarmed and drew up a petition. He gave it to me to give to Caesar, as I was going to have dinner with him that day.] I gave Caesar the document. [He approved the case.]’ It is clear that not only the constituent libellum, but also Caesari is already known: Atticus’s request was to “give the document to Caesar”. Thus the sentence does not answer ‘to whom?’ but the Focus is on the accomplishment of the action: dedi in the perfect tense (‘what did I do?’). The Focus may concern not a whole constituent but only a part of it. This is especially the case with noun phrases: for example the qualifying adjective aculeatas ‘stinging’ in (26) bears the Focus. The context makes it clear that the fact of having written a letter to Dolabella is not new. (26) [Atque ego ad eum (=Dolabellam) VIII Id. litteras dederam bene mane... Sed, ut ad te eo ipso die scripseram,] satis aculeatas ad Dolabellam litteras dedi; rather stinging-acc to Dolabella letter-acc I gave



Constituent order in Classical Latin prose

[quae si nihil profecerint...] (Cic. Att. 14.18.1) ‘[And, in fact, I had written to him very early in the morning of the 8th... But, as I wrote to you that same day,] I sent Dolabella a fairly stinging letter. [Even if that does no good...]’ There is no rigorous correlation between the contextual independence of constituents and Focus function; contextually dependent elements can also bear Focus, as is nicely shown in (18), quoted above, concerning the constituent Caesar. There are sentences which have no one salient constituent: the information in them is complex and is conveyed by a pragmatic unit with several constituents. We have, for example: (27)

Qua opinione adductus Varus postero die mane by this view persuaded Varus-nom next day morning legiones ex castris educit. Facit idem Curio… legions-acc from camp leads out does the same Curio-nom ‘Under the influence of these views Varus leads his legions out of camp early the next day. Curio does the same...’ (Caes. Civ. 2.27.3)

Varus leads his legions out of camp and disappears from the scene for a certain period. The author does not focus on ‘what’ Varus leads out, nor ‘from where’; he is merely relating what Varus does. The constituents legiones ex castris educit function as a pragmatic unit (Pinkster 1995:€232), and the clause answers the underlying question ‘what does Varus do?’ Note that his action is summarized by facit idem ‘(Curio) does the same’ in the immediately subsequent context. Finally, an example to illustrate sentences that contain contextually unbound elements: (28) Nunc neglegentiam meam cognosce. De gloria librum now carelessness-acc my-acc notice-imp On Glory book-acc ad te misi et in eo prohoemium id quod est to you I sent and in it preface-acc it rel is in Academico tertio. Id evenit [ob eam rem quod habeo in Academic third it happened volumen prohoemiorum. Ex eo eligere soleo cum aliquod sύggramma institui.]  (Cic. Att. 16.6.4) ‘Now notice my carelessness. I have sent the book “On Glory” to you and in it a preface which is in Book III of the “Academics”. This happened [because I have a volume of prefaces from which I am in the habit of selecting when I have put a work in hand.]’ Cicero is telling Atticus a story: he mixed up his prefaces by mistake and accompanied his book De gloria with the preface of the third book of Academica. The sentence nunc neglegentiam meam cognosce ‘now notice my carelessness’ has an introductory function and serves to draw attention to the story that follows. De gloria... tertio answers



Chapter 2.╇ Pragmatic functions 

the question ‘what happened?’, which is explicitly referred to by id evenit ‘it happened’ signalling that the whole event is meant. However, I suggest assuming a second underlying question for the clause De gloria librum ad te misi. Indeed, the subsequent context concerns more particularly the constituent De gloria librum (note the anaphora in eo); the secondary question is ‘what did I send you?’

2. Focus In the Introduction, Focus was defined as the element conveying salient information. The aim of this section is to present globally the concept of Focus. As detailed analysis of the corpus will follow in Chapters 3, 4, and 5, this section does not present figures concerning numbers of occurrences. I will first concentrate on different types of Focus: Focus related to newness (Section 2.1), presentative and what-happens sentences (Section 2.3), and complex information (Section 2.2), then on related phenomena: contrast and its types (Section 2.4), focusing particles (Section 2.5), emphasis (Section 2.6), connective particles (Section 2.7), and cleft constructions (Section 2.8). Hyperbaton or discontinuity of noun phrases represents a means of pragmatic highlighting; it is not included in this chapter, but was briefly presented in Chapter 1 (Section 7), and is discussed in detail in Chapter 6 (Section 9).

2.1

Focus related to newness

Focus is the most important or salient piece of information in a clause (Dik 1997, I: 326 ff.); there is only one Focus per clause. The pragmatic function of Focus can be assigned to entities and to actions. Focus function is closely, but not exclusively, related to the unbound contextual status of entities. There are a few formal indicators of salience: enumeration, repetition, and coordination. Enumeration of several constituents, as in (1) below, shows clearly that the underlying question is ‘from where?’ and not ‘what?’ Repetition can be linked with saliency but also with emphasis (see below, p. 47). In (2), the contextually given constituent filiam is repeated with insistence and stands in clause-initial position due to the contrast with the parts of the body (membra and others) that were enumerated before. In this passage, Cicero asks Decianus who had kidnapped a young woman and mistreated her, to restore her to her father. (1)

Frumenti vim maximam ex Thessalia, Asia, Aegypto, of grain quantity-acc large-acc from Thessaly Asia Egypt Creta, Cyrenis reliquisque regionibus comparaverat. Crete Cyrene other=and districts he had collected ‘He had collected a very large quantity of grain from Thessaly, Asia, Egypt, Crete, Cyrene, and other districts.’ (Caes. Civ. 3.5.1)



Constituent order in Classical Latin prose

(2)

[reddat misero patri filiam; membra... restituere non potest;] filiam, filiam, inquam, aerumnoso patri, Deciane, redde. daughter-acc I say sorrowing-dat father-dat Decianus-voc restore-imp ‘[let him restore the daughter to her miserable father; he cannot restore... the limbs...;] the daughter, restore the daughter, I say, Decianus, to her sorrowing father.’ (Cic. Flac. 73)

Especially in the case of actions21 – that are often not easy to recognise as Focus constituents – coordination helps us to determine their saliency. A good example is given in (3); discedunt is the only contextually independent constituent. Coordination with domum revertuntur (a pragmatic unit providing complex information) confirms that discedunt functions as Focus and that the underlying question is ‘what do they do?’ Coordination without any coordinator involved also counts here (Pinkster 1990:€12), see Example (4). Note that these verbs are used in an absolute way (scripsi), or with ellipsis of usually obligatory constituent(s): misi.22 (3) Milites in itinere ab eo discedunt ac domum revertuntur. soldiers-nom on way from him depart and home return ‘His soldiers desert him on the way and return home.’ (Caes. Civ. 1.12.2) (4) Iucundissimas tuas legi litteras. Ad Plancum scripsi, misi. most delightful your-acc I read letter-acc to Plancus I wrote I sent ‘I have read your most delightful letter. I have written to Plancus and sent it off.’  (Cic. Att. 16.16.1) Focus can be involved in the case of expressions of tense or modality of a verb (cf. Firbas 1992:€17), as illustrated in (5) with the present iudicant and the perfect iudicaverunt, coordinated by et... et. (5)

Itaque viri fortissimi atque honestissimi, qui ordines and so men very valiant and very honest-nom who classes-acc duxerunt, res et iudicant et iudicaverunt. led matter-acc and judge and have judged ‘Accordingly men of the greatest valour and integrity who have been in command still act as judges and have hitherto acted.’ (Cic. Phil. 1.20)

21. Cf. Daneš (1968:€56 f.) who observes that verbs of action are strong candidates for Focus function, due to their semantic force. 22. For ellipsis, see Section 5.1 below p. 96.



Chapter 2.╇ Pragmatic functions 

2.2

Focus-first strategy

A speaker or writer may start a clause or sentence with the most salient element. I will call such a strategy “Focus first”.23 A good example is in (6): whereas oppido Gadibus is contextually bound and the action of selecting somebody as chief is not unexpected (this was the usual practice in similar situations), Gaius Gallonius is a contextually independent constituent to be developed later on, that functions as Focus and occupies the clause-initial position. (6)

2.3

[Pecuniam omnem... in oppidum Gadis contulit...] Gaiumque Gallonium, equitem Romanum, Gaius-acc=and Gallonius-acc equestrian Roman-acc familiarem Domiti... oppido Gadibus praefecit... relative-acc of Domitius town-dat Gades-dat he put in charge ‘[Varro collected all the money... into the town of Gades....] and he put in charge of the town of Gades Gaius Gallonius, a Roman equestrian, a relative of Domitius...’ (Caes. Civ. 2.18.2)

Presentative and what-happens sentences

Presentative sentences,24 which introduce new discourse participants, generally convey completely new information. However, they can also contain elements anchored into the context, such as a temporal or spatial setting. There is another, non-presentative, type of sentence that also contains new information; I will call these “what-happens” sentences. Although in both cases the whole message that is communicated is new, such sentences are usually focused on one particular constituent that is important for the further development of the discourse. A second underlying question can thus be determined in view of the saliency of this constituent: ‘who?’, ‘what?’, and so forth. I will start with presentative sentences that are anchored in the context; Examples (7) and in (8) concern an animate and an inanimate entity respectively. (7)

Erat comes eius Rubrius quidam, homo factus ad was companion of his Rubrius certain-nom man-nom made to istius libidines, qui miro artificio quocumque his vices who with wonderful ingenuity wherever venerat haec investigare omnia solebat. Is he went this-acc investigate all-acc he used to he-nom

23. On priority for pragmatically marked constituents, see H. Dik (2007:€39), and Givón (1983:€20). 24. See Hannay (1991:€146), and for Latin Bolkestein (1998a: 198) and Rosén (1998).



Constituent order in Classical Latin prose

ad eum rem ita defert:... to him news-acc thus brings ‘There was a companion of his by the name of Rubrius, a man made for such vices as his, who used to find out all these things for him wherever he went, with wonderful ingenuity. This fellow brings him the following news...’  (Cic. Ver. 1.64) (8)

Et forte in eo loco grandis ilex coaluerat inter and by chance in that place great oak tree-nom had grown between saxa... Cuius ramis... nisus Ligus... rocks whose by branches supporting Ligurian-nom ‘And by chance amongst the rocks in that place a great oak tree had taken root... Supporting himself by its branches... the Ligurian...’ (Sal. Jug. 93.4)

The sentence in (7) contains contextually bound elements: eius, a possessive genitive referring to the central character in Cicero’s speech, Verres, and comes ‘companion’ deducible from the context (having a companion was quite normal in the given circumstances). Rubrius is a new character entering on the scene, explicitly marked by quidam ‘a certain’ as contextually unbound. He plays a role in the subsequent sentence, which is signalled by an anaphoric continuation (is). In (8), the most salient constituent, grandis ilex ‘a great oak tree’, is also new and has an indefinite reading. The sentence opens with a contextually dependent element, the Topic in eo loco ‘in this place’. The verb coaluerat ‘had grown’, contextually unbound in the same way as erat in (7), expresses a process applicable to trees and can thus be considered as a means for introducing grandis ilex into the discourse. A presentative sentence need not exhibit any relation with the previous context. Such a sentence conveys entirely new information. As we will see in detail (Chapter 3, Section 7.3 p. 188 ff.), they often but not systematically start with the verb. In (9) there is a proper name in initial position that reflects a Focus-first strategy for presenting information. It is important to point out such occurrences, for they should not be confused with other types of sentences containing the verb sum; in these est is best translated in English as “there is”. (9)

est quidam, cui cum praefecturam Gavius25 Gavius-nom is certain to whom when governorship-acc detulissem Bruti rogatu... Is me nec I had offered of Brutus at request he-nom me-acc neg proficiscentem Apamea prosecutus est nec... setting out-acc from Apamea followed neg ‘There is a certain Gavius, to whom I had offered a governorship at Brutus’s request... This fellow neither came with me when I was setting out from Apamea, nor...’ (Cic. Att. 6.3.6)

25. A variant Gaius is attested in some manuscripts.



Chapter 2.╇ Pragmatic functions 

I turn now to sentences answering primarily the question ‘what happened?’ They are not anchored in the preceding context, and appear in two main situations. Firstly, when the author reports events, for example in Cicero’s correspondence (10): (10)

[Redeo enim ad miseram seu nullam potius rem publicam.] M. Antonius ad me scripsit de restitutione Sex. Cloeli; M. Antonius-nom to me wrote about recall of Sex. Cloelius [quam honorifice, quod ad me attinet, ex ipsius litteris cognosces.] ‘[For now I return to the unhappy or rather non-existent republic.] Mark Anthony has written to me about the recall of Sextus Cloelius; [in how complimentary a style so far as concerns me personally you will see from his own letter.]’  (Cic. Att. 14.13.6) Just like presentative ones, what-happens sentences allow a secondary question; in (10), the sentence focuses on the action: ‘what did Mark Anthony do?’ There is, however, one difficulty: how to consider the subject M. Antonius, and how to explain its initial position. Whereas de Jong (1989:€536) considers that initial placement of subjects in this type of sentence is unmarked, I would interpret M. Antonius as a Topic that is introduced without further commentary because Antonius belongs to knowledge shared between Cicero and Atticus.26 It stands in initial position because it was selected as the Topic.27 Then, an action is ascribed (scripsit) to him. Secondly, we find what-happens sentences in Cicero’s philosophical works used in support of his argumentation, in order to introduce examples of a discussed phenomenon. In (11), Cicero aims to show that the passage of time has an alleviating effect on distress; it becomes less poignant and disappears in number of cases: (11)

Karthaginienses multi Romae servierunt, Macedones Carthaginians-nom many at Rome were slaves Macedonians-nom rege Perse capto; vidi etiam in Peloponneso, cum king Perses being captured I saw also in Peloponnese when essem adulescens, quosdam Corinthios. Hi poterant I was young some Corinthians-acc these could omnes eadem illa de Andromacha deplorare... all-nom same these-acc in Andromache lament-inf ‘Many Carthaginians were slaves at Rome, many Macedonians after the capture of King Perses. I have seen too some Corinthians in the Peloponnese when I was young. All of them could have made the same lament as that in the Andromache...’ (Cic. Tusc. 3.53)

26. For an initial subject in a what-happens sentence with a different function, see Appendix, p. 316. 27. There is disagreement about the interpretation as Topics of such not previously mentioned initially placed subjects. Several scholars consider them as Topics (cf. H. Dik 1995:€26), some do not (de Jong 1989:€536), including Firbas (1992:€64); see his remarks about President Kennedy has been assassinated.



Constituent order in Classical Latin prose

Here the multiple subjects Karthaginienses and Macedones are also contextually independent, but they do not have the same function as M. Antonius in (10). They are not treated as Topics but are salient. They are summarized by hi omnes afterwards (note that quosdam Corinthios has the same pragmatic function, even if it is syntactically the object). In other words, this sentence secondarily focuses on the subject (question ‘who?’). The subject is placed initially according to the Focus first strategy. What-happens sentences can also present another ordering of constituents. Cicero several times uses the pattern exemplified in (12) and (13), with contextually independent entities, starting with the verb. Such sentences serve to introduce support for the author’s argumentation. It is as if he were saying: ‘there is/was situation X’. Example (12) is the starting point for a demonstration that we have to avoid desire; Example (13) opens a series of errors that various nations commit. (12)

Laudat Africanum Panaetius quod fuerit abstinens. praises Africanus-acc Panaetius-nom that he was self-denying Quidni laudet? why not he should praise ‘Panaetius praises Africanus for being self-denying. Why should he not praise him?’ (Cic. Off. 2.76)

(13) Condiunt Aegyptii mortuos et eos servant domi. embalm Egyptians-nom dead-acc and them keep at home Persae... Persians-nom ‘The Egyptians embalm their dead and keep them in the house. The Persians...’  (Cic. Tusc. 1.108)

2.4

Complex information

There are sentences that do not exhibit one salient constituent; new information is conveyed by a pragmatic unit. In such cases, I prefer to talk about “complex information” rather than “complex Focus” (the latter term is used for example by H. Dik 1995:€71 f.).28 This phenomenon can be illustrated by a passage taken from Caesar: (14)

(vulnerati) Qui omnes discessu Curionis... ex castris in wounded these all-nom after departure of C. from camp in oppidum propter timorem sese recipiunt. Qua re animadversa... town because fear refl retreat this thing being noticed ‘On Curio’s departure all these wounded men... retreat from the camp into the town because of fear. Seeing this...’ (Caes. Civ. 2.35.5)

28. In my view, Focus is a salient element; even if there are several elements conveying new information, the sentence can focus on only one of them.



Chapter 2.╇ Pragmatic functions 

Curio’s soldiers leave the camp and retreat into the town. This sentence answers not ‘where do the soldiers retreat?’ but ‘what do the soldiers do?’ This is summarized by qua re animadversa ‘noticing this’ indicating that the action itself (and not the destination) is important. The constituent in oppidum forms a pragmatic unit with sese recipiunt and the information conveyed is complex.

2.5

Contrast

Contrast is not a pragmatic function but a pragmatic feature.29 It is a type of confrontation between two elements that are brought into contact with each other. These elements either share some property or differ in some property (de Jong 1983:€132). Contrast represents a priority factor for constituent ordering (de Jong 1989:€525). 2.5.1 Contrastive Topic and Focus Contrast can concern both Topic and Focus constituents. Example (15) illustrates a contrastive Topic: Caesar is contrasted with Varro, Topic of the preceding sentence. Contrastive Topic typically goes together with Topic shift (H. Dik 1995:€27; cf. Section 3.3 below p. 64). Example (16) illustrates a contrastive Focus: two chiefs are designated, one for the command of the navy (D. Brutum) and one for the siege of Massilia (C. Trebonium). (15)

Tradita legione Varro Cordubam ad Caesarem handed over legion-abl Varro-nom to Corduba to Caesar venit... Caesar contione habita Cordubae omnibus comes Caesar-nom assembly being held at Corduba all-dat generatim gratias agit. by classes thanks-acc gives ‘When the legion was handed over, Varro comes to Caesar at Corduba... Caesar called a public meeting at Corduba and expressed his thanks to all classes separately.’ (Caes. Civ. 2.21.1)

(16)

[naves longae... Quibus effectis armatisque diebus XXX... adductisque Massiliam] his D. Brutum praeficit, C. Trebonium legatum these-dat D. Brutus-acc puts in command G. Trebonius legate-acc ad oppugnationem Massiliae relinquit. for siege of Massilia leaves ‘[warships... These were finished and equipped in thirty days...; when they had been brought to Massilia,] he put Decimus Brutus in command of them, and left his deputy, Gaius Trebonius, to conduct the siege of Massilia.’  (Caes. Civ. 1.36.5)

29. By pragmatic feature, I mean prominence given to a constituent bearing a pragmatic function (Topic or Focus).



Constituent order in Classical Latin prose

Both can be put together in one sentence, as in (17). Contrastive Topics Caesar and Pompeius stand in clause-initial position; contrastive Focus concerns the nature of their actions: clam et noctu and palam atque interdiu. (17) [Haec eodem fere tempore Caesar atque Pompeius cognoscunt... eodemque die uterque eorum ex castris stativis a flumine Apso exercitum educunt:] Pompeius clam et noctu, Caesar palam atque interdiu. Pompey-nom secretly and by night Caesar-nom openly and by day [Caesar and Pompey discover this at almost the same time... and the two of them lead their armies from the permanent camps on the Apsus on the same day,] Pompey secretly by night, Caesar openly by day.’ (Caes. Civ. 3.30.1–3) There can be contrast not only between explicitly expressed elements, as in the examples quoted above, but also with implicit elements, as in Example (18). The constituent tuas litteras, placed in sentence-initial position, implies a contrast with meae litterae, deducible from the verb scribo ‘I write’ that refers to the author of the letter. (18)

[Nunc quae scribo, scribo ex opinione hominum atque fama.] Tuas litteras avide exspecto. your letter-acc greedily I await ‘[As it is, what I write is only based on general opinion and hearsay.] I am greedily awaiting a letter from you.’ (Cic. Fam. 12.4.2)

As we can see from this example, in interactive texts, such as dialogues or correspondence, contrast is often established between the participants of the actual, reproduced or imagined speech: the speaker and the addressee. This is why personal and possessive pronouns are often contrastive.30 (19) Tu Atticae salutem dices. Nos Piliam diligenter tuebimur. you Attica-dat salutation-acc will give we Pilia-acc diligently will take care ‘You will give my regards to Attica. I will take good care of Pilia.’  (Cic. Att. 14.19.6) 2.5.2 Verum Focus Verum Focus (Bolkestein 1996b: 17; cf. Kühner and Stegmann 1914, II: 600) is a type of contrastive Focus concerning actions. It answers the implicit question: ‘is it really the case?’ and distinguishes itself by the initial position of the verb.31 Two examples are given in (20) and (21) with feci ‘I really did’ and fuisti ‘you really were’. Verum Focus is

30. On personal pronouns, see below, Section 4.6., p. 92. 31. Verum Focus covers the cases that Marouzeau (1938:€49 ff.) characterises as mise en relief de la notion et de la fonction verbales ‘prominence given to the idea expressed by the verb and by the function of the verb’.



Chapter 2.╇ Pragmatic functions 

frequently accompanied by affirmative or conclusive particles, such as mehercule in (20) and igitur in (21). (20) [Evolve diligenter eius eum librum, qui est de animo; amplius quod desideres nihil erit.] – Feci mehercule, et quidem saepius. I did by Hercules and so repeatedly ‘[Read carefully that book of his, which is about the soul; there will be nothing more you could wish for.] – I have done so, I assure you, and done so many times.’ (Cic. Tusc. 1.24) (21)

[Dico te priore nocte venisse... in M. Laecae domum. Num negare audes?...] Fuisti igitur apud Laecam illa nocte, Catilina, distribuisti you were then at Laeca’s that night Catiline-voc you divided partis Italiae... parts-acc of Italy ‘[I say that you came the night before... to the house of Marcus Lecca... Do you dare to deny it?...] You were, then, Catiline, at Laeca’s that night; you divided Italy into sections.’ (Cic. Cat. 1.9)

It is worth pointing out that the initial position does not indicate per se that the verb is focal. An initial verb may also be pragmatically unmarked or contextually bound.

2.6

Emphasis

The word “emphasis” is often used in a vague way with several possible senses.32 What is frequently labelled as “stressed”, “emphatic”, “highlighted” or in French “mis en relief” in studies of Latin constituent order, could be more properly analysed either as contrastive (when there is a confrontation of two elements, see Section 2.5) or as emphatic. I will use this term in the sense defined by de Jong (1989:€528): “emphasis is a means that serves the speaker or writer to express his personal evaluation of information”. As well as contrast, emphasis is a pragmatic feature, not a pragmatic function.33 Words that are eligible for emphasis are in particular evaluative expressions, for example intensive adjectives and adverbs, tantus, talis ‘such’, multus ‘numerous’, magnus ‘big’ and the like. Also negative words, for example, nullus ‘not any, none’ and nemo ‘nobody’ may be used with emphasis. Consequently, they are often placed in the prominent initial or final positions, as in (22) to (24).

32. Such as “importance”, “phonetic stress”, “prominence”, “intensity”. See Dover’s (1960:€32 f.) remarks on misuse of this term. 33. This means that emphasis as pragmatic feature (or prominence) can be given to a constituent bearing a pragmatic function (Topic or Focus) but also to a constituent that is not assigned a pragmatic function. An emphatic constituent is not necessarily Topic or Focus constituent.



Constituent order in Classical Latin prose

(22)

Multi iam menses erant34 et hiems praecipitaverat many now months-nom were and winter-nom was advanced [neque Brundisio naves legionesque ad Caesarem veniebant]. ‘Many months had now passed and winter was far advanced, [yet his ships and legions from Brundisium had not reached Caesar].’ (Caes. Civ. 3.25.1)

(23) Tales igitur amicitiae sunt remissione usus such so friendships-nom are by dismissing of familiarity eluendae. to be phased out ‘So that kind of friendship is best phased out by a dismissing of familiarity.’  (Cic. Amic. 76) (24) Sed hunc laborem recusabat nemo, quod... but this toil-acc refused nobody as ‘But no one shirked this toil, as...’

(Caes. Civ. 1.68.3)

Emphasis may cause discontinuity of noun phrases. Evaluative expressions do not necessarily appear in a special sentence-position, nor are they necessarily separated from their head noun. On the other hand, these evaluative expressions often occupy an extreme position in a noun phrase: the beginning or the end of them, such as talem in (25). (25)

[Vos autem, Adherbal et Hiempsal, colite,] observate talem hunc virum, imitamini virtutem... respect-imp such this man-acc imitate-imp prowess-acc ‘[As for you, Adherbal and Hiempsal, pay attention and] respect to such a man as this, imitate his prowess...’ (Sal. Jug. 10.8)

The above examples have shown emphatic elements that belong to the most informative part of their sentences. This need not always be the case. In (26), an emphatic word, praeclare ‘excellently’, opens a what-happens sentence, but the Focus function is carried by Alexandrum filium. This sentence is used in a passage concerning the duties of young people. Praeclare in epistula quadam Alexandrum filium Philippus excellently in letter certain Alexander-acc son-acc Philip-nom accusat quod largitione benivolentiam Macedonum consectetur. reprimands that by bribery good will-acc of Macedonians he seeks after ‘Excellently in one of his letters Philip reprimands his son Alexander for trying to gain the good will of the Macedonians through bribery.’  (Cic. Off. 2.53)

(26)

34. With a textual variant transierant.



Chapter 2.╇ Pragmatic functions 

2.7

Particles

2.7.1 The function of focusing particles I will use the term “focusing particles”35 for words such as also, even, only, as is common in modern linguistic studies (for example, König 1991). However, this is not to say that focusing particles mark Focus constituents, for Focus (in the sense of Functional Grammar) is not formally marked in Latin, except in the case of interrogative sentences.36 The role of focusing particles is to signal a constituent that contrasts with another, expressed or understood constituent, when this is contrary to expectations or presuppositions.37 Constituents modified by such particles may or may not represent the most informative part of the sentence. This explains why it is possible to find more than one focusing particle in the same sentence, as is shown in (27) with both hi quoque ‘these too’ and etiam nunc ‘even now’. Hi modified by the additive quoque functions as the Topic, and etiam nunc volitant is complex information. The use of focusing particles (quoque and etiam) is due to the fact that, contrary to expectations, these men, who were also involved in the sedition of Clodius, should have been punished already. (27)

[omitto iam Numerium, Serranum...] Sed tamen hi quoque etiam nunc volitant. but still these-nom too even now flutter about ‘[I say nothing here of Numerius, Serranus...] But still, even now these too flutter about.’ (Cic. Sest. 94)

Focusing particles can differ as to their scope: they can modify one constituent or may have scope over a whole clause or sentence. The particle modifying a single constituent is normally adjacent to that constituent, which theoretically may occupy any position in the sentence.38 Particles modifying the whole sentence are expected to be in an early position in the sentence. Focusing particles are either paired or single. When they function in pairs, they signal a relationship between two explicitly expressed constituents. Paired focusing particles are mainly used for substitution and extension of information (Dik 1997, I: 333). In the case of substitution, the speaker or author assumes that the addressee has incorrect information and replaces it by correct information. Substitution is signalled by expressions such as non… sed… or non… verum… ‘not... but’. Incorrect 35. Cf. Panhuis’s (1982:€89 ff.) section on “rhematising factors”. 36. On the enclitic particle -ne that accompanies Focus or a word of the focal part of interrogative sentences, see Chapter 4, Section 3.1.1. p. 199. 37. The use of focusing particles goes together with the prosodic prominence of a constituent, as is common in modern spoken languages. See König (1991:€11 f.) for discussion. For focusing particles in Latin, see also Rosén (2009:€323–326). 38. Cf. König’s remarks on additive and restrictive particles (1991:€164 f.) in modern languages. Cf. also Kroon (1995:€282).



Constituent order in Classical Latin prose

information such as lex€in (28) is replaced by correct libido. These are Focus constituents; the sentence answers the question ‘what?’ Substitution concerns two verbs in (29); dignitatem is contextually given. (28) Vtriusque cladis non lex, verum libido eorum finem fecit. of each disaster neg law-nom but whim-nom their end-acc put ‘To each disaster an end was put, not by law but by their whim.’  (Sal. Jug. 31.7) (29) Quorum quidem tu non contempsisti, sed pertimuisti dignitatem. of which certainly you neg despised but dreaded worthiness-acc ‘Actually you did not make light of their worthiness, rather you were terrified of it.’ (Cic. Dom. 132) Extension consists of completing some information by supplying a missing element. Extension of information by an additional element is expressed by the pair non solum… sed (etiam)… ‘not only... but (also)...’; for example in (30) addition concerns tota urbe ‘the whole city’. (30)

Nam nunc quidem, pontifices, non solum domo de for now indeed priests-voc not only house-abl about qua cognostis, sed tota urbe careo, which you inquire but whole city-abl I am deprived [in quam videor esse restitutus]. ‘Because now, your reverences, I am not only deprived of my house, which is what you were looking into, but of the whole city, [to which people see me as being restored].’ (Cic. Dom. 146)

Single focusing particles such as nisi ‘except’,39 solum, tantum ‘only’ are used to indicate a restriction: the speaker or author corrects information that the addressee is supposed to have by restricting the content to only pertinent element or elements. Two examples are given in (31) and (32); nisi stands before the modified constituent (caedem, incendia, rapinas), tantum prefers postposition. (31) Nihil cogitant nisi caedem, nisi incendia, nisi rapinas. nothing they think but carnage-acc but arson-acc but pillage-acc ‘They think of nothing but carnage, arson, and pillage.’ (Cic. Cat. 2.10) (32)

Inter bina castra Pompei atque Caesaris unum flumen between two camps of Pompey and of Caesar one river-nom tantum intererat Apsus. only laid between Apsus-nom ‘Between the two camps of Pompey and Caesar there was just the one river Apsus.’  (Caes. Civ. 3.19.1)

39. For the use of nisi, see Orlandini (2001:€151 ff.).



Chapter 2.╇ Pragmatic functions

Latin single focusing particles are anteposed (et, etiam ‘also’) or postposed (quoque ‘also’, quidem). In the following sections, I will briefly discuss the additive particles such as et, etiam, quoque, and the particle quidem. They can be used with respect to an already expressed element, but also may refer to an implicit content.

2.7.1.1 Et, etiam and quoque ‘also’ The particles et, etiam and quoque40 signal the addition of an element€to a set (‘also’). Whereas et and etiam are anteposed, quoque always stands after the word it modifies. In the case of noun phrases, quoque comes after the first element, as illustrated by tuum nomen in (33). Addition not only operates inside a sentence, but can also go beyond it and refer to the immediately preceding context. An example is given in (34) with an etiam marking the inclusion of Stesichoros’s statue (Stesichori poetae statua) in the set of all statues found in the city of Himera. When an entire sentence is concerned, etiam can stand sentence-initially (cf. Example (29), p. 21) or, as in (34), after the first word with discontinuity of a noun phrase. (33) ...ad eum numerum clarissimorum hominum tuum to this number of distinguished men your-acc quoque nomen adscripsit. also name-acc he added ‘...he added your name also to that number of distinguished men.’  (Cic. Ver. 2.110) (34)

[Erant signa ex aere complura...] Erat etiam Stesichori poetae statua... was also of Stesichorus poet statue-nom ‘[There were many statues of bronze...] There was also a statue of the poet Stesichorus...’ (Cic. Ver. 2.87)

In principle, constituents focused by a particle, as in (33) with quoque, can stand anywhere in the sentence. When the modified constituent occurs last, postposed quoque occupies the sentence-final position (35). (35)

[Quod si perseveras me ad matris tui cenam revocare], feram id quoque. I will do it too ‘But if you persist in inviting me back to one of your mother’s dinners, I will do that too.’ (Cic. Fam. 9.16.8)

40. On et and etiam, see Kroon (1995:€63); on etiam and quoque, see Orlandini (2001:€211 ff.).





Constituent order in Classical Latin prose

2.7.1.2 Quidem The particle quidem has very diverse uses,41 and I will present only those that are interesting for the present topic. Just like quoque, quidem is a postpositive particle that never appears in initial position.42 When quidem modifies a constituent it signals a contrast with another element, for example tibi as against nobis in (36). Servitutem in (37) contrasts with the implicit idea that the servitude would have been a necessary consequence for the Roman people if Jugurtha had won the war. (36) [Nobis ob merita sua carus est…] Tibi quidem pro nostra amicitia gratulor. to you part for our friendship I congratulate ‘[To us he is dear for his services...] I congratulate you on behalf of our friendship.’  (Sal. Jug. 9.2) (37)

[Nisi forte nondum etiam vos dominationis eorum satietas tenet...] Nam servitutem quidem quis vostrum recusare audebat? For slavery-acc part who of you to reject dared ‘[Unless perhaps you have not even yet had enough of their domination...] For which of you dared to reject slavery?’ (Sal. Jug. 31.20)

Quidem occurs in combination with the coordinator et (also ac and atque) to form a discontinuous particle et... quidem ‘and that’ which gives prominence the constituent it modifies.43 The constituent modified by et... quidem belongs syntactically to the sentence but is presented as additional.44 Et... quidem incorporates single constituents as in (38), but never a verb.45 (38) Itaque et ego cum illo locutus sum et saepius so also I with him spoke and so often

41. See Orlandini (2005 and 2001:€211 ff.) and, in particular for the status quaestionis, Kroon (2005); the use of quidem described as “emphatic” (Solodow 1978:€94; cf. Kühner and Stegmann 1914, I: 802) was not included in Kroon’s study. 42. Quidem is, like quoque, a postpositive focusing particle (cf. Berger 1942:€342), and not an enclitic, pace Marouzeau (1949:€98 ff.). 43. For the non-discontinuous combination of et quidem ‘and that’, used for corrective or additive coordination, see Section 7, p. 113 on Tail constituents. 44. See Kroon (2005:€583), and Rosén (1990:€35) on detachment by coordinators. In numerous cases, we have to deal with “argumentative gradation” (Orlandini 2005:€164 f.). 45. In Cicero (BTL), there are 133 instances of et… quidem that integrates (unlike ne… quidem, see below) the first word of a noun phrase. In 50% of cases, this element is anaphoric. Cf. Orlandini (2005:€165).



Chapter 2.╇ Pragmatic functions 

quidem, ut etiam accusarer... part that even I was reproached ‘So I spoke to him about it, and that indeed so often that I was even reproached...’ (Cic. Att. 16.16a.3) The discontinuous particle ne… quidem ‘not even’ is a focusing particle that occurs in a negative context, signalling the exclusion of an element from a preset group. It can incorporate various constituents, including verbs, as in (39).46 Only the first element of the noun phrase usually figures in between. From this point of view, it is not surprising to find several instances with the anaphor is, as in (40), that itself does not bear the Focus.47 (39)

Non possunt omnes esse patricii; si verum quaeris, neg they can all-nom be patricians-nom if truth you seek ne curant quidem. not they care even ‘Not everyone can be patricians and, to tell the truth, they do not even care about it.’ (Cic. Sul. 23)

(40) Ne eo quidem tempore quisquam loco cessit... not at that even time anybody from place yielded ‘Not even then did any man yield his ground...’ (Caes. Gal. 7.62.7) 2.7.2 The function of connective particles Connective particles do not mark Focus constituents; they need not even accompany them. Their function is to indicate a semantic relationship, for example causal or adversative, between a sentence and the preceding context. As opposed to focusing particles, which mark an important element (Section 2.7.1), connective particles contribute to the cohesion of the text. This property explains their placement in the sentence: connective particles are expected to appear in an early position. Example (41) shows an initial igitur that is not in contact with the Focus of the sentence, decemviri ‘decemvirs’. The information conveyed in (42) is complex; the conclusive particle igitur ‘accordingly’ separates the modifier frugi from its head noun hominem; frugi bears emphasis and is placed initially (note its repetition and the presence of the parenthetical inquam).€

46. Ne… quidem can incorporate complex constituents, for example possum and its infinitive: Ego autem ne irasci possum quidem iis quos valde amo. (Cic. Att. 2.19.1) ‘For myself, I am positively incapable of being angry with those I love deeply.’ For an incorporated subordinate clause, see Cic. Q. fr. 3.1.21. Cf. also the examples quoted by Devine and Stephens (2006:€267). 47. In my view, there is no special emphasis on the pronoun (pace Solodow 1978:€98); the fact that the pronoun figures inside the particle is a result of discontinuity of the noun phrase (for example, ne M. quidem Seio, Cic. Off. 2.58).



Constituent order in Classical Latin prose

(41)

[pecuniam referre in aerarium lex vetat...] Igitur pecuniam omnem decemviri tenebunt, vobis so money-acc all-acc decemvirs-nom will keep for you ager non emetur. land-nom neg will be bought ‘[the law forbids bringing the money into the treasury...] So the decemvirs will keep all the money, land will not be bought for you.’ (Cic. Agr. 2.72)

(42)

Frugi igitur hominem, iudices, frugi, inquam, et in honest then man-acc judges-voc honest I say and in omnibus vitae partibus moderatum ac temperantem... videtis... all of life parts moderate-acc and self-controlled-acc you see ‘You see then an honest man, an honest, I say, gentlemen, moderate and selfcontrolled in every detail of life...’ (Cic. Font. 40)

The following examples show the use of the particles tamen€‘nevertheless’ and at ‘but’: (43)

[... in Africam mittendum... enisum, ne tale decretum fieret...] Legantur tamen in Africam maiores natu nobiles... are sent nevertheless to Africa senior-nom in age nobles-nom ‘[...an army should be send to Africa... to prevent such a decree from being passed...] Nevertheless, legates were chosen for Africa who were nobles more senior in age.’ (Sal. Jug. 25.3)

(44) (Epicurus) Ille a virtute summum bonum separavit. he from virtue the highest good-acc separated – At laudat saepe virtutem. but he praises often virtue-acc ‘Epicurus separated the highest good from virtue. – Yes, but he often praises virtue.’ (Cic. Tusc. 3.48) Example (43) is taken from a passage where Sallust is writing about a debate in the Senate€concerning dispatching troops to Africa; some senators are strongly opposed and wish to prevent it. Effective action is thus ruled out. However, a deputation is finally sent. The verb legantur is deducible from the preceding context (mittendum); the concessive particle tamen, ‘however’, signals that the action was contrary to expectations.€ Salient information is conveyed by the contextually independent constituent maiores natu nobiles ‘senior nobles’ (answering the question ‘who?’). According to this analysis, the initial, contextually given verb legantur does not bear either Focus or emphasis.€The informational status of the verb is different in (44). If Epicurus separates the highest good from virtue, one might expect that he neglects virtue. Such a conclusion would be incorrect – this is what at ‘but’ signals, for Epicurus praises (laudat) it; this verb carries the Focus.



Chapter 2.╇ Pragmatic functions 

2.8

Cleft constructions

Cleft constructions are a means to mark Focus constituents in many languages.48 The prototypical structure of these constructions is: an identificatory or presentative neuter pronoun (ce in French, it in English) followed by the verb ‘to be’; then comes the Focus: lui in (45a) and his wife in (45b), and finally a relative clause. (45) a. C’est lui qui l’a fait. it is he who it aux did b. It is his wife who decides. Even if only sporadically, such cleft constructions are found in Latin of all periods. Löfstedt (1966) distinguishes two types. The first type, cleft constructions without identificatory pronoun, is exemplified in (46), (47), and (48). When examining Löfstedt’s examples, we observe that the focused constituent is always the predicative noun, in the nominative, or in the accusative (Clodium) in the case of AcI clauses; it is never in another case. It stands in sentence- or clause-initial position and a relative clause follows.49 Cleft constructions in Latin thus make it possible to focus on the noun that in a non-cleft construction would function as subject; for Example (48) Charilaus ad Publilium Philonem venit might be used equally well but Charilaus should be understood as Focus first. (46) Temperantia est enim, quae... ut rationem sequamur monet. moderation-nom is indeed which that reason-acc we follow warns ‘For it is moderation that warns us to be guided by reason.’ (Cic. Fin. 1.47) (47) ut nescires Clodium esse qui contra leges faceret that you did not know Clodius-acc be-inf who against laws would do ‘that you did not know that it was Clodius who was to act illegally’  (Cic. Dom. 48) (48) [Charilaus et Nymphius... inter se consilio partes ad rem agendam divisere, ut alter ad imperatorem Romanorum transfugeret, alter subsisteret...] Charilaus fuit, qui ad Publilium Philonem venit. Charilaus-nom was who to Publilius Philo came ‘[Charilaus and Nymphius... schemed together and arranged the part that each should play in order to bring this about. One was to go over to the Roman general, the other to remain behind...] It was Charilaus who went to Publilius Philo.’  (Liv. 8.25.10) The second type of cleft constructions contains a personal or demonstrative pronoun that is the subject of the verb sum, as in (49). A demonstrative pronoun of identification 48. See Lambrecht (1988), Hannay (1985a: 76) and Miller (2006:€171). 49. However, Löfstedt (1966:€259) quotes an example of postposition found in poetry: Quae iurat, mens est. (Ov. Her. 21.135) ‘It is the mind that swears.’



Constituent order in Classical Latin prose

(hic; haec) is used for the third person in (50) and, in conformity with Latin usage, agrees in gender and number with the predicative noun. (49) Tu es qui pretia diligentissime exquisisti. you are who values-acc very diligently researched ‘You are the man who researched with the greatest care the value of these things.’  (Cic. Ver. 3.71) (50) Hic est danista, haec illa est autem quam emi de praeda. this is money-lender-nom this she is however whom I bought from spoil ‘This is the money-lender, and this is the girl I bought at the sale of the spoils.’  (Pl. Epid. 621) To these two types, it is also worth adding the expressions quis est qui ‘who’ and quid est quod ‘what’ used quite frequently in interrogative sentences: (51) Nunc enim quis est qui non probet, qui non laudet? now indeed who is who neg approves who neg praises ‘For now, as it is, who is there who does not approve of what has been done? who does not praise it?’ (Cic. Mil. 77)

3. Topic The pragmatic function of Topic is defined as what the sentence is about (see Introduction, p. 6). In this section, I will look at the way it functions in my corpus 1, in particular in historical narrative. Historical narratives, being complex and coherent stories, provide excellent illustrations of the functioning of Topics. Similar strategies also appear in shorter stories, some narrated for example by Cicero in his speeches or correspondence. Future Topic and the introduction of new discourse participants will be examined in Section 3.1; the workings of referential chains will be discussed in Section 3.2. Sections 3.3, 3.4 and 3.5 are devoted to various kinds of Topics: Discourse Topic, Sub-Topic and Sentence Topic. The positioning of two prototypical Discourse Topics, Iugurtha and Caesar, will be examined in detail in Section 3.6. As pronouns frequently – but not exclusively – function as Topics, I devote special sections to pronominal anaphora and topicalisation (3.7), and to personal pronouns (3.8).

3.1

Future Topic

3.1.1 Introduction of new discourse participants Every story is about somebody or something, and the more complex the story is, the more participants have a role to play. Sallust’s (The Jugurthine War) and Caesar’s (The Civil War) historical narratives include a number of episodes, changes of scenes, and entities that the authors have to introduce or to reintroduce after a certain time.



Chapter 2.╇ Pragmatic functions 

The various characters each have their own status, and they are accessible to readers in various ways. Protagonists like Caesar or Pompey who would have been known to all of Caesar’s contemporaries do not need special introduction (cf. Caes. Civ. 1.1.1). The first mention of other characters is usually accompanied by some specification. By using apposition or a relative clause, the author gives indications about the person: his function, origin, deeds. Certain characters who are to play a significant role later are introduced by a presentative sentence. Presentative sentences, which serve to introduce a new discourse participant, contain the verb sum ‘to be, to exist’ or a similar one (ago ‘to deal, to pass one’s time’), or a verb meaning ‘to come’, ‘to appear’.50 The entity introduced, animate or inanimate, functions as a Future Topic, i. e. it becomes the Topic in the subsequent context. Two typical examples are given in (1) and (2), where the Future Topics are continued with huic and quem, respectively.51 (1)

Erat ea tempestate Romae Numida quidam nomine was at that time in Rome Numidian certain-nom by name Massiva, Gulussae filius, qui... profugus ex patria Massiva of Gulussa son who as fugitive from country abierat. Huic Sp. Albinus... persuadet... had gone away him-dat Sp. Albinus-nom persuades ‘There was at this period in Rome a certain Numidian named Massiva, a son of Gulussa, who... had been exiled from his country. This was the man whom Spurius Albinus... persuaded...’ (Sal. Jug. 35.1–2)

Ibi cum Boccho Numida quidam Aspar nomine there with Bocchus Numidian certain-nom Aspar by name multum et familiariter agebat, praemissus much and friendly was conducting sent in advance ab Iugurtha... Quem Bocchus... mittit... by Jugurtha him-acc Bocchus-nom sends ‘There a certain Numidian named Aspar, sent by Jugurtha, was conducting regular and friendly relations with Bocchus. Bocchus... sent this man...’  (Sal. Jug. 108.1)

(2)

In these examples two completely unknown persons are introduced who are inaccessible to the reader; Sallust uses here an explicit expression of indefiniteness (Numida quidam ‘a certain Numidian’; see Rosén 1998). Appositions, relative or participial clauses specify the origin or the relationship with other people or other circumstances. In spite of their name, Future Topics convey salient information in their sentences that 50. See Hannay (1985a) and (1991:€144); Dik (1997, I: 315–318); Pinkster (1995:€237–238) and Rosén (1998). See also Toth (1994:€ 178) and Wehr (1984:€ 7). Presentative sentences are discussed in detail in Chapter 3, Section 7.3. p. 187 ff. 51. On the obligatory anaphoric continuation of Future Topics, see Section 3.2.



Constituent order in Classical Latin prose

answer the underlying question ‘who?’52, and thus they function as Focus. As to the position of Future Topics, in my corpus from Caesar and Sallust, they are found in a late position in the sentence, as in (1) above, in 42% of the instances, especially with the verb sum ‘to be’; in 58% of the instances, they come early in the sentence and stand initially or after the Topic, as in (2). This is to say that rightwards placement in the sentence is not systematic in Latin, as it is in other languages.53 The same strategies are used for the introduction of inanimate entities that mainly serve as the setting for the development of further events. These are for example towns, rivers, valleys and other entities, as in (3). Unlike animate Future Topics, inanimate entities appear more frequently in final or pre-verbal position (10 instances out of 12); an early position is possible but less frequent. (3)

Erat haud longe ab eo itinere... oppidum Numidarum was not far from the route town-nom of Numidians nomine Vaga, forum rerum venalium totius regni by name Vaga market of merchandise of whole kingdom maxime celebratum... Huc consul... praesidium imposuit. most frequented there consul-nom garrison-acc established ‘Not far from the route... there was a Numidian town by the name of Vaga, the most frequented market for merchandise in the whole kingdom... Here the consul... established a garrison...’ (Sal. Jug. 47.1–2)

The delimitation of presentative sentences poses some problems, especially because the verbs meaning ‘to be’ or ‘to appear’ do not constitute a homogeneous group (cf. Firbas 1992:€60). Furthermore, the phenomenon of Future Topic does not seem to be restricted to presentative sentences, for a similar strategy that uses the same means (a new entity with an anaphoric continuation), and establishes a Topic, can be observed with other verbs. Consider for example the introduction of Scaurus in (4), with the help of the verb censeo ‘to consider’: (4)

At contra pauci... Hiempsalis mortem severe vindicandam however a few-nom of Hiempsal death-acc severely to be punished censebant, sed ex omnibus maxime Aemilius Scaurus, argued but from all most Aem. Scaurus-nom homo nobilis impiger factiosus... Is postquam videt... man noble energetic factious he-nom when sees ‘A small number, however... argued that the death of Hiempsal ought to be severely punished – most of all Aemilius Scaurus, a noble, energetic and factious man... When he had seen...’ (Sal. Jug. 15.4–5)

52. See Hannay (1991:€138), Spevak (2007a); cf. Bolkestein (1998a: 195). Thus it is not the verb but the entity that is the Focus. For initial verbs, see Spevak (2004). Cf. also H. Dik (1995:€221 ff.) who considers the verb ‘to be’ in presentative sentences a Topic. 53. See note 52 for references; see also Spevak (2008) for Czech.



Chapter 2.╇ Pragmatic functions 

Furthermore, people can be introduced without a special introductory formula. A typical example is given in (5) with centuriones Marsi duo, mentioned here for the first time. The information conveyed is complex and, so to speak, condensed: we are informed about (the existence of) two centurions as well as about their action (perfugiunt ‘they desert’), as if the author has skipped the introductory formula: “there are two centurions... they desert...” (cf. H. Dik 1995:€26). Centuriones Marsi duo are picked up by hi in the subsequent sentence that tells us what happened to them. Proxima nocte centuriones Marsi duo ex castris Curionis... following night centurions Marsic two-nom from camp of Curio ad Attium Varum perfugiunt. Hi... confirmant... totius to Attius Varus desert these assert of whole exercitus animos alienos esse a Curione. army-gen minds distant-acc be-inf from Curio ‘The following night two Marsic centurions... desert from Curio’s camp to Attius Varus. They assert that Curio does not have support of any of his army...’  (Caes. Civ. 2.27.1)

(5)

Without any introductory formula, inanimate entities can be presented as destinations of a movement verb such as pervenio ‘to arrive’ or mitto ‘to send’, as for example in (6). Anaphoric continuation is effected by the anaphoric adverbs huc, ibi or eo ‘there’, anaphoric pronouns (the connecting relative quod referring back to oppidum in 6), nouns, or prepositional phrases such as ad eum locum ‘to that place’. (6) (Aulus) magnisque itineribus hieme aspera pervenit ad by forced=and marches in winter harsh he arrived at oppidum Suthul, ubi regis thesauri erant. Quod... town Suthul where of king treasures-nom were that [neque capi neque obsideri poterat…] ‘By forced marches through harshness of winter Aulus arrived at the town of Suthul, where the king’s treasures were. It... [could be neither taken nor besieged].’  (Sal. Jug. 37.3–4) However, a first mention of a person or another entity does not necessarily mean that this element will be important afterwards. In historical narrative, there are also secondary people who appear on the scene but disappear immediately after. This is illustrated in (7):

(7) [Qua in fuga Fabius Paelignus quidam ex infimis ordinibus de exercitu Curionis... magna voce Varum nomine appellans requirebat...] Vbi ille saepius appellatus aspexit ac... when he-nom often being called looked and

 Constituent order in Classical Latin prose

quaesivit, umerum apertum gladio Ø appetit... asked shoulder exposed-acc with sword strikes ‘[In this rout, Fabius, a Pelignian from the lowest ranks of Curio’s army... kept shouting out Varus’s name and asking for him...] Varus, hearing his name constantly called, looked at him and asked... Fabius struck at his exposed shoulder with a sword...’ (Caes. Civ. 2.35.1–2) There is a contrast between Varus (the established Topic) and a certain Fabius who nearly kills him and leaves the scene after this episode. The information is complex here just as in (5) above. However, Fabius is not an important character for the further development of events, which explains the reference to him by means of zero anaphora (appetit) in the following sentence instead of anaphoric resumption (by is, hic, etc.) In other words, secondary or episodic persons that do not play a role afterwards are not treated as Future Topics. 3.1.2 Reintroduction of discourse participants Characters that have already been introduced into the discourse and played a role can disappear from the scene for a longer or shorter period. When they come back, there is no need to specify once more the identity of a main protagonist (for example, Jugurtha or Marius in Sallust’s narrative), but characters who are less well established and less accessible to the reader have to be recalled, and sometimes reintroduced. In these cases, the same presentative sentences are used as for the first introduction of Future Topics; both animate and inanimate entities can be treated in this way.54 In order to make a clear connection with what has already been said, explanations and reminders are used, especially formulas like ‘as mentioned above’. An example is given in (8) with Varus; the relative clause refers to the passage in Civ. 1.35. Erat in exercitu Vari Sextus Quintilius Varus, quem was in army of Varus Sex. Quintilius Varus-nom who-acc fuisse Corfinii supra demonstratum est. Hic dimissus be-inf at Corfinium above was mentioned this dismissed a Caesare in Africam venerat... by Caesar to Africa had come ‘In the army of Varus was Sextus Quintilius Varus, who had been at Corfinium, as mentioned above. Dismissed by Caesar, he had come to Africa...’  (Caes. Civ. 2.28.1–2)

(8)

To sum up, Future Topics, i. e. newly introduced or reintroduced entities that will play a role afterwards, represent a special type of discourse participants. Anaphoric continuation is typical for them but, as we will see below, is not exclusive to them, because it is also applicable to other constituents functioning as the Focus. 54. See the example from Caesar (Civ. 2.34.1), quoted above in Section 1.4, p. 33 as Example (13).



Chapter 2.╇ Pragmatic functions 

3.2

Referential chains

In the preceding section, I have demonstrated that newly introduced entities are focal and require anaphoric continuation in the subsequent sentence. In other words, zero anaphora is not admitted in such cases. This statement accords with observations made by de Jong (1989:€532 f.) concerning subjects; however, as we will see, there is no syntactic restriction involved. Bolkestein and van de Grift (1994:€290) when examining various means of expressing anaphoric reference to subjects (noun, noun phrase, zero anaphora, anaphor is and the demonstratives hic and ille),55 argue that references to entities reoccurring in the subsequent context as subjects depend on what they call “the thematic chain”. This concept can be developed in the following way. For the first reference to a new entity, a full noun or noun phrase is used; is, hic or connecting relative qui56 appear for the second reference; zero anaphora, ille, or a noun, are usual for any other reference, depending on whether there is absence or presence of contrast with another person. These referential means vary in strength: a (proper or generic) noun or a noun phrase represents an anaphoric expression more strongly than a pronoun; zero anaphora is the weakest means (Givón 1983:€6). Figure 1: Pattern of a referential chain noun (phrase) R1 first reference R2 second reference is, hic, or qui R3… other reference zero anaphora, ille, or a noun This referential chain affects constituent ordering: the second reference is an instance of topicalisation, i. e. the selection of a constituent as Topic (see Section 4.2, p. 75 f. below), and therefore the anaphoric pronoun stands in sentence-initial position. The third reference, when expressed by ille or by a noun, is related to a change of Topic (see the following Section 3.3). A typical example is given in (9). The new entity need not be the subject of its sentence and need not become the subject afterwards, as is seen in (10). (9)

homo Venerat ad eum illo biduo Laetilius1 quidam, had come to him within two days Laetilius certain-nom man non alienus a litteris... Is2 epistulas complures not unversed from letters he letters-acc several-acc attulerat... Repente Ø coepit3 dicere se omnia brought suddenly he started say refl everything-acc

55. Cf. also Orlandini (1995:€87 f.). 56. The connecting relative was not included in the quoted article (Bolkestein 1994); however, it can be added on the basis of another article by Bolkestein (2000:€128 f.), as well as on my own observations.



Constituent order in Classical Latin prose

Verris causa velle. of Verres on account want ‘Within two days of that time he had had a visit from one Laetilius, a person not unversed in letters... Laetilius had brought several letters... He began forthwith to say that he was ready to do anything for Verres.’ (Cic. Ver. 2.64) (10)

est Hieronis filius Syracusanus, homo... Heraclius1 Heraclius-nom is of Hiero son Syracusan-nom man-nom ad sestertium facile nobilis... Huic2 hereditas noble-nom this-dat inheritance-nom up to sesterces at least triciens venit testamento propinqui sui Heraclii. three millions came by testament of relative his Heraclius-gen ‘There is a Syracusan Heraclius, son of Hiero, a man of the highest rank... By the will of a close relative also named Heraclius, he received a legacy amounting to at least three millions of sesterces.’ (Cic. Ver. 2.35)

Laetilius in (9) and Heraclius in (10) are introduced into the discourse: these are the first mentions of them. The second reference (R2) to Laetilius is made by is, to Heraclius by the dative huic. These resumptive pronouns occupy the sentence-initial position. Zero anaphora is possible starting from the third reference (R3 coepit). Other material may intervene between R1 and R2 or/and R2 and R3, for example indirect speech or explanatory remarks. These units do not interfere with the identification of the referent (see p. 78, Example 5). Whereas pronominal anaphora is quite common (such as is and huic in 9 and 10), a stronger form of anaphora, especially resumption with a proper name, is used for avoiding ambiguity. Such a situation is illustrated in (11) with Bomilcar coming after the mention of Massiva. (11) (Iugurtha) Bomilcari... imperat pretio... insidiatores Bomilcar-dat commands by payment ambushers-acc Massivae paret... Bomilcar mature regis mandata exequitur. for Massiva procure Bomilcar-nom quickly of king orders-acc carries out ‘He commanded Bomilcar... to procure by payment... ambushers for Massiva. Bomilcar hastened to carry out king’s orders.’ (Sal. Jug. 35.4–5) The above-indicated referential chain seems to apply systematically; in my corpus of historical narratives, I note only one exception for Future Topics.57 On the other hand, it is worth mentioning an apparent exception in (12) where zero anaphora appears after a presentative sentence.

57. Erat eo tempore Antonius Brundisii; is virtute militum confisus… (Caes. Civ. 3.24.1); ‘Antonius was at that time at Brundisium; and having confidence in the valour of his soldiers...’ However, this is was added by editors.



Chapter 2.╇ Pragmatic functions 

(12)

Est etiam Asinius quidam, senator voluntarius, lectus is also Asinius certain-nom senator volunteer-nom chosen ipse a se. Apertam curiam Ø vidit post Caesaris himself by himself open-acc court-acc he saw after of Caesar mortem; mutavit calceos, pater conscriptus repente factus est. death he changed shoes father conscript-nom suddenly became ‘There is also a certain Asinius, a volunteer Senator, chosen by himself. He saw the Senate house open after Caesar’s death; he changed his shoes, and suddenly became a conscript Father.’ (Cic. Phil. 13.28)

Asinius is newly mentioned but he does not play a role in the discourse afterwards. Cicero does not talk about his further action but only explains the strange manner in which this Asinius became a Senator. This is not an instance of a Future Topic. The notion of referential chain is also relevant for other already known entities that are treated as Focus.58 An example is given in (13): C. Marius bears the Focus function (parallel to equitatum omnem); the second reference is made with eum in clause-initial position.59 Ille in the third reference contrasts with Metellus and marks at the same time a change of Topic to Marius. The discourse then switches to Jugurtha. (13) (Metellus) Igitur equitatum omnem ad castra propere misit therefore cavalry-acc all-acc to camp instantly he sent ac statim C. Marium1 cum cohortibus sociorum, and immediately G. Marius-acc with cohorts of allies lacrumans... obsecrat, ne... hostis inultos-acc eum2que him-acc=and weeping he begs not to enemy unpunished abire sinat. Ille3 brevi mandata efficit. // At Iugurtha... go away let he soon orders-acc executes but Jugurtha-nom ‘He instantly, therefore, dispatched the whole of the cavalry to the camp, and immediately afterward Gaius Marius, with the cohorts of the allies, and begged him with tears... not to let the enemy escape with impunity. Marius soon executed his orders. // However, Jugurtha...’ (Sal. Jug. 58.5) To conclude, the examples presented above show the strategy with which one can promote a newly introduced – or reintroduced – entity to the Topic (not necessarily as subject), as well as the treatment of focal entities. These facts have their effect on 58. See again de Jong (1989:€532) on the different treatment of focal subjects (with anaphoric continuation) and non-focal subjects (without anaphoric continuation). However, there is no need to postulate a syntactic restriction on subjects. 59. An apparent exception is: At rex postero die Asparem, Iugurthae legatum, appellat dicitque... (Sal.€Jug. 112.1) ‘Now, on the following day the king summoned Aspar, Jugurtha’s envoy, and said that...’ The choice of dicitque with ellipsis of the pronoun and not eique dicit can be explained by the fact that Aspar is not the real addressee; he is only a messenger, and what the king says is addressed to Sulla. For ellipsis, see Section 5.3. below p. 99 ff..



Constituent order in Classical Latin prose

constituent ordering: the anaphoric continuation for the second reference has its place in the initial position; the form of the third or subsequent reference is closely connected with the presence or absence of contrast, which is a priority factor. In the absence of contrast or change of Topic, zero anaphora is used for the third reference.

3.3

Discourse Topic

A complex text such as a historical narrative presents several Topics – so-called Discourse Topics – with variable degrees of importance (Dik 1997, I: 314). Discourse Topics mainly refer to people, due to the general nature of this type of text. They are accessible to the reader because they have already been introduced into the discourse, or because they belong to the shared knowledge of the author and the readers; for example, Caesar and Pompey. If need be, established Discourse Topics are recalled and reinserted in the narrative without a special signal. The re-appearance of a Discourse Topic manifests itself generally as a shift from one Topic to another; for example: Interim Albinus renovato bello commeatum, meanwhile Albinus-nom being resumed war supplies-acc stipendium aliaque... maturat in Africam portare... pay-acc other-acc=and hurries to Africa transport At contra Iugurtha trahere omnia... in response Jugurtha-nom protracted everything-acc ‘Meanwhile Albinus, resuming the war, hurried to transport to Africa supplies, pay, and other things... Jugurtha, in response, protracted everything...’  (Sal. Jug. 36.1–2)

(14)

Sallust is describing the situation on the Roman side, and then he switches to the Numidians. Discourse Topics Albinus and Iugurtha are contrasted; the contrast is underlined here by adversative (at ‘however’)60 and continuative (interim ‘meanwhile’) connectors.

3.4

Sub-Topic

A Sub-Topic is a Topic that is deducible from a previously mentioned entity. For example, as cities are usually inhabited, oppidani in (15) are deducible from ad Thalam ‘to (the city of) Thala’. Oppidani is thus a contextually dependent constituent that goes to the sentence-initial position. It is possible to deduce several Sub-Topics from one entity, for example princeps and reliquos legatos are deducible from decem legati in (16). (15) (Milites) Deinde postero die contra opinionem Iugurthae then next day against expectation of Jugurtha 60. For at ‘however’ marking a change of Topic, cf. Kroon (1995:€350) and Orlandini (1995:€90).



Chapter 2.╇ Pragmatic functions 

ad Thalam perveniunt. Oppidani... nihilo segnius bellum parare. to Thala they arrive inhabitants-nom without delay war-acc prepare ‘(Soldiers) Then, on the next day, contrary to what Jugurtha had thought, they arrived at Thala. The inhabitants... without delay prepared for war.’  (Sal. Jug. 75.9–10) Decretum fit, uti decem legati regnum... inter decree was passed that ten legates-nom kingdom-acc between Iugurtham et Adherbalem dividerent. Cuius legationis Jugurtha and Adherbal should divide of which delegation princeps fuit L. Opimius... Reliquos legatos chief-nom was L. Opimius-nom other-acc legates-acc eadem via aggressus plerosque capit... in same way assailing several he takes ‘A decree was passed that the kingdom... should be divided by ten legates between Jugurtha and Adherbal. The chief of the delegation was Lucius Opimius... Assailing the other legates in the same way, Jugurtha made several his captives.’  (Sal. Jug. 16.2–4)

(16)

The relationship between a Sub-Topic and its superordinate concept is usually based on an extra-linguistic fact: a town has its inhabitants (15); an embassy (legati) usually has a leading person (princeps) and other participants (16). However, the deduction can also take the form of an enumeration. For example, from a group of persons (Romani), Sub-Topics such as plerique ‘the great number’, multi ‘several others’, alii ‘others’ can be deduced.

3.5

Sentence Topic

The concept of Topic and its different sub-types is not without difficulties in the framework of Functional Grammar (Bolkestein 1998a: 195). For the needs of my analysis, I will adopt Hannay’s (1991:€141)61 approach and his definition of Topic. According to him, the pragmatic function of Topic consists in the selection of an element, accessible to the addressee, that will serve as basis for the construction of the message.62 The element selected as Topic is generally treated in a specific way: it is placed in sentenceinitial position. This definition leads up to the concept of Sentence Topic, which has priority over the Discourse Topic and can co-occur with it in the same sentence. The Sentence Topic, representing the point of departure – this term goes back to Weil (1844:€25) – for the message, stands in sentence-initial position (eum in 17), whereas the Discourse Topic is placed after it (Iugurtha). These Topics have a different discourse status: the Discourse Topic belongs to a higher, discourse level (H. Dik 1995:€20 ff.) whereas the Sentence Topic belongs to the sentence level. 61. Cf. Hannay (1985b: 61); see also Lambrecht (1994:€118) and Maslova and Bernini (2006:€68). 62. See also Knoth (2006:€1) with further references concerning the selection of a starting point.



Constituent order in Classical Latin prose

Cuius legationis princeps fuit L. Opimius, homo of which delegation chief-nom was L. Opimius-nom man-nom clarus et tum in senatu potens... Eum Iugurtha famous and then in Senate powerful him-acc Jugurtha-nom tametsi Romae in amicis habuerat, tamen although at Rome among friends had nevertheless accuratissime recepit. very formally received ‘The chief of the delegation was Lucius Opimius, a distinguished man who was then powerful in the Senate... Although Jugurtha had already counted him among his friends at Rome, yet he received him in a most formal manner.’  (Sal. Jug. 16.2)

(17)

Sentence Topic can also be represented by elements expressing a spatial setting, as illustrated in (18). The central part of a range of mountains serves as the setting for the description of a hill that will be occupied by Iugurtha. (18) (mons) Sed ex eo medio quasi collis oriebatur, but from its middle sort of hill-nom rose in immensum pertingens... Igitur in eo colle... Iugurtha... consedit. immensely stretching out so on this hill Jugurtha-nom set down ‘(mountain) But from the middle of it, reaching an immense distance, there sprang a kind of hill... On this hill..., Jugurtha took up position...’  (Sal. Jug. 48.3 – 49.1) Temporal settings have a different character: they usually do not pick up a referent already presented in the context, but refer to the situation of the discourse. Nevertheless, we can assume that expressions such as eo triduo ‘three days after’ are also a kind of Sentence Topic. The sentence in (19) answers the underlying question ‘what happened next?’ (19) Eo triduo legio VIII ad eum venit… three days after legion eighth-nom to him comes ‘Three days after, the Eighth legion joins him.’

3.6

(Caes. Civ. 1.18.5)

Discourse Topics Iugurtha and Caesar

In this section, I will examine the position of two prototypical Discourse Topics in my corpus of historical narratives: Iugurtha and Caesar, the central protagonists of Sallust’s The Jugurthine War and Caesar’s The Civil War respectively. My aim is to find out the relative positioning of Discourse Topics, Sentence Topics and setting elements. However, I have to state at the outset, firstly, that Latin manifests null subjects, i. e. a subject is not repeated when it is accessible.63 Secondly, we must not equate subjects with 63. This point is treated in detail in Section 5.2 below p. 98.



Chapter 2.╇ Pragmatic functions 

Topics. Topic can have other syntactic functions and can be expressed in a case other than the nominative, as in (20) with the Topic Caesari in the dative being the indirect object. The fact that Topics are often selected as subjects can be explained by their semantic-referential properties: animate, human entities bearing a high degree of individuation64 are more likely to be assigned subject function than other entities. (20)

Interim Caesari nuntiatur Sulmonenses... cupere ea meanwhile to Caesar is announced Sulmoneians-acc wish it-acc facere, quae vellet... do what he wants [Mittit eo M. Antonium cum legionis XIII cohortibus V.] ‘The news was meanwhile brought to Caesar that the inhabitants of Sulmo... are ready to carry out his wishes... [He sends Mark Anthony there with five cohorts of the Thirteenth legion.]’ (Caes. Civ. 1.18.1–2)

3.6.1 Data Iugurtha and Caesar in the nominative in syntactically independent sentences65 are mostly used to mark a change of Topic, which is related to contrast.66 In Table 1 I present a classification with a distinction between absolute initial position (column a) and non-absolute initial position (column b). The latter concerns cases in which the proper name is preceded by a connector (sed, at, igitur...), an adverb such as interim ‘meanwhile’, a setting constituent (prepositional phrase, participial clause, ablative absolute, temporal subordinate clause), or a Sentence Topic. Column c concerns every other position of Iugurtha and Caesar, i. e. cases when they neither stand initially nor are preceded by constituents enumerated in column b. Table 1 shows that there are differences between our two authors: whereas the proper name in the nominative is often found in absolute initial position (25 occ.) or Table 1.╇ Positions occupied by the nouns Iugurtha and Caesar in the nominative (corpus 1)

Protagonist Iugurtha Caesar Total

a Absolute initial

connector

â•⁄ 4 25 29

21 â•⁄ 3 24

b After temporal setting

c Other Sentence Topic

â•⁄ 6 22 28

â•⁄ 5 â•⁄ 7 12

Total â•⁄ 1 17 18

â•⁄ 37 â•⁄ 74 111

64. See Hopper and Thompson (1980:€253) and Givón (1983:€22). 65. In my corpus, there are 39 such occurrences of Iugurtha in the nominative, and 76 occurrences of Caesar. Instances in subordinate clauses as well as four other cases were disregarded. 66. Of course, the constituents Iugurtha and Caesar can also have Focus function.



Constituent order in Classical Latin prose

preceded by a setting constituent (22 occ.) in Caesar, Sallust often puts a connector before Iugurtha, but does not use setting constituents as frequently as Caesar does. It is remarkable that Caesar stands in an “other” position 17 times in Caesar, but Iugurtha only once in Sallust. This point will be commented on in Section 3.6.4, p. 72. The position of connectors and other connectives (columns a and partially b) is either fixed, for example at Iugurtha ‘however Jugurtha’ and Caesar enim ‘therefore Caesar’ with no possible variation, or it does not seem to be significant, for example igitur Iugurtha ‘thus Jugurtha’ and Iugurtha interim ‘Jugurtha meanwhile’. Consequently, I will restrict myself to the positioning of Discourse Topics with respect to temporal setting constituents and Sentence Topics. 3.6.2 Sentence Topic and Discourse Topic Anaphoric expressions can pick up a preceding Focus, a preceding content, or a part of preceding complex information, in order to transform them into a Topic. This procedure of topicalisation, described below in Section 4.2, p. 75 ff, establishes a Sentence Topic with respect to the (immediately) preceding context, and this has priority over the Discourse Topic. Thus, we have ad ea Caesar respondit and not Caesar ad ea respondit (21): (21) [...loquitur Afranius:... ne ad ultimum supplicium progredi necesse habeant. Haec quam potest demississime et subiectissime exponit.] Ad ea Caesar respondit:... to this Caesar-nom replies ‘[...Afranius speaks:... that they should not think it necessary to proceed to the extreme punishment. He argues this as humbly and submissively as possible.] To this Caesar replies:...’ (Caes. Civ. 1.84.5 – 85.1) Ad ea summarizes the content of the direct speech pronounced by Afranius. The sentence starting with ad ea belongs to the narrative line, in contrast with the one starting with haec (with the same referent and topicalising function as ad ea); the latter presents an author’s comment. Ad ea is the Sentence Topic, standing in initial position, whereas Caesar, referring to the principal protagonist, is a Discourse Topic belonging to a higher level; it provides a Topic shift from Afranius to Caesar. 3.6.3 Setting constituent and Discourse Topic Setting constituents can take various syntactic forms, such as noun phrases, prepositional phrases, ablative absolutes, and temporal subordinate clauses. They indicate the temporal or spatial background of the action expressed in the main clause (de Jong 1989:€523). From a pragmatic point of view, these constituents may or may not function as Topic. Compare the two ablative absolutes in (22) and (23): (22) Recepto Firmo expulsoque Lentulo Caesar Firmum being recovered expelled=and Lentulus Caesar-nom



Chapter 2.╇ Pragmatic functions 

conquiri milites... delectumque institui iubet. be rounded up soldiers-acc recruitment-acc=and begin orders ‘After the recovery of Firmum and the expulsion of Lentulus, Caesar gives orders for the soldiers... to be sought for and recruitment to begin.’  (Caes. Civ. 1.16.1) (23)

[Tuberonem portu atque oppido prohibet...] His rebus confectis Caesar... milites in proxima these things being done Caesar-nom soldiers-acc to nearby municipia deducit. towns withdraws ‘[He prevents Tubero from approaching the port and the town...] When these matters had been dealt with, Caesar withdraws his men into the nearby towns...’  (Caes. Civ. 1.32.1)

The ordering of constituents is the same but the adverbial clauses do not have the same pragmatic function. The ablative absolute recepto Firmo expulsoque Lentulo does not pick up preceding information, but conveys new information. His rebus confectis refers to events that were mentioned in the preceding section (preventing his enemy from entering Utica), as is signalled by the anaphoric his (cf. Bolkestein 2002). This difference permits us to formulate several tendencies in the placement of setting constituents. 1. When the adverbial clause, in particular an ablative absolute or a conjoined participle, or another adverbial expression, contains an anaphoric element, it has priority, and the Discourse Topic comes after it (24).67 (24) Ea fuga Iugurtha... cum perfugis... Thalam pervenit. by this retreat Jugurtha-nom with deserters to Thala arrived ‘By this retreat Jugurtha... arrived with the deserters... at Thala.’  (Sal. Jug. 75.1) Phrases and clauses containing the connecting relative qui, which must stand initially, also belong to this category, for example: qua re animadversa Caesar (Caes. Civ. 1.83.5) ‘noticing this’. Sometimes the relationship with the preceding context is not explicitly signalled by an anaphoric pronoun, in which case also the adverbial expression comes first. Re in (25), summarising the preceding content, has thus a definite reading, ‘the/this event’. (25) [Caesar... iuxta murum castra posuit.] Re cognita, thing being learnt

67. Cf. also the constructions given by Chausserie-Laprée (1969:€ 61 ff.) and considered as “clichés de liaison, de reprise et d’enchaînement”.



Constituent order in Classical Latin prose

Domitius ad Pompeium... peritos... mittit... Domitius-nom to Pompey experts-acc sends ‘[Caesar... pitched camp close to the wall.] Learning what had occurred, Domitius... sends some experienced men... to Pompey...’ (Caes. Civ. 1.17.1) 2. When an adverbial clause does not pick up any preceding content and contains no anaphoric pronoun, it is not given priority. The Discourse Topic appears more often in initial position instead (9 occurrences out of 12) or it is inserted inside the adverbial clause. For example, exploratis regionibus does not summarize preceding information but denotes an action that happens before educit. (26)

Caesar exploratis regionibus, albente caelo omnes Caesar-nom being explored regions at first light all-acc copias castris educit... forces-acc from camp leads out ‘Caesar after reconnoitring the district leads his entire force out of camp at first light...’ (Caes. Civ. 1.68.1)

3. Time expressions such as postero die ‘next day’, sub vesperum ‘towards evening’, or eo tempore ‘at that time’, do not refer to an element expressed in the preceding context but to a moment relative to the situation of the discourse. These expressions of temporal setting are in strong competition with Discourse Topics as to positioning. There is a slight preference (6 occurrences out of 10) for temporal settings in initial position, as is shown in (27). (27) Eo tempore Iugurtha per collis sequi... at that time Jugurtha-nom along the hills follows ‘At that time, Jugurtha would follow along the hills...’

(Sal. Jug. 55.8)

4. The same holds true for temporal subordinate clauses containing a finite verb form. They can appear first in a complex sentence, but more frequently they are preceded by a Discourse Topic. In the traditional view, this phenomenon is explained as “an anteposition of the subject common to the subordinate and the main clause” (Kühner and Stegmann 1914, II: 625).68 In my view, however, the subject stands in initial position not because it is a subject but because it is a Discourse Topic, mostly contrastive, that signals a change of Topic. Iugurtha in (28) contrasts with consul, and Caesar in (29) with his adversary Lentulus Spinther. Whereas the Example (28) could illustrate the above-mentioned “school” rule about the subject shared by both subordinate and main clauses (Iugurtha), the Example (29) weakens it because Caesar is not the subject of luxit. 68. Kühner and Stegmann (1914, II: 625) mention “anteposition” of subjects and objects, but not pronominal expressions. Cf. also Marouzeau (1953:€83) on “postposition” of subordinators. As it belongs syntactically to the sentence, I would not qualify such a subject as “détaché” or extraposed in sentence-initial position, as Chausserie-Laprée (1969:€150) does, among many others.



Chapter 2.╇ Pragmatic functions 

(28) (Consul...) Iugurtha ubi Metelli dicta cum Jugurtha-nom when of Metellus words-acc with factis composuit..., statuit armis certare. facts compared decided with arms to struggle ‘(The consul...) When Jugurtha compared what Metellus said with what he did, he decided on armed struggle.’ (Sal. Jug. 48.1) (29) (Lentulus Spinther...) Caesar, ubi luxit, omnes Caesar-nom when it dawned all-acc senatores... ad se produci iubet. senators-acc to him be brought orders ‘(Lentulus Spinther...) At dawn Caesar orders all senators... to be brought before him.’ (Caes. Civ. 1.23.1) A closer look at temporal (cum, ubi, postquam, ut, dum) and concessive (etsi, tametsi) subordinators in my corpus shows that they mainly follow the Discourse Topic (15 occurrences out of 17), as in (28) and (29). One of the two instances where this is not the case is quoted in (30). There is no Topic shift to Iugurtha, because animus, the subject of the main clause, refers to the maintained Topic Calpurnius. (30) (Calpurnius...) Sed ubi Iugurtha per legatos pecunia but when Jugurtha-nom by legates with money temptare... coepit, animus aeger avaritia facile conversus est. tempt began mind-nom ill by avarice easily was turned ‘(Calpurnius...) But, when Jugurtha by means of legates began to tempt him with money..., Calpurnius’s mind, diseased by avarice, was easily turned.’  (Sal. Jug. 29.1) We have seen that, on the one hand, a Discourse Topic stands after a contextually bound constituent, and on the other hand, before a temporal subordinate clause. Instances such as (31) with a temporal clause placed first would seem to contradict this. Actually, this temporal clause is anchored in the preceding context (note sic), which was not the case in the other examples quoted in this section. Referring to what has just happened, the dum-clause serves as a transition from one scene to another one. An adverb such as interim ‘meanwhile’ could be used instead. Its initial position is fully justified and the Discourse Topic comes after it. (31)

Dum apud Zamam sic certatur, Iugurtha ex inproviso while at Zama thus struggle-pas Jugurtha unexpectedly castra hostium cum magna manu invadit. camp-acc of enemies with large unit attacks ‘While the struggle was going on in this way at Zama, Jugurtha unexpectedly attacked the enemy camp with a substantial unit.’ (Sal. Jug. 58.1)



Constituent order in Classical Latin prose

Table 2.╇ Placement of sentence constituents Sentence Topic eum, quem, hos montes, in eo loco… Setting re cognita eo tempore dum… sic certatur

Discourse Topic Caesar Discourse Topic Caesar

Adverbials exploratis regionibus ubi… cognovit Other elements

Other elements

To sum up, I have sought to demonstrate in this section that contextually dependent constituents providing an explicit link with the preceding context, and at the same time representing a point of departure for the current sentence, have priority and stand in sentence-initial position. These are Sentence Topics or setting elements. Established Discourse Topics are also contextually dependent but belong to a higher, discourse level. This explains why they come after these Sentence Topics or setting elements. Adverbials that are not linked to the preceding context are not placed in initial position. The positioning of the constituents in question is summarized in Table 2. 3.6.4 Discourse Topic in a late position In Table 1 (p. 67) we saw that there are a considerable number of instances (17 in total) of Caesar in a position other than initial. Some of these instances involve the “postverbal placement” of the subject. Traditionally, this phenomenon is explained as caused by prominence given to the verb (Kühner and Stegmann 1914, II: 599).69 It is worth considering a well-known example taken from The Civil War: (32) [...facultates pons Ilerdae praebebat et loca trans flumen integra, quo omnino Caesar adire non poterat. Hae permanserunt aquae dies complures.] Conatus est Caesar reficere pontes, sed nec attempted Caesar-nom rebuild bridges but neg magnitudo fluminis permittebat... size-nom of river allowed ‘[...access to all this was provided by the bridge at Ilerda and the safe country across the river, which Caesar was quite unable to reach. The floods lasted for several days.] Caesar attempted to rebuild the bridges, but the swollen river did not allow it.’ (Caes. Civ. 1.50.1) A violent storm has broken two bridges (see Civ. 1.48.2) and Caesar has difficulties with obtaining supplies. At the same time, he knows there are provisions across the river. The repair of the bridges is not completely new information: Caesar’s action is 69. See also Marouzeau (1953:€45 ff.); recently de Jong (1989:€535) and Bolkestein (1995:€38, and 1996b: 12 f.)



Chapter 2.╇ Pragmatic functions 

fully expected in such a situation. The verb conatus est in sentence-initial position is the most informative element (it has verum Focus), and furthermore, it contrasts with nec… permittebat. What is more interesting than the fact that the subject comes after the verb is that it is expressed at all. It could be treated as a maintained Topic because the description of unfavourable conditions (hae aquae... complures) does not interfere with it (see Section 5.2 below, p. 98). The expression of the subject is to be explained as a characteristic of the accurateness of Caesar (cf. Odelman 1972:€156); he aims at giving the most exact information possible at the cost of repetitions. Another example of a similar nature is given in (33). This is also an instance of Topic reminder.70 The constituent Caesar does not stand initially because there is no Topic shift. (33) [Quibus rebus neque tum respondendum Caesar existimavit, neque nunc, ut memoriae prodantur, satis causae putamus.] Postulabat Caesar, ut legatos sibi asked Caesar-nom that envoys-acc him-dat ad Pompeium sine periculo mittere liceret. to Pompey without danger to send be allowed ‘[Caesar thought no reply necessary at the time, nor do we consider there is sufficient reason now to set one down in the record.] Caesar asked to be allowed to send envoys to Pompey under a safe conduct.’ (Caes. Civ. 3.17.1–2)

4. The placement of pronouns Pronouns have a specific function: they pick up an entity retrievable from the preceding context, from shared knowledge, or from the situation of the discourse. The aim of this section is to examine the position of anaphoric and demonstrative pronouns (4.1) as well as of personal pronouns (4.2).

4.1

Pronominal anaphora

Latin possesses several pronouns that serve for anaphoric referencing: one anaphoric (is), three demonstratives (hic, iste,71 ille), and one connecting relative (qui). These pronouns refer to contextually given entities, and, with the exception of the connecting relative qui, they can theoretically stand anywhere in the sentence. As I will argue

70. These Topic reminders are not to be analysed as Tails; see Section 7 below, p. 111. 71. Iste used for references concerning the addressee will be ignored here because it does not appear in historical narrative.



Constituent order in Classical Latin prose

in this section, these pronouns may or may not function as the Topic, and the pragmatic function they fulfil affects their position in the sentence.72 Some recent studies devoted to anaphoric and demonstrative pronouns concern the referential properties of the entity that is referred to (Orlandini 1995:€63 ff.), the distribution of pronouns, and also the position they occupy. I briefly summarize here the main observations that have been made, which serve as a point of departure for my analysis: Pennell Ross (1996) states that in Caesar’s historical narrative (The Civil War) hic tends to occur in the sentence-initial position, whereas is favours an internal position; ille does not show a clear preference. This point was confirmed by Bolkestein (2000:€118). De Jong (1996:€502) noticed that ille appears in the nominative more frequently than is or hic, which are mainly attested in oblique cases. Although statistics concerning Caesar’s The Civil War have already been compiled – especially by Pennell Ross (1996), and Bolkestein (2000) – , it is important for my discussion to make a comparison with Sallust’s historical narrative.73 The data presented in Table 3 and 4 are organised as follows. In the column “initial position” are the figures for occurrences in absolute initial position in a sentence or a clause as well as after priority connectors such as at and sed ‘but’. A distinction is made between pronouns used alone (is ‘he’) and in a phrase as attributes (eo tempore ‘at that time’).74 Anaphoric adverbs such as ibi, eo, huc and the like are indicated in a separate row. The results are shown in two tables, one for Caesar, the other for Sallust; high numbers of occurrences are in bold. A first glance at the tables permits us to conclude that Caesar prefers using hic, whereas Sallust prefers is. It is remarkable that in spite of the less frequent use of anaphoric pronouns in Sallust, the two authors agree in the proportion of them placed in the initial (49% + 18/19%) and in an internal (33/32%) position. The initial position is globally the most represented one (67/68%). We note that the majority of hic and ille used alone stand in initial position in both authors.75 Anaphoric adverbs are mostly found in initial position. Attributive uses mainly concern hic in Caesar (32%) and is in Sallust (28%); attributive ille in initial position is absent in Caesar, and very rare in 72. See Spevak (2008). Several articles have been devoted to anaphora but no clear and systematic conclusions have been drawn. De Jong (1989:€524 f.), limiting himself to subjects, observed that the antecedent of an anaphoric pronoun functioning as subject picks up a preceding Focus constituent. Bolkestein and van de Grift (1994:€287 ff.) investigated the pragmatic value of a constituent which is selected as a subject afterwards. Cf. also Jones (1991:€86). 73. The two samples, Caesar’s and Sallust’s, have a comparable length, i. e. 21,000 words. 74. Correlative uses were not taken into consideration. Furthermore, occurrences appearing in non-declarative sentences or presenting textual problems were not counted. 75. As for concrete forms, ille in the nominative is used in initial position. Cf. the analysis by Longrée (2004:€ 171) who observes that unlike other pronouns, ille in a case other than the nominative does not appear in initial position. Furthermore hic and hi in the nominative are absent in Sallust.



Chapter 2.╇ Pragmatic functions 

Table 3.╇ Pronominal anaphora in Caesar (corpus 1) Pronoun

Is Hic Ille Adverbs Total

Initial position Pronoun alone Attributive use 76 Occ. % Occ. % â•⁄ 44 102 â•⁄ 33 â•⁄ 41 220

29% 53% 69% 75% 49%

21 60 â•⁄ 0 – 81

14% 32% â•⁄ 0% – 18%

Other position (alone + attribute) Occ. % 88 28 15 14 145

57% 15% 31% 25% 33%

Total (100%) 153 190 â•⁄ 48 â•⁄ 55 446

Table 4.╇ Pronominal anaphora in Sallust (corpus 1) Pronoun

Is Hic Ille Adverbs Total

Initial position Pronoun alone Attributive use Occ. % Occ. % â•⁄ 74 â•⁄ 30 â•⁄ 29 â•⁄ 21 154

40% 58% 52% 78% 49%

51 10 â•⁄ 1 – 62

28% 19% â•⁄ 2% – 19%

Other position (alone + attribute) Occ. % 58 12 26 â•⁄ 6 102

32% 23% 46% 22% 32%

Total (100%) 183 â•⁄ 52 â•⁄ 56 â•⁄ 27 318

Sallust. It is worth adding that a position other than initial is mainly occupied by nonnominative forms of is, in particular eius and eum. Furthermore, certain inflected forms of ille are found in an internal position in Sallust, especially illis and illi. From these data we can conclude that there are distributional differences between Caesar’s and Sallust’s works that have to be ascribed to the personal style of these authors.77

4.2

Topicalisation

As Pinkster (1995:€222) observed, Topic constituents have their place in initial position. The anaphoric and demonstrative pronouns under examination are strong candidates for the Topic function because they refer to contextually given entities. However, 76. Percentages are calculated horizontally and indicate the proportion of occurrences in initial position and the proportion in an internal position. The distinction between pronouns used alone and as attributes does not affect results; this distinction is given simply for information. 77. Pennell Ross (1996) and Bolkestein (2000) have only examined Caesar. Cf. Longrée (2004:€140) who observes that there are also differences in distribution in the works of a single author.



Constituent order in Classical Latin prose

as we will see, anaphora is not a priority factor per se and anaphoric constituents need not be selected as Topics. Compare the examples given in (1) and (2); the first one illustrates an anaphoric pronoun in an internal position (32/33% in Tables 3 and 4), the second one the initial position (67/68%). (1)

[Illi (Massilienses)... minus libere, minus audacter vagabantur...] Interim Oscenses et Calagurritani... meanwhile inhabitants of Osca and Calagurris-nom mittunt ad eum legatos seseque imperata facturos pollicentur. send to him envoys-acc refl=and orders-acc do-fut.ptc promise ‘[(They (Massiliots)... roamed less freely and less boldly...] Meanwhile the inhabitants of Osca, and those of Calagurris... send envoys to him and promise to obey his wishes.’ (Caes. Civ. 1.59.2 – 60.1)

Simul... P. Atti equitatus omnis et una levis armaturae at once of P. Attius cavalry all-nom and together of light-armed interiecti complures... cernebantur. Ad eos Curio equitatum interspersed many-nom were seen to them Curio-nom cavalry-acc et duas Marrucinorum cohortes mittit. and two of Marrucini cohorts-acc sends ‘All Attius’s, cavalry together with a number of interspersed light-armed were observed... To meet them Curio sends his cavalry and two cohorts of the Marrucini.’  (Caes. Civ. 2.34.2–3) (2)

The difference between the placement of the anaphoric pronouns in (1) and (2) resides in the informational status of the antecedent and in the pragmatic function of the anaphoric pronoun itself. The pronoun (ad) eum refers back to a well established Discourse Topic (Caesar) that does not have a special pragmatic function here, for the current paragraph talks about his enemies. By contrast, (ad) eos in (2) summarizes two coordinated salient elements of the preceding sentence (equitatus et complures) and makes them Sentence Topic in the subsequent sentence.€I will name this procedure “topicalisation”,78 i. e. resumption, by means of pronominal anaphora, of a preceding Focus constituent and its selection as Sentence Topic.79 The sentence in (1) tells us what the inhabitants of Osca and Calagurris do: they send envoys to him (Caesar), whereas in the second one (2) we are informed about what happened to the cavalry and infantry: somebody is sent against them. The topicalisation signals that the sentence will inform us about what happens to this entity, and it is effected by anaphoric pronouns 78. I am aware of the fact that the term “topicalisation” is not ideal, because Generative Grammar uses it for another concept. Cf. also Firbas (1992:€125). On the other hand, this term is unavoidable in order to keep the relationship between Topic and Topicalisation as “selection of an element as Topic”. 79. Absence of resumption does not mean that a constituent does not function as the Focus; but the constituents that are continued are more readily identified as the Focus.



Chapter 2.╇ Pragmatic functions 

standing, as Sentence Topics, in initial position, such as (ad) eos in (2). But anaphoric pronouns in an internal position do not have this special pragmatic function; they only maintain the coherence of the text by referring to something that is already known. Topicalisation can intervene between two independent sentences (2) as well as between two clauses of a complex sentence (see Example 3 harenam, quoted below). It is important to underline that picking up a salient element in order to make it the point of departure (Sentence Topic) of the subsequent sentence corresponds to the progression of information in a discourse (cf. Firbas 1992:€8).80 4.2.1 Resumption of Focus

4.2.1.1 Initial is and hic Initial is and hic pick up a salient element given in the preceding clause or sentence. Apart from several exceptions (see Section 4.2.1.4 below, p. 81), this principle functions in a systematic way in Caesar and Sallust (corpus 1). Future Topics, discussed in detail in Section 3.1 above, p. 56 ff, are prototypical examples of where topicalisation takes place. However, topicalisation is not restricted to new discourse participants; it also works for contextually bound entities that are treated as the Focus. Animate, human entities are usually picked up by an anaphoric pronoun, inanimate ones either by an anaphoric pronoun, or by a phrase containing one. The syntactic function of the antecedent and of the pronoun plays no role in topicalisation. Two typical examples are given below. In (3), the Focus is on harenam (question ‘what?’), picked up by the anaphoric pronoun ea in the nominative. Here, topicalisation is combined with a change of the subject, because otherwise there would be continuation of the subject ventus. The same question, ‘what?’, underlies the first clause given in (4); the subsequent clause with the anaphoric pronoun in the accusative (has) informs us about what happens with the ships. When the pronoun stands in the clause-initial position, the clause is juxtaposed to the preceding one or is coordinated by means of -que,81 atque or et. (3)

Nam ubi... ventus coortus harenam humo excitavit, for when wind-nom risen sand-acc from ground dislodges ea magna vi agitata ora oculosque implere solet. it with great force whipped up mouth eyes=and fill is used to ‘For when the wind rises and dislodges the sand from the ground..., the sand is whipped up with great force and is accustomed to fill the mouth and eyes.’  (Sal. Jug. 79.6)

80. It would be a mistake to confuse topicalisation with the principle called Tail-Head linking in Functional Grammar (Dik 1997, II: 438 f.), although this principle is not very well described within this framework and remains vague. For me, Tail-Head linking is only suitable for repetitions and redundancies, as presented in Section 7, p. 114. 81. As Bolkestein (2000:€134, note 7) signalled, combinations of words ending in -c with the enclitic -que are excluded (*hicque, *huncque); cf. also Kühner and Stegmann (1914, II: 13).



Constituent order in Classical Latin prose

(4)

Certas sibi deposcit naves Domitius atque determined-acc for him he asks ships-acc Domitius-nom and has colonis pastoribusque, quos secum adduxerat, these-acc with tenants shepherds=and rel with him had brought complet. fill up ‘Domitius asked for ships of his own and manned them with tenants and shepherds he had brought with him.’ (Caes. Civ. 1.56.3)

Between the sentence containing a focal antecedent and the sentence with a resumptive pronoun, explicative or descriptive information can intervene. This type of information does not belong to the same level of discourse and has no influence on the identification of the referent that is picked up by the anaphor.82 A good example is given in (5) with hos (oppressos et dispersos) resuming (imprudentes atque inopinantes) hostes ‘enemy’; the explicative sentence (Numidae enim… consederant) clarifies what made the attack possible. (5) [Equites missi nocte iter conficiunt, imprudentes atque inopinantes hostes aggrediuntur. //Numidae enim quadam barbara consuetudine nullis ordinibus passim consederant.//] Hos oppressos somno et dispersos adorti these-acc overcome by sleep and dispersed attacking magnum eorum numerum interficiunt. great-acc of them number-acc they kill ‘[The cavalry complete their journey by night and attack the enemy taken off guard and unawares. The Numidians, according to some barbarous custom of their own, had camped indiscriminately, in no sort of formation.] Attacking them when overcome by sleep and dispersed, the cavalry kill a great number of them.’  (Caes. Civ. 2.38.4–5)



One Focus constituent may be referred to anaphorically several times, as is shown in (6) where montem ‘mountain’ is topicalised twice by hunc. (6) [(Afranius) Ex eo loco IIII caetratorum cohortes in montem, qui erat in conspectu omnium excelsissimus, mittit.] Hunc magno cursu concitatos iubet occupare... this-acc at full speed urged on-acc he orders to occupy Hunc cum obliquo itinere caetrati peterent, this-acc when by oblique route cohorts-nom approached

82. See Pennell Ross (1996:€517 ff.), Bolkestein (2000:€128), and Kroon (1995:€129 f.).



Chapter 2.╇ Pragmatic functions 

conspicatus equitatus Caesaris in cohortes impetum fecit. having observed cavalry-nom of Caesar to cohorts attack-acc did ‘[(Afranius) From this spot he sends four light-armed cohorts towards the highest mountain in sight.] He orders them to hurry at full speed and occupy it... As these cohorts approached this hill from the side, Caesar’s cavalry saw them and attacked.’ (Caes. Civ. 1.70.4–5)

4.2.1.2 Is versus hic The examples presented above raise a question: what conditions determine the choice between is and hic? Examining the choice of these two pronouns, de Jong (1996:€504 and 506) separates nominative forms from oblique forms and concludes that hunc has an introductory function whereas eum has a continuative one. Consequently, he would reject atque hunc in (7) and eum in (8). (7)

Pauci... ad T. Labienum legatum in hiberna perveniunt a few-nom to T. Labienus legate to winter camp they arrive atque eum de rebus gestis certiorem faciunt. and him-acc about events they inform ‘A few... make their way to the legate Titus Labienus at winter camp... and inform him of these events.’ (Caes. Gal. 5.37.7)

Caesar... Brutum adulescentem his copiis praeficit; Caesar Brutus-acc young these forces-dat puts in command hunc monet, ut in omnes partes equites quam this-acc instructs that in all directions horsemen-nom as latissime pervagentur. extensively should range ‘Caesar... places young Brutus in command of these forces; he gives him instructions that the cavalry should range as extensively as possible in all directions.’  (Caes. Gal. 7.9.1)

(8)

However, this is more likely a mere authorial preference for using is where there is coordination and hic where there is not (cf. note 81 above). At any rate, is and hic have the same function in (7) and (8): they pick up a preceding Focus constituent. Is cannot be seen as referring only to an established Topic, pace de Jong (1996:€504), because it can pick up a Future Topic, as we have seen above.83 The conclusions drawn by de Jong are, I suspect, influenced by the fact that he has only examined the distribution of anaphors in Caesar (The Gallic War). However, it is different in Sallust, who, unlike Caesar,

83. See p. 58, Example (4), and p. 66, Example (17).



Constituent order in Classical Latin prose

does not use hic very much.84 The choice between is and hic seems to depend, among other things, on the author’s personal style and his expressiveness.85

4.2.1.3 Is and hic in an internal position Is and hic placed in a sentence-internal or clause-internal position do not pick up a Focus constituent of the preceding sentence; they simply refer to an already known entity, as illustrated in (9) and (10). (9)

[De quibus rebus Servilius consul ad senatum rettulit] senatusque Caelium ab re publica removendum censuit. Senate-nom=and Caelius-acc from public duties to be suspend decided Hoc decreto eum consul senatu prohibuit. by this decree him-acc consul-nom from Senate barred ‘[Servilius as consul referred the episode to the Senate,] which passed a motion suspending Caelius from public duties. On the strength of this decree the consul barred him from the Senate.’ (Caes. Civ. 3.21.3)

(10)

Igitur Marius... in Africam profectus paucis diebus Vticam so Marius-nom to Africa setting out in few days to Utica advehitur. Exercitus ei traditur a P. Rutilio legato. arrives army-nom him-dat is handed over by P. Rutilius legate ‘Setting out, accordingly, for Africa... Marius arrived in a few days at Utica. The army was handed over to him by Publius Rutilius, the legate.’  (Sal. Jug. 86.5)

In these instances, the information given is not what happens to Caelius and to Marius. Caelius, referred to with eum in (9) is not the Focus of the sentence. The information is complex (Caelium ab re publica removendum) and is summarized by the noun phrase containing hic (hoc decreto) that topicalises it and stands sentence-initially. The sentence in (10) informs us about the arrival of Marius at Utica; ei does not refer to the new piece of information but simply to the maintained Discourse Topic Marius. The reference to an established Topic exemplified above is normally made by the anaphor is, especially in an oblique form; by contrast, hic does not refer to established Topics (de Jong 1996:€504).86 Used in such a way, is placed in an internal position and referring to an established entity has no special pragmatic function.

84. Compare the numbers in Tables 3 and 4, indicated above, p. 75. 85. Cf. Bolkestein (2000:€121) who, at least for the continuation of Future Topics, does not see any clear difference between the choice of is and hic. 86. The functional difference between is and hic, observed by de Jong (1996:€504), is perfectly applicable here but not when it is a question of picking up a Focus constituent (see preceding section).



Chapter 2.╇ Pragmatic functions 

Hic placed in an internal position can also pick up a previously mentioned element, as for example hanc fossam in (11).87 However, hanc fossam is not an instance of topicalisation: it is a part of complex information and is picked up afterwards by ibi. In other words, unlike is, hic in an internal position is pragmatically pertinent and often has Focus function, on its own or as a part of a pragmatic unit. (11) [Caesar... pedum XV fossam fieri iussit. Prima et secunda acies in armis, ut ab initio constituta erat, permanebat...] Sub vesperum Caesar intra hanc fossam legiones towards evening Caesar-nom within this ditch legions-acc reducit atque ibi sub armis proxima nocte conquiescit. withdraws and there under arms following night he rests ‘[Caesar... ordered a fifteen-foot ditch to be dug. The first and the second ranks stayed under arms, as had been the case from the beginning...] Towards evening Caesar withdraws the legions behind this ditch and rests there the following night under arms.’ (Caes. Civ. 1.41.5)

4.2.1.4 Special cases In my corpus of historical narrative there were several occurrences that do not conform to the principle of topicalisation described above.88 I will present here only three of them. The first type concerns participial constructions (especially predicative participles).89 In (12), navem is evidently a focal constituent, and thus I would expect in eam placed before duabus. (12)

Conspicataeque naves triremes duae navem spotting=and ships-nom triremes two-nom ship-acc D. Bruti... duabus ex partibus sese in eam incitaverant. of D. Brutus two from parts refl to it had hastened [Sed tantum re provisa Brutus celeritate navis enisus est…] ‘Two triremes, spotting Decimus Brutus’s ship... raced from opposite direction towards it. [But Brutus, seeing what was coming, made so vigorous an effort, thanks to the speed of his ship...]’ (Caes. Civ. 2.6.4)

The second type of exception is exemplified in (13). The first sentence answers the question ‘what?’ but the placement of eas, resuming litteras Graecas, is explained by the

87. Note that fossam in the first sentence conveys new information, but is not treated as Topic in the immediately following context. 88. 13 occurrences in total (5 in Caesar, 8 in Sallust); four of them concern resumption of Focus, three resumption of content, and four anaphoric adverbs. For more detail, see Spevak (2007b). 89. For the moment, I have no explanation for this placement. Participial constructions and their pragmatic value need further investigation.



Constituent order in Classical Latin prose

priority placement of an emphatic constituent: parum goes with placebat, and together they express subjective evaluation. This example is taken from Sallust’s direct speech. (13)

Neque litteras Graecas didici; parum placebat eas neg letters Greek-acc I learned little pleased them-acc discere, [quippe quae ad virtutem doctoribus nihil profuerant]. to learn ‘Nor have I studied Greek; I had little inclination to study it, [since it had done nothing for those who taught it.]’ (Sal. Jug. 85.32)

I have mentioned some cases where anaphors picking up a preceding Focus do not stand in initial position. On the other hand, there are a few occurrences of initial anaphors that would be expected in an internal position. This is the case with eos referring to quinque ‘five persons’ in (14), which has already been topicalised in a preceding sentence. A possible interpretation would be that eos repeats the preceding instance of eos and topicalises the same entity again. (14) [Bocchus... quinque delegit... Eos... ire iubet. Illi mature ad hiberna Romanorum proficiscuntur..., ad Sullam profugiunt...] Eos ille non pro vanis hostibus, uti meriti erant, sed them-acc he neg for enemy liars as they deserved but accurate ac liberaliter habuit. considerately and generously he treated ‘[Bocchus... selected... five persons... He ordered these to go... They immediately set out for the Roman winter camp..., they fled to Sulla...] The latter did not treat them like enemy liars, as they deserved, but considerately and generously.’  (Sal. Jug. 103.4–5) 4.2.2 Resumption of larger content

4.2.2.1 Is and hic The examples discussed in the preceding section have shown topicalisation of a constituent that bears the Focus function or is part of complex information. It is possible to pick up a more complex constituent, represented for example by a subordinate or infinitive clause. In such cases topicalisation is achieved with attributive is or hic, or with a neuter pronoun, singular or plural. They stand in sentence- or clause-initial position,90 with possible discontinuity of the noun phrase. Two examples are given in (15) and (16).

90. For a few exceptions that I encountered in my corpus, see above, note 88.



Chapter 2.╇ Pragmatic functions 

(15) [Igitur bello Numantino Micipsa... sperans... facile eum occasurum, praefecit Numidis, quos in Hispaniam mittebat.] Sed ea res longe aliter, ac ratus erat, evenit. but this affair very differently than he supposed turned out ‘[So during the Numantine war..., Micipsa gave him the command of the Numidians whom he was sending into Spain, hoping that he would certainly perish...] But the affair turned out very differently from that which he had calculated.’  (Sal. Jug. 7.2–3) (16) [At illa multo optima rei publicae doctus sum: hostem ferire, praesidia agitare..., laborem tolerare.] His ego praeceptis milites hortabor. by these I precepts soldiers-acc I will encourage ‘[But I have learned much the best thing for our Republic: to strike down the foe, to keep watch and ward... to bear hard work.] It is with these maxims that I shall encourage the soldiers.’ (Sal. Jug. 85.33–34) The resumption concerns Micipsa’s action in (15) and the enumeration of Marius’s virtues in (16). The resumptive pronoun is accompanied by, respectively, a generic noun (res ‘thing’) and a noun that designates what Marius has learnt (praecepta ‘maxims’). These noun phrases could be replaced by a neuter pronoun: singular id could summarize one clause or sentence, or plural haec with a larger scope (de Jong 1996:€507), as for example the three infinitive phrases in (16).

4.2.2.2 Summarising function of hic Hic can pick up several sentences and even a whole paragraph. This property permits it to end whole sections by introducing a summary. Is cannot be used in this way (Bolkestein 2000:€121). For example, haec in (17) summarizes the content of the preceding decree of the Senate; his mandatis in (18) picks up some indirect speech. Summarizing hic91 often figures in an ablative absolute construction. (17) [Decurritur ad illud extremum atque ultimum senatus consultum...: dent operam consules... ne quid res publica detrimenti capiat.] Haec senatus / consulto perscribuntur a.d. VII Id. Ian. these by decree of Senate are recorded on January 7 ‘[Recourse is had to that last and final decree of the Senate...: “The consuls... are to take care that the State suffers no harm.”] These resolutions are recorded by decree of the Senate on January 7.’ (Caes. Civ. 1.5.3–4)

91. Is is less frequent in the ablative absolute construction, as observed by Bolkestein (2002).



Constituent order in Classical Latin prose

(18)

[Quem Caesar ad eum remittit cum mandatis:...] His datis mandatis Brundisium cum legionibus VI pervenit. these being given instructions to Brundisium with legions six he arrives ‘[Caesar sends him back to Pompey with instructions:...] After dispatching this message, he arrived at Brundisium with six legions.’  (Caes. Civ. 1.24.5 – 25.1) The summarising hic, used alone (19) or in noun phrases (20), marks the end of a speech, a letter, a message, or a description. It always stands first in sentence-initial position. (19) Hic fuit oppugnationis exitus. this was of siege end-nom ‘Such was the end of the siege.’ (20)

(Caes. Civ. 3.9.7)

[‘...si ulla apud vos memoria remanet avi mei Masinissae.’] His litteris recitatis fuere qui exercitum in Africam this letter being read were who army-acc to Africa mittendum censerent. to be sent proposed ‘[“...if the memory of my grandfather Masinissa is still cherished by you.”] When this letter had been read out, there were those who proposed that an army should be send to Africa.’ (Sal. Jug. 25.1)

Hic may also open a new section. It announces some content that will follow and marks the beginning of a new textual unit, for example a letter, a speech, a description of a place, or a battle. In this case, hic has no summarising function but is instead cataphoric, and functions as the Focus, and so it is not confined to the sentence-initial position (21).92 Furthermore, the following examples show clearly the difference between the topicalising pronouns earum in (21) and id in (22), and the introductory function of the predicative haec and the noun phrase hac ratione; these are the Focus of their sentences. (21)

[Sic locutus cum litteris eum, quas Micipsae redderet, dimisit.] Earum sententia haec erat:... of them purport-nom this was ‘[Having spoken thus, Scipio dismissed him with a letter to take to Micipsa,] the purport of which was this:...’ (Sal. Jug. 9.1)

(22) [... si haec esset in altitudinem turris elata.] Id hac ratione perfectum est. it in this way was done 92. Similarly, when hic refers to an extra-textual element it does not stand in initial position, for example in one of Sallust’s speeches: Nunc, quoniam…, per hanc dexteram, per regni fidem moneo obtestorque te… (Sal.€Jug. 10.3) ‘And now, since... by this right hand, by the loyalty due to the kingdom, I conjure and implore you to...’



Chapter 2.╇ Pragmatic functions 

[Vbi turris altitudo perducta est ad contabulationem...] ‘[... if this tower were raised to a height.] This was done in the following way. [When the tower has been built sufficiently high for the first floor to be put in...]’  (Caes. Civ. 2.8.3 – 2.9.1) 4.2.3 Selection of a constituent Having examined the resumption of Focus and the resumption of content, I come to the description of a third type of topicalisation: topicalisation of a selected single constituent. It is indeed possible to pick up just one constituent from a preceding sentence that presented complex information. Consider the following example: (23)

Nostrae naves duae... contra Lissum in ancoris our ships two-nom opposite Lissus in anchors constiterunt. Has scaphis minoribusque navigiis stopped these-acc by boats smaller=and vessels compluribus immissis Otacilius Crassus... expugnare parabat. several sent against Otacilius Crassus-nom capture prepared ‘Two of our ships... anchored off Lissus. Otacilius Crassus... sent a collection of ship’s boats and smaller vessels against them and made preparations to take them by boarding.’ (Caes. Civ. 3.28.1)

Several sections of his narrative had already been devoted to ships, but here Caesar starts a new episode that will concern two ships in particular. Nostrae naves duae ‘two of our ships’ represents a new piece of information, and what happened to them is also new (‘overtaken by night’, ‘anchored’). In the subsequent sentence, only one part of this complex information is selected for continuation and topicalised by has. Such cases of topicalisation can be compared with the introduction (or reintroduction) of an entity into the discourse without a special introductory formula (see above, p. 59, Example 5). 4.2.4 The adverbs ibi, eo, huc, etc. Topicalisation can be achieved not only by pronouns but also by anaphoric adverbs such as eo, ibi, huc, quo ‘there’. They pick up a locality that becomes the point of departure for further events. There is no essential difference between topicalisation by anaphoric pronouns or noun phrases (24) and by anaphoric adverbs (25); it is just a different way of expressing it. (24)

Ex magnis rupibus nactus planitiem in hac contra hostem from great rocks finding plain-acc in this opposite enemy aciem instruit. battle-line-acc draws up ‘Finding a plain after crossing the great rocks, he draws up a battle-line there facing the enemy.’ (Caes. Civ. 1.70.3)



Constituent order in Classical Latin prose

(25)

Pompeius... ex eo loco discedit omnibusque copiis ad Pompey-nom from this place withdraws with all=and forces to Asparagium Dyrrachinorum pervenit atque ibi idoneo Asparagium of Dyrrachii arrives and there in suitable loco castra ponit. place camp-acc places ‘Pompey... abandons his position, and marches with all his forces towards Asparagium, in the territory of Dyrrachium, where he encamped in a suitable spot.’  (Caes. Civ. 3.30.7)

Like anaphoric pronouns and phrases containing them, anaphoric adverbs topicalise a salient element from the preceding sentence or clause. The difference between the initial and internal position is shown in (26). The first ibi, placed in an internal position, picks up already known information that was not presented as salient in the immediately preceding sentence (the complement clause depending on iubet is the Focus). On the other hand, the initial eo picks up Corfinium and the initial ibi refers back to pontem. Sentences containing the adverbs eo and ibi that topicalise a locality answer the question ‘what does he do there?’ (26)

[Caesar conquiri milites, qui ab eo discesserant, delectumque institui iubet;] ipse unum diem ibi rei frumentariae causa moratus Corfinium himself one day there of provision for sake pausing to Corfinium contendit. Eo cum venisset, cohortes V praemissae a Domitio sets off there when came cohorts five-nom sent forward by Domitius ex oppido pontem fluminis interrumpebant qui erat from town bridge-acc of river were breaking down which was ab oppido milia passuum circiter III. Ibi cum from town miles of paces about three there with antecursoribus Caesaris proelio commisso... advance guard of Caesar battle being joined ‘[Caesar gives orders for the soldiers who had deserted Lentulus to be rounded up, and recruitment to begin.] He himself pauses for one day to provision, and sets off for Corfinium. On his arrival there five cohorts dispatched from the town by Domitius were breaking down the bridge over the river, about three miles from the town. There, when battle had been joined with Caesar’s advance guard...’ (Caes. Civ. 1.16.1–3)

My corpus offered four occurrences contrary to the principle of topicalisation.93 I give only one here: it concerns a salient element (idoneum locum) contained in a participial clause (27). However more frequently, the ordering exemplified in (28) is found; a forward slash (/) marks a domain boundary. 93. See above, note 88.



Chapter 2.╇ Pragmatic functions 

(27) …ipsi idoneum locum nacti / reliquam noctis themselves suitable place-acc finding remaining-acc of night partem ibi confecerunt. part-acc there finished ‘...themselves finding a suitable place, they finished the rest of the night there.’  (Caes. Civ. 3.28.5) (28) …idoneum locum nactus / ibi copias collocavit… suitable place-acc finding there forces-acc stationed ‘...finding a suitable place, he stationed his forces there... (Caes. Civ. 3.30.5)

4.3

The connecting relative qui

Up till now, I have discussed topicalisation by means of the pronouns is and hic; in this section, I will consider the connecting relative qui that shares several functions with them.94 The connecting relative qui always stands in sentence-initial position. According to Bolkestein (1996a: 559), qui refers in the majority of cases to inanimate entities and its antecedent figures in the immediately preceding sentence or section.95 Furthermore, in 29% of the cases, the connecting relative is used in an ablative absolute. In my corpus of historical narrative, the connecting relative, both used alone and in its attributive use, may summarize the Focus of the preceding sentence. An example is given in (29), with an inanimate referent (mons saxeus), which functions as a Future Topic. (29) [Namque haud longe a flumine Muluccha... erat inter ceteram planitiem mons saxeus...] Quem locum Marius... summa vi capere intendit. this place-acc Marius-nom with utmost efforts to take aimed ‘[Not far from the river Muluccha... there stood, in the midst of a plain, a rocky hill...] This place... Marius directed his utmost efforts to take.’  (Sal. Jug. 92.5)

94. For problems of identification of the connecting relative qui, see Bolkestein (1996a), and Rosén (1999:€165 ff.). Sentences containing a connecting relative qui belong to the narrative line, in contrast with sentences containing background information (Bolkestein 1996a: 557). 95. As for its frequency, according to Pennell Ross (1996:€514) and her analysis of the first two books of Caesar’s The Civil War, qui accounts for 12% of all anaphors. It is thus as frequent as ille (11%) and anaphoric adverbs (13%), but less frequent than hic (27%) and is (37%). Cf. also Longrée (2002).



Constituent order in Classical Latin prose

The connecting relative is also convenient for summaries. In (30), the noun phrase quibus rebus containing the generic noun res ‘thing’ picks up the whole preceding content (cf. Chausserie-Laprée 1969:€39 on complex initial sequences): (30) [Massilienses arma tormentaque ex oppido, ut est imperatum, proferunt..., pecuniam ex publico tradunt.] Quibus rebus confectis Caesar... duas ibi legiones these things being done Caesar-nom two-acc there legions-acc praesidio relinquit… as garrison leaves ‘[The Massiliots, as instructed, brought their weapons and artillery out of the town..., and hand over their money from the treasury.] When this was done, Caesar... leaves two legions there as a garrison...’ (Caes. Civ. 2.22.5–6) Participial clauses containing qui often serve as transitions. A typical example is given in (31) where cuius adventus expresses the result of the preceding action (Caesar... proficiscitur) and establishes the temporal setting for the following action. Note the Topic shift from Caesar to Staberius. However, this instance is different from (32). (31)

[Recepto Caesar Orico nulla interposita mora Apolloniam proficiscitur.] Cuius adventu audito L. Staberius... aquam comportare his approach being heard L. Staberius-nom water-acc to carry in arcem... coepit. into citadel started ‘[After taking Oricum, Caesar immediately set out for Apollonia.] When Lucius Staberius... heard the news of his approach, he began to take water up to the citadel.’ (Caes. Civ. 3.12.1)

(32)

[Caesar... ab urbe proficiscitur atque in ulteriorem Galliam pervenit.] Quo cum venisset, cognoscit missum a Pompeio Vibullium Rufum... there when he arrived learns was sent by Pompey Vibullius Rufus-acc ‘[Caesar... sets out from the city and arrives in Further Gaul.] On his arrival there he learns that Vibullius Rufus... has been sent by Pompey...’  (Caes. Civ. 1.33.4 – 34.1)

Quo in (32) is used in a subordinate clause repeating, in a sense, what has already been said. The resumption concerns not the place itself but the whole action of coming there (cf. Bolkestein 2000:€129 f.). This use of qui, involving more or less literal repetitions (pervenit… venisset), is an instance of what is called “Tail-Head linking” in Functional Grammar:96 the resumption of a preceding action with the help of a subordinate clause placed at the beginning of the next clause.97 Tail-Head linking is a kind of semantic redundancy that contributes to textual cohesion. By contrast, there is no semantic redundancy in (31). 96. Ibi cognoscit would be possible as well. For Tail-Head linking, cf. also Section 7, p. 114. 97. Cf. H. Dik (1995:€241), Pinkster (1995:€319) and Bolkestein (1998a: 199 f.) on ablative absolutes.



Chapter 2.╇ Pragmatic functions 

Unlike is and hic, qui can select and continue a constituent that is not focal. A good example is given in (33): qui picks up legatos, and not the new piece of information which is represented by Jugurtha’s requests. (33) [Inter haec negotia Iugurtha impensius modo legatos supplices mittere, pacem orare...] Quos item uti priores consul illectos ad proditionem these-acc like as former consul-nom enticed-acc to betrayal domum dimittebat. home sent ‘[While this was going on, Jugurtha, with greater insistence sent envoys, begged for peace...] These, like the former ones, the consul first enticed to defect, and then sent home.’ (Sal. Jug. 47.3)

4.4

Ille

The pronoun ille has very different functions: unlike is, hic, and qui, ille does not pick up preceding Focus constituents.98 In the nominative, singular and plural, the typical function of ille is to mark a change of Topic.99 The referent of ille is either mentioned in the immediately preceding context, or is accessible from shared knowledge. In Caesar’s and Sallust’s historical narratives,100 ille typically serves to shift attention to “the other side”, as pertinently formulated by de Jong (1996:€502), and introduces what happens on the side of the adversary. In my corpus, ille always has an animate referent and denotes the “other” protagonist. It is interchangeable with a proper name (Caesar, Iugurtha…) or with a common noun referring to a specific person (such as rex ‘king’, imperator ‘general’, consul ‘consul’). Bearing the function of contrastive Topic, ille comes at the beginning of a sentence. Two examples are given in (34) and (35). (34) [(Iugurtha) Igitur ex improviso finis eius (Adherbalis) cum magna manu invadit..., existimans Adherbalem... iniurias suas manu vindicaturum...] At ille... legatos ad Iugurtham de iniuriis questum misit. but he envoys-acc to Jugurtha about outrages to complain sent ‘[Therefore, Jugurtha suddenly invaded his territory with a large force..., supposing that Adherbal... was about to avenge his wrongs by force...] Adherbal, however... sent envoys to Jugurtha to complain of the outrages.’  (Sal. Jug. 20.3–5)

98. Pace Bolkestein and van de Grift (1994:€288). They argue that ille pick up focal entities, but without giving an example. 99. See Pinkster (1987), Bolkestein and van de Grift (1994), de Jong (1996), and Bolkestein (2000:€125 ff.). In the oblique cases, ille is not used in this way. 100. For the use of ille in another type of text, such as biographic description, see Pinkster (2005a).

 Constituent order in Classical Latin prose

(35) [Acceptis mandatis Roscius cum L. Caesare Capuam pervenit ibique consules Pompeiumque invenit; postulata Caesaris renuntiat.] Illi deliberata re respondent... they being discussed matter reply ‘[Having received his instructions, Roscius arrives at Capua with L. Caesar, and there finds the consuls and Pompey, and delivers Caesar’s demands.] Having discussed the matter, they reply...’ (Caes. Civ. 1.10.1–2) The first example concerns a confrontation between Jugurtha and Adherbal.€The verb invadit has as its – maintained – subject Jugurtha; ille, preceded here by the adversative connector at refers to Adherbal and marks a change of Topic. The second example concerns negotiations between Caesar and his adversaries: illi refers to the consuls and Pompey. The use of ille produces an effect of contrast; it is noticeable that in Latin, changes of Topic are (generally) marked, by ille, or by a proper name, even when, as in (35), there is no ambiguity. The subject of the first clause is in the singular, and the plural subject could be easily deduced from the verb in plural: respondent. When marking a change of Topic, ille stands in initial position, just as we have seen for the Discourse Topics Iugurtha and Caesar (see Section 3.6 above, p. 66 ff.). However, if the sentence has a Sentence Topic or a topical setting element or an adverbial expression, ille comes after it, as in (36). (36) [Hic servo spe libertatis magnisque persuadet praemiis, ut litteras ad Caesarem deferat.] Has ille iaculo inligatas effert... this-acc he on javelin tied on-acc carries out ‘[He persuades his slave, by the hope of freedom, and by great rewards, to convey a letter to Caesar.] This he (the slave) carries out tied on to his javelin...’  (Caes. Gal. 5.45.3) Ille referring to “the other one” or “this one (as opposed to another one)” enters into contrastive structures. This is illustrated in (37) where there is a contrast between illi and vos. In such cases, ille is not necessarily placed in initial position, because it does not mark a change of Topic. (37) Dominari illi volunt, vos liberi esse; facere illi iniurias, to dominate they want you free to be to do they injury-acc vos prohibere. you to prevent ‘They want to dominate, you to be free: they want to do harm, you to prevent it.’  (Sal. Jug. 31.23) Oblique cases of ille have also another use, exemplified in (38). Deserters take valuable objects from the king’s palace (topicalised afterwards by ibi). Illaque is not an instance



Chapter 2.╇ Pragmatic functions 

of topicalisation, but part of the Focus of the sentence, represented by the coordinated constituents illaque et domum et semet; the underlying question is ‘what do they do?’ (38) [Ii... aurum atque argentum et alia, quae prima ducuntur, domum regiam comportant.] Ibi vino et epulis onerati illaque et domum et there with wine and food gorged those-acc=and and palace-acc and semet igni corrumpunt… themselves-acc by fire they destroy ‘[They... carried the gold, silver, and other valuables to the royal palace.] There, gorged with food and wine, they burned the treasure, the palace, and themselves...’  (Sal. Jug. 76.6) The best example offered by my corpus showing ille with Focus function is the following one (39). In the final sentence of The Jugurthine War, ille refers to Marius, who has already been topicalised by means of ei and is. The sentence containing in illo conveys information answering the question ‘who were the hopes placed in?’ and in illo, the Focus of its sentence, is placed in pre-verbal position: (39) [Sed postquam bellum in Numidia confectum et Iugurtham Romam vinctum adduci nuntiatum est, Marius consul absens factus est, et ei decreta prouincia Gallia, isque Kalendis Ianuariis magna gloria consul triumphavit.] Et ea tempestate spes atque opes civitatis and at that time hopes-nom and resources-nom of city in illo sitae. in him rested ‘[But after it was reported that the war in Numidia was concluded and that Jugurtha was being led to Rome in chains, Marius was elected consul in his absence and Gaul was decreed as his province, and on the Kalends of January he triumphed as consul with great glory.] And at that time the hopes and resources of the community rested in him.’ (Sal. Jug. 114.3–4)

4.5

Conclusions

In the preceding Sections (4.1 – 4.4), I have sought to show the difference between anaphors placed in the initial and an internal position. Anaphors placed sentence- or clause-initially pick up a salient constituent of the preceding sentence or clause and provide a point of departure (Sentence Topic) for the subsequent sentence or clause. Anaphors in an internal position refer to an established entity. To topicalise an entity, both animate and inanimate, means to say what happens to it next; to topicalise a locality



Constituent order in Classical Latin prose

is to inform us about what happens there afterwards. The pragmatic status of the element in question thus affects the positioning of anaphoric pronouns and adverbs. There are three main types of resumption of Focus: full resumption of a Focus constituent, resumption of content, and selection of a single constituent. Is and hic are the principal means of topicalisation. The connecting relative qui fulfils the same functions but can also pick up a non-focal constituent. Unlike is, hic fulfils a summarising function, and can also be contrastive and focal. Ille is not used for picking up Focus constituents. In the nominative, its typical functions are marking a change of Topic and (other) contrastive uses. Oblique case forms of ille can bear the Focus function. Anaphoric adverbs, consistent with their semantic value, refer to localities and topicalise, just like the corresponding anaphoric pronouns, salient elements contained in preceding sentences or clauses.

4.6

Personal pronouns

First and second person pronouns101 have a particular status: they refer to the speaker or writer and to the addressee or reader. They are situationally bound elements; their referent is thus always accessible. A distinction has to be made between the uses in the nominative and in oblique cases because of their functional differences. In the nominative, personal pronouns are normally not expressed; they only appear under particular pragmatic conditions that were described by Pinkster (1987), especially€contrast, emphasis and subjectivity.102€In oblique cases, they are normally expressed, but can be omitted by ellipsis. This section will be devoted to the pragmatic aspects of the use of personal pronouns.103 4.6.1 Personal pronouns in the nominative

4.6.1.1 Contrast and emphasis First and second person pronouns used in the nominative often involve an idea of contrast: between the speaker or writer and the addressee or reader or a third person. The typical function of ego is to mark a change of Topic (Pinkster 1987:€374), as illustrated in (40). The contrast can be only partially marked; in (41) ego contrasts not with 101. There is no third person pronoun in Latin, only an anaphoric pronoun (is) and a set of demonstrative pronouns. 102. As for their frequency, ego is expressed in 8% of cases, tu in 16%. Pinkster (1987:€370) also mentions the identifying function of tu in imperative sentences (see Chapter 5, p. 209, Examples (13) and (15)). 103. Traditional analyses talk about “emphatic” uses as opposed to “weak”, “non-emphatic” ones (Szantyr 1972:€173). I will not use this two-fold distinction because it is defined in too vague a way (cf. p. 47, note 32). In addition, I do not consider personal pronouns enclitic (Spevak 2006b). On expressions of ego, see also Adams (1999). Cf. also H. Dik (2003) on personal pronouns in Ancient Greek.



Chapter 2.╇ Pragmatic functions 

a pronoun but with the third person expressed by the verb contemnunt. Note that ego compensates for the ellipsis of the verb (contemno). Contrastive pronouns mostly go in the initial position. (40) Tu id semper facis, quia semper potes; ego in hac causa faciam. you it always do because always can I in this case will do ‘You always do so, because you always can. I will do so in this case.’  (Cic. Quinct. 35) (41) Contemnunt novitatem meam, ego illorum ignaviam. they scorn newness-acc my-acc I their laziness-acc ‘They scorn my humble birth, I their laziness.’ (Sal. Jug. 85.14) Personal pronouns in the nominative may bear the pragmatic function of Focus, as in (42) with tu placed initially with emphasis, and repeated in the subsequent clause. (42) Tu M. Bibulum in contionem, tu augures produxisti. you M. Bibulus-acc in assembly you augurs-acc brought forward ‘You brought forward Marcus Bibulus and the augurs into the assembly.’  (Cic. Dom. 40)

4.6.1.2 Subjectivity The pronoun ego is used when the speaker or writer wants to express his personal judgement, experience, opinion, or involvement (Pinkster 1987:€369). In such cases, the expression of ego is a mark of subjectivity. This is confirmed by the distribution of ego and tu in my corpus 1: whereas ego typically (44% of the total) occurs with verbs of thinking (cogito, puto) and saying (dico, nego), tu does not show such a correlation and is used more frequently with verbs of action. For example, in (43) the presence of ego is justified by the fact that the writer indicates what he intended to do. In (44), there is a personal involvement in the action expressed by the verb: the speaker makes a promise. (43)

[De Herode tibi adsentior. Saufei legisse vellem.] Ego ex Pompeiano VI Id. Mai. cogitabam. I from Pompei on 10th May was thinking ‘[I agree about Herodes. I should like to have read Saufeius’s letter.] I was thinking of leaving Pompeii on the 10th of May.’ (Cic. Att. 14.18.4)

Ego flumen Muluccham, quod inter me et Micipsam I river-acc Muluccha-acc which between me and Micipsa fuit, non egrediar neque id intrare Iugurtham sinam. was neg I will pass and not it come within Jugurtha-acc I will allow ‘The river Muluccha, which was the boundary between Micipsa and me, I shall neither pass myself, nor allow Jugurtha to come within it.’  (Sal. Jug. 110.8) (44)



Constituent order in Classical Latin prose

4.6.1.3 The placement of ego Ego usually stands in sentence-initial position (Adams 1999:€103). This placement is fully expected because ego is a situationally bound element. It can be preceded by a priority constituent: a connecting relative, an emphatic (45) or contrastive (46) constituent, or a Sentence Topic (47). (45) Parvum ego te, Iugurtha,... in meum regnum accepi. little-acc I you Jugurtha-voc in my kingdom accepted ‘When you were little, Jugurtha,... I accepted you into my kingdom...’  (Sal. Jug. 10.1) (46) Rerum ego vitia collegi, non hominum. of things I faults-acc collected neg of men ‘It was the weak points in the situation, not in the individuals, that I reviewed.’  (Cic. Att. 14.14.2) (47) Huic ego litteras ipsius arbitratu dabo. him-dat I letter-acc according to his decision will give ‘I will give him a letter such as he himself thinks fit.’ (Cic. Att. 16.1.6) There are also instances where ego stands after a verb placed sentence-initially (Adams 1999:€105), usually a verb of thinking. I do not consider such an ordering an instance of enclitic placement of ego (pace Adams) but a form of contrast consisting of the rapprochement of two persons, as ego and Quirites are in my Example (48), which is the opening of Marius’s speech.€ Another probable explanation is that starting such a speech with ego would be perceived as impolite. (48)

Scio ego, Quirites, plerosque non isdem artibus know I citizens-voc most men-acc neg by same methods imperium a vobis petere et, postquam adepti sunt, gerere. power-acc from you ask and when have got exercise ‘I know, fellow citizens, that it is not by the same methods that most men ask for power at your hands and exercise it after they have got it.’ (Sal. Jug. 85.1)

4.6.2 Personal pronouns in an oblique case Unlike the nominative forms, personal pronouns in an oblique case are normally expressed.104 I have discussed their properties in detail elsewhere (Spevak 2006b, and forthcoming a) and concluded that there is no real support for claiming an enclitic status for Latin personal pronouns (pace Salvi 2004, among others).105 Where do personal pronouns go then, if there are no restrictions on positioning? Their placement 104. The reflexive pronouns se and sibi also belong to this category. For the reflexive se, see p. 170. 105. In the Tusculan Disputations and in Cicero’s correspondence in my corpus 1, there are 15% of pronouns in the first position, 37% in the second, 25% in the third, 12% in the fourth, and 11% in another position.



Chapter 2.╇ Pragmatic functions 

is variable: in general, contrastive (or emphatic) pronouns stand in the sentence-initial position, whereas non-contrastive pronouns avoid it and occupy an internal position. I will first exemplify the contrastive use (49). (49)

[“Me facere magnam πρ ξιν Dolabellae.”] Mihi mehercule ita videtur, non potuisse maiorem tali re me-dat honestly so seems neg could bigger-acc in such situation talique tempore... Tibi vero adsentior maiorem πρ ξιν at such a time you-dat however I agree greater-acc coup-acc eius fore si mihi quod debuit dissolverit. his will be if me-dat what owed he pays [‘You say I am “making a big thing of Dolabella’s coup”.] Well, I think honestly that in such circumstances and at such a time nothing could have been bigger... However, I agree with you that it will be a greater coup of his if he pays me what he owed.’ (Cic. Att. 14.19.5)

However, the initial position is not exclusive for contrastive pronouns. They can stand anywhere in a sentence. It is important to underscore that Latin has a predilection for bringing constituents referring to people close together,106 especially when there is some particular personal involvement. There is contrast created between Leonides and me (50), but not in (51), where the author does not seem to be personally concerned with Cleopatra’s flight. Contrast is particularly frequent between the speaker and the addressee (52). Note that pronouns me (50) and tibi (52) produce discontinuity of noun phrases. These are instances of contrast and not instances of enclitic placement. In English, non-contrastive expressions are sufficient in all these cases. (50) Leonidae me litterae non satis delectarunt. of Leonides me-acc letter-nom neg enough pleased ‘I did not much like Leonides’s letter.’ (Cic. Att. 14.18.4) (51) Reginae fuga mihi non molesta est. of queen flight-nom me-dat neg troublesome is ‘The queen’s flight does not bother me.’

(Cic. Att. 14.8.1)

(52) Meae tibi litterae non deerunt. my-nom you-dat letters-nom neg will be lacking ‘My letters to you will not be lacking.’

(Cic. Att. 15.27.2)

Non-contrastive (and non-emphatic) pronouns avoid the prominent sentence-initial position and occupy an internal position, in principle the same one as a full noun would occupy. Compare the placement of the pronouns in Examples (53) to (55), with

106. Cf. remarks by Kühner and Stegmann (1914, II: 617), and Herman (1985:€349).



Constituent order in Classical Latin prose

the underlying question ‘what did I send?’, or eventually ‘what did I do?’, but certainly not ‘to whom?’ (53) Epistulam tibi misi. letter-acc you-dat I sent ‘I have sent you the letter.’

(Cic. Att. 15.5.1)

(54) ‘De gloria’ misi tibi. On Glory I sent you-dat ‘I have sent you my On Glory.’

(Cic. Att. 16.2.6)

(55) Librum tibi celeriter mittam ‘De gloria’. book-acc you-dat soon I will send On Glory ‘I will send soon you my book On Glory.’

(Cic. Att. 15.27.2)

When not contrastive, personal pronouns in a sentence-internal position do not have a pragmatic function. This is the case in (56) and (57): idque is the first word of the clause and the Focus is on the action. I consider the variable placement of te as pragmatically insignificant. (56)

Hoc mihi gratius facere nihil potes idque ut than this me-dat more favour do nothing you can it=and that facias te vehementer rogo. (Cic. Fam. 13.74.1) you do you-acc strongly I ask

(57) Hoc mihi gratius facere nihil potes idque ut facias vehementer te rogo.  (Cic. Fam. 13.44.1) ‘You cannot oblige me more than by doing this; I earnestly beg you to do it.’

5. Ellipsis Latin is a language that allows, under certain conditions, ellipsis or omission of any obligatory constituents (subject, objects, verb). From a pragmatic point of view, ellipsis signals that the omitted constituent has no pragmatic function in the given sentence. This section will be devoted to ellipsis of the verb, the first argument (subject) and the second argument (direct object).

5.1

Ellipsis of the verb

Traditionally, ellipsis of the verb is thought of as a way of being economical (Szantyr 1972:€424; Kühner and Stegmann 1914, II: 551 f.). Grammars usually list various verbs of action, particularly verbs of movement, saying, writing, and the verb ‘to be’. I, however, will aim to show the pragmatic implications of this phenomenon using examples collected from Cicero’s correspondence.



Chapter 2.╇ Pragmatic functions 

(1) De meo itinere variae sententiae; multi enim ad me. about my journey various opinions-nom many-nom indeed to me ‘Views about my journey vary; for many people came to see me.’  (Cic. Att. 15.25) (2) Ego paulisper hic, deinde in Tusculanum. I for a short while here then to Tusculum ‘I am here for a short while, then to my place at Tusculum.’(Cic. Att. 13.52.2) (3) Ego igitur ad eum litteras. I so to him letter-acc ‘So I am writing to him.’

(Cic. Att. 15.4a.5)

In these examples of ellipsis the omitted verbs are ones meaning ‘to be’ and ‘to come’ in (1), ‘to stay’ and ‘to go’ in (2), and ‘to send’ (or ‘to write’) in (3). Omission of the verb is possible because it is not the most informative element. Thus, the above-quoted sentences can be interpreted as answering the questions ‘what?’ and ‘who?’ in (1) with the Focus on subjects, ‘where?’ in (2) with the Focus on the locative and directional complements, and on the direct object in (3). Furthermore, Panhuis (1980) examined non-expression of identical verbs in coordinated clauses. His conclusions are that Latin allows both rightwards and leftwards gapping, depending on the structure of the sentence. When there is a sentence-final verb, gapping (Ø) appears leftwards, as illustrated in (4); on the other hand, in a sentence without a final verb, rightwards gapping takes place (5).107 In both cases, the omitted verbs are not the Focus of their clause. (4)

Gallos ab Aquitanis Garunna flumen Ø, a Belgis Gauls-acc from Aquitani Garunna river-nom from Belgae Matrona et Sequana dividit. Matrona-nom and Sequana-nom separates ‘The river Garonne separates the Gauls from the Aquitani; the Marne and the Seine separate them from the Belgae.’ (Caes. Gal. 1.1.2)

(5)

Iam intelleges multo me vigilare acrius ad salutem now you will know more me-acc watch sharply for safety quam te Ø ad perniciem rei publicae. than you-acc for ruin of State ‘Now you will know that I watch much more vigilantly for the safety of the State than you for its ruin.’ (Cic. Cat. 1.8)

107. According to Panhuis (1980), the proportion is 62,5% for leftwards gapping, and 37,5% for rightwards gapping.



Constituent order in Classical Latin prose

5.2

Non-expression of the first argument

Latin is a so-called subject pro-drop language: the subject need not be expressed in every sentence when it is accessible. In this section, I will examine in detail the conditions under which the subject is not expressed, and other associated matters. In Section 3.2, p. 61 devoted to the referential chain, we have seen that a subject co-referential with a newly introduced entity must be expressed and that zero anaphora is not admitted (de Jong 1989:€532).108 On the other hand, a subject co-referential with an established Discourse Topic is normally not repeated; it remains unexpressed, and stays so until a change is marked by a proper or common noun, for example rex ‘king’ or consul, or by the pronoun ille (see Section 4.4., p. 89).109 Intervening material such as descriptive sentences (mostly using the imperfect tense), explicative sentences marked by enim or nam, sentences with an abstract subject, passive constructions (6), indirect speech or subordinate clauses do not interfere with the maintained subject. In the following example, double forward slashes mark intervening material: (6) ...imperat militibus Caesar, ut naves faciant... orders soldiers-dat Caesar-nom that ships-acc they make [//Carinae ac prima statumina ex levi materia fiebant; reliquum corpus navium viminibus contextum coriis integebatur.//] Has perfectas... Ø devehit... these-acc finished-acc he conveys ‘... Caesar orders the soldiers to build ships... [//The keel and the main frames were made of light timber; the rest of the hull of the ships was woven from osier and covered with skins.//] When they are finished, he conveys...’  (Caes. Civ. 1.54.1–3) It is possible – although not very frequent in historical narrative – to promote a constituent expressed in the preceding sentence to the unexpressed subject of the subsequent sentence. Authors have recourse to this strategy when no ambiguity threatens, for example when the subjects differ in grammatical number, as in (7). (7) (Curio) e captivis quaerit, quis castris ad Bagradam from captives inquires who camp-dat at Bagrada praesit: Ø respondent Saburram. commands they reply Saburra-acc 108. It is worth noting that zero anaphora is excluded under the same conditions in Slavic languages of the eastern group (for example in Russian); the western group (such as Polish and Czech) admits it even for the second reference (Siewierska and Uhlířová 1998:€106). 109. According to Bolkestein and van de Grift (1994:€ 293), zero anaphora is the preferred means of continuation (45%) in the case of identical subjects; then come hic (27%), is/ille (10%), and a noun (8%).



Chapter 2.╇ Pragmatic functions 

[Reliqua studio itineris conficiendi quaerere praetermittit...] ‘(Curio) He inquires of the captives, who is in command of the camp at the Bagrada. They reply, “Saburra”. [In his eagerness to complete his journey, he omits other questions...]’ (Caes. Civ. 2.39.1–2) The verb quaerit has as its (maintained) subject Curio; the subject of respondent is abstracted from e captivis, a constituent expressed in the preceding sentence. Praetermittit refers back again to Curio. We can also find an expressed subject in situations where no ambiguity would be produced. This is the case in (8): the interposing of a plural subject (nonnulli) would not confuse the reader’s identification of the subject of habuit. Nevertheless, ille is used for reference to Caesar in order to underline his generosity. This is fully understandable, for we have already seen that Latin has a predilection for every kind of contrastive expression. Caesar... milites adversariorum... conquiri et Caesar-nom soldiers-acc of opponents be sought for and remitti iubet. Sed ex numero tribunorum militum sent back orders but from number of junior officers centurionumque nonnulli sua voluntate apud eum centurions=and several-nom of their own will with him remanserunt. Quos ille postea magno in honore habuit. stayed these-acc he afterwards with great respect treated ‘Caesar gives orders that the men of the other side... should be sought for and sent back. But some of the groups of junior officers and centurions stayed with him of their own accord. These he afterwards treated with great respect.’  (Caes. Civ. 1.77.1–2)

(8)

5.3

Ellipsis of the second argument

Latin allows the omission of the second and third argument of bivalent and trivalent verbs.110 Recently, two studies have been devoted to the omission of the direct object: Luraghi (1997) sought to determine the syntactic conditions for this phenomenon, and Sznajder (1998) provided us with a global survey based on a well-defined corpus. I will concentrate only on second arguments, assuming that the conditions for omission are also valid for third arguments, and I will pay special attention to the pragmatic role of the unexpressed object in the subsequent sentence. The ellipse of a direct object is sometimes difficult to distinguish from the absolute use of a verb. Sznajder (1998:€792), with reference to Touratier (1987), defines the absolute use of verbs in the following way: “we have to do with absolute use of the verb when the unexpressed object is interchangeable with whatever element belonging to a 110. This fact does not change the valency of the verb (Touratier 1987 and 1991:€111). On transitive verbs used without object from a diachronical perspective, see Johnson (1991).

 Constituent order in Classical Latin prose

set or a sub-set of possible objects associable to the verb; a precise or unique object cannot be restored”. On the other hand, “we have to do with ellipsis when the unexpressed object is a precise or unique object that could be restored with the help of the context, both linguistic and extra-linguistic”. The difference between absolute uses and ellipses appears very well in the following examples of the verb scribo ‘to write’. In (9), no precise object can be restored because the verb concerns the mere action of writing. One could paraphrase it as ‘he manifested himself by writing to me’, and not: ‘he wrote me something’. By contrast, the object is absent (zero anaphora) with scripseram (and also delere) in (10) because a precise object is meant, i. e. words that had been written.111 (9)

Theophanes quid velit nescio; scripserat enim Theophanes-nom what wants I don’t know he had written indeed ad me. cui rescripsi ut potui. to me him-dat I answered as I could ‘I don’t know what Theophanes wants; he had already written to me. I answered him as best I could.’ (Cic. Att. 15.19.1)

(10) Rubeo, mihi crede, sed iam Ø scripseram; delere nolui. I blush me-dat believe but already I had written erase I didn’t want ‘I am blushing, believe me; but I had already written it, and did not want to erase it.’ (Cic. Att. 15.4.3) 5.3.1 Zero anaphora in complex sentences According to Sznajder (1998:€799), instances of non-expression of the second argument can be classified in two groups: one in which the unexpressed object is present in the preceding sentence or clause and one in which it is not expressed but can be deduced from it. An examination of instances in my corpus confirms Sznajder’s (1998:€801) findings: almost half the ellipses occur within a compound sentence that exhibits the coordination of two clauses by means of -que, atque (ac) ‘and’, neque ‘and not’, or sed ‘but’. The absence of the second argument (Ø) occurs rightwards in the sentence and concerns the second verb: second argument > verb > coordinator > Ø > verb This pattern is exemplified in (11). The other solution, non-expression of the second argument leftwards, is presented in (12). (11) Milites... acta edocent atque Ø audiunt. soldiers-nom deeds-acc relate and listen ‘The soldiers... relate their deeds, and listen (to those of others).’  (Sal. Jug. 53.8) 111. In my corpus of Classical Latin prose, ellipses of the direct object with a bivalent verb represent 15% of all instances. There are variations between authors, ranging from 8% in Sallust up to 26% in Cicero’s correspondence.



Chapter 2.╇ Pragmatic functions 

(12)

Ad haec bona me si revocas, Epicure, Ø pareo, Ø to these good things me if you recall Epicurus-voc I obey sequor, utor te ipso duce. I follow I take you-abl as only guide ‘If you recall me to good things like this, Epicurus, I obey, I follow, I take you as my only guide.’ (Cic. Tusc. 3.37)

It is time now to examine the contextual and informational status of unexpressed direct objects. Generally, the conditions for the expression or non-expression of direct objects can be brought together with the concept of referential chains and the nonexpression of the subject (see p. 61 ff and p. 98). New, contextually independent entities are topicalised by anaphoric means when they play an important role in the subsequent context, and with these entities it is only starting from the third time they are referred to that they can remain implicit as objects of a verb. An example is given in (13) with an inanimate entity; this treatment also holds true for animate objects. (13) [His perfectis conlocatisque alias deinceps pari magnitudine rates iungebat. Has terra atque aggere integebat ne aditus atque incursus ad defendendum impediretur.] A fronte atque ab utroque latere cratibus ac pluteis at front and on both sides with wickerwork and penthouses Ø protegebat. he protected ‘[When these had been completed and put in place, he joined to them other pontoons of equal size, covering them with earth and filling material so that it would not be difficult to move rapidly on to them to defend them.] At the front and on both sides he protected them with wickerwork and penthouses.’  (Caes. Civ. 1.25.8–9) On the other hand, contextually given entities of any kind, be it a Discourse Topic or merely some item that has been mentioned in the preceding context, can be unexpressed straight away. For example, a Discourse Topic like Marius in (14) need not be pronominalised and the direct object can be omitted starting from the verb edocet; these clauses focus on the action of the Ligurian with the underlying question ‘what does he do?’, and not ‘who does he come to?’112 (14) (Ligus) Itaque Marium propere adit, acta Ø so Marius-acc quickly comes to deeds-acc edocet, Ø hortatur... castellum temptet, Ø pollicetur relates urges fort-acc that he attack promises 112. It would be possible to express this and to topicalise Marius (eum... edocet), cf. Example (19) quoted below p. 103. The absence of topicalisation helps us to interpret the sentence as focussing on the actions of the Ligurian.

 Constituent order in Classical Latin prose

sese itineris periculique ducem. refl of route of danger=and as guide ‘(Ligus) He then hastened to Marius, acquainted him with what he had done, and urged him to attack the fort..., and offered himself to lead the way and the attempt.’ (Sal. Jug. 93.6–7) Absence of a direct object thus mainly concerns contextually dependent entities; however, contextually independent ones can be involved as well, as in (11) with acta edocent, quoted above. They have no pragmatic function afterwards. 5.3.2 Zero anaphora overlapping a sentence Ellipsis of direct objects is not restricted to compound sentences: it can also occur across the boundary of two independent sentences, as illustrated in (15) with a contextually bound constituent (Marcus Antonius). The sentence without an explicit object focuses on the action of D. Brutus; if Mark Anthony were topicalised, the information would be ‘what does he do with him?’ Example (16) shows the ellipsis of a contextually independent referent, the book De gloria. Here again, only the action (custodies) as such is envisaged.113 (15) Quid igitur D. Brutus, de M. Antonio iudicavit? what then D. Brutus-nom about M. Antonius decided Ø excludit provincia, exercitu Ø obsistit. he shuts out province-abl with army he resists ‘What then is the judgment of Decimus Brutus on Mark Anthony? He shuts him out of his province, with an army he resists him.’ (Cic. Phil. 4.8) (16) ‘De gloria’ misi tibi. Ø Custodies igitur, ut soles... On Glory I sent you-dat you will keep then as you are used to ‘I sent you my On Glory. Please keep it as usual....’ (Cic. Att. 16.2.6) This point, i. e. concentration on the action itself, is well illustrated in (17). The ellipsis is of a pronoun that would pick up the preceding infinitive clause. The interrogative sentence focuses on the verb (das-ne) and that the minimal answer is given by repeating the word that was questioned (do). (17)

Hoc dasne aut manere animos post mortem aut this you accept=q either remain souls-acc after death or morte ipsa interire? – Ø Do vero. at death itself perish I accept indeed ‘Do you accept that either souls survive after death, or they perish at death itself? – Yes, I accept that.’ (Cic. Tusc. 1.25)

113. I would interpret the following instance in this sense: Milites imperat: mittunt. (Caes. Civ. 1.15.3) ‘He requisitions soldiers; they send them’, i. e. they do so. Luraghi (1997:€245) holds that the ellipsis concerns here a pronominal indefinite object and she translates as ‘they send some’; I do not find any evidence for this.



Chapter 2.╇ Pragmatic functions 

5.3.3 Zero, or pronominal anaphora? The case of coordination The conditions for the use of zero anaphora presented in preceding Section call for additional verification. My claim is that zero and pronominal anaphora are not in free variation, but are pragmatically motivated. Such a statement is in contradiction with the generally shared view on the case of coordination: “when two coordinated verbs share the same object, Latin uses ellipsis and not pronominal anaphora”, Sznajder says (1998:€800). Also Luraghi (1997:€255) arrives at the conclusion that the non-expression of the object is obligatory in the case of coordination, with reference to Szantyr (1972:€824 f.) who describes the use of anaphoric pronouns as emphatic.114 However, compare Examples (18) and (19). (18) (P. Scipio)... Iugurtham in praetorium abduxit ibique Jugurtha-acc into tent led there=and secreto Ø monuit, ut… privately advised to ‘P. Scipio led Jugurtha... into his tent and there, he privately advised him to...’  (Sal. Jug. 8.2) (19)

Elephantis et parti copiarum pedestrium elephants-dat and part-dat of forces of infantry Bomilcarem praefecit eumque edocet quae ageret. Bomilcar-acc he put in command him=and he instructs what he should do ‘He put Bomilcar in command of the elephants and of part of the infantry, and explained to him what to do.’ (Sal. Jug. 49.1)

These are instances of two coordinated verbs sharing the same direct object, Iugurtham in (18) and Bomilcarem in (19). I argue that there is a difference between the two instances and that it is due to the informational status of the antecedents. Iugurtha is an established Discourse Topic that allows zero anaphora, and is not given as salient in its sentence (Focus is on in praetorium ‘into his tent’). Bomilcar, on the other hand, is treated as a new, salient element, important for further events, and hence, pronominalised by eum. It is important to verify whether this difference applies systematically and what is the proportion between zero and pronominal anaphora. Instances collected from corpus 1 are indicated in Table 5; they concern complex sentences presenting coordination by et, atque, ac, -que ‘and’, and nec, neque ‘and not’. Only syntactically independent clauses, not subordinate ones, were taken into consideration. Instances of zero anaphora correspond to the type in (18) above whereas pronominal anaphora is of the type exemplified by (19) above.

114. It is worth mentioning that Szantyr (1972:€824 f.) treats ellipsis in a chapter on stylistics, and that for him, the use of is, rare in Cicero, has a colloquial value.

 Constituent order in Classical Latin prose

Table 5.╇ Zero anaphora versus pronominal anaphora (corpus 1)€ Author Sallust Caesar Cicero Total Percentage Total without quasi-synonyms Percent. without quasi-synonyms

Zero anaphora 14 15 11 40 68% 29 60%

Pronominal anaphora Total â•⁄ 5 10 â•⁄ 4 19 32% 10 40%

19 25 15 59 100% 39 100%

As we can see, both types appear in Caesar’s and Sallust’s historical narratives. Cicero offers only a few occurrences of this type of coordination in his three works together. In general, zero anaphora (68%) is more frequent than pronominal anaphora (32%). However, there are number of recurring expressions involving quasi-synonymous pairs of verbs, such as oro obtestorque ‘I beg and entreat’ that never present pronominal anaphora. After eliminating these, the proportion for zero and pronominal anaphora is 60% and 40%. In any case we find a significant number of instances of pronominal anaphora. The tendencies observed in this corpus may be summarised in the four points below. 1. When the second argument that the coordinated verbs share is expressed by a pronoun (quem, has, id...), the pronoun is not repeated in the next clause. This pattern typically occurs when a former salient constituent is picked up, for example: (20) (Sex naves) Has superioribus diebus refecerat atque these-acc during preceding days he repaired and omnibus rebus Ø instruxerat. with all things he equipped ‘(Six ships) He had repaired and fully equipped them during the preceding days.’  (Caes. Civ. 2.5.1) 2. Coordinated quasi-synonymous verbs such as orat atque obsecrat ‘he begs and beseeches him’ (Caes. Civ. 1.22.3) or fugant funduntque ‘they routed them and put to flight’ (Sal. Jug. 21.2) have no pronoun with the second verb. Zero anaphora is fully justified because in this case only verbs, not clauses, are coordinated (cf. Pinkster 1990:€8).115 This point allows us to refine Sznajder’s (1998:€800) typology, which takes no account of such a case. In addition, coordinated successive actions, for example castigat atque incitat (Caes. Civ. 1.3.1) ‘he reproves and

115. These pairs are mainly coordinated by means of -que; atque and ac are possible, et is infrequent.



Chapter 2.╇ Pragmatic functions 

stimulates’, and similar ones, for example in (21), also represent instances of coordination of two verbs: (21) Quare primum maximas gratias et ago et therefore first great thanks-acc both I express and Ø habeo Pisoni. acknowledge to Piso ‘Therefore in the first place, I both feel and acknowledge great obligations to Piso.’  (Cic. Phil. 1.15) 3. Zero anaphora is linked with non-topicalisation of the entity in question, both animate and inanimate. When zero anaphora is used, the entity is not assigned Topic function in the subsequent context, as is shown in (22): (22)

Quo facto conventus civium Romanorum... Antonium after this event assembly-nom of citizens Roman Antonius-acc recepit omnibusque rebus Ø iuvit. Otacilius... admitted with all=and things assisted Otacilius-nom ‘After this event, the corporation of Roman citizens... admitted Antonius and assisted him in every way. Otacilius...’ (Caes. Civ. 3.29.1)

On the other hand, there is pragmatic function assignment for signum ‘statue’ in (23); idque... collocarat answers the question ‘what happened to this statue next?’ This entity persists in the discourse afterwards (hoc signum).116 Tigna bipedalia ‘two-foot beams’ in (24) has a shorter life in the subsequent context, but the underlying question associated with this sentence is the same (‘what happened next to the beams?’). (23)

Q. Marcius censor signum Concordiae fecerat idque Q. Marcius censor-nom statue-acc of Concord had made it=and in publico conlocarat. Hoc signum C. Cassius censor in public had placed this statue-acc G. Cassius censor-nom cum in curiam transtulisset... when in court had transported ‘Quintus Marcius the censor had made a statue of Concord, and had erected it in a public place. When Gaius Cassius the censor had transported it into the Senate-house...’ (Cic. Dom. 130)

(24) Eo super tigna bipedalia iniciunt eaque laminis these over beams two-foot-acc they lay them=and with plates

116. In my corpus 1, the type persisting in the discourse (23) is more frequent (13 occurrences out of 19) than the other one (24).

 Constituent order in Classical Latin prose

clavisque religant. bolts=and fasten ‘Over these (rafters) they lay two-foot beams and fasten them with plates and bolts.’ (Caes. Civ. 2.10.3) 4. Pronominal anaphora also concerns the whole contents of a sentence; in this case, the pronoun id or a noun phrase such as ea res ‘this thing’ (25) is used. Note that the subsequent clause (infensi...) brings information that is contrary to expectations. (25)

5.4

Deinde signo dato undique simul clamor ingens oritur then at given signal on all sides at once shout huge-nom arises neque ea res Numidas terret; and not this thing-nom Numidians-acc terrifies [infensi intentique sine tumultu manent, proelium incipitur.] ‘Then, at a given signal, a loud shout arose simultaneously on all sides; and yet this did not terrify the Numidians; [in ferocious concentration they remained unpertubed, the battle began].’ (Sal. Jug. 57.3)

Conclusions

Ellipsis of the verb or of an obligatory argument signals that the constituent in question fulfils no pragmatic function. Conditions for null subjects and null objects are roughly the same: ellipsis is not allowed for new discourse participants that play a role afterwards. In Latin the subject of a verb, once given, remains the subject of subsequent verbs without having to be repeated or picked up by a pronoun, until there is a change of subject. Pronominalisation of objects in the subsequent sentences has to do with Topic function assignment. The choice between zero and pronominal anaphora is thus pragmatically motivated.



Chapter 2.╇ Pragmatic functions 

6. Theme 6.1

Definition. Problems of identification

Theme117 is one of the pragmatic functions described by Functional Grammar.118 The role of Theme constituents is to specify an entity with respect to which the subsequent clause is going to present some relevant information. From a syntactic point of view, these constituents do not belong to the clause: they are typically extra-clausal, leftdisplaced constituents.119 (1) a. That guy, is he a friend of yours? b. As for the students, they won’t be invited. By using an extra-clausal constituent such as that guy€in (1a) and as for the students in (1b), speakers signal what they are going to produce some information about, in other words, what is the domain of applicability of the content of the subsequent clause.120 In spoken languages, Theme constituents are followed by an intonation break and are referred to with an anaphoric expression (he, they in examples above). In Latin, identification of Theme constituents poses a problem, mainly because we do not know the prosodic segmentation of sentences. I will consider non-integration in a clause as a criterion of identification for Theme constituents and in this section, I will argue that integration and nonintegration are what distinguish Topic constituents from Theme constituents: the first belong syntactically to the clause whereas the second stand outside it.

6.2

Typology of Theme constituents

As was pointed out by Somers (1994:€161), Theme constituents do not appear in every type of text. They are found in Plautus’s comedies, Cicero’s correspondence and in Cato’s treatise. They characterise everyday speech121 as well as instructional texts that progress point by point. A good typology of Theme constituents in Latin was presented by Hoffmann (1989:€188 ff.). I will concentrate on only two types, exemplified in (2) and (3). According to Somers (1994:€151), prepositional phrases with de represent the most frequent type of Theme constituents. 117. The concept of theme described by Functional Grammar is different from Firbas’s (1992) concept of theme (see Introduction, p. 4). 118. Dik (1997, II: 389), cf. Bolkestein (1981:€65 f.) and Pinkster (1995:€46 and 222 f.). See also Salvi (2004:€49 f.). 119. Cf. Touratier (1994:€709), Toth (1994:€188), and Amacker (2001). Cf. also Chausserie-Laprée (1969:€39 and passim) on initial complex sequences and on “mots-titres”. 120. Thus it is not a ‘movement’ leftwards, see Bolkestein (1981:€67). 121. Note that the English examples in (1a) and (1b) are of the same nature.

 Constituent order in Classical Latin prose

(2) Amicos domini, eos habeat sibi amicos. friends-acc of master them-acc he should have for himself friends-acc ‘Master’s friends, he must consider them his own friends.’ (Cato Agr. 5.3) (3) De domo et Curionis oratione, ut scribis ita est. as for house and of Curio speech as you write so is ‘As for my house and Curio’s speech, it is just as you say.’

(Cic. Att. 3.20.2)

Theme constituents amicos domini ‘his master’s friends’ (2) and de domo et Curionis oratione ‘concerning my house and Curio’s speech’ (3) express what the subsequent sentence relates to. They correspond to the speaker’s or writer’s intention that could be paraphrased in the following way: ‘now, I will tell you something about X; it is...’ Theme constituents – those not introduced by de – can be referred to with an anaphoric pronoun (eos in 2).

6.3

Theme versus Topic constituents

Although Theme constituents have their specific function, permitting introduction of new subject matter, in a number of cases it is difficult to distinguish them from Topic constituents, as has been pointed out by Bolkestein (1981:€ 67 f.), Somers (1994), Hoffmann (1989)€ and Cabrillana (1999). The difference between Theme and Topic constituents is clear in the following French Example (4), borrowed from Somers (1994:€152). The Theme constituent (quant à Paris) concerns the whole content expressed by the subsequent sentence whereas the Topic constituent (la tour Eiffel) only relates to the predicate (est spectaculaire). (4) Quant à Paris, la tour Eiffel est vraiment spectaculaire. ‘As for Paris, the Eiffel Tower is really spectacular.’ As for the nature of the information conveyed by Theme constituents, Hoffmann (1989) argues that they do not bring completely new information. I would specify that Theme constituents may contain both elements deducible from the preceding context, because they form a part of an enumeration,122 and elements belonging to shared knowledge. Thus, pater tuus ‘your father’ in (5) is deducible from the context because mater ‘mother’ has been mentioned before. Signa ‘the statues’ (6), determined by a relative clause with the explicit mention of the addressee, belongs to the knowledge shared between the writer and the addressee. In both cases, these constituents must be extra-clausal, because they are referred to with anaphoric pronouns in the main clause (cf. Cabrillana 1999:€423 f.). Is and ea function as the Topics of their respective sentences. (5) (mater) Pater tuos, is erat frater patruelis meus. father your-nom he was brother paternal uncle’s my-nom ‘(mother) Your father, he was the son of my father’s brother.’ (Pl. Poen. 1069) 122. Cf. also Pinkster (2005c) on Pliny the Elder.



Chapter 2.╇ Pragmatic functions 

(6) Signa quae nobis curasti, ea sunt statues-nom which for us you obtained these-nom be-aux ad Caietam exposita. at Caieta landed ‘The statues which you obtained for me have been landed at Caieta.’  (Cic. Att. 1.3.2)123 However, unlike Hoffmann (1989:€192), I do not consider a constituent like Caesari in (7) a Theme. (7)

Caesari cum id nuntiatum esset eos per provinciam Caesar-dat when it had been reported they-acc through province nostram iter facere conari, maturat ab urbe proficisci... our way-acc make try he hastens from town set out ‘When it had been reported to Caesar that they were trying to make their way through our province, he hastens to set out from the city...’ (Caes. Gal. 1.7.1)

Although placed in sentence-initial position, Caesari forms a part of the subordinate clause (it is the indirect object of nuntiatum esset) and is by no means extra-clausal. Caesari functions as a contrastive Topic, marking a change of Topic. Furthermore, Caesar is a well-established Discourse Topic; Theme constituents usually do not have such a status in the discourse.124

6.4

Theme versus prepositional phrases with de

Theme can be expressed by a prepositional phrase with de; the best example I know is found in (8), also quoted by Somers (1994:€159) and Cabrillana (1999:€418). (8)

De Tadiana re, mecum Tadius locutus est as for matter of Tadius with me Tadius-nom spoke be-aux te ita scripsisse... you so wrote ‘In the matter of Tadius’s property, he tells me that you have written (to him) that...’ (Cic. Att. 1.5.6)

De Tadiana re ‘as to the business of Tadius’s€is an extra-clausal Theme constituent, and this for two reasons. Theoretically, the verb loquor ‘to talk’ could be completed by a prepositional phrase with de but its second argument is expressed by the infinitive clause (te scripsisse). Furthermore, considering de Tadiana re as a complement of the 123. This example is quoted by Amacker (2001:€189). 124. Theoretically, a well-established character such as Caesar could be selected as a Theme constituent, for example, when the author would enumerate a number of characters including Caesar and talking about what happened to them; he could say: Quod ad Caesarem attinet, … ‘As for Caesar, ...’

 Constituent order in Classical Latin prose

verb loquor would be unacceptable from the semantic point of view: ‘Tadius told me about the business of Tadius’ would make no sense because the person is the same (Cabrillana 1999:€418). When using a Theme constituent, the author starts a new topic, belonging to the knowledge shared by himself and Atticus. As Pinkster (1995:€47, and notes p. 340) rightly observed, Theme constituents expressed by prepositional phrases should not be confused with prepositional arguments of verbs such as scribo ‘to write’ or cogito ‘to think’. These complements fill an argumental position of the verb, and in several cases they alternate with a direct object. They may stand anywhere in the sentence, as is shown in (9) and (10). Both are fully integrated into the sentence. (9)

Post horam VIII in balneum. Tum audivit de Mamurra, after hour 8th to bath then he heard about Mamurra vultum non mutavit. expression-acc neg he changed ‘After two he went to the bath. Then he heard about Mamurra without changing his facial expression.’ (Cic. Att. 13.52.1)

(10) De Antonio iam antea tibi scripsi non esse eum about Antonius already before to you I wrote neg be him-acc a me conventum. by me met-acc ‘As for Antonius, I have told you before that he has not met me.’  (Cic. Att. 15.1.2) Prepositional phrases functioning as Theme, however, must stand in initial position. Unlike the verbs of perception such as audio ‘to hear’ and scribo ‘to write’ quoted in (9) and (10), the verb operam do ‘to try’ in (11) does not need a prepositional phrase with de at all. Puto ‘to think’ can have a de-phrase, but in (12) its argumental position is filled by an infinitive clause. (11) De geographia, dabo operam ut tibi satis faciam. as to geography I will give effort-acc that you I satisfy ‘As to geography, I will try to satisfy you.’ (Cic. Att. 2.4.3) (12)

De Publilio autem, quod perscribi oportet, moram as for Publilius but what be recorded should delay-acc non puto esse faciendam. neg I think be to be made-acc ‘As for Publilius, however, I think there ought to be no delay in paying him the amount for which a draft is due.’ (Cic. Att. 16.2.1)



Chapter 2.╇ Pragmatic functions 

6.5

Subordinate clauses

Some subordinate clauses, especially the formula quod ad me attinet€‘as for me’, can function as Theme constituents (13). Such clauses need not stand sentence-initially, but can also be placed as a parenthesis inside the sentence. (13) Equidem quod ad me attinet, quo me vertam nescio. truly what to me concerns where refl I should turn I don’t know ‘As far as I myself am concerned. I hardly know which way to turn.’  (Cic. Clu. 4) We also find autonomous sentences that serve to articulate the discourse; for example, to announce the transition from one point to another (14). By definition, we cannot regard them as Themes for they do not affect constituent ordering. They belong to the means and strategies for structuring a text.125 (14)

6.6

Redeo enim ad miseram seu nullam potius rem publicam. I return indeed to crushed or non-existent rather republic [M. Antonius ad me scripsit de restitutione Sex. Cloeli.] ‘For now I return to the crushed or rather non-existent republic. [Mark Anthony has written to me about the recall of Sextus Cloelius.]’(Cic. Att. 14.13.6)

Conclusions

Themes are constituents placed leftwards in the sentence. They indicate in what respect the whole sentence is to be interpreted. They typically represent elements deducible from the context or belonging to shared knowledge. Prepositional phrases with de are the most common Theme constituents. Themes are rare in narrative texts; they are used in instructional texts, in correspondence, or in interactive texts such as dialogues.

7. Tail 7.1

Tail constituents

We have seen that a clause can be preceded by a Theme constituent that sets the scene for its interpretation. Similarly, constituents can be added at the end of the clause as a further specification of a constituent that is already contained in the clause (1a), or as a specification of the whole content of the clause (1b). They are a kind of afterthought that specifies, clarifies or corrects something that has already been said. Such constituents that are added to complete the clause are called “Tails” in Functional Grammar 125. Cf. also the structures found by Naas (Le projet encyclopédique de Pline l’Ancien, école française de Rome, 2002), that serve to link of chapters and books.

 Constituent order in Classical Latin prose

(Dik 1997, II: 401 f.).126 These are extra-clausal constituents, placed rightwards, which can be identified by their prosodic contour in spoken languages, and are mainly used in unplanned spoken conversation. For example: (1) a. I didn’t like it very much, that book of yours. b. John gave the book to a girl, in the library. I will start with a discussion of the first category (1a) of Tail constituents. A good example of this type that specifies a constituent contained in the preceding clause is quoted by Panhuis (1982:€85). Here, viginti minae specifies argentum: (2) Hercle te hau sinam emoriri, nisi mi by Hercules you neg I will let die unless me-dat argentum redditur, viginti minae. money-nom is given back twenty minae-nom ‘I’m damned if I’m going to let you die unless I get my money back, twenty minae.’  (Pl. Ps. 1222–1223) As we are dealing with a dead language, the identification of extra-clausal Tail constituents poses problems. Recently Amacker (2001:€196 f.) tried to define them on the basis of a hypothetical – and in my view unfortunate – prosodic structure of Latin sentences. I will comment on two of his examples, concerning subjects in sentence-final position. At127 studuit Catilinae, cum iam aliquot annos esset but supported Catilina-dat when already several years was in foro Caelius; in public life Caelius-nom [et multi hoc idem ex omni ordine atque ex omni aetate fecerunt.] ‘Yes, Caelius did support Catiline, after he had had several years’ training in public life; [and many of all ranks and ages, have done the same].’  (Cic. Cael. 12) (3)

Cicero had already said that Caelius supported Catiline in the preceding section; he repeats this idea once more with insistence. However, Caelius was not the only one to do so, and this is the information that Cicero wants to communicate. Caelius is in this way put together with multi ‘many others’ (I would prefer a comma after Caelius) and contrasted with them. Consequently, it is difficult to see Caelius as a Tail constituent. The second example is even more explicit. Cicero is pleading in favour of Sextus Roscius, who is accused of having killed his father of the same name. A relative of his, Titus Roscius, is also involved in the case:

126. Lambrecht (1994:€203) talks about the “Antitopic”. For Tails in ancient Greek, see H. Dik (2007:€35). 127. In some editions, at is deleted.



Chapter 2.╇ Pragmatic functions 



(4) [Nam cum hic Sextus Roscius esset Ameriae, T. autem iste Roscius Romae, cum hic filius assiduus in praediis esset..., iste autem frequens Romae esset,] occiditur ad balneas Pallacinas rediens a cena Sex. Roscius. is killed at baths Palacine returning from diner Sex. Roscius-nom ‘[For when this Sextus Roscius (son) was at Ameria, but that Titus Roscius at Rome; while the former, the son, was diligently attending to the farm... but the latter was constantly at Rome,] Sextus Roscius (father), returning home from a dinner party, is slain near the Palacine baths.’ (Cic. S. Rosc. 18) The subject of occiditur must be expressed because there is no way of maintaining it (see p. 98); in the preceding context Cicero has been talking about the two Roscii. The complex sentence has a setting (two temporal subordinate clauses with cum) and a main clause that answers the question ‘what happened?’ In such cases, opening a clause with a verb is not unusual (see p. 44), and the subject can be placed in final position. If the main clause started with Sex. Roscius, the underlying question would be ‘what happened to Sex. Roscius?’ but this is not the case. I would even interpret this clause as answering a secondary question ‘who was killed?’ and consider Sex. Roscius as the Focus. The analysis as a Tail constituent is thus, in my view, incorrect. In any case, I propose to use the concept of Tail constituents in Latin rather carefully, and in particular to avoid the mistake of applying it to contextually given entities (often subjects of their clauses) just because they happen to be in sentence-final position. In Classical Latin prose, Tail constituents are not likely to occur except in correspondence, speeches, and dialogues. I have no clear-cut examples of the (1a) type. However, one could consider as Tails appositions, exemplified in (5), that represent a kind of afterthought. (5) Nobis iter est in Asiam, maxime Cyzicum. us-dat way-nom is to Asia in particular to Cyzicus ‘My way lies to Asia, Cyzicus for preference.’

(Cic. Att. 3.6)

The second category of Tails, ones which complete the whole sentence content, as exemplified above in (1b), is impossible to recognise without information about prosodic segmentation. However, this type can be signalled by particles such as et quidem,128 videlicet, scilicet ‘that is’ and other similar expressions. Two examples are given in (6) and (7). (6)

[Hic mecum Balbus, Hirtius, Pansa.] Modo venit Octavius, et quidem in proximam villam Philippi. just came Octavius-nom and that in next house of Philippus ‘[Balbus, Hirtius, and Pansa are here with me.] Octavius has just come to stay, and that, in the very next house, Philippus’s place.’ (Cic. Att. 14.11.2)

128. On et quidem, see Rosén (forthcoming). See also Rosén (2009:€413–416) on “epitaxis”, assymetric coordination.

 Constituent order in Classical Latin prose

(7)

7.2

Sed erant permulti alii ex quibus id facillime scire but were many others-nom from who it very easily know posset, omnes scilicet Lanuvini. he could all-nom for instance Lanuvinians-nom ‘But there were many others from whom he might very easily have found out about it, for instance, all the people of Lanuvium.’ (Cic. Mil. 46)

Tail-head linking

Tail-head linking is the term in Functional Grammar for chaining sentences. This phenomenon is not extensively discussed in Dik (1997, II: 438 ff.), and thus I only retain the following type:

(8) They crossed the river... After crossing the river...

In Latin,129 we have for example: (9)

Caesar... ab urbe proficiscitur atque in ulteriorem Galliam Caesar from city sets out and in Further Gaul pervenit. Quo cum venisset, cognoscit... arrives there when had arrived he learns ‘Caesar... sets out from the city and arrives in Further Gaul. When he gets there, he learns...’ (Caes. Civ. 1.33.4 – 34.1)

Tail-head linking consists in the repetition of content that has been explicitly given in the preceding sentence. Since it involves repetition of several constituents, Tailhead linking is therefore different from what I described as topicalisation (in Section 4.2, p. 75 ff.).

129. For ancient Greek, see H. Dik (1995:€249) and her example Herod. 7.27.1–2.

chapter 3

Declarative sentences The aim of this chapter is to describe the placement of the verb and its obligatory arguments in declarative sentences1 in relation to their pragmatic values. As declarative sentences cover various syntactic and semantic patterns, the valency2 of the verb will serve as a criterion for gathering together verbs that share similar semantic and syntactic properties. The number of obligatory arguments increases the number of positions that have to be occupied in a sentence. For example, a monovalent verb such as nascor ‘to be born’ requires only one argument (the first argument, the subject), whereas the trivalent verb do ‘to give’ needs three arguments in order to build up a meaningful sentence: the subject (first argument), the direct object (second argument) and the indirect object (third argument). A similar methodological approach based on gathering together comparable verbs has already been used by Panhuis (1982), whose analysis only concerns the verbs do ‘to give’ and mitto ‘to send’, and more recently by Devine and Stephens (2006), although from a different interpretative point of view. In this chapter, I will successively examine bivalent transitive verbs (Section 1), trivalent verbs (Section 2), verbs in the passive voice (Section 3), verbs of thinking and saying (Section 4), bivalent intransitive verbs (Section 5), monovalent verbs (Section 6), and finally the verb sum ‘to be’ (Section 7).

1. Bivalent transitive verbs Studies devoted to Latin constituent order concentrate on transitive verbs, and aim to formulate typological theories on this basis. I will not embark on a discussion about Greenbergian typology nor try to determine the so-called “basic order” of the Latin sentence, for it is difficult – indeed impossible – to reduce the complex facts to a simple formula. Furthermore, the typological discussion is mainly oriented towards syntax, and pays attention to the relative ordering of the verb, the subject, and the object. My aim is 1. From a statistical point of view, declarative sentences are the most frequent sentence type. In my corpus 1, they represent 92% of sentences in Sallust, 96% in Caesar, 79% in Cicero’s correspondence, 68% in his Tusculan Disputations and 58% in his speeches. 2. Valency theory was applied to Latin by Happ (1976). For the concept of valency and transitivity, see Touratier (1991:€105). Transitivity is defined as “a syntactic property due to which a verb is constructed with one or more complements”, and valency as “a property requiring a certain number of arguments for building up a meaningful utterance.”

 Constituent order in Classical Latin prose

to describe which pragmatic values correspond to the syntactic patterns one finds. In fact, there is no one-to-one correspondence between syntactic patterns and pragmatic values; in other words, one syntactic pattern can encompass several pragmatic values. In order to explain constituent ordering, one has to consider also some semantic aspects. As we have seen in the Introduction (p. 8 ff.), from a more general point of view, languages can be sensitive to the semantic role and personal hierarchy. If an argument of the verb is the agent of the process, it is more likely to be assigned the syntactic function of subject than another function. The same holds true for patients that are often selected as objects. I reproduce below the eligibility of constituents bearing a particular semantic role for the syntactic function of subject and object, following Siewierska (1988:€47): subject: agent > patient > recipient > benefactive > instrumental > spatial > temporal object: patient > recipient > benefactive > instrumental > spatial > temporal As agents stand higher on the scale of hierarchy than patients, it is understandable that – if no other priority factors apply – they occur earlier in a sentence. The semantic role hierarchy is an important factor to be considered in Latin where sentence-positions do not signal the syntactic function of a constituent. A sentence such as (1) with an agentive, contextually bound subject (L. Tarquinius), and a contextually unbound patient object (primores patrum), is to be interpreted as a result of two interplaying hierarchies: the given > new hierarchy and the agent > patient hierarchy. By contrast, the pragmatic function of Sentence Topic (see p. 65) is the priority factor responsible for the initial placement of hunc in (2) which has the semantic role of patient and yet comes before the agent (Marius). Similarly, Adherbalem and omnes ‘all’ in (3) are semantic patients, but they are contextually given and function as Sentence Topics. Subject function is assigned here to metus ‘fear’, an inanimate contextually independent entity that functions as Focus. The ordering in both sentences (2) and (3) corresponds to the given > new hierarchy. There is no need to introduce syntactic factors in order to understand these patterns. (1) 

Inde L. Tarquinius regnare occepit... primoresque then L. Tarquinius-nom to reign began leading-acc=and patrum... interfecit. of fathers he killed ‘Lucius Tarquinius now began his reign... and killed the leading nobles...’ (Liv. 1.49.1)

(2) (Gauda) Hunc Marius anxium adgreditur atque this-acc Marius-nom worried-acc approaches and hortatur, ut... urges that ‘(Gauda) This was the distressed figure whom Marius approached and urged that...’ (Sal. Jug. 65.3)



Chapter 3.╇ Declarative sentences 

(3) Adherbalem omnesque... metus invadit. Adherbal-acc all-acc=and dread-nom overcomes ‘Adherbal and all who... were overcome with dread.’

(Sal. Jug. 13.1)

In Latin, subjects are formally differentiated from objects by case marking (the nominative and the accusative case respectively). There is one situation in which this distinction is neutralised: in the accusative plus infinitive (AcI) construction. In theory, an ambiguous sentence may result, which Quintilian (Inst. 7.9.6, and 8.2.16) suggests can be avoided by using a passive construction (4). However, when no priority factor applies, the interpretation of the first name Lachetem as agent and the second Demean as patient is obvious due to the agent > patient hierarchy.3 (4)

[Accusativi geminatione facta amphibolia solvitur ablativo, ut illud] Lachetem audivi percussisse Demean fiat Lachetes-acc I heard struck Demeas-acc becomes a Lachete percussum Demean. by Lachetes was struck Demeas-acc ‘[The ambiguity resulting from two accusatives is remedied by using the ablative:] “I heard that Lachetes struck Demeas” becomes “I heard that Demeas was struck by Lachetes.”’ (Quint. Inst. 7.9.10)

For the analysis of transitive sentences, it is important to take into account other aspects that have been discussed by Hopper and Thompson (1980). They define transitivity as a property of a sentence in which “an activity is carried over or transferred from an agent to a patient” (p. 251). This property depends on components such as the nature of the participants (animate or inanimate), their individuation, agency, etc. Transitivity has various degrees; for example, verbs of action convey a higher degree of transitivity than verbs expressing a state. Objects with a high degree of individuation contribute to a high degree of transitivity, i. e. objects that refer to a proper, human, animate, concrete, singular, count, referential and/or definite entity. On the other hand, objects referring to a common, inanimate, abstract, plural, mass, non-referential and/or indefinite entity are less individuated, and they lower the degree of transitivity of the sentence. From this perspective, we can expect differences of placement between highly individuated objects like Adherbal in (5a), and objects with a low degree of individuation, like arms in (b). (5) a. Jugurtha kills Adherbal. b. Romans take arms.

3.

For extensive discussion, see Bortolussi (2007). Cf. also Kühner and Stegmann (1914, II: 699).

 Constituent order in Classical Latin prose

In order to examine the effect of transitivity in this sense, I have selected bivalent4 verbs of action (Section 1.1), and verbo-nominal constructions functioning as pragmatic units (Section 1.2) for a detailed analysis.

1.1

Verbs of action

In this section, I will examine highly transitive sentences containing various verbs admitting an animate or inanimate second argument such as accuso ‘to accuse’, capio ‘to take’, interficio ‘to kill’, and sequor ‘to follow’. 1.1.1 Data Table 1 presents statistical data for verbs of action collected from my corpus 1 containing Cicero’s, Caesar’s and Sallust’s works. A distinction is made between animate and inanimate second arguments. The patterns A1 (first argument, subject), A2 (second argument, object), and V(erb) are indicated in a simplified way: other elements may intervene between them. For example, the pattern {A2 > V} does not necessarily mean that the second argument is placed immediately before the verb. According to these data, there is a remarkable congruence between the ordering {(first argument) > second argument > verb} of animate and inanimate objects (20/20 and 58/59 occurrences). Apart from several instances of second arguments placed post-verbally (found in Cicero’s Tusculan Disputations), we observe that animate objects have a greater mobility (see the last three rows of Table 1) and enter into more Table 1.╇ Verbs of action with animate and inanimate second argument (corpus 1) Pattern A 1 > A2 > V A2 > V A1 > V > A2 V > A2 A2 > A1 > V V > A2 > A1 A2 > V > A1 Total

Animate

Inanimate

Total

20 58 â•⁄ 1 â•⁄ 7 11 â•⁄ 1 â•⁄ 1 99

20 59 â•⁄ 4 11 â•⁄ 2 â•⁄ 0 â•⁄ 0 96

â•⁄ 40 117 â•⁄â•⁄ 5 â•⁄ 18 â•⁄ 13 â•⁄â•⁄ 1 â•⁄â•⁄ 1 195

Percentage 21% 60% 3% 9% 7% 0% 0% 100%

4. Bivalent verbs do not form a semantic group; however, they share several properties, especially the possibility of admitting an argument in the accusative and of being passivized (Touratier 1994:€247). In my analysis, I include also the so-called deponent verbs that have their second argument in the accusative but cannot be passivized. This last property does not seem to affect constituent ordering.



Chapter 3.╇ Declarative sentences 

Table 2.╇ Placement of bound and unbound animate and inanimate second arguments (corpus 1) Position Pre-verbal Post-verbal Other Total Percentage

Animate bound unbound 19 â•⁄ 3 â•⁄ 9 31 42%

16 â•⁄ 3 24 43 58% 100%

Inanimate bound unbound â•⁄ 5 â•⁄ 5 â•⁄ 3 13 15%

41 â•⁄ 5 30 76 85%

Total 81 16 66 163

50% 10% 40% 100%

100%

patterns than inanimate objects, especially in {second argument > first argument > verb} sequences.5 I come now to the examination of the contextual status of second arguments expressed by a full noun (phrase) in order to see whether there is a correlation between contextual dependency and the position in a sentence. In Table 2, three positions are considered: immediately pre-verbal, immediately post-verbal, and elsewhere. These data show several interesting points. The pre-verbal position represents half the occurrences, other positions take 40%, and only 10% of second arguments go in the post-verbal position. Whereas animate second arguments are contextually bound as well as unbound (42 and 58%), inanimate second arguments are typically unbound (85%). This type of argument stands in the immediately pre-verbal or another position. 1.1.2 The pattern first argument > second argument > verb The pattern {(first argument >) second argument > verb} is the most frequently attested one for both animate and inanimate objects. I will now examine which pragmatic arrangements correlate with this pattern, starting with contextually independent constituents. Typical examples are given in (6) and (7). (6) (Metellus) Igitur Bomilcarem, qui Romae cum Iugurtha therefore Bomilcar-acc who at Rome with Jugurtha fuerat... multis pollicitationibus aggreditur... Is had been with many promises he addresses he-nom [ubi primum opportunum fuit, Iugurtham... accedit.] ‘It was therefore Bomilcar – who had been at Rome with Jugurtha – whom he addressed himself with many promises... [As soon as there was an opportunity, the man approached Jugurtha...]’ (Sal. Jug. 61.4 – 62.1)

5.

Animate second arguments appear notably in Caesar and Sallust, inanimate ones in Cicero.

 Constituent order in Classical Latin prose

(7)

[Id ubi vident, mutant consilium:] cupas taeda ac pice refertas incendunt barrels-acc of resin and pitch full-acc they set fire easque de muro in musculum devolvunt. Involutae labuntur... them-acc=and from wall on gallery rolled spun-nom they rolled ‘[Seeing this, they changed their plan:] they set light to barrels full of pitch and pine-shavings and rolled these from the wall on to the gallery. The barrels spun along it and rolled off...’ (Caes. Civ. 2.11.2)

Contextually independent second arguments with a relatively high degree of individuation are strong candidates for Focus function. Although the verb, contextually unbound, conveys some new information (Firbas 1992:€46), it is not salient; Focus function is fulfilled by the second argument. We can also say that the sentence focuses on the object that usually exhibits anaphoric continuation afterwards. Such sentences usually answers the question ‘who, what?’; for example, the Future Topics Bomilcarem, continued by means of is in (6), and cupas ‘barrels’ in (7) by means of eas. Such second arguments seem to be free as far as their position is concerned. Sentences containing contextually independent, especially inanimate, entities with a very low degree of individuation usually do not answer the question ‘what?’, but form a pragmatic unit with the verb. A typical example from historical narrative is arma associated with the verb capio in (8). (8)

Cognita re properantes arma capiunt ac being learnt thing hurrying-nom arms-acc they take and pro castris... consistunt. before camp stand ‘When they realized the situation, they hurriedly seize their arms and stood before the camp...’ (Sal. Jug. 53.1)

The clause containing arma capiunt ‘they seize their arms’ answers the question ‘what do they do?’, and not ‘what do they take’? This interpretation is confirmed by the fact that the whole sentence gives a succession of actions: arma capiunt and consistunt. Unlike the expression castra moveo ‘to move camp’ that will be examined in Section 1.2, p. 125 ff the second argument (arma) cannot be omitted. In pragmatic units such as arma capio ‘to take arms’ or portas aperio ‘to open the doors’ and the like, the object regularly appears in the immediately pre-verbal position and is rarely qualified or determined by a modifier.6 I come now to contextually dependent second arguments, exemplified in (9) – (11).

6. See BTL. For arma, qualified by an adjective, see for example: Quid est enim dementius quam, cum rei publicae perniciosa arma ipse ceperis, obicere alteri salutaria? (Cic. Phil. 2.19) ‘For what is madder when you yourself have taken up arms to destroy the State, than to reproach another for taking them up to save it?’



Chapter 3.╇ Declarative sentences 

(9) At Caesar milites Attianos conlaudat, Pupium dimittit. but Caesar soldiers-acc of Attius praises Pupius-acc dismisses ‘But Caesar praised Attius’s soldiers, dismissed Pupius.’ (Caes. Civ. 1.13.5) (10) Ita... repente magna vi murum adgreditur. so suddenly with great force wall-acc he attacks ‘Then... he suddenly attacked the wall with great violence.’

(Sal. Jug. 60.6)

(11) Iugurtha ex inproviso castra hostium cum magna manu invadit. Jugurtha unexpectedly camp-acc of enemies with large unit attacks ‘Jugurtha unexpectedly attacked the enemy camp with a substantial unit.’  (Sal. Jug. 58.1) Both milites Attianos and Pupius in (9) are known from the preceding context. Murum in (10) has not been mentioned as such but can be inferred from the preceding moenia; the same holds true for castra in (11). The contextual status in combination with an verb of action (conlaudat, dimittit and adgreditur) leads to the conclusion that these sentences answer the underlying question ‘what did he do?’ The object does not necessarily stand in the pre-verbal position, as is shown in (11) where castra comes after inproviso ‘suddenly’ (Torrego 2005). Such an arrangement seems to reflect a frame formed by castra and invadit that delimit the focal part of the sentence. Information is complex in all these cases; the verb and the second argument function together as a pragmatic unit. The pattern {first argument > second argument > verb} is also found with other pragmatic values; I will mention only three of them. (i) Firstly, the subject is contextually independent and conveys salient information (answering the question ‘who, what?’). This is exemplified in (12) with legio XII (note the resumptive anaphoric continuation cum his duabus); the second argument (Caesarem) is contextually dependent, and the verb consequitur is deducible from the context. Indeed, Caesar wishes to form a new army and the inhabitants of Cingulum offer him their help. Milites imperat; mittunt. Interea legio XII soldiers-acc he orders they send meanwhile 12th legion-nom Caesarem consequitur. Cum his duabus Asculum Caesar-acc follows with these two to Ausculum Picenum proficiscitur. in Picenum he sets out ‘He demanded troops, and they sent them. Meanwhile the Twelfth legion caught up with him. With the two he now had he set out for Asculum in Picenum.’  (Caes. Civ. 1.15.3)

(12)

(ii) When conveying new information, the second argument may be contextually given, as is illustrated in (13), which is an instance of enumeration (primo ‘first’, dein ‘afterwards’). (iii) Another pragmatic value is shown in (14) where both arguments are

 Constituent order in Classical Latin prose

contextually dependent: ego and vos are situationally given and put close together by contrast; Focus is on the verb. (13) (Ti. et C. Gracchus...) et primo Tiberium, dein paucos and first Tiberius-acc then after post annos eadem ingredientem Gaium... ferro (nobilitas) necaverat. a few years the same going in-acc Gaius-acc by sword killed ‘(Tiberius and Gaius Gracchus...) And first the nobility executed Tiberius by the sword; then, after a few years, Gaius embarked as he was on the same course...’  (Sal. Jug. 42.1) (14) Neque ego vos, Quirites, hortor, ut malitis... neg I you-acc citizens-voc encourage that you prefer ‘I do not encourage you, citizens, to prefer...’ (Sal. Jug. 31.27) 1.1.3 The pattern first argument > verb > second argument It is quite rare to find objects that follow the verb in Caesar’s or Sallust’s historical narratives. An example is given in (15). The post-verbal placement of exercitum is explained not only by the fact that it is a syntactically heavy constituent (LIPOC rule) but also by the fact that it is contextually independent and functions as Focus. In addition, the qualifications of the army (ampliorem, hebetem...) are picked up by id in the following sentence. ibique cogebat exercitum numero hominum there=and he was collecting army-acc in number of men ampliorem, sed hebetem infirmumque, agri ac pecoris larger-acc but dull-acc weak-acc=and of land and cattle magis quam belli cultorem. Id ea gratia eveniebat... more than of war expert-acc it for this reason happened ‘and there he was collecting an army, which was larger in numbers, but dull and weak, more expert in land and livestock than in war. This came about because...’  (Sal. Jug. 54.3)

(15)

In Cicero, the post-verbal placement of a direct object appears more frequently. I compare now instances of pre-verbal and post-verbal placement: both sentences given below answer the question ‘who?’ and both second arguments, Volcacium and Vettienum, are contextually unbound without an anaphoric continuation. However, there is a Sentence Topic (in qua re) in (16); Vettienum in (17) can be interpreted as contrastive with respect to Sestius. (16) In qua re accusat Volcacium. it this matter he blames Volcacius-acc [Ita mihi videtur bellum illud instare.] ‘He blames Volcacius in the matter. [So it looks to me as though a war is imminent.]’ (Cic. Att. 14.9.3)



Chapter 3.╇ Declarative sentences 

Vettienum accusat (tricatur scilicet ut monetalis), Vettienus-acc he blames he shuffles of course like coiner-nom sed ait totum negotium Sestium nostrum suscepisse... but says whole business-acc Sestius-acc our-acc has taken ‘He blames Vettienus (who is shuffling, of course, in the manner of his profession); but he says that our friend Sestius has taken the whole business upon himself.’  (Cic. Att. 15.13a.1) (17)

The following examples are also worth comparing. The expression exspecto tuas litteras, with a possessive pronoun referring to the addressee, is frequently found at the end of Cicero’s€letters. Different orderings may occur: with post-verbal placement of tuas litteras (18) or with its initial placement (19). The second case can, however, be explained as due to contrast with scribo, which implies meae litterae ‘my letter’. In sum, the clauseor sentence-final placement of such objects corresponds to the given > new hierarchy whereas the clause- or sentence-initial placement covers the cases of contrast. (18)

[Quod censueris faciam. Sed quam primum!] Avide exspecto tuas litteras. eagerly I await your letter-acc ‘[I will do what you advise. But don’t put it off!] I am eagerly awaiting a letter from you.’ (Cic. Att. 16.10.2)

(19)

[Nunc quae scribo, scribo ex opinione hominum atque fama.] Tuas litteras avide exspecto. your letter-acc eagerly I await ‘[As it is, what I write is founded on common opinion and rumour.] I am anxiously longing for a letter from you.’ (Cic. Fam. 12.4.2)

Objects placed in post-verbal position are usually contextually independent. However, it is possible to find contextually bound constituents in this position, as in (20). Aerarium with the definite reading of ‘the treasure’, already mentioned in the cumclause, is not salient. The clause focuses on verbis and answers the question ‘how’? (20)

Et quidem C. Gracchus, cum largitiones maximas indeed G. Gracchus-nom when distributions massive-acc fecisset et effudisset aerarium, verbis tamen had done and had squandered treasury-acc by words still defendebat aerarium. defended treasury-acc ‘Gaius Gracchus, after squandering the state treasury on massive handouts to the public, still defended that same treasury in words.’ (Cic. Tusc. 3.48)

 Constituent order in Classical Latin prose

1.1.4 The pattern second argument > first argument > verb The pattern {second argument > first argument > verb} mainly concerns second arguments that are contextually bound. This arrangement covers instances with an anaphoric pronoun functioning as direct object, that picks up a salient element of the preceding sentence, as does hunc in (21); anxium is predicative. This clause answers the question ‘what happens to him?’ The same pattern is observed for sentences answering the question ‘who?’ The contextually given second argument stands in sentence-initial position; the first argument is in the pre-verbal position in (22). (21) (Gauda) Hunc Marius anxium adgreditur atque hortatur, ut... this-acc Marius-nom worried-acc approaches and urges that ‘(Gauda) This was the distressed figure whom Marius approached and urged that...’ (Sal. Jug. 65.3) (22)

Adherbalem omnesque, qui sub imperio Micipsae Adherbal-acc all-acc=and who under command of Micipsa fuerant, metus invadit. had been dread-nom overcomes ‘Adherbal and all who had been under Micipsa’s command were overcome with dread.’ (Sal. Jug. 13.1)

In addition, metus is without modifier in (22); when it is qualified by an adjective expressing subjective evaluation (like tantus ‘such’ or ingens ‘enormous’), either the whole noun phrase or only the adjective are placed with emphasis in an early position, as in (23). (23)

Quod postquam auditum est, tum vero ingens metus this when was heard then part mighty dread-nom nostros invadit. our men-acc overcomes ‘When they heard this, a mighty dread overcame our men.’ (Sal. Jug. 106.6)

However, a second argument occupying the initial position can be interpreted as Focus (note the anaphoric continuation), as in (24). This placement is explained as ‘Focus-first’ strategy. (24)

Domitium Massiliensesque de suo adventu Domitius-acc Massiliots-acc=and about his arrival facit certiores eosque magnopere hortatur ut... he informs these-acc=and strongly urges that ‘He informs Domitius and the Massiliots of his arrival and urges them most strongly to...’ (Caes. Civ. 2.3.3)



Chapter 3.╇ Declarative sentences 

1.1.5 The pattern verb > second argument > first argument This pattern is extremely rare in my sample of examined verbs (see Table 1). It concerns the well-commented sentence laudat <promptos>7 Pompeius (Caes. Civ. 1.3.1) ‘Pompey praises the zealous’, that exhibits a contextually given subject and Focus on the action (Bolkestein 1996b: 14). However, in his Tusculan Disputations, Cicero offers one instance of this arrangement in a different context. Example (25) opens a demonstration that no sensation remains in the body. There is no question about what Achilles is doing or who he is dragging. It is an all-new sentence whose function is to set a frame that we can paraphrase as: ‘imagine the following situation: Achilles...’ (25) [Sed plena errorum sunt omnia.] Trahit Hectorem ad currum religatum drags Hector-acc to chariot tied-acc Achilles: lacerari eum et sentire, credo, putat. Achilles-nom be mangled him-acc and feel I suppose he thinks ‘[But false ideas abound everywhere.] Achilles drags Hector tied to his chariot: he thinks, I suppose, that Hector is being mangled and is aware of it.’  (Cic. Tusc. 1.105) 1.1.6 Conclusions The examination of verbs of action in highly transitive sentences shows that the most frequent pattern, {(first argument) > second argument > verb}, mainly appears in Caesar’s and Sallust’s historical narrative; it is less frequent in the works of Cicero. This pattern covers several pragmatic values and concerns constituents with different contextual status: bound as well as unbound ones. Objects with a low degree of individuation typically occur in the immediately pre-verbal position and form a pragmatic unit with the verb. Sentences containing them answer the underlying question ‘what does he do?’ Another interesting conclusion is that animate entities manifest more mobility in the sentence than inanimate ones, especially when the sentence focuses on the direct object with the underlying question ‘who, what’?

1.2

Verbo-nominal constructions

It is time now to turn to sentences of low transitivity and to discuss verbo-nominal constructions. These are formed by a verb with a weak semantic value and a noun bearing a low degree of individuation (de Jong 1989:€533 f.); for example bellum gero ‘to wage war’.8 Neither the verb nor the noun is a candidate for Focus; they function 7.

Promptos is added by editors; I did not include this instance in Table 1.

8. Verbo-nominal constructions of the bellum gero ‘to wage war’ type have drawn much attention from scholars adopting the typological perspective. The compound verbs like belligero ‘to wage war’ or nuncupo ‘to denominate’ are considered proofs of Latin {object > verb} order

 Constituent order in Classical Latin prose

together as a pragmatic unit (as dolores habet does in 25). According to the analysis proposed by Pinkster (1991:€77, and 1995:€231), who has especially examined this phenomenon, the sentence exemplified in (26) does not answer the question ‘what has Terentia got?’ but ‘how is Terentia?’ The adjective magnos can be viewed as intensifying the state expressed by the pragmatic unit dolores habet; this analysis is confirmed by the coordination of saepe and maximas9 in (27). (26) Terentia magnos articulorum dolores habet. Terentia-nom great-acc of joints pain-acc has ‘Terentia has a bad attack of rheumatism.’

(Cic. Att. 1.5.8)

(27) Terentia tibi et saepe et maximas agit gratias. Terentia you-dat both often and great-acc gives thanks-acc ‘Terentia thanks you frequently and very warmly.’ (Cic. Att. 3.5) Table 3.╇ Verbo-nominal constructions in Caesar, Sallust, and Cicero (corpus BTL) Construction

NV

N...V

VN

V...N

Total

terga verto castra pono castra moveo iter facio verba facio mentionem facio finem facio operam do gratias ago Total Percentage

â•⁄ 7 13 â•⁄ 6 â•⁄ 6 â•⁄ 9 â•⁄ 4 â•⁄ 9 â•⁄ 5 32 91 50%

â•⁄ 0 â•⁄ 3 â•⁄ 4 â•⁄ 4 â•⁄ 2 â•⁄ 1 â•⁄ 8 â•⁄ 2 11 35 21%

â•⁄ 0 â•⁄ 0 â•⁄ 2 â•⁄ 1 â•⁄ 2 â•⁄ 2 â•⁄ 1 18 17 43 24%

0 0 0 0 0 0 0 2 7 9 5%

â•⁄â•⁄ 7 â•⁄ 16 â•⁄ 12 11 13 â•⁄â•⁄ 7 â•⁄ 18 â•⁄ 27 â•⁄ 67 178 100%

(Oniga 1988:€155, note 22). Similarly, Giannecchini (1982:€93 f.) aimed to explain the transition from the {object > verb} order to {verb > object}. He formulated a hypothesis that the {verb > object} order (or the anticipation of the verb in the sentence) can be viewed as a result of the weak semantic value of the verb and the “dynamic” character of the noun. According to him (Giannecchini 1982: 106–107), the tendency to anticipate the verb is observed in Cicero: the {object > verb} order represents 41% of instances in his works – and thus there are 59% of the instances of the anticipate verb in the order {verb > object}. In Caesar, the {object > verb} order makes up 54%, in Sallust 59%. However, Pinkster (1995:€231, note 27) points out that an examination of expressions such as mentionem facio ‘to mention’ in Cicero does not lead to the same conclusion. Furthermore, I should add that Giannecchini, using the typology of verbo-nominal constructions proposed by Renský (1966:€292 ff.), also includes the forms of the perfect tense of deponent verbs and the passive voice in his corpus. In Renský there is no support for doing so. 9. For coordination of members that belong to different categories, see Pinkster (1990:€9).



Chapter 3.╇ Declarative sentences 

A detailed examination of several verbo-nominal constructions taken from Caesar, Sallust, and Cicero will show the ordering of their components. The selected verbonominal constructions differ in their ability to admit other complements: terga verto ‘to flee’ is monovalent, castra pono ‘to encamp’ and castra moveo ‘to move camp’ as well as iter facio ‘to march’ may be accompanied by a complement of space or direction; verba facio ‘to speak (in an assembly)’ may be completed by an expression indicating the content of the speech; mentionem facio ‘to mention’ and finem facio ‘to terminate’ admit genitive complements; the expression gratias ago ‘to give thanks’ takes a dative complement, and operam do ‘to make effort, to give attention’ appears with a dative or a complement clause.10 The main question to be answered is whether these constructions exhibit a fixed order and how mobile they are in their sentence. I will pay particular attention to the possibility of inverting the order of the components, of inserting other elements between them, and of finding modifiers of the nouns. With the help of BTL, I examined 178 instances. Table 3 shows the results: the placement of the noun (N) and the verb (V) in continuous sequences, or their separation (...) in syntactically independent declarative clauses or sentences. Table 3 brings to light at least two interesting points. The first one is that the ordering {noun > verb} is not exclusive but represents 71% (50 + 21%) of all cases, whereas the ordering {verb > noun} is found in 29% of verbo-nominal expressions.11 The other point is that the more the number of complements increases, the greater the amount of variation that can be observed. For example, monovalent terga verto appears only as {noun > verb} without discontinuity but gratias ago allows more variability, and for operam do the ordering {verb > noun} is even prevalent (20 occ.). These variations, as well as properties of each construction, will be illustrated in what follows. 1.2.1 Terga verto ‘to flee’ The expression terga verto ‘to flee’ (properly ‘to turn one’s back’) appears only as {noun > verb}, which may be partially due to the low number of occurrences in Classical prose (7 in total).12 An example is given in (28), which shows clearly that terga verto 10. For verbo-nominal constructions formed by a semantically weak verb (habeo, facio…) and a verbal noun (mentionem…), see Rosén (1981, in particular 130 ff.). According to her, components of these constructions – as well as other analytic formations – do not manifest a fixed order (Rosén 1999:€87). 11. These results therefore differ noticeably from those presented by Giannecchini (1982:€106 f.) and confirm Pinkster’s (1995) observations; see above, note€8. 12. However, separation produced by an adverb is not excluded, cf. terga haud dubie vertunt (Liv. 28.13.8) ‘they faced an unmistakable retreat.’ I also noticed the order {verb > noun}, for example in Seneca, for an action situated in the future; however here, terga has its full meaning: referet pedem, non vertet terga, sed sensim recedet in tutum (Sen. Ep. 22.8) ‘he will withdraw, not turning his back on the foe, but falling back little by little to a safe position’. I have no example of terga accompanied by a modifier.

 Constituent order in Classical Latin prose

functions as a pragmatic unit; the underlying question is ‘what did the army do?’ and not ‘what did the army turn?’ (28)

Omnis Vari acies terga vertit seque whole-nom of Varus formation-nom back turned refl=and in castra recepit. to camp retreated ‘Varus’s whole formation turned round and retreated to camp.’(Caes. Civ. 2.34.6)

1.2.2 Castra pono ‘to encamp’, castra moveo ‘to move camp’ and iter facio ‘to march’ The expressions castra pono ‘to encamp’, castra moveo ‘to move camp’13 and iter facio ‘to march’ (properly ‘to make one’s journey’) appear with or without an adverbial complement indicating the place, the departure or arrival point, or the time (for example, noctu ‘at night’). Castra and iter are rarely accompanied by a modifier.14 An example is given in (29); the sentence answers the question ‘what do they do?’, the adverb maturius ‘earlier’ has scope over the whole pragmatic unit. However, these expressions do not seem to be fixed or inseparable units: it is possible to insert a constituent between them, for example an adverb or a prepositional phrase, as in (30). (29) Illi necessario maturius quam constituerant castra ponunt. these of necessity earlier than had intended camp-acc pitch ‘Of necessity they pitched camp earlier than they had intended.’( Caes. Civ. 1.65.3) (30)

Caesar... tribus horis noctis exercitui ad quietem datis Caesar three hours-abl of night army-dat to rest given-abl castra ad Gergoviam movit. camp-acc to Gergovia moved ‘Caesar... gave over three hours of the night for his army to rest, then struck camp and moved out to Gergovia.’ (Caes. Gal. 7.41.1)

The less frequent ordering {verb > noun} is found without any perceptible difference of meaning; see example (31) with the underlying question ‘where does he go?’ However, feci iter in (32) with an initial verb can be explained as a verum Focus in the sense of ‘I (really) did’.

13. With the same meaning, the verb moveo can be used without the noun castra;€see Lebreton (1901:€152), Happ (1976:€234 f.) and Touratier (1987:€409). 14. For castra, cf. altera castra… ponit (Caes. Civ. 1.18.5)€‘he forms a second camp...’; for iter, cf. ut illud iter familiarius facere vellent (Cic. Inv. 2.14) ‘that they were ready to make the trip together as close friends’.



Chapter 3.╇ Declarative sentences 

(31)

satis ad... militum animos confirmandos factum enough to of soldiers morale-acc be improved done existimans in Haeduos movit castra thinking-nom to Aedui he moved camp-acc ‘... judging that enough had been done to improve his soldiers’ morale, he struck camp and moved out to the Aedui’ (Caes. Gal. 7.53.3)

(32)

Moeragenes certe periit; feci iter per eius Moeragenes-nom certainly is dead I did route-acc through his possessionem, in qua animal reliquum nullum est. land in which living soul-nom left-nom none is ‘Moeragenes is certainly dead. I marched through his country, and there is not a living thing left.’ (Cic. Att. 6.1.13)

1.2.3 Verba facio ‘to speak’ The expression verba facio ‘to speak’ (properly to ‘make words’) typically announces indirect speech, or is completed by a prepositional phrase introduced by de. Here also, the orders {noun > verb} and {verb > noun} seem to be variants; compare the examples (33) and (34). (33) Ad ea rex satis placide verba facit: sese pacem cupere... to this king-nom enough gently words-acc makes refl peace-acc desire ‘The king’s words in response were calm enough: he desired peace...’( Sal. Jug. 83.2) (34) Pro his Diviciacus... facit verba: Bellovacos… for them Diviciacus-nom makes words-acc Bellovaci-acc ‘On their behalf, Divitiacus... intercedes: The Bellovaci...’ (Caes. Gal. 2.14.1) The pragmatic unit verba facio may be qualified by an adverb, for example placide in (33), or the noun may be accompanied by a modifier. In (35) verba is quantified by multa, expressing subjective evaluation, which therefore stands with emphasis in sentence-initial position. The arrangement in (35) should be regarded as an instance of discontinuity of multa verba (multa stands initially with emphasis) caused by the intervening feci, and not as an instance of postposition of verba with respect to feci. (35) mvlta feci verba de toto furore latrocinioque P. Clodi. many-acc I made words-acc about all frenzy robbery=and of P. Clodius ‘I spoke at great length on all the insane and murderous acts of Publius Clodius.’  (Cic. Q. fr. 2.1.3) 1.2.4 Mentionem facio ‘to mention’ and finem facio ‘to end’ Mentionem facio ‘to make mention’ and finem facio ‘to make end’ are usually completed by a genitive that syntactically depends on the noun (Rosén 1981:€ 142); the genitive can precede or follow the noun and be separated. The noun also admits

 Constituent order in Classical Latin prose

attributes such as emphatic nullus ‘none, not... any’. Focus is usually on the verbonominal expression that behaves as a pragmatic unit (36); note that Tusculanarum disputationum is a contextually dependent, non-focal constituent. (36)

Quintus hic dies, Brute, finem faciet fifth this day-nom Brutus-voc end-acc will make Tusculanarum disputationum. of Tusculan of disputations ‘This fifth day, Brutus, will conclude the discussions at Tusculum.’(Cic. Tusc. 5.1)

1.2.5 Operam do ‘to do one’s best’ The expression operam do ‘to do one’s best’ (properly ‘to give effort’) and ‘to do a favour’ used especially by Cicero, is completed by a dative or a complement clause with ut or ne. In comparison with other constructions (see Table 3), operam do presents frequently (76%, including separations) the ordering {verb > noun}. This pattern is regularly used when a complement clause with ut or ne, as a modifier of operam do, is expressed (37); the subordinate clause conveys salient information. (37) Nunc... dabo operam ne mea valetudo tuo labori desit. now I will give effort-acc neg my health-nom your work-dat fails ‘I shall now do my best to prevent my health from failing to further your efforts.’  (Cic. Fam. 14.1.2) Emphatic modifiers may be separated from their head noun; they go in a privileged sentence-initial or final position, as egregiam in (38). (38) Nicanor tuus operam mihi dat egregiam. Nicanor your-nom service-acc me-dat gives great-acc ‘Your Nicanor is a great help to me.’

(Cic. Att. 5.3.3)

1.2.6 Gratias ago ‘to thank’ Gratias ago ‘to give thanks’ can be expanded by a dative and/or by a subordinate clause with quod. It shows the {noun > verb} (64%) as well as {verb > noun} (36%) ordering and allows the separation by (an)other constituent(s), as illustrated in (39): (39) Ego... egi ei per litteras gratias idque ut I gave him-dat by letter thanks-acc it=and that maturaret hortatus sum. he hasten I urged ‘I expressed my gratitude to him by letter and urged him to hasten.’  (Cic. Fam. 15.4.5) Unlike operam do, the ordering {verb > noun} is not preferred in the case of gratias ago + complement clause. An example is given in (40).



Chapter 3.╇ Declarative sentences 

(40)

Princepsque decima legio per tribunos militum ei first=and tenth legion-nom through military tribunes him-dat gratias egit, quod de se optimum iudicium fecisset. thanks-acc gave that about refl excellent opinion-acc had formed ‘And the Tenth legion was the first to express its thanks to him through the military tribunes for the excellent opinion he had formed of it.’ (Caes. Gal. 1.41.1)

It is also worth quoting an example with {verb > noun}, which exhibits Focus only on the verb, on the coordinated forms egi and agam that share the same object gratias. (41)

Vobis vero, patres conscripti, singulis et egi you-dat part conscript fathers-voc each one-dat both I gave et agam gratias. and will give thanks-acc ‘And to each one of you, conscript fathers, I have already expressed, and shall continue to express, my gratitude.’ (Cic. Red. sen. 30)

1.2.7 Conclusions This brief examination of several verbo-nominal expressions shows that the presence of other complements increases the number of possible orderings. Expressions such as terga verto or castra moveo seem fixed, but in reality they have only a limited chance to show variation, for they are rarely modified or expanded. On the other hand, operam do requires a complement, and thus it enters into more complex arrangements. In relation to their meaning, they appear in sentences with various pragmatic values. The verb often forms a pragmatic unit with its noun, but another constituent (adjective, dative or directional complement) may carry Focus. In sum, it would be inappropriate to try to establish a relative ordering of the verb and the noun of verbo-nominal constructions because their behaviour is in a direct relationship with their syntactic capacities and their semantic properties.

2. Trivalent verbs Unlike other sections in this chapter, the examination of trivalent verbs only concerns single verbs, firstly because they each offer a sufficient number of occurrences and secondly because they do not have the same combinatorial properties. The following verbs will be examined: mitto ‘to send’ (Section 2.1), duco ‘to lead’ (Section 2.2), and do ‘to give’ (Section 2.3).

 Constituent order in Classical Latin prose

2.1

The verb mitto ‘to send’

The verb mitto ‘to send’, in particular in the expression ‘to send a delegation to someone’ is very frequent in historical narrative, and Panhuis (1982:€117 ff.) analysed its pragmatic values in Caesar’s The Gallic War. His analysis is based on the concept of increasing communicative dynamism (theme > rheme, verb). From this perspective, the most informative constituent should be found before the verb, which – in the case of mitto – is not itself sensitive to the communicative dynamism. Some of the instances can be explained in this way (1), but the others (2) cannot. (1)

Caesar... statim nuntium in Bellovacos ad Marcum Caesar at once messenger-acc to Bellovaci to Marcus Crassum quaestorem mittit, cuius hiberna Crassus quaestor sends whose winter camp-nom aberant ab eo milia passuum XXV. Iubet... was about from him miles of paces 25 he orders ‘Caesar... at once sends a messenger to the territory of the Bellovaci, to his quaestor Marcus Crassus, whose winter camp was about 25 miles away. He orders him...’ (Caes. Gal. 5.46.1)

(2) Helvetii... legatos ad eum mittunt. Cuius legationis Helvetii-nom envoys-acc to him send of this embassy Divico princeps fuit, qui... Divico-nom chief was who ‘The Helvetii... send envoys to him. At the head of this embassy was Divico, who...’  (Caes. Gal. 1.13.2) In the first example, (1), the sentence actually answers the question ‘to whom?’ and ad Marcum Crassum quaestorem standing in pre-verbal position represents the most salient element. This constituent is contextually unbound and the relative clause in cuius specifies this person. On the other hand, ad eum in (2) cannot be interpreted in the same way (cf. Pinkster 1995:€236). It is a contextually bound constituent, expressed by an anaphoric pronoun (eum) referring to Caesar, who is a well-established Topic. If ad eum were salient, the sentence should answer the question ‘to whom?’ and, as a bound element, it should imply contrast with another person. This is not the case, for is is never contrastive. Instead, as is suggested by the following context that concerns the head of the delegation, legatos seems to carry the most important information. However, the constituent legatos is not found in the immediately pre-verbal position. 2.1.1 Data The verb mitto ‘to send’ requires three obligatory arguments: a first argument (subject), a second argument (direct object) and a third argument, either in the dative, or a directional



Chapter 3.╇ Declarative sentences 

complement introduced by the preposition ad,15 or represented by an adverb. In Caesar’s and Sallust’s historical narrative, the verb mitto is always used with a directional complement whereas Cicero in his correspondence has both directional (ad te) and dative (tibi) complements. Table 4 shows the patterns offered by my corpus 1. Other elements can intervene between the obligatory constituents. Second arguments refer to persons or to objects in an almost equal proportion, whereas directional complements (D, including dative arguments) mainly refer to persons; references to localities are infrequent. From the statistical point of view, the pattern {(first argument >) second argument > directional complement > verb} is the most frequent (50%). However, it competes with the pattern {(first argument >) directional complement > second argument > verb} (25%), and with other arrangements (25% in total). Many sentences containing the verb mitto inform us that someone sends a delegation, a letter, or a book to someone. The second argument and the directional complement are therefore the strongest candidates for Focus function. As the most informative elements, they should appear rightwards in the sentence or clause, according to the given > new hierarchy. In order to confirm this point, I will examine their position (pre-verbal, post-verbal or other) as well as their contextual dependency and continuation by an anaphoric constituent (+ and –An(aphora)). Table 5 presents the results for directional complements; Table 6 concerns second arguments. In both cases, only nominal (not pronominal) constituents were taken into account. Table€4.╇ The verb mitto (corpus 1) Pattern A1 > A 2 > D > V A2 > D > V A1 > D > A2 > V D > A2 > V A2 > V > D A1 > V > D > A2 V > D > A2 D > A1 > A2 > V A2 > A1 > D > V D > V > A2 Total

Total

Percentage

â•⁄ 6 25 â•⁄ 7 â•⁄ 8 â•⁄ 5 â•⁄ 2 â•⁄ 5 â•⁄ 1 â•⁄ 1 â•⁄ 1 61

50% 25% â•⁄â•⁄ 8% 11% â•⁄â•⁄ 2% â•⁄â•⁄ 2% â•⁄â•⁄ 2% 100%

15. It is worth noting a co-occurrence of two directional complements arranged according to increasing specification (cf. Pinkster 1972:€95); ad flumen Bagradam further specifies ad castra hostium: ...equitatum omnem prima nocte ad castra hostium mittit ad flumen Bagradam, quibus praeerat Saburra… (Caes. Civ. 2.38.3) ‘...about the beginning of the night, he sent all his cavalry towards the enemy’s camp, which was upon the river Bagradas, and where Saburra... was commander in chief.’

 Constituent order in Classical Latin prose

Table 5.╇ Directional€complements of the verb mitto: contextual dependency and anaphoric continuation (corpus 1) Position Pre-verbal Post-verbal Other Total Percentage

Bound + An.

– An.

1 0 1 2 7%

â•⁄ 6 â•⁄ 0 10 16 55%

Unbound + An. – An. 0 1 3 4 14%

1 1 5 7 24%

Total

Percentage

â•⁄ 8 â•⁄ 2 19 29 100%

28% â•⁄ 7% 65% 100% x

Table 6.╇ Second arguments of the verb mitto: contextual dependency and anaphoric continuation (corpus 1) Position Pre-verbal Post-verbal Other Total Percentage

Percentage

– An.

Unbound + An. – An.

Total

+ An.

Bound 1 0 2 3 5%

2 0 9 11 18%

6 1 11 18 30%

16 3 41 60 100%

27% 5% 68% 100% x

7 2 19 28 47%

It appears from Table 5 that directional complements expressed by a noun (phrase) are contextually bound in the majority of cases (7 + 55%) and go in pre-verbal position (28%), rarely in post-verbal (7%), but mainly in another position (65%). Anaphoric continuation is not very well represented (2 + 4 occ.). Second arguments (Table 6) are mainly contextually unbound (30 + 47%); additionally, a third of them are continued by anaphoric means (3 + 18 occ.), as illustrated in (5) below.16 2.1.2 Analysis I will start with the statement that constituents expressed by anaphoric pronouns or noun phrases containing an anaphoric modifier stand in sentence- or clause-initial position when they connect with the context and pick up a preceding salient constituent (cf. p. 77 ff.), as is shown in (3); eum libellum picks up causam. Since tibi is contextually bound and non-contrastive, the focus is here on the action (misi). This sentence answers the underlying question ‘what did I do with this document?’

16. Furthermore, there are also instances with a strong form of anaphora, such as a full noun – see legatos picked up by legati in Sal.€Jug. 13.6–7.



Chapter 3.╇ Declarative sentences 

(3)

[Silius ad me non venerat.] Causam composui; eum libellum tibi misi. pleading-acc I wrote the document-acc you-dat I sent ‘[Silius has not arrived.] I have drawn up a pleading; I sent the document to you.’ (Cic. Att. 15.24)

From the preceding tables we have seen that directional complements are typically bound whereas second arguments are more often contextually independent. An example is given in (4) with ad consulem (bound) and legatos (unbound). A contextually independent second argument with anaphoric continuation afterwards is illustrated in (5). (4)

Igitur legatos ad consulem cum suppliciis mittit, therefore envoys-acc to consul with petitions he sends [qui tantummodo ipsi liberisque vitam peterent.] ‘He therefore sends legates with supplicatory offerings to the consul [to beg only for the lives of himself and his children.]’ (Sal. Jug. 46.2)

(5) Ad Dolabellam Tironem misi cum mandatis et litteris. to Dolabella Tiro-acc I sent with message and letter Eum ad te vocabis... him-acc to you you will call ‘I sent Tiro to Dolabella with a message and a letter. Send for him and...’  (Cic. Att. 15.4.5) I have also observed that second arguments are found in immediately pre-verbal position in only 27% of cases; more frequently, they stand elsewhere in the sentence. Indeed, they are mostly placed in an initial position: either absolute initial or after a connector, or after a Topic (6, Bocchus), or after a participial clause that forms the setting (7): (6)

Nam Bocchus nuntios ad eum saepe miserat: for Bocchus-nom messengers-acc to him often had sent [velle populi Romani amicitiam…] ‘For Bocchus had been sending messengers to him frequently: [he wanted the friendship of the Roman people...]’ (Sal. Jug. 88.5)

(7)

Dein brevi Vticam adpulsi litteras ad Iugurtham mittunt: then soon to Utica landing-nom letter-acc to Jugurtha they send [quam ocissume ad provinciam adcedat.] ‘Then landing soon at Utica, they send a letter to Jugurtha: [he should come to the province as speedily as possible.]’ (Sal. Jug. 25.5)

In view of these results, I will proceed to an additional detailed examination of two recurring expressions: ‘to send a delegation to someone’, frequent in historical narrative, and ‘to send a letter or book to someone’, used by Cicero in his correspondence.

 Constituent order in Classical Latin prose

2.1.3 Legatos ad Caesarem mittunt ‘they send a delegation to Caesar’ The stereotyped expression ‘to send a delegation to someone’ in Caesar (cf. Odelman 1972:€176) has already been examined by Panhuis (1982:€117–149) from the communicative sentence perspective. For my part, I use the criterion of contextual dependency in order to investigate where legatos or nuntios ‘messengers’ and directional complements go. I assume that in principle, this recurrent expression has a similar informative value; variations of placement – if any – are to be considered as pragmatically more or less equivalent. Results of this examination are presented in a schematised way in Table 7. Subjects were not taken into consideration. Ad Caesarem signifies that the person is expressed by a (whatever) proper name, ad eum by a pronoun. Seven occurrences were found in Sallust, twenty-eight in Caesar. Other constituents can intervene between the constituents under examination. Sending messengers to somebody is closely connected with the transmission of a message. This is why a modifier follows: indirect speech, a supine (rogatum, postulatum…), a prepositional phrase announcing indirect speech (cum mandatis), or a final or relative clause. The pattern legatos ad Caesarem/ad eum mittunt (rogatum) appears in 42% of cases, the pattern ad Caesarem/ad eum legatos mittunt (rogatum) in 26%. When the goal of the mission is expressed by a prepositional phrase such as de pace ‘about peace’ or de deditione ‘about surrender’ (see the second half of Table 7), it precedes the verb, or follows legatos. The question to be answered now is to what extent these variations concerning comparable informative content are connected with particular pragmatic values. Table€7.╇ Positions of complements of the verb mitto in Caesar and Sallust (corpus BTL) Pattern legatos

ad Caesarem

mittunt

legatos ad Caesarem ad eum mittunt legatos legatos legatos ad Caesarem ad eum

ad eum legatos legatos ad eum ad Caesarem de pace de deditione legatos legatos

mittunt mittunt mittunt legatos de pace ad Caesarem ad eum de pace cum his mandatis

Total

(ind. disc., rogatum) (ind. disc.) (ut…, rogatum) (qui…) (ind. disc.) mittunt mittunt mittunt mittunt mittunt

Occ.

%

10

42%

5 8 1 1 5 2 1 1 1

3% 14% 9%

35

100%

26%

6%



Chapter 3.╇ Declarative sentences 

I will analyse first sentences without prepositional phrases expressing the goal of the mission (first half of Table 7). When the person – always contextually dependent – to who a delegation is sent is expressed by a proper name, the order {legatos > ad Caesarem} (example 8) is slightly more frequent than {ad Caesarem > legatos} (example 9). The informative value of these sentences is the same: they answer the underlying question ‘who?’ (quem?), and legatos represents the salient element in both cases. It announces the message that has to be communicated. Sometimes we could envisage a contrast between the subject and the recipient of the message (cf. Ariovistus and ad Caesarem in 9) but this does not seem to have a consequence for the placement of constituents. (8)

Indutiomarus veritus, ne ab omnibus desereretur, Indutiomarus-nom afraid neg by everyone be abandoned legatos ad Caesarem mittit: sese idcirco ab suis envoys-acc to Caesar sends refl for this reason from his men discedere... go away ‘Indutiomarus was afraid of being abandoned by everyone and sent envoys to Caesar: he had been reluctant to leave his own people...’ (Caes. Gal. 5.3.5)

(9) Biduo post Ariovistus ad Caesarem legatos misit: velle se... two days later Ariovistus-nom to Caesar envoys-acc sent wish refl ‘Two days later, Ariovistus sent envoys to Caesar: he wished...’(Caes. Gal. 1.47.1) When the person is expressed by an anaphoric pronoun, there is mostly the ordering legatos ad eum mittit, with ad eum in pre-verbal position. Compare example (10) with (11), which contains in addition an expression of cause. (10)

Qua re nuntiata Pirustae legatos ad eum this thing being reported Pirustae-nom envoys-acc to him mittunt, qui doceant... send who would explain ‘When news of his action came the Pirustae send him envoys who were to explain...’ (Caes. Gal. 5.1.7)

(11) Vbii... purgandi sui causa ad eum legatos mittunt, Ubii-nom justifying refl for sake to him envoys-acc send qui doceant... who would explain ‘The Ubii..., to justify themselves, send him envoys who were to explain...’  (Caes. Gal. 6.9.6) The second half of Table 7 shows prepositional phrases such as de pace ‘about peace’ standing after legatos or in pre-verbal position. However, they do not carry special

 Constituent order in Classical Latin prose

saliency. These constituents can be considered as forming a pragmatic unit with legatos from which they can be separated, as in (12). (12) (Atuatuci) commoti legatos ad Caesarem de pace troubled-nom envoys-acc to Caesar about peace miserunt, qui ad hunc modum locuti: non se existimare they sent who in this way spoke neg refl consider ‘(the Atuatuci)... troubled by..., send envoys to Caesar to ask for peace; they spoke to the effect that in their opinion...’ (Caes. Gal. 2.31.1) Supines placed after the verb mitto have another function. They make explicit the act done by the delegation – for example, oratum in (13) expresses a request – and thus convey the salient information. (13)

Segni Condrusique... legatos ad Caesarem miserunt Segni Condrusi-nom=and envoys-acc to Caesar sent oratum, ne se in hostium numero duceret... to beg neg refl in of enemies number he would consider ‘The Segni and Condrusi... sent envoys to Caesar to beg him not to consider them among the number of his enemies.’ (Caes. Gal. 6.32.1)

In sum, the expression ‘to send a delegation to someone’ in Caesar’s and Sallust’s historical narrative permits us to conclude that, in the majority of cases, the sentences containing them answer the underlying question ‘who?’ (quem?), and legatos is the most salient element. Different arrangements can be viewed as variants without a special pragmatic value involved. The pre-verbal position is not necessarily filled with legatos; another constituent can be found there, even ad eum, that carries no pragmatic function. 2.1.4 Ad te litteras misi ‘I sent you a letter’ The expression ‘to send a letter to someone’ is recurrent in Cicero’s correspondence. I will examine the attested patterns, making a distinction between nominal and pronominal forms that refer to the addressee expressed in the dative (tibi) or in a prepositional phrase (ad te). Second arguments are represented by litteras, epistulam and epistulas ‘letter(s)’. Table 8 indicates, in a schematic way, the results of the examination; other elements can intervene between the obligatory constituents. Only syntactically independent sentences or clauses were taken into account. As we can observe, the pattern ad te/Brutum litteras misi is the most frequently attested (46%), but it competes with litteras ad te misi (29%). The verb comes first several times: misi ad eum/Brutum litteras (16%). The main patterns are exemplified below. Epistulam in (14), occupying the pre-verbal position, is a new piece of information, expanded afterwards. However, epistulam ad Ciceronem ‘a letter for Cicero’ in€(15), also placed initially, is more difficult to analyse. This constituent could have been mentioned in the preceding context, and thus could represent a piece of shared



Chapter 3.╇ Declarative sentences 

Table€8.╇ Placement of complements of the verb mitto in Cicero, Ad Atticum (BTL) Pattern

Occurrences Percent.

ad te/tibi Quinto/ad Brutum litteras ad te litteras misi misi Total

litteras litteras ad te/tibi misi misi ad Brutum ad eum

misi misi misi litteras ad Brutum/ad te litteras litteras

11 â•⁄ 3 â•⁄ 9 â•⁄ 1 â•⁄ 2 â•⁄ 1 â•⁄ 4 31

â•⁄ 46% â•⁄ 29% â•⁄â•⁄ 3% â•⁄â•⁄ 6% â•⁄ 16% 100%

knowledge. Nevertheless, without a counter-indication, I consider it as new information with an indefinite reading (as also the translator does). In all these cases, the sentences answer the underlying question ‘what’? Consequently, ad me in (14) as well as tibi in (15), situationally bound, do not have a special pragmatic value. (14)

Varro autem noster ad me epistulam misit sibi Varro-nom but our-nom to me letter-acc sent him-dat a nescio quo missam – nomen enim delerat – from I don’t know who sent-acc name-acc indeed had erased in qua scriptum erat... in which was written ‘Our friend Varro however has sent me a letter from I don’t know who (he had crossed out the name) in which it was asserted that...’ (Cic. Att. 15.5.3)

(15)

[Heri enim misi qui videret; cui etiam ad te litteras dedi.] Epistulam ad Ciceronem tibi misi. letter-acc for Cicero you-dat I sent ‘[I sent someone to see yesterday; I also gave him a letter for you.] I sent you a letter for my son Cicero.’ (Cic. Att. 12.49.3)

When a second argument is expressed by a noun phrase, it can be discontinuous. Discontinuity confirms saliency of the whole noun phrase or of a part of it. In (16), the underlying question is not ‘whose?’, but ‘what?’; the following context indicates to us that it is considered as a unit (note the ellipsis of the object of legeris). In (17), however, the question ‘whose?’ is implied, with an initial verb misi; Focus (Antoni) stands at the end of the sentence.

 Constituent order in Classical Latin prose

(16)

Bruti ad te epistulam misi. [Di boni, quanta ἀμηχανία!] of Brutus to you letter-acc I sent Cognosces cum legeris. you will see when you will read ‘I sent you Brutus’s letter. [Heavens above, what helplessness!] You will see what I mean when you read it.’ (Cic. Att. 15.29.1)

(17)

[Me caecum qui haec ante non viderim!] Misi ad te epistulam Antoni. Ei cum ego scripsissem... I sent to you letter-acc of Antonius him-dat when I had written ‘[Blind that I was not to have foreseen this!] I sent you Antony’s letter. I had written to him...’ (Cic. Att. 10.10.1)

To sum up, second arguments such as litteras ‘letter(s)’ mostly represent salient constituents; they are often further expanded by prepositional phrases, genitives, adjectives or determiners, from which they can be separated. They are found in various positions. As I pointed out above concerning example (15), the contextual status of constituents is sometimes difficult to establish in Cicero’s correspondence. An additional example is given in (18), with orationem: (18)

[Duas a te accepi litteras...] Orationem tibi misi. Eius custodiendae et speech-acc to you I sent of it to be kept back and proferendae arbitrium tuum. to be put out decision your-nom ‘[I received two letters from you...] I sent you the speech, to be kept back and put out at your discretion.’ (Cic. Att. 15.13.1)

The speech (oratio), evidently a focal constituent, was not mentioned in the preceding context, but it must concern Cicero’s Second Philippic.17 It belongs to shared knowledge between Cicero and Atticus, and it is obvious that Cicero had promised it before. However, oratio has neither a possessive modifier nor an anaphoric pronoun marking explicitly that it is a contextually given element. Such constituents may occur in initial position (6 occurrences out of 10), as in (19) with epistulam which is deducible from the preceding context and means ‘his/this letter’ (Cassius’). Both sentences in (18) and (19) answer the question ‘what?’ At any rate, it is not possible to give mechanically a definite reading to constituents placed leftwards and an indefinite reading to constituents standing rightwards in the sentence. Furthermore, we cannot decide about saliency of a constituent judging from the position it occupies in a sentence.

17. See Beaujeu, Cicéron, Correspondance, tome IX, Paris (1988:€83 and note, p.€64).



Chapter 3.╇ Declarative sentences 

(19) [Cassius vero vehementer orat ac petit ut Hirtium quam optimum faciam. Sanum putas? ὁ γναϕεὺς ἄνθρακας.] Epistulam tibi misi. letter-acc you-dat I sent ‘[However, Cassius begs and requests me pressingly to make Hirtius as good a republican as I can. Is he in his right mind, do you think? You can’t wash a blackamoor white.] I sent you his letter.’ (Cic. Att. 15.5.1)

2.2

The verb duco ‘to lead’

The verb duco (ad) ‘to lead someone somewhere’, together with its compounds such as adduco (in) ‘to lead (to)’ or deduco (in/ex) ‘to lead away (to/from)’, requires a directional complement (D) indicating the place where or from where somebody is led. In my corpus 1, directional complements refer to an inanimate entity in the majority of cases. Patterns collected from this corpus are indicated in Table 9. The verb duco, attested especially in Caesar (29 occurrences out of 34), is mainly found with the {(first argument >) second argument > directional complement > verb} ordering (73%). This ordering appears in sentences such as (20). (20) Thermus cohortes ex urbe reducit et profugit. Thermus-nom cohorts-acc from town withdraws and flees ‘Thermus withdrew his cohorts from the town and fled.’ (Caes. Civ. 1.12.2) Obligatory complements of this verb are usually contextually dependent: the subject (Thermus) is a Discourse Topic; the object (cohortes) is deducible from the context, as well as the directional complement (ex urbe). Consequently, Focus is on the action, expressed by the verb. We can also envisage that Focus is on the verb forming a pragmatic unit with the directional complement that further specifies the action. In any case, the underlying question is ‘what did he do?’, which is confirmed by the coordination with another action (reducit et profugit). Table€9.╇ The verb duco and its compounds (corpus 1) Pattern A1 > A 2 > D > V A2 > D > V A1 > D > A2 > V D > A2 > V A2 > V > D V > A2 > D V > D > A2 Total

Total

Percentage

15 10 â•⁄ 4 â•⁄ 1 â•⁄ 1 â•⁄ 2 â•⁄ 1 34

â•⁄ 73% â•⁄ 15% â•⁄â•⁄ 3% â•⁄â•⁄ 6% â•⁄â•⁄ 3% 100%

 Constituent order in Classical Latin prose

Sentences containing duco can also cover other pragmatic values. For example, the direction where somebody is led can be Focus, as in (21) with the underlying question ‘where?’ (note the anaphoric continuation by ibi). The directional complement need not stand directly before the verb; legiones is a contextually dependent constituent. (21)

Sub vesperum Caesar intra hanc fossam legiones towards evening Caesar-nom within this ditch legions-acc reducit atque ibi sub armis proxima nocte conquiescit. withdraws and there under arms following night rests ‘Towards evening Caesar withdraws the legions behind this ditch and rests there the following night under arms.’ (Caes. Civ. 1.41.6)

To sum up, the verb duco mainly appears in sentences answering the question ‘what does he do?’ Its obligatory complements are contextually dependent in the majority of cases.

2.3

The verb do ‘to give’

The trivalent verb do ‘to give’18 and its compounds addo ‘to add’ and reddo ‘to give back’ appear with three obligatory arguments: first argument (A1), second argument (A2) in accusative and third argument (A3) expressed in the dative or by a prepositional phrase with ad.19 Results are presented in Table 10. Table€10.╇ The verb do (corpus 1) Pattern A1 > A2 > A3 > V A 2 > A3 > V A1 > A3 > A2 > V A 3 > A2 > V A3 > A1 > A2 > V A3 > A1 > V > A2 A2 > A1 > A3 > V A3 > V > A2 V > A3 > A2 Total

Total

Percentage

â•⁄ 2 12 â•⁄ 1 â•⁄ 7 â•⁄ 2 â•⁄ 1 â•⁄ 1 â•⁄ 3 â•⁄ 2 31

â•⁄ 46% â•⁄ 26% â•⁄â•⁄ 6% â•⁄â•⁄ 3% â•⁄â•⁄ 3% â•⁄ 10% â•⁄ 6% 100%

18. Panhuis (1982:€31–57) analysed sentences containing the verbs do and reddo in Plautus’s comedies, which is a different type of text in comparison with my corpus 1. According to him, these sentences mainly reflect the theme > rheme organisation that I presented in Introduction, p. 5. 19. Expressions like operam do ‘to make effort, to give attention’ were not taken into account.



Chapter 3.╇ Declarative sentences 

Table 11.╇ The verb do: placement of second and third arguments (corpus 1) Position Pre-verbal Post-verbal/final Other Total Percentage

A2 unbound

A2 bound

A3 unbound

A3 bound

Total

Percentage

â•⁄â•⁄ 8 â•⁄â•⁄ 6 â•⁄â•⁄ 6 â•⁄ 20 32%

â•⁄â•⁄ 2 â•⁄â•⁄ 0 â•⁄â•⁄ 9 â•⁄ 11 18%

â•⁄â•⁄ 3 â•⁄â•⁄ 0 â•⁄â•⁄ 3 â•⁄â•⁄ 6 10%

â•⁄ 12 â•⁄â•⁄ 0 â•⁄ 13 â•⁄ 25 40%

â•⁄ 25 â•⁄â•⁄ 6 â•⁄ 31 â•⁄ 62 100%

â•⁄ 40% â•⁄ 10% â•⁄ 50% â•⁄ 100%

The verb do, mainly attested in Cicero (19 occurrences out of 31), manifests most frequently the ordering {(first argument >) second argument > third argument > verb} (46%); however, the third argument can precede the second one (26%). It is important to look now at the contextual status of second and third arguments. Table 11 indicates their placement in pre-verbal, post-verbal or other position. Direct objects are typically contextually independent (32%), whereas indirect objects are situationally or contextually bound (40%). Both types of arguments can occupy any position, except for bound constituents that avoid the post-verbal position. In addition, there is no special tendency for unbound constituents to go in the preverbal position. Sentences containing the verb do show various pragmatic arrangements. Firstly, they can be focused on the second argument, and answer the question ‘who, what did he give?’, as is illustrated in (22) with a discontinuous second argument; parvulos carries emphasis, and igniculos is placed post-verbally. Secondly, the verb itself (it could also be envisaged as a pragmatic unit together with its object) can function as a new piece of information. Sentences such as (23) answer the question ‘what did he do?’; the verb dedidit figures at the same level as other actions in this complex sentence (aperuit, conservatus est). (22) (Natura) Nunc parvvlos nobis dedit igniculos, quos now very small-acc us-dat gave lights-acc which [celeriter malis moribus opinionibusque depravati sic restinguimus...] ‘Nature has given us some faint glimmering of insight, which, [under the corrupting influence of bad habits and beliefs, we speedily quench so completely that...]’ (Cic. Tusc. 3.2) (23) (Torquatus) portas aperuit et se atque gates-acc he opened and refl-acc and oppidum Caesari dedidit incolumisque ab eo conservatus est. town-acc Caesar-dat surrendered safe-nom=and by him was kept ‘Torquatus opened the gates, and surrendered himself and the town to Caesar, who spared him unharmed.’ (Caes. Civ. 3.11.3)

 Constituent order in Classical Latin prose

Thirdly, Focus can be on the third argument with the underlying question ‘to whom did he give it?’ This is illustrated in (24): the second argument (signum) is contextually dependent, isti referring to Clodius depends on the situation but seems to be the most salient element. We could also consider that isti forms a pragmatic unit with dedit, with the underlying question ‘what did he do with the statue?’ (24)

[simulacrum... quidam homo nobilis...] Signum de busto meretricis ablatum isti (Clodio) statue-acc from pedestal of courtesan taken-acc him-dat dedit, quod esset signum magis istorum quam publicae libertatis. he gave because was statue-nom rather of these than of public liberty ‘[a statue... a certain nobleman...] He took the statue of the courtesan from its pedestal and presented it to Clodius, that it might symbolize the “liberty” of Clodius and his like rather than of the State.’ (Cic. Dom. 112)

As Focus on the third argument is not very well represented in my corpus, I give below two additional examples. Whereas the constituent Brutus in (25) is contextually bound, properantibus tabellariis alienis in (26) conveys new information and stands with emphasis in sentence-initial position. Cicero explains why his letter is too short. (25)

Sed totum hoc Bruto dedi; qui de me ad te but all this-acc Brutus-dat I gave who about me to you humanissimas litteras scripsit… very kind-acc letters-acc wrote ‘I allowed all this to Brutus; and he has written about me so kindly to you...’  (Cic. Att. 6.1.7)

(26)

Properantibus tabellariis alienis hanc epistulam dedi. hurrying-dat couriers-dat sb else’s-dat this letter-acc I gave [Eo brevior est, et quod eram missurus nostros.] ‘The couriers to whom I am giving this letter are not my own and they are in a hurry. [That is why it is so short, and also because I was going to send my own men.]’ (Cic. Att. 11.17.1)

To sum up, sentences containing the verb do show various pragmatic values that can hardly be reduced to one dominant pattern. In particular, we cannot say that the second argument necessarily precedes the third one. The contextual status permits us to identify contextually independent constituents that are the strongest candidates for Focus function. On the other hand, the pre-verbal position does not systematically receive Focus constituents.



Chapter 3.╇ Declarative sentences 

3. Passive sentences 3.1

Introduction

The passive voice consists of the promotion of the second argument to the syntactic function of the subject, and can concern bivalent as well as trivalent transitive verbs. Adams (1976b: 121 ff., and. 1976a: 95) pointed out the frequency of the ordering {verb > subject} with passive verbs in Late Latin texts and related it to the pragmatic organisation of the sentence where the subject of the passive verb conveying new information is placed rightwards. Panhuis’s (1984b) investigations go in the same direction: in historical narrative, the passive voice is used for preparing changes of Topic, and the subject, as a new element of the narration, may occupy a late position in the sentence. According to Bolkestein (1996b: 19), the order {verb > subject} is found in passive sentences more often than in active ones in Classical Latin.20 In the same article, she describes the {verb > subject} pattern as a result of its “introductory” and “annunciatory” function. However, this does not mean that subjects in the {subject > verb} pattern cannot convey new information (Bolkestein 1995:€ 39 f.). To sum up, previous studies have shown that, on the one hand, the passive voice is a means that permits speakers or authors to highlight newness of the subject. On the other hand, sentences containing the passive voice enter, in some cases, into the group of presentative sentences that introduce a new element into the discourse. Therefore, I argue that the passive voice itself does not have an “introductory” function and that it can be used in sentences with specific pragmatic value, for example in presentative sentences. A similar observation will be also formulated for monovalent verbs (see Section 6 below, p. 172 ff.). In this section, after presenting the data (Section 3.2), I will concentrate on the pragmatic values of sentences containing a passive as well as on the contextual status of the subject in passive sentences (Section 3.3–4). Verbs with different semantic values will be considered here, although a distinction between verbs of action (like interfectus est ‘he was killed’) and verbs of event (inventus est ‘he was found’) would bring more refined statistical results.21 The following Section 3.5 will be devoted to analytic forms of the perfect passive tense, and accompanied by a brief comparison with the perfect of deponent verbs (Section 3.6).

20. Bolkestein (1996b: 19) argues that in the active voice, the order {subject > verb} appears in 93% of sentences, the order {verb > subject} 7%, whereas in the passive voice, the proportion is 75% for {subject > verb} and 25% for {verb > subject}. Cf. also Risselada (1991). For the passive voice, see also Pinkster (1985). 21. See Spevak (2005a: 251 f.) for the passive in the Itinerarium Egeriae, and cf. Adams (1976b: 122).

 Constituent order in Classical Latin prose

3.2

Data

Table 12 presents data concerning the position of nominal, animate and inanimate second arguments (A2)22 with respect to the verb (V) in my corpus 1. Table€12.╇ The passive voice: animate and inanimate nominal second arguments, contextually bound and unbound (corpus 1)23 Pattern

A2 > V V > A2 Total

Animate A2

dependent independent dependent independent

Inanimate A2

Total

occ.

%

occ.

%

occ.

%

22 28 â•⁄ 3 16 69

â•⁄ 32% â•⁄ 41% â•⁄â•⁄ 4% â•⁄ 23% 100%

â•⁄â•⁄ 52 â•⁄â•⁄ 85 â•⁄ 5 â•⁄â•⁄ 28 170

â•⁄ 31% â•⁄ 50% â•⁄â•⁄ 3% â•⁄ 16% 100%

â•⁄ 74 113 â•⁄â•⁄ 8 â•⁄ 44 239

â•⁄ 31% â•⁄ 47% â•⁄â•⁄ 3% â•⁄ 19% 100%

Inanimate second arguments are globally more frequent than animate ones (170 versus 69 occurrences). The subject of a verb in the passive voice is mostly contextually independent. It shows a tendency to precede the verb (41 and 50%), slightly more for inanimate entities, but it can follow it as well (23 and 16%), more for animate entities. Contextually dependent subjects privilege the leftwards position in the sentence (32 and 31%), they appear rarely after the verb (4 and 3%).

3.3

Contextually independent second arguments

Table 12 shows that contextually independent entities enter into both patterns with a clear preference for {second argument > verb}, and not for {verb > second argument}, as might be expected in view of the considerations presented in Section 3.1. Two typical examples are given in (1) and (2), with contextually independent subjects, one animate (L. Pupius), the other inanimate (duae trabes). These sentences answer the underlying question ‘who, what?’ In the first example, the subject is placed pre-verbally, in the second, sentence-initially (with discontinuity). Both verbs, adducitur and conlocantur, are deducible from the context.

22. The subject of passive constructions is named the second argument (see Pinkster 1995:€25 f.). 23. Authors participate in these figures in the following way: Caesar 36 occ. for animate and 73 for inanimate arguments, Cicero 25 animate, 86 inanimate, and Sallust 8 animate and 11 inanimate.



Chapter 3.╇ Declarative sentences 

(1)

[...reliqui ad Caesarem perveniunt] atque una cum iis deprensus L. Pupius, primi and together with them prisoner-nom L. Pupius-nom of first pili centurio, adducitur qui rank centurion-nom is brought who [hunc eundem ordinem in exercitu Cn. Pompei antea duxerat]. ‘[...the rest made their way to Caesar,] bringing as a prisoner their highestranking centurion, Lucius Pupius, who had previously held this same post in Pompey’s army.’ (Caes. Civ. 1.13.4)

(2)

Duae primum trabes in solo aeque longae two-nom first baulks-nom on ground equally long-nom distantes inter se pedes IIII conlocantur inque eis distant between refl feet four are laid in=and them columellae pedum in altitudinem V defiguntur. uprights-nom of feet in height five are inserted ‘First two baulks of equal length were laid on the ground four feet apart, and onto them were inserted uprights five feet high.’ (Caes. Civ. 2.10.2)

Cicero, especially in his Tusculan Disputations and in his speeches, gives various statements in order to support his argumentation. Reminding the reader of previous cases, he shows what happened to someone before. He uses what-happens sentences with contextually independent subjects in initial position. The subject is thus introduced without any special formula and the conveyed information is ‘what happened to him?’ (3)

M. Flaccus quia cum C. Graccho contra salutem M. Flaccus-nom because with G. Gracchus against safety rei publicae fecerat, ex senatus sententia est of State had acted by of Senate resolution was interfectus, eius domus eversa et publicata est. killed his house-nom destroyed and confiscated was ‘Marcus Flaccus, because he had acted with Gaius Gracchus in a manner opposed to the safety of the republic, was put to death by a resolution of the Senate; his house was destroyed and confiscated.’ (Cic. Dom. 102)

Contextually independent second arguments can appear rightwards in the sentence: this is typical of enumerations, as in (4), but also occurs in other cases, such as in (5). The passive verbs involved are not only semantically weak verbs (like facio ‘to do’ or habeo ‘to have’); verbs of action occur as well, such as creantur ‘are chosen’ or datur ‘is given’. The underlying question is ‘who?’

 Constituent order in Classical Latin prose

(4)

[Dictatore habente comitia Caesare,] consules creantur Iulius Caesar et P. Servilius. consuls-nom are chosen Julius Caesar-nom and P. Servilius-nom ‘[Caesar, as dictator, holding the Comitia,] Julius Caesar, and P. Servilius, were chosen consuls.’ (Caes. Civ. 3.1.1)

(5)

[Vbi id a Caesare negatum et palam si conloqui vellent concessum est,] datur obsidis loco Caesari filius Afrani. is given of hostage in place Caesar-dat son-nom of Afranius ‘[While Caesar refused this request, but allowed them, if they wished, to negotiate in public,] Afranius’s son was delivered to Caesar as a hostage.’  (Caes. Civ. 1.84.2) Subjects modified by an apposition or a relative clause are also found in post-verbal position. For example, the subject in (6) is presented as an addition (note etiam ‘also’); the verb subicitur is contextually bound. (6)

3.4

Subicitur etiam L. Metellus tribunus plebis is put up also L. Metellus-nom tribune of people ab inimicis Caesaris qui hanc rem distrahat. by enemies of Caesar who this affair should disrupt ‘Also the plebeian tribune Lucius Metellus was put up by Caesar’s enemies to disrupt this business.’ (Caes. Civ. 1.33.3)

Contextually dependent second arguments

Whereas contextually independent entities enter into both patterns, {second argument > verb} and {verb > second argument}, although the former is more frequent, dependent constituents strongly prefer a position leftwards in the sentence. They function as Topic, for example M. Cato in (7) and hoc consilium in (8), which summarizes preceding information: (7) Sic M. Cato invisus quasi per beneficium Cyprum relegatur. so M. Cato-nom hated as if through favour to Cyprus is relegated ‘So the hated Marcus Cato is banished to Cyprus under show of having a favour bestowed upon him.’ (Cic. Dom. 65) (8) Hoc consilium Caesaris plerisque non probabatur. this plan-nom of Caesar to many-dat neg was approved ‘This plan of Caesar’s met with general disapproval.’ (Caes. Civ. 1.72.4) Besides nominal subjects, such as all those presented up till now, passive sentences can contain a pronominal subject. These mainly function as Sub-Topics (for example, omnes ‘all’, multi ‘many’, plures ‘several’) or as Sentence Topics, and occupy the initial position. An example is given in (9), with a resumptive pronoun:



Chapter 3.╇ Declarative sentences 

(9)

[Dolabella me sibi legavit a. d. III Non.] Id mihi heri vesperi nuntiatum est. this-nom me-dat yesterday evening announced be-aux ‘[Dolabella appointed me to his staff on the 3rd.] I was informed of this yesterday evening.’ (Cic. Att. 15.11.4)

When contextually dependent subjects are placed rightwards, the verb is usually Focus (question ‘what happens?’). The expression of the subject in (10), Varus, is a simple reminder, because this character is the Topic of the current section. In (11) we have an instance of a focal verb placed initially. This event is contrary to expectation, as signalled by tamen (Spevak 2006d). (10)

[Varus... profugit. Hunc... Caesaris milites consistere coegerunt.] Commisso proelio deseritur a suis Varus. being joined battle is deserted by his men Varus-nom ‘[Varus... fled. Caesar’s soldiers... forced him to stop.] When battle was then joined, Varus was deserted by his men.’ (Caes. Civ. 1.13.4)

(11)

Producitur tamen res is prolonged however affair-nom [aciesque ad solis occasum continentur, inde utrique in castra discedunt.] ‘The confrontation was nevertheless prolonged [and the men remained in battle order until sunset; then both sides went back to camp.]’(Caes. Civ. 1.83.3)

It is worth noting that in sentences containing a passive verb Focus can concern another constituent than the subject or the verb. A good example is given in (12) with (contrastive) Focus on the agent of the process: a P. Rutilio legato ‘by the legate Publius Rutilius’. This information is surprising because one would expect that Metellus, the general, himself will hand over his army to Marius: (12)

3.5

Exercitus ei traditur a P. Rutilio legato; army-nom him-dat is handed over by P. Rutilius-abl legate-abl [nam Metellus conspectum Mari fugerat...] ‘The army was handed over to him by Publius Rutilius, the legate: [for Metellus had fled from seeing Marius...]’ (Sal. Jug. 86.5)

Analytic forms of the passive voice

The passive forms of the perfect and the pluperfect tenses are analytic and consist of a passive past participle and the auxiliary sum. The auxiliary usually does not bear Focus

 Constituent order in Classical Latin prose

except in special cases in which the time of the event is salient information (cf. Firbas 1992:€17 and Marouzeau 1938:€24).24 Both forms function together as a unit but they do not exhibit a fixed order, as has been pointed out by Marouzeau (1938:€ 23), for whom the group participle + sum obeys the same principles as the group predicative + sum (see also Hoff 1996).€Traditionally, the order factus est ‘he was made’ is considered as “normal”, whereas the order est factus as “marked” (Marouzeau ibid.; Szantyr 1972:€403). The two elements may also be separated by other words. For Adams (1994a: 40 ff.), est is attracted, in the case of separation, to a “strong” (or “emphatic”) word, for example to a member of an antithesis, a negative word, or a demonstrative. The variations between factus… est and est… factus remain, however, difficult to explain. Hoff (1996), who has examined the status quaestionis in detail,25 concentrated on the sequence factus… est that appears to be particularly infrequent, compared to other arrangements.26 Unlike Hoff ’s analysis, mine will examine not only possible orderings of the two components, but also the ability of the participle and the auxiliary to occupy the initial and the final sentence-position. Data presented in this section were collected with the help of LASLA; only occurrences of the perfect in the indicative mood and in main clauses were taken into consideration. 3.5.1 Data In Table 13 placement of the participle and the auxiliary is presented in continuous (factus est and est factus, see examples (13) and (15) below) and discontinuous (factus… est and est… factus, see (17) and (20)) sequences. It concerns various verbs in the passive, the verb factus est ‘he was done’ serves as an example. The number of occurrences in initial and final position is noted. For example, an indication that factus est is in initial position means that the sentence containing it starts with the participle (Factus est..., see example (13)); when it is final, the sentence ends with the auxiliary (...factus est, see example (21)).

24. Marouzeau gives as an example: Est istuc datum profecto, ut grata mihi sint quae facio omnia. (Ter. Eun. 395) ‘It is a gift of mine that everything I do wins me favour.’ However, we cannot consider every auxiliary placed initially as Focus (see the following section). Similarly, the participle in initial position need not be focal.€Thus, Marouzeau’s (1910:€113) statement that “factus est is chosen when an action is defined, and est factus when the reality of an action is affirmed or confirmed” holds true for only some instances. 25. Hoff (1996:€373) refers to the study by Vogel (1938), who looked at the positioning of sum with respect to monosyllabic or plurisyllabic forms of the verb, and of final and non-final positions. However, there is no direct correspondence between the form and the position. 26. This sequence is infrequent, but that does not mean that it is “prohibited”.



Chapter 3.╇ Declarative sentences 

Table 13.╇ The passive perfect: placement of the participle and the auxiliary (LASLA)27 Sequence

total

percent.

where initial

factus est factus… est est factus est… factus Total

150 â•⁄ 26 â•⁄ 55 â•⁄ 34 265

â•⁄ 56% â•⁄ 10% â•⁄ 21% â•⁄ 13% 100%

â•⁄ 5 â•⁄ 4 â•⁄ 3 â•⁄ 1 13

final â•⁄ 3% 15% â•⁄ 5% â•⁄ 3% â•⁄ 5%

41 â•⁄ 5 11 10 67

27% 19% 20% 29% 25%

Overall, these results are in accordance with Hoff ’s (1996:€374) findings. The sequence participle + auxiliary of the factus est type is the most frequent (56%). Yet the order est factus is well represented (21%). It is interesting to observe that in sentence-final position, the sequence...factus est is more frequent than...est factus (41€versus 11 occurrences). Whereas it is not surprising to find sentences starting with the participle (factus... est), we do not expect many occurrences with the initial auxiliary of the est... factus type; however, there are four of them in my sample. Separations like factus… est and est… factus, found in Caesar and Cicero, represent 10% and 13% of cases. To sum up, all these combinations confirm that there is no fixed order and that both the participle and the auxiliary can figure in initial position. In addition, the passive may be expressed by the participle with ellipsis of the auxiliary (as factus), and this is quite frequent (103 occurrences in total, not indicated in Table 13). As for individual authors, the order factus est is the majority in every work examined, except for Sallust, who frequently has ellipsis of the auxiliary. We also observe a tendency to uniformity in Sallust:28 his choice is practically limited to factus est, which never stands in initial position, and to factus alone. Other authors show more variation. 3.5.2 Analysis I will now recapitulate the tendencies for the placement of the participle and the auxiliary in perfect passive constructions, in several points. 1. The verb placed in initial position is not necessarily Focus but can be deducible from the context, such as interfectus est ‘was killed’ in (13). The author gives here his statement about a battle and enumerates the losses. All the same, the appearance of pugnatum est at the beginning of the sentence (14) is to be explained not as a signal that a description of a battle is starting (pace Luraghi 1995:€372; cf. Marouzeau 1938:€ 78), but as a contextually dependent event. Confligunt ‘they join battle’ implies pugnatum est ‘it was fought’, as has already been observed by Kühner and Stegmann (1914, II: 600). Salient elements of these sentences are Crastinus 27. This corpus includes: Caes. Civ., Sal.€Jug., Cic. Off., Dom., and Phil. 1–4. 28. For differences between Caesar and Sallust, see Vogel (1938:€66).

 Constituent order in Classical Latin prose

(question ‘who?’) and fortissime atque acerrime (question ‘how?’), respectively. By contrast, sentences with an initial auxiliary present completely new content,29 as illustrated in (15) with est inventus ‘was found’.30 Here, Focus is on the subject tribunus plebis and the sentence is of the presentative type. (13)

[In eo proelio non amplius ducentos milites desideravit...] Interfectus est etiam fortissime pugnans Crastinus, cuius... slain be-aux also very bravely fighting Crastinus-nom whose ‘[In this battle he lost no more than two hundred soldiers...] Also Crastinus, who..., was slain while fighting with utmost bravery.’ (Caes. Civ. 3.99.2)

(14)

[...cum Massiliensibus confligunt.] Pugnatum est utrimque fortissime atque acerrime... fought be-aux on both sides very bravely and fiercely ‘[...and engaged the Massiliots.] The fight was fierce and bitter on both sides...’  (Caes. Civ. 1.57.3)

Est inventus tribunus plebis qui, consularibus copiis be-aux found tribune-nom of people who by consular forces instructus, omni impetu furoris in eum civem inruerit quem... assisted with all force of frenzy to this citizen rushed at whom ‘There has been found a plebeian tribune, who, assisted by the forces of the consuls, has launched the full fury of his madness upon a citizen, whom...’  (Cic. Dom. 119)

(15)

The verbs exemplified in (13) – (15) are not focal or emphatic; however, an initially placed participle (or less frequently the copula, see note 24) may bear contrast (cf. Marouzeau 1953:€38), such as excisa in (16) with respect to evulsa: (16) 

Excisa enim est arbor, non evulsa; felled indeed be-aux tree-nom neg uprooted [itaque quam fruticetur vides]. ‘The tree was felled, but not uprooted, [and you see how it is sprouting].’ (Cic. Att. 15.4.2)

2. For Marouzeau (1938:€26), separations such as factus… est highlight the idea expressed by the participle. This sequence was examined in detail by Hoff (1996:€377 f.), who concentrated on elements that intervene between the participle and the auxiliary. He observes that these can be adverbs and subjects, or other constituents, 29. See also est animadversum (Caes. Civ. 2.8.1) ‘it was noticed’. 30. However, cf.€inventus est: [Erant in magna potentia qui consulebantur...] Inventus est scriba quidam Cn. Flavius qui cornicum oculos confixerit... (Cic. Mur. 25) ‘[Those were possessed of great power who were consulted...] ‘A certain notary was found, by name Gnaeus Flavius, who could deceive the most wary...’ The participle seems to be emphatic here in order to stress the craftiness of the recorder (scriba).



Chapter 3.╇ Declarative sentences 

which can number up to three or four. Even so, I cannot see an evident difference between factus… est and est… factus except for the fact that factus… est is slightly – but not extremely, as Hoff states – less frequent than est… factus. In my corpus 1, separation produced by the subject31 (8 occurrences) concerns in particular contextually independent entities such as tabula in (17) and genus radicis€in (18), both conveying new information. Unlike discontinuity in noun phrases (see Chapter 6, Section 9, p. 275 ff.), it is not the discontinuous analytic verb form (fixa est) but the intervening element that bears Focus. (17)

Nuper fixa tabula est qua civitates recently posted tablet-nom be-aux by which cities-nom locupletissimae Cretensium vectigalibus liberantur. very rich-nom of Crete from taxes are exempted ‘A document has recently been posted exempting the extremely wealthy communities of Crete from paying tax.’ (Cic. Phil. 2.97)

(18)

Est autem genus radicis inventum ab eis, qui be-aux but kind-nom of root found by these who fuerant vacui ab operibus, quod appellatur chara… were exempt of labor which is called chara ‘There is also a kind of root, which was found by the men stationed on the fortifications, and is called ‘chara’...’ (Caes. Civ. 3.48.1)

Another element than the subject, or even several constituents, may be inserted between the participle and the auxiliary, as illustrated in (19); Focus is on the direction (ad templum Concordiae), and not on the action itself. Finally, the example quoted in (20) shows that between the auxiliary and the participle, only one element carries Focus: sex ‘six’; signa militaria are contextually dependent, as the clause answers the question ‘how many?’ (19)

[Nam cum ingravesceret annona...,] concursus est ad templum Concordiae factus, rush-nom be-aux to temple of Concord become [senatum illuc vocante Metello consule]. ‘[For when prices were rising steadily...,] there was a rush towards the Temple of Concord, [whither the consul Metellus was summoning the Senate].’  (Cic. Dom. 11)

(20) ...signaque sunt militaria sex relata. standards-nom=and be-aux military-nom six captured ‘...and six military standards were captured.’ (Caes. Civ. 3.53.1) 31. This separation by the subject, also signalled by Kühner and Stegmann (1914, II: 603), does not seem to me to be exceptional, pace Hoff (1996:€376 f.). The subject can be inserted inside factus… est as well as in est… factus.

 Constituent order in Classical Latin prose

3. The sequences factus est (41 occurrences in Table 13) and est factus (11 occ.) are both used in sentence-final position; the question is, what difference, if any, is involved. Although it is difficult to explain, the two following examples may be helpful: (21)

Quibus submotis omnes sagittarii funditoresque these being removed all-nom archers-nom slingers-nom=and destituti inermes sine praesidio interfecti sunt. defenceless unarmed without protection killed be-aux ‘When they had gone, all the archers and slingers, left unarmed and defenceless, were killed.’ (Caes. Civ. 3.93.7)

(22)

L. Domitius ex castris in montem refugiens, cum L. Domitius-nom from camp to hill fleeing when vires eum lassitudine defecissent, ab equitibus forces-nom him-acc from tiredness had abandoned by horsemen est interfectus. be-aux killed ‘Lucius Domitius fled from the camp to the hills, but when he succumbed to exhaustion he was killed by the cavalry.’ (Caes. Civ. 3.99.5)

The sentence in (21) tells us what happened: the archers and slingers were slain (interfecti sunt), and Focus is on the action.32 The sentence in (22), however, informs us how or by whom L. Domitius was killed. In other words, when aspects of the action are focused, the analytic passive is found in the order {auxiliary > participle}.33 In (23) and (24), I give two more examples for illustration. (23) ... atque ibi reliqua pars exercitus dimissa est. and there remaining part-nom of army discharged be-aux ‘... and there the remainder of the army was discharged.’ (Caes. Civ. 1.87.5) (24)

Pompeio, nostro familiari, cum graviter aegrotaret Pompey-dat our friend-dat when seriously was ill Neapoli, melius est factum. at Naples better be-aux become ‘Our dear friend, Pompey, on the occasion of his serious illness at Naples, got better.’ (Cic. Tusc. 1.86)

32. Cf. also the following instance with coordination: (frater) se hostibus obtulit atque interfectus est (Caes. Gal.€4.12.6) ‘(his brother) threw himself upon the enemy, and was killed’. 33. Cf. Marouzeau (1938:€26) about est natus as a result of emphasis on the quality scitus: Per ecastor scitus puer est natus Pamphilo. (Ter. An. 486) ‘Lord knows that’s a lovely baby Pamphilus has got!’



Chapter 3.╇ Declarative sentences 

3.6

Analytic forms of deponent verbs

Deponent verbs in the perfect and pluperfect tenses show morphological affinities with the passive voice. Therefore, I will examine them from the same point of view. The data are indicated in Table 14 that concerns deponent verbs in the perfect tense; usus est ‘he used’ serves as an example. Table 14.╇ Deponent verbs in the perfect tense: placement of the participle and the auxiliary (LASLA)34 Sequence usus est usus… est est usus est… usus Total

Total (also %) â•⁄ 70 â•⁄â•⁄ 5 â•⁄ 19 â•⁄â•⁄ 6 100

where

where

initial

final

4 2 0 0 6

â•⁄ 6% 40% â•⁄ 0% â•⁄ 3% â•⁄ 6%

18 0 7 2 27

26% ╆0% 37% 33% 27%

The order of the usus est type (70%) is dominant, and even more so than in the case of the passive voice, where it was 56% (Table 13); est usus is used less frequently (19%). The participle alone, usus (not included in Table 14), appears 26 times, in Sallust, who never uses it in initial position, as with the sequence usus est. Separations are less frequent than in the case of the perfect passive voice: usus… est represents 5%, and est… usus 6% of instances. Both€usus est and est usus may stand at the end of a sentence. Even if this corpus does not offer instances of est usus with the auxiliary placed in initial position, such a placement is not impossible, but it is rare; it is exemplified in (25).35 Focus is on ita that announces a consecutive clause. (25) Es enim ita ingressus, ut... be-aux indeed so approached that ‘For you have approached your subject as if...’

(Cic. Rep. 2.22)

Deponent verbs in the perfect tense show globally the same tendencies as the perfect passive voice, but to a lesser extent. I give two examples for illustration: a separation produced by the subject (contextually independent) is shown in (26), and focus on a complement of the verb, beneficio tuo ‘your favour’ in (27):

34. The same corpus was used as for the passive voice, see above note 27. 35. Cf. also Est enim... profectus, quoted in Chapter€1, example (11) p. 16. Such instances with the auxiliary placed in sentence-initial position permit us to reject the theory of the enclitic character of the auxiliary. See Spevak (2006b).

 Constituent order in Classical Latin prose

(26)

[Omnes autem P. Lentulus me consule vicit superiores.] Hunc est Scaurus imitatus. this-acc be-aux Scaurus-nom emulated ‘[Publius Lentulus, however, in the year of my consulship, eclipsed all that had gone before him.] Scaurus emulated him.’ (Cic. Off. 2.57)

(27) At beneficio sum tuo usus. – Quo? but favour-abl be-aux your-abl used what-abl ‘But I was done a favour by you. – What favour?’

(Cic. Phil. 2.5)

In the case of deponent verbs also, the question arises of what is the difference between profectus est (28) and est profectus (29). As expected – given the variation observed for the passive voice – , the first ordering appears in a sentence that informs us about ‘what does Domitius do?’, whereas the second ordering concerns the time when the departure took place (paucisque post diebus). (28)

[Sed paucis ante diebus L. Domitius... unam (navem) ipse conscenderat...,] nanctus turbidam tempestatem profectus est. obtaining wild-acc weather-acc set out be-aux ‘[However, a few days previously Lucius Domitius... embarked on the third ship himself,] and took advantage of some wild weather to set out...’  (Caes. Civ. 2.22.2)

(29) [His litteris acceptis quos advocaverat dimittit; ipse iter in Macedoniam parare incipit] paucisque post diebus est profectus. few=and after days be-aux set out ‘[On receipt of the letter he dismissed the men he had summoned; he himself began to make preparations for a march to Macedonia,] and set out a few days later.’ (Caes. Civ. 3.33.2)

3.7

Conclusions

It is well known that the components of the analytic forms of perfect tenses of passive and deponent verbs do not manifest a fixed order: they may be presented in either order as well as separated. The parameter of contextual dependency permits us to throw more light on the ordering of their components. In the passive voice, verbs deducible from the context are placed in initial position in the order factus est, whereas the order est (...) factus is found in presentative sentences. The initial position can also be filled with a contrastive participle. A contextually independent constituent, which is very often the subject in passive constructions, typically takes a non-initial position and may separate the participle from its auxiliary. When the action itself represents Focus, the uninterrupted analytic form, in the order factus est, is found in the final sentence or clause position. In other words, discontinuity is produced when the verb



Chapter 3.╇ Declarative sentences 

is not focal. Deponent verbs show globally the same tendencies of placement as passive verbs, they just seem to be less flexible. However, this can be ascribed to their semantic value. Indeed, a passive verb like est inventus ‘he was found’ occurs more frequently in initial position, because it can introduce a new entity, whereas a verb of action like profectus est ‘he set out’ has practically no chance for such a use. Nevertheless, sentences starting with the auxiliary do exist; this happens mainly in an explicative context when focus is on another constituent than the verb.

4. Verbs of thinking and saying In this section, verbs of thinking (Section 4.1) and of saying (Section 4.2) will be examined. They can present various constructions and admit various complements, but I will concentrate on only one type: the Accusative and Infinitive (henceforth AcI) clause (Section 4.2.1). Finally, I will add some remarks concerning verbs of saying that accompany direct or indirect speech (Section 4.2.2) and the parenthetical verb inquam (Section 4.2.3).

4.1

Verbs of thinking

Verbs of thinking, such as arbitror€‘to judge’, censeo ‘to think’, existimo ‘to consider’ and puto ‘to think’, are found, in 52% of cases, in the first person singular, referring to the speaker or author. They are particularly frequent in Cicero (81% of all instances offered by corpus 1). Table 15 shows the placement of the verb with respect to the Accusative and Infinitive clause (following it or preceding: {AcI > verb} or {verb > AcI}). The verb can also be inserted into the AcI clause. Table 15.╇ Verbs of thinking (corpus 1) Pattern AcI > verb verb > AcI verb inserted total

Occurrences

Percentage

â•⁄ 73 â•⁄ 25 â•⁄ 39 137

â•⁄ 53% â•⁄ 18% â•⁄ 29% 100%

As we can see, verbs of thinking often follow the AcI clause (53%); they precede it in only 18% of cases. However, this ratio is not in direct correlation with the verbal person:36 the verb in the first person, which refers to the author and is therefore 36. Cf. Jones (1991:€95) who remarks that verbs in the first and the second person singular appear in interactive contexts and that this explains their placement leftwards in the sentence.

 Constituent order in Classical Latin prose

situationally given, may follow as well as precede the AcI clause. It is anteposed in 23% of cases (this figure is not in Table 15), which is 5% more than the average (18% in Table 15). Two examples are given in (1) and (2). They show, among other things, that the verbs arbitror and puto are not the Focus of their sentences but the complement clauses are. In addition, the constituents Albianum (1) and Lanuvium (2) are contextually independent. (1) Albianum te confecisse arbitror. Albius’s affair-acc you-acc have settled I suppose ‘I suppose you have settled Albius’s affair.’ (2)

(Cic. Att. 14.18.2)

[Ibi te quam primum per videre velim.] Puto enim nobis Lanuvium eundum et quidem non I think indeed us-dat to Lanuvium must go and this neg sine multo sermone. without much talk ‘[I should like very much to see you there as soon as possible.] For I think we shall have to go to Lanuvium, which will need much talk.’ (Cic. Att. 15.4.2)

The AcI clause usually conveys new information and thus carries Focus function. However, not all the instances allow such an interpretation. In (3) we have an example of an AcI construction that reproduces a suggestion made by Atticus (‘go and see Brutus’). Note the parenthesis ut scribis and the passive construction referring to the author (a me). The AcI clause is not Focus, but the verb puto, expressing an intention of the author, is. The postposition of the verb can be explained by the fact that the content of the AcI clause is contextually given. (3) Brutum, ut scribis, visum iri a me puto. Brutus-acc as you write seen-acc is going by me I think ‘Brutus, as you say, I think I shall see him.’

(Cic. Att. 15.25)

Table 15 provides us also with interesting information about the insertion of a verb of thinking in the AcI clause. The insertion represents more than one fourth of all occurrences (29%), which is a relatively high figure. Such arrangements can be explained by pragmatic reasons. In particular, constituents that are placed first function as Topic; for example, Sestium in (4). Other orderings are found, especially when the sentence contains emphatic constituents. These tend to occupy extremities of the sentence, as is shown in (5) with contrastive hominem and emphatic neminem: (4) De Buciliano, Sestium puto non moleste ferre. as for Bucilianus Sestius-acc I think neg badly take ‘As for Bucilianus, I don’t think Sestius is annoyed about him.’(Cic. Att. 15.17.2)



Chapter 3.╇ Declarative sentences 

(5)

Vtrum autem sit melius, dii inmortales sciunt, which one but is better gods immortal-nom know hominem quidem scire arbitror neminem. man-acc part know I think no one-acc ‘Which of the two, however, is better the immortal gods know; no human being, I think, does know.’ (Cic. Tusc. 1.99)

To sum up this section, the placement of verbs of thinking before or after the AcI clause they govern is difficult to explain. In a more general way, there is a tendency for postposition of the verb, even if it does not have Focus function. There is not a direct correlation between the placement of the verb and the first or third person. More detailed research in the domain of complex sentences needs to be done in order to formulate more nuanced conclusions. In addition, instances of insertion of the verb in the AcI clause represent a case of violation of the rule of domain integrity: one domain (AcI clause) is being interrupted by a constituent (verb) belonging to another domain. Such arrangements are chosen for pragmatic reasons: priority is given to the Topic and contrastive constituents.

4.2

Verbs of communication

This section is devoted to several verbs of communication: dico, aio ‘to say’ and nuntio ‘to announce’, nego ‘to deny’, inquam ‘to say’, quaero, rogo ‘to ask’ and respondeo ‘to answer’. The AcI clause that can expand some of them is examined in Section 4.2.1. The two following sections deal with some particular questions, especially direct and indirect speech and the parenthetical verb inquam. 4.2.1 Verbs of saying + AcI clause Verbs of saying are constructed with an Accusative and Infinitive clause. Just as in the case of verbs of thinking (see preceding section), the verb takes several positions with respect to the governed infinitive clause: it precedes or follows it, or is inserted inside it. Table 16 presents the result of this examination: Table 16.╇ Verbs of saying and thinking (corpus 1) Verb dico ‘to say’ nego ‘to deny’ aio, nuntio ‘to announce’ Total

AcI + verb numbers percent.

verb + AcI numbers percent.

inserted verb numbers percent.

19 â•⁄ 4 â•⁄ 6

41% 14% 21%

17 21 14

37% 75% 48%

10 â•⁄ 3 â•⁄ 9

22% 11% 31%

29

28%

52

51%

22

21%

 Constituent order in Classical Latin prose

Table 16 shows that there are striking differences between the verbs under consideration: dico can stand either before or after its infinitive clause; nego mainly precedes it; aio and nuntio prefer anteposition. Insertion of the verb is relatively frequent with the last two mentioned verbs. I will now formulate the tendencies of their positioning. The verb dico ‘to say’ stands before or after its subordinate clause in almost equal proportions (41% and 37%). Its anteposition with respect to the AcI clause may be due to the contextual givenness of the verb itself, as in (6), where dico is deducible from quaesivit. Indeed, if one is questioned, one is expected to say something. Also when dico is accompanied by a contextually bound argument, for example the indirect object ei in (7), it does not come after its AcI clause. (6)

[Quaesivit ex me pater qualis esset fama.] Dixi nihil sane me audisse... nisi de ore et patre. I said nothing really me-acc had heard except about face and father ‘[His father asked me what sort of a reputation she has.] I said... that I had heard nothing... except about her looks and her father.’ (Cic. Att. 15.29.2)

(7)

[Tabellarius quem ad Brutum miseram ex itinere rediit VII Kal.;] ei Servilia dixit eo die Brutum him-dat Servilia-nom told that day Brutus-acc hora II S profectum. at hour first and half set out ‘[The courier whom I sent to Brutus turned back on the 25th.] Servilia told him that Brutus had set out at half-past six that morning.’ (Cic. Att. 15.24)

Except for these two explanations, I find the question about anteposition and postposition of dico with respect to the AcI clause difficult to answer; I leave it for further research. In addition, the verb dico can also be inserted inside the AcI clause; and not only single constituents (for example Topics) may precede the verb, but also more than one, as is shown in (8). (8)

Cum ϕιλοϕρόνως ex eo quaererem quid opus esset, when in a friendly way from him I asked what need was Atticum se dixit quaerere. Atticus-acc refl-acc he said look for ‘And when I asked him in a friendly way what I could do for him he replied that he was looking for Atticus.’ (Cic. Att. 15.15.2)

The verb nego ‘to deny, to say that... not’ precedes its AcI clause in the majority of cases, as in (9); Epicurus is a contextually bound constituent. The anteposition is due to its semantic force: it is a negative verb bearing emphasis. Unlike other verbs of saying, the verb nego also stands before the AcI clause – and not in sentence-final position – in the case of complex sentences, for example in (10).



Chapter 3.╇ Declarative sentences 

(9)

negat Epicurus iucunde posse vivi, nisi cum virtute denies Epicurus-nom pleasantly can live-pass unless with virtue vivatur, negat ullam in sapientem vim esse fortunae... live-pass denies any-acc over wise man power-acc be of fortune ‘Epicurus says that it is not possible to live pleasantly except by the exercise of virtue, that fortune has no power over the wise person.’ (Cic. Tusc. 3.49)

(10)

tum ego, etsi ἐβδελυττόμην, tamen negavi putare then I even if disgusted still I said not think me illa esse vera. me-acc these-acc be true-acc ‘On that, disgusted as I was, I said I did not believe the stories.’(Cic. Att. 15.29.2)

The verbs aio ‘to say’ and nuntio ‘to announce’ slightly privilege anteposition, which is illustrated in (11).37 They are often inserted inside their AcI clause, especially when a Topic constituent has priority, such as Sextum in (12). Postposition is not very frequent.€This variability in placement could be ascribed to the fact that these verbs do not convey new information themselves: this is contained in the AcI clause, as is shown in (13) with Focus in initial position. (11)

[Is hinc VII Id.;] ait enim attributionem in Idus, se autem urgeri acriter. he says indeed assignment-acc on Ides refl but be pressed strongly ‘[He leaves on the 9th,] for he says he has a transfer of claim due on the Ides and is hard pressed. (Cic. Att. 16.1.6)

(12) Sextum autem nuntiant cum una solum legione fuisse ad Carthaginem. Sextus-acc but they report with one only legion was at Cathage ‘They further report that Sextus had been gone to New Carthage with only one legion.’ (Cic. Att. 16.4.2) (13) Quid quaeris? Perisse omnia aiebat. what you ask perished everything he said ‘In short, he said everything was finished.’

(Cic. Att. 14.1.1)

In sum, the examination of verbs of saying shows that there are remarkable differences between the verbs themselves, and that the tendency for positioning of the AcI clause is difficult to explain. The question of the placement of these verbs seems to be related to the pragmatic value of the infinitive clause as well as to other factors, such as the contextual dependency of the verb and the presence or absence of a Topic constituent. There is another problem, pointed out by Bolkestein (1989); namely, the relationship between the complexity of constituents and their placement in the sentence. According to the theory of Functional Grammar, syntactically complex constituents are 37. Cf. also ThLL, s. v. (1455–1456); the anteposition of aio appears to be the most frequent order.

 Constituent order in Classical Latin prose

expected to be placed rightwards in the sentence, and to do this independently of their pragmatic status. This tendency, called the LIPOC rule, might apply to AcI clauses. However, according to Bolkestein (1989:€ 24 ff.), AcI clauses show placement rightwards in the sentence less often than subordinate clauses introduced by a subordinator and containing a finite verb. In addition, there are different tendencies in individual authors. I do not believe that the LIPOC rule can explain the placement of AcI clauses in Latin complex sentences; in contrast, pragmatic value and contextual status seem to be more determinative. All these aspects call for further systematic investigation. 4.2.2 Verbs of saying and direct and indirect speech or questions The verbs dico, aio ‘to say’, but also respondeo ‘to reply’, quaero and rogo ‘to ask’, may accompany direct or indirect speech or introduce a direct or indirect question. In such situations, they do not bear Focus function. When accompanying direct speech, the verb dico usually precedes it.38 (14) Cum igitur ‘nosce te’ dicit, hoc dicit: ‘nosce animum tuum’. when therefore know you-acc he says this he says know soul-acc your-acc ‘When Apollo says, “Know thyself ”, he says this “Know thy soul”.’(Cic. Tusc. 1.52) The verbs quaero and rogo ‘to ask’ typically introduce an indirect question, which usually comes after the governing verb. The sentence in (15) answers the underlying question ‘what happened?’, and this explains the initial placement of quaesivit. Respondeo ‘to reply’ in (16), preceded by a Sentence Topic (ad ea) and a Discourse Topic (Caesar), introduces indirect speech. (15) [Ait hic sibi Tutiam ferre...] quaesivit ex me pater qualis esset fama. asked from me father-nom what was reputation ‘[He says that Tutia is making him proposals...] His father asked me what sort of a reputation she has.’ (Cic. Att. 15.29.2) (16)

Ad ea Caesar respondit nulli omnium has partes to this Caesar replied no one-dat of all these parts-acc vel querimoniae vel miserationis minus convenisse. either of complaints or of compassion less suited ‘To this Caesar replied that no one was less well suited to make these complaints and these appeals for pity.’ (Caes. Civ. 1.85.1)

However, in the case of indirect questions when the speaker reproduces words pronounced by someone else, the verb may appear in postposition, as in (17):€

38. Cf. ThLL, s. v. (982 ff.).€According to the author of the article in ThLL, when dico follows the direct speech, it is a fictive speech like nosce te dicit (Cic. Tusc. 1.52) in the example (14). The verb dico can also be inserted in direct speech.



Chapter 3.╇ Declarative sentences 

(17) Rusne iam redierim quaeris. to country=q already I have come back you ask ‘You ask whether I have come back yet to the country.’

(Cic. Att. 15.6.2)

4.2.3 The verb inquam as a parenthetical expression The verb inquam, especially in the first and third person singular, frequently functions as a parenthetical expression. It is rare in historical narrative, but quite frequent in Cicero’s speeches and in his Tusculan Disputations.39 Inquam or inquit accompany direct speech, an interrogative or a declarative sentence. With inquam the explicit subject ego40 is rare, as is a personal pronoun referring to the addressee.41 Both inquam and inquit are inserted; in the sentences in which they appear, one or several constituents can precede them (18). Insertion is not obligatory; these verbs can also be placed after the interrogative or declarative sentence that they accompany, as in (19), but they never precede it (Berger 1942:€346).42 (18) ‘Quas tu mihi,’ inquit, ‘intercessiones? Quas religiones?’ what-acc you me-dat he says vetoes-acc what sanctities-acc “What vetoes,” he says, “what sanctities are you telling me of?”(Cic. Phil. 1.25) (19) Cum ille diceret ita futurum, ‘quid igitur festinamus?’ inquam. when he said so will be why therefore we hurry I say ‘When he replied that it would be so, I said, “What is our hurry then?”’  (Cic. Att. 16.15.3)

39. My corpus offers 12 occurrences of inquam, 41 of inquit, and one of inquiunt. 40. The subject of inquit in the third person is usually a maintained one. When the subject is made explicit (by a proper name, ille or a common noun), it is postposed to the verb; for example: inquit Calenus (Cic. Phil. 12.4), or in a separated way: ‘Licet,’ inquit, ‘rogare?’ Philippus. (Cic. De orat. 2.245) ‘May I examine him?, said Philippus’. When the subject comes first, as in particular in Cicero’s De oratore, it signals a change of speaker; for example: ‘Nos vero et valde quidem’, Catulus inquit, ‘putamus’ (Cic. De orat. 2.31) ‘Assuredly we think so most decidedly, said Catulus’. 41. It is rare to find an instance like the following one: Tum Quintus ‘en’, inquit mihi, ‘haec ego patior cottidie’. (Cic. Att. 5.1.3) ‘Thereupon Quintus said to me, “There, that is what I have to put up with every day!”’ However, the use of mihi is fully justified because the speaker addresses his words to only one person and not to the whole audience. 42. Cf. ThLL, 1787–1797. See also inquam and inquit in Cicero (BTL). ThLL (1795 f.) gives a few examples with an inquam or inquit introducing the quoted sentence from non-classical and mostly Late authors.

 Constituent order in Classical Latin prose

In my corpus, inquam and inquit stand after an emphatic word or a Focus constituent, for example magnam in (20), in 33% of the cases;43 note the discontinuity of the noun phrase (magnam... religionem). (20) ‘Dedicatio magnam,’ inquit, ‘habet religionem.’ dedication-nom grave-acc he says has binding force-acc ‘“A dedication” says he, “has a grave binding force.”’ (Cic. Dom. 127) In the first person singular, inquam has a specific use: it accompanies a repeated word or a noun phrase after a digression (21). The repetition of a word (or a group of words) is linked with emphasis. (21)

Hanc tv, P. Dolabella, hanc tv, inquam, potuisti this-acc you P. Dolabella-voc this-acc you I say you could aequo animo tantam dignitatem deponere? with calm mind such-acc dignity-acc lay aside ‘Could you, Dolabella, could you, I say, lay this dignity aside without emotion?’ (Cic. Phil. 1.31)

However, it would be incorrect to say that inquam or inquit always follows a Focus constituent and thus permits us to identify it as such. This parenthetical verb is also found after a subordinator or a connector, which can hardly be focal. An example is given in (22) with Focus on iratus est that indicates the reason for worry; neither quod nor mihi are Focus. (22)

[At ille: ‘sed me maxime angit avunculus’. ‘Quidnam?’ inquam.] ‘Quod mihi’, inquit, ‘iratus est’. because me-dat he said angry is ‘[He went on: “What distresses me most is my uncle.” “How so?” said I.] “Because he is angry with me,” said he.’ (Cic. Att. 13.42.1)

5. Bivalent intransitive verbs This section is devoted to bivalent intransitive verbs such as proficiscor ‘to set out’, which requires an expression of direction. I will concentrate on the placement of the directional complements (Section 5.2) as well as on cases of complex information (Section 5.3) that can be conveyed by such sentences. Finally, I add some remarks concerning the reflexive verb se recipere ‘to retire’ (5.4).

43. Cf. Bolkestein (1998b); she claims that parentheses are mainly followed by focal words.



Chapter 3.╇ Declarative sentences 

5.1

Data

My corpus 1, especially the historical narratives, offers a great number of verbs of movement (103 in total). The prototypical one is proficiscor ad or in ‘to set out to’, accompanied by a complement of direction referring to a person or to a place, or occasionally a combination of both. Other verbs in this category are for example pergo ‘to proceed’, venio ‘to come’, and redeo ‘to return’.44 Directional complements denoting a locality are represented by towns, by countries, or by common nouns such as castra ‘camp’. These make up the majority of complements (74%). Persons to whom one goes are less frequently chosen as a destination (20%). A combination of both types (locality and person) occurs in only 5% of cases. The relative positions of the subject (A1), the directional complement (D), and the verb (V) are indicated in Table 17. There are two interesting points to be mentioned: on the one hand, the most frequent pattern {(first argument >) directional complement > verb} represents 68% of all instances (compare it with the results for duco ‘to lead’ in Table 9 above, p. 141). On the other hand, all possibilities of placement are attested, even if some patterns occur only once. It is also remarkable that directional complements figure after the verb in a quarter of the cases (24%). This calls for another complementary examination, because the placement of directional complements in post-verbal position might correlate with their contextual status. At the same time, I will pay attention to the topicalisation of Table 17. ╇ Verbs of the proficiscor type (corpus 1) Pattern A1 > D > V D>V A1 > V > D V>D V > D > A1 V > A1 > D D > A1 > V D > V > A1 Total

Caesar

Sallust

Cicero

Total

Percent.

21 24 â•⁄ 8 â•⁄ 1 â•⁄ 1 â•⁄ 0 â•⁄ 5 â•⁄ 0 60

19 11 â•⁄ 3 â•⁄ 6 â•⁄ 0 â•⁄ 0 â•⁄ 1 â•⁄ 0 40

â•⁄ 4 â•⁄ 4 â•⁄ 0 11 â•⁄ 0 â•⁄ 1 â•⁄ 0 â•⁄ 1 21

â•⁄ 44 â•⁄ 39 â•⁄ 11 â•⁄ 18 â•⁄â•⁄ 1 â•⁄â•⁄ 1 â•⁄â•⁄ 6 â•⁄â•⁄ 1 121

â•⁄ 68% â•⁄ 24% â•⁄â•⁄ 1% â•⁄â•⁄ 1% â•⁄â•⁄ 5% â•⁄â•⁄ 1% 100%

44. Instances without explicit directional complements, as well as abstract uses (for example, ad rem venio Cic. Att. 16.15.1) and expressions of finality (ad exploranda loca proficiscitur ‘he sets forth to explore the district’, Caes. Civ. 1.66.3), were disregarded here. The placement of source expressions (ex castris proficiscuntur ‘they depart from the camp’, Caes. Civ. 1.78.3) were not examined either.

 Constituent order in Classical Latin prose

Table 18.╇ Directional complements of proficiscor: contextual dependency and pronominal anaphoric continuation (corpus 1) Placement

Bound + An. – An.

Pre-verbal Post-verbal Other Total Percentage

10 â•⁄ 2 â•⁄ 0 12 13%

32 â•⁄ 2 â•⁄ 5 39 41%

Unbound + An. – An. 13 â•⁄ 7 â•⁄ 2 22 23%

11 â•⁄ 8 â•⁄ 3 22 23%

Total 66 19 10 95€

â•⁄ 69% â•⁄ 20% â•⁄ 11% 100% 100%

these constituents in the subsequent sentence. The results are presented in Table 18; anaphoric continuation is indicated as + An. and its absence as – An.45 Directional complements expressed by a noun or a noun phrase may be either contextually dependent or independent. As we can see in Table 18, bound constituents (10 + 32 occ. out of 66) are mostly found in immediately pre-verbal position, whereas unbound constituents (7 + 8 occ. out of 19) occupy the post-verbal position. As for pronominal anaphoric continuation, contextually bound constituents are usually not topicalised afterwards (39 occ. out of 51), but unbound ones are more frequently (22 occ. out of 44).

5.2

Placement of directional complements

I will first analyse sentences with a single Focus constituent, and will start with the positioning of directional complements that are the strongest candidates for Focus function in sentences containing a verb of movement. Directional complements placed in pre-verbal position are usually proper names referring to persons or localities (such as contextually independent Narbonem and Massiliam in 1), or, less frequently, common nouns; these usually have a definite reading, as does castellum ‘the fortress’ in (2). Although contextually dependent, these directional complements can represent salient information; in that case they are developed further in the following context.

45. Table 18 does not include directional complements expressed by anaphoric pronouns (14 occ.) that are per se contextually bound, as well as personal pronouns (ad me, 3 occ.), that are situationally given. It does not include the co-occurrence of two directional complements and two other special cases.



Chapter 3.╇ Declarative sentences 

(1) (Caesar...) Tarracone discedit pedibusque Narbonem from Tarraco he departs by land=and to Narbo atque inde Massiliam pervenit. Ibi... cognoscit. and thence to Massilia arrives there he learns ‘(Caesar...) left Tarraco, and travelled by land to Narbo, and thence to Massilia. There he heard that...’ (Caes. Civ. 2.21.5) (2) [castellum temptet...] Igitur diu multumque fatigati so long time much=and tired tandem in castellum perveniunt, desertum ab ea parte... at last to fortress they arrive deserted from that side ‘[to attack the stronghold...] So, after a considerable period of great tiredness, they at last reached the stronghold, deserted on that side...’ (Sal. Jug. 94.3) Post-verbal placement of directional complements is mainly found in Sallust, when he talks about localities that are new in the story.46 He uses common nouns such as locus ‘place’, as in (3), with an indefinite reading; sometimes accompanied by an indefinite pronoun like quidam ‘a certain’. However, such a tendency is also observed in Caesar and Cicero, who may also place directional complements post-verbally, as in (4). Dein tertia multo ante lucis adventum pervenit in then on third day much before of light arrival he arrives in locum tumulosum ab Capsa non amplius duum milium place hillocky from Capsa not more of two miles intervallo, ibique... opperitur. distance there=and waits ‘Then, on the third night, considerably before sunrise, Marius arrived at a hillocky area, not more than two miles’ distance from Capsa, and there he waited...’  (Sal. Jug. 91.3)

(3)



(4) [in Cumanum... Ibi bene valentem videram Piliam, quin etiam paulo post Cumis eam vidi;] venerat enim in funus, cui funeri ego quoque operam dedi. she had come indeed to funeral which funeral-dat I also attended ‘[near Cumae... There I found Pilia in good health. Indeed, I saw her soon afterwards at Cumae,] for she had come to a funeral, which I also attended.’  (Cic. Att. 15.1a.1) Directional complements in another sentence-position are less frequent but neither exceptional nor impossible. Two different pragmatic values are concerned. Firstly, the directional complement can function as Focus of the sentence, as Tarraconem in (5). This town, not recently mentioned, represents the scene of the subsequent events. This sentence thus focuses on Tarraconem and not on the time expression (paucis diebus) in 46. The Jugurthine War is set in Africa, in places that were hardly known to Roman readers.

 Constituent order in Classical Latin prose

pre-verbal position. Secondly, the Focus function can be carried by another constituent. Such a situation is illustrated in (6) with a contextually bound directional complement (Brundisium). On the basis of the following context, the constituent indicating who he arrives with (cum legionibus VI) can be identified as Focus. (5) Ipse iis navibus... Tarraconem paucis diebus pervenit. Ibi... himself with these ships to Tarraco in few days arrives there ‘He himself reached Tarraco in a few days with the ships... There...’  (Caes. Civ. 2.21.4) (6) [ipse Brundisium sit venturus...] His datis mandatis Brundisium cum legionibus VI these instructions being given to Brundisium with legions-abl six pervenit, veteranis III et reliquis quas... he arrives veteran three-abl and other-abl that [Domitianas enim cohortes protinus a Corfinio in Siciliam miserat.]  (Caes. Civ. 1.25.1) ‘[He himself was on the way to Brundusium...] After dispatching this message, he arrived at Brundisium with six legions, three of them veteran, the remainder those which... [As for Domitius’s cohorts, he had sent them directly from Corfinium to Sicily.]’ Directional complements can refer to persons as well as to localities. It is possible to find them both together in one sentence. In such cases, there is no fixed order, for example {locality > person} or {person > locality}. The ordering depends on their contextual status, as is shown in the following examples. In (7) the destination of the Massiliots, the fortress of Taurois, is contextually independent, whereas Nasidius is well established in the discourse. This makes it possible to identify Tauroenta as the salient element and ad Nasidium as its further specification; the reader already knows Nasidius but he does not know where he was based. This interpretation is confirmed by the anaphoric continuation of the locality (ibi). In (8), the person (ad Domitium Ahenobarbum) is mentioned first, the locality (Corfinium) comes after, they are both contextually independent. As the following context concerns Domitius in particular, I would say that here the person represents Focus and the place is its further specification. Note also that neither Tauroenta in (7), nor ad Domitium Ahenobarbum in (8) figures in immediately pre-verbal position. (7)

[Massilienses... ex portu exeunt] et Tauroenta, quod est castellum Massilensium, and to Taurois which is fortress-nom of Massiliots ad Nasidium perveniunt ibique naves expediunt... to Nasidius they arrive there=and ships-acc get ready ‘[The Massiliots... leave the port] and reach Nasidius at Taurois, a fortified outpost of Massilia, and there get their ships ready...’  (Caes. Civ. 2.4.5)



Chapter 3.╇ Declarative sentences 

(8)

5.3

Cum his ad Domitium Ahenobarbum Corfinium magnis with these to Domitius Ahenobarbus to Corfinium by forced itineribus pervenit Caesaremque adesse cum legionibus marches he arrives Caesar-acc=and be present with legions duabus nuntiat. Domitius... two announces Domitius-nom ‘With these he made his way by forced marches to Domitius Ahenobarbus in Corfinium and broke the news that Caesar was close upon them with two legions. Domitius...’ (Caes. Civ. 1.15.6)

Complex information

In the preceding section, I analysed sentences in which directional complements are evidently salient; in this section, I will pay attention to instances of complex information conveyed by a directional complement together with the verb. A good example of such a pragmatic value is given in (9): the clause containing the verb pervenit answers the question ‘what does Otacilius do next?’, and not ‘where does Otacilius go?’ There is no continuation of the place nor of Otacilius; this person disappears completely from the scene and the narration continues with another Topic, Antonius. (9)

Otacilius sibi timens ex oppido fugit et ad Pompeium pervenit. Otacilius-nom refl-dat afraid from town flees and to Pompey arrives [Expositis omnibus copiis Antonius...] ‘Otacilius, in fear for his life, fled the town and made his way to Pompey. [When Antonius had lauded all his forces...]’ (Caes. Civ. 3.29.1)

Apart from clear instances such as (9), there are cases that are more difficult to judge; especially if they contain a contextually given directional complement that does not exhibit a direct anaphoric continuation afterwards. Consider the following example: (10) Recepto Caesar Orico nulla interposita mora being taken Caesar-nom Oricum-abl with no intervening delay Apolloniam proficiscitur. Cuius adventu audito to Apollonia sets out whose arrival being heard [L. Staberius, qui ibi praeerat, aquam comportare in arcem...] ‘After taking Oricum, Caesar immediately set out for Apollonia. [When Lucius Staberius, who was in command there, heard the news of his approach, he began to take water up to the citadel.]’ (Caes. Civ. 3.12.1) The information conveyed is that Caesar goes to Apollonia, which is picked up by cuius adventu audito ‘hearing of his approach’. Two explanations can be envisaged here. Firstly, cuius adventu can be interpreted as resuming the preceding information and thus Apolloniam proficiscitur functions as a pragmatic unit. Secondly, Apollonia can be

 Constituent order in Classical Latin prose

viewed as a salient element permitting a shift of the setting from Oricum to Apollonia accompanied by a Topic shift from Caesar to Staberius. In historical narratives, departure to some place often signals a change of scene (cf. Chausserie-Laprée 1969:€21). For example, the indication that the enemy is approaching Rutilius’s camp permits the author to start a description of the situation on the Roman side in (11). Romani, deducible from the directional complement (ad Rutili castra), is selected as Topic afterwards. In such cases, it is difficult to decide whether the directional complement (ad Rutili castra) is the most salient information and marks a change of scene, or forms a pragmatic unit with the verb. (11) (Bomilcar) aciem... latius porrigit eoque modo line-acc broadly he extends in this=and way ad Rutili castra procedit. Romani to of Rutilius camp-acc advances Romans-nom [ex improviso pulveris vim magnam animadvertunt.] ‘(Bomilcar) extended his line along a broader front... and in that manner he advanced towards Rutilius’s camp. The Romans [noticed unexpectedly a great mass of dust.]’ (Sal. Jug. 52.6 – 53.1)

5.4

The reflexive verb se recipere ‘to retire’

The semantic group of verbs such as proficiscor also includes one frequent reflexive verb, se recipere ad or in ‘to retire (somewhere)’, that mainly appears in Caesar’s and Sallust’s historical narrative. As reflexive verbs are not very numerous in Latin, se recipere is worth examining in particular for the placement of the pronoun se, which is also used in its reduplicated form sese. My corpus 1 offers 20 occurrences of this verb with the following patterns; the first one occurs 16 times, the second one four times: (first argument >) se(se) > directional complement > verb (first argument >) directional complement > se(se) > verb Possible placements of se(se) are exemplified in (12) – (15), with an indication of the colon subdivision (/) in (13). (12)

[Legio quae in eo cornu constiterat locum non tenuit] atque in proximum collem sese recepit. [Caesar...] and to nearest hill refl withdrew ‘[The legion which had taken up station on that wing did not maintain its position] and withdrew to the nearest hill. [Caesar...]’ (Caes. Civ. 1.44.4)

(13) Reliquae Caesaris naves / cognita eius fuga / remaining-nom of Caesar ships-nom being learnt his flight-abl se Hadrumetum receperunt. refl to Hadrumetum withdrew ‘On Caesar’s flight, the rest of his ships withdrew to Hadrumetum.’  (Caes. Civ. 2.23.4)



Chapter 3.╇ Declarative sentences 

(14) Diebusque post paucis se in regnum cum omnibus copiis recepit. days=and after few refl to kingdom with all forces withdrew ‘A few days later he withdrew with his entire army to his kingdom.’  (Caes. Civ. 2.44.3) (15)

Omnis Vari acies terga vertit seque whole-nom of Varus formation-nom turned round refl=and in castra recepit. to camp withdrew ‘Varus’s whole formation turned round and retreated to camp.’(Caes. Civ. 2.34.6)

The pronoun se does not exhibit a fixed position in the sentence or clause (cf. Spevak forthcoming a). Not only is it not attracted by the verb but it can also occupy any position, including the first position in its colon (13), after temporal setting (14) and the clause-initial position (15; note that se hosts here the enclitic coordinator -que). How to explain these placements of se? Sentences containing the verb se recipere usually answer the question ‘what does he do?’ because directional complements are often contextually given and lack anaphoric continuation. When no Topic is present, which is typically the case with clause-initial position (15), se is the least informative element, co-referential with the subject of the preceding clause. Furthermore, in a number of cases, the reflexive se, together with the verb recipere, seems to frame the focal part of the sentence, represented by the verb and its complements; see the examples in (13) – (15).47

5.5

Conclusions

The analysis of verbs of the proficiscor type has shown that obligatory directional complements usually occur in the pre-verbal position; but they can also appear after the verb, or in another position. Their placement is not an indication of their pragmatic function. More generally, contextually unbound complements with anaphoric continuation in the subsequent sentence can be considered as Focus; the underlying question is in particular ‘where?’ Information conveyed by these sentences is often complex, especially when directional complements are contextually given without anaphoric continuation. In this case, the underlying question is often ‘what does he do?’

47. In proximum collem in (12) may be the salient element; the immediately following context presents a change of Topic.

 Constituent order in Classical Latin prose

6. Monovalent verbs 6.1

Introduction

Monovalent verbs together with the existential verb sum (see p. 187 ff.) have drawn the attention of many scholars because they quite frequently exhibit the ordering {verb > subject}. Furthermore, the verb in initial position is commonly considered as “emphatic” (Kühner and Stegmann 1914, II: 590, among others)48 and this ordering is usually called “inversion” of the subject and the verb. Several explanations of the ordering {verb > subject} have been proposed. Firstly, it is considered a phenomenon that has to do with the text structure, as a mark of discontinuity or “rupture” in the discourse.49 This use is exemplified in (1). Pulchellus puer ‘this pretty boy’ is contextually given50 and the sentence focuses on the verb. Secondly, the order {verb > subject} is found in descriptions and expressions of accompanying circumstances (Luraghi 1995:€367 f.), as in (2). Finally, the phenomenon of {verb > subject} ordering is attributed to the intransitive character of the verbs. This point was signalled in particular by Adams (1976b: 126). More recently, Bolkestein (1995:€35) showed that semantic and syntactic properties of intransitive verbs are in correlation with the {verb > subject} ordering, since it appears more frequently (in 27% of cases) than in the case of transitive verbs (13%). (1)

[Redeo ad altercationem.] Surgit pulchellus puer, obicit mihi me ad Baias fuisse. gets up very pretty boy-nom reproaches me-dat me-acc at Baiae had been ‘[I return to the dispute.] Up gets this pretty boy and reproaches me with spending my time at Baiae.’ (Cic. Att. 1.16.10)

(2) Augebatur auxiliorum cotidie spes. was increasing of reinforcements daily hope-nom ‘Their hope of seeing the reinforcements was increasing daily.’(Caes. Gal. 6.7.6)51 However, as I will argue in this section, the {verb > first argument} ordering is not an inherent property of monovalent verbs but a result of their semantic and syntactic values. Neither would I attribute to monovalent verbs functions such as “interruption” or “discontinuity” (cf. p. 145). The sentences containing monovalent verbs examined 48. For a detailed discussion about the initial verb in Latin, see Spevak (2005b). 49. See Chausserie-Laprée (1969:€347 ff.), and Luraghi (1995:€370). The discursive functions of the {verb > subject} ordering proposed by Bolkestein (1996b) are based on the distinction of “thetic” and “categorical” sentences, established by Sasse (1995); however, see the remarks by Rosengren (1997). 50. Pulchellus puer refers back to Clodius; Cicero plays on his cognomen Pulcher ‘Pretty’. This example is quoted and commented on by Bolkestein (1996b: 12). 51. Example quoted by Luraghi (1995:€368).



Chapter 3.╇ Declarative sentences 

in this section correspond partially52 to what are called “presentative sentences” (Hannay 1985a: 9), which means sentences introducing new entities into the discourse. They contain a verb meaning ‘to be’, ‘to appear’, ‘to arrive’, ‘to remain’, and other similar verbs. Although it is impossible to give an exhaustive list of the verbs concerned, presentative sentences distinguish themselves by certain common properties, in particular by a contextually independent subject placed rightwards in the sentence (Dik 1997, I: 315). A typical example is given in (3a), in Italian, that can be compared with (4a), which exhibits a definite subject. The post-verbal placement is only possible in the case of indefinite subjects. In French, the pre-verbal position is filled with a dummy pronoun il in (3b).53 (3) a. è partito un treno. be-aux left a train b. Il est parti un train. it be-aux left a train (4) a. Il treno è partito. the train be-aux left b. Le train est parti. the train be-aux left The post-verbal placement of subjects is also observed in Latin, as has been stated by Pinkster (1991:€78 and 1995:€237), by Bolkestein (1996b: 10), and by Rosén (1998); for example, una via in (5).54 (3) Relinquebatur una per Sequanos via, qua… remained-pass one-nom through Sequani way-nom by which ‘Only one way remained through the land of the Sequani, by which...’  (Caes. Gal. 1.9.1) In this chapter, the monovalent verbs under examination are divided into three semantic categories: verbs of the type accurro ‘to run’ (Section 6.3), nascor ‘to be born’ (Section 6.4), and permaneo ‘to remain’ (Section 6.5). Particular attention will be paid to the contextual status and the positioning of their subjects.

52. See also Section 7.3 below p. 188 ff.. 53. See Vet (1981:€151) and cf. Vincent (1988:€60) Cf. also Perrot (1978:€97) on il reste trois places. He identifies the noun phrase (trois places) as the salient element of the sentence.€ 54. See also Jones (1991:€91).

 Constituent order in Classical Latin prose

6.2

Data

Corpus 1 offered 154 monovalent verbs with an overt subject. As monovalent verbs – or verbs behaving as monovalent – have different semantic values, I divided them into three categories: 1. monovalent verbs expressing dynamic controllable actions; for example, accurro ‘to run’, discedo ‘to go away’, profugio ‘to run away’, venio ‘to come’; 2. monovalent verbs expressing dynamic uncontrollable processes, such as fio ‘to become’, morior ‘to die’, nascor ‘to be born’, vivo ‘to live’; 3. monovalent verbs expressing permanent or temporary states: desum ‘to fail’, permaneo ‘to remain’, valeo ‘to be well’ and other similar verbs. Table 19 present the results of my examination. The ordering of the verb and the first argument only concerns verbs in the third person (sg. and pl.) with an expressed nominal or pronominal subject. It is noteworthy that with monovalent verbs in the third person the first argument is expressed in 84% of all instances, which is a considerable proportion.55 The table shows that the dominant ordering is {first argument > verb}, but {verb > first argument} is also a well represented alternative. However, there are variations between authors: whereas Sallust uses almost uniformly {first argument > verb}, Cicero slightly prefers the ordering {verb > first argument}. Table 19.╇ Monovalent verbs: third person sg. and pl. (corpus 1) Pattern

Caesar

Sallust

Cicero

Total

Percentage

A1 > V V > A1 Total

23 11 34

32 â•⁄ 5 37

20 24 44

75 40 115

65% 35% 100%

Table 20.╇ Nominal first arguments with monovalent verbs (corpus 1) Pattern A1 > V V > A1 Total

bound unbound bound unbound

animate

inanimate

Total (also %)

29 12 â•⁄ 5 â•⁄ 9 55

â•⁄ 5 16 â•⁄ 9 12 42

34 28 14 21 97

55. A similar observation was made by Bolkestein (1995:€33).



Chapter 3.╇ Declarative sentences 

A complementary examination will inform us about the contextual status of subjects; in this case only those expressed by a noun (phrase), because pronominal subjects are per se situationally or contextually dependent. Bound and unbound entities are considered separately, as well as animate and inanimate ones. What we can read from Table 20 is that the results – which are not influenced by literary genre – do not lead to clear conclusions.56 In the {first argument > verb} pattern, animate contextually bound subjects are found; inanimate subjects in this position are mostly unbound. In the {verb > first argument} pattern, inanimate entities are predominant, with a slight preference for unbound ones. More generally, contextually bound subjects privilege leftwards placement (about 34% versus 14%) but rightwards placement for unbound subjects is not necessary (21% versus 28%). This can only be explained by the different pragmatic values of sentences in which they occur. I will present below the main characteristics of each type, based on the corpus examined.

6.3

The accurro ‘to run’ type: dynamic controllable processes

Dynamic monovalent verbs implying movement generally have an agentive subject. The most frequent ordering, {first argument > verb}, appears when the sentence answers the underlying question ‘what happens?’ (note the presence of repente ‘suddenly’ in example 6). However, example (7) can be interpreted as answering the question ‘what do the Romans do?’ as well. Both subjects (Maurus, Romani) are contextually dependent. (6)

[Post ubi castra locata et diei vesper erat,] repente Maurus incerto voltu pavens ad Sullam adcurrit dicitque… suddenly Moor-nom with unsure face fearing to Sulla runs says=and ‘[After, when the camp had been sited and the day had turned to evening,] the Moor with an ambiguous expression suddenly ran panicking to Sulla, and said that...’ (Sal. Jug. 106.2)

(7) Denique Romani... advorso colle, sicuti praeceptum fuerat, evadunt. finally Romans-nom in opposite hill as instruction was take refuge ‘Finally the Romans... made their way uphill as they had been instructed.’  (Sal. Jug. 52.3) The sentence exemplified in (8) with the ordering {verb > first argument} has another pragmatic value: it informs us about the men who ran to Sulla (question ‘who?’). The contextually dependent subjects, legati, centuriones and tribuni, are placed after the verb. One might also explain this ordering as due to the heaviness of a complex constituent. Tabellarius ‘messenger’ in the presentative sentence given in (9) is contextually independent and Focus of its sentence. This sentence has a temporal setting as its 56. This statement is in accordance with observations made by Hoffmann (1991:€377) concerning the verb excipio in the sense of ‘to follow after’.

 Constituent order in Classical Latin prose

starting point. By contrast, the sentence in (8) lacks a Topic, so that the verb lands in the initial position. (8)

Concurrebant legati, centuriones tribunique ran legates-nom centurions-nom tribunes-nom=and militum, ne dubitaret proelium committere. of soldiers neg he would hesitate battle-acc to join ‘Senior commanders, centurions, and officers ran to him, urging him not to hesitate to join battle.’ (Caes. Civ. 1.71.2)

(9)

VIIII Kal. hora VIII fere a Q. Fufio venit tabellarius. on 24th hour 8th about from Q. Fufius came courier-nom [Nescio quid ab eo litterularum.] ‘About four o’clock in the afternoon of the 24th, a courier arrived from Quintus Fufius. [He brought me some sort of a note from him.]’ (Cic. Att. 15.4.1)

Table 20 has shown that it is impossible to view the leftwards placement of subjects as a result of their contextual boundness, and the rightwards position as reserved for – or at least strongly privileged by – contextually independent entities. The following examples further illustrate this point, and also show another aspect of intransitive sentences. L. Manlius praetor Alba cum cohortibus sex profugit, L. Manlius praetor-nom from Alba with cohorts six fled Rutilius Lupus praetor Tarracina cum tribus; quae Rutilius Lupus praetor-nom from Tarracina with three which [procul equitatum Caesaris conspicatae…] ‘The praetor Lucius Manlius fled from Alba wi th six cohorts, the praetor Rutilius Lupus from Tarracina with three; [when these sighted Caesar’s cavalry...]’  (Caes. Civ. 1.24.3)

(10)

(11) (Illi...) Quorum cognita voluntate clam profugit of whom being learned will secretly fled Apollonia Staberius. from Apollonia Staberius-nom ‘(They...) On realizing their sympathies, Staberius fled secretly from Apollonia.’  (Caes. Civ. 3.12.3) The sentence in (10) answers the question ‘what happens?’, with Focus on non obligatory arguments (cum cohortibus sex; cum tribus). L. Manlius as well as Rutilius Lupus in (10) are contextually independent constituents, secondary characters that have no special role further on in the discourse. These subjects are so to speak “set” as contrastive Topics, and occur in initial position (cf. de Jong 1989:€536; cf. above, p. 43). The sentence in (11) focuses on the action; Staberius is a contextually dependent constituent. This subject might either remain unexpressed, for there is no ambiguity, or be



Chapter 3.╇ Declarative sentences 

selected as Topic. In final position, I consider it as a sort of reminder.57 Such a postverbal or even sentence-final placement of contextually bound subjects is noteworthy, as was pointed out by Bolkestein (1995:€38).

6.4

The nascor ‘to be born’ type: dynamic uncontrollable processes

Monovalent verbs expressing uncontrollable processes are non-agentive. The following examples show a sentence with a contextually dependent subject placed initially (Micipsa in 12)58 answering the question ‘what happens to Micipsa?’, and a sentence with an independent subject placed rightwards answering the question ‘who?’ (13). (12) Micipsa paucis post diebus moritur. Micipsa-nom few after days dies ‘A few days after, Micipsa died.’ (13)

(Sal. Jug. 11.2)

V Id. cum... venissem diluculo ad pontem Tirenum... on 9th when I had come at daybreak to bridge Tirenian obviam mihi fit tabellarius, qui towards me-dat becomes messenger-nom who [me offendit δολιχὸν πλόον ὁρμαίνοντα.] ‘On the 9th, when, at daybreak... I had reached the Tirenian bridge..., I met a messenger, who [found me pondering a lenthy voyage.]’ (Cic. Att. 16.13.1)

Inanimate entities often appear with this semantic type of monovalent verbs. A sentence with Focus on an inanimate entity, placed pre-verbally and exhibiting anaphoric continuation, is shown in (14). Another ordering is illustrated in (15) with Focus on the verb; note that the subject haec sententia is contextually bound and discontinuous. Sed ex eo medio quasi collis oriebatur, in inmensum but from its middle sort of hill-nom rose vastly pertingens, vestitus oleastro... Igitur in eo colle… extending covered with wild olive so on this hill ‘(mountain) But from the middle of it, reaching an immense distance, there sprang a kind of hill, which was covered in wild olive... On this hill...’  (Sal. Jug. 48.3)

(14)

57. For similar sentence-structures, see p. 72 ff. I would not consider it a Tail constituent, by any means. 58. The death of Micipsa was expected, cf.: cum sibi finem vitae adesse intellegeret (Sal.€Jug. 9.4) ‘when he perceived that the end of his life was at hand’. However, considering the context, I do not believe that this sentence answers the question ‘when?’; it informs us that the process was accomplished.

 Constituent order in Classical Latin prose

(15)

Haec vincit in consilio sententia [et prima luce postridie this-nom wins in council opinion-nom constituunt proficisci.] ‘This view prevailed in the council, [and they decided to start out at first light next day.]’ (Caes. Civ. 1.67.6)

A verb like fio ‘to become’ almost always stands leftwards in the sentence (8 occ. out of 10) or at least precedes the subject (cf. example 13). This can be explained by its weak semantic value; consequently, fio does not bear Focus, and the underlying question is ‘what?’ Furthermore, sentences that contain it usually lack a Topic, as in (16): (16) Fit celeriter concursus in praetorium. becomes quickly crowd-nom to commander’s tent ‘A crowd quickly gathers in front of the commander’s tent.’(Caes. Civ. 1.76.2) Sentences expressing that a season or another similar circumstance approaches, such as in (17), answer the question ‘what?’ and show {first argument > verb} ordering (cf. Spevak 2004). The same order is chosen to indicate that something begins; Focus is on the action itself in (18). By contrast, we have a verum Focus in (19) that concerns prediction and can be paraphrased as ‘it will really come’. (17)

Iamque hiems adpropinquabat now=and winter-nom was approaching [et... Dyrrachium sese ad Pompeium recepit.] ‘Winter was now approaching, [and... he withdrew to Pompey at Dyrrachium.]’ (Caes. Civ. 3.9.8)

(18)

[Numidae ab Iugurtha pro tempore parati instructique,] dein proelium incipitur. then battle-nom starts ‘[The Numidians were prepared and drawn up by Jugurtha to suit the moment;] then the battle began.’ (Sal. Jug. 74.2)

(19)

Veniet tempus, et quidem celeriter, will come time-nom and this quickly [sive retractabis sive properabis; volat enim aetas]. ‘The time will come, and quickly too, [whether you hold back or hasten; life passes swiftly].’ (Cic. Tusc. 1.76)

6.5

The permaneo ‘to remain’ type: permanent or temporary states

Monovalent verbs expressing a permanent or temporary state admit animate as well as inanimate subjects. Sentences containing them answer the question ‘who’ or ‘what?’ (20) or focus on the verb itself, as in (21), with the contextually dependent subject



Chapter 3.╇ Declarative sentences 

musici. Another constituent may be focal, for example the time complement (dies complures) in (22). (20)

[...deinde multis audientibus, Servilia, Tertulla, Porcia, quaerere quid placeret.] Aderat etiam Favonius. was present also Favonius-nom ‘[...then before a large company, including Servilia, Tertulla, and Porcia; he asked me for my opinion.] Favonius too was present.’ (Cic. Att. 15.11.1)

(21) [Summam eruditionem Graeci sitam censebant in nervorum vocumque cantibus...] Ergo in Graecia musici floruerunt discebantque id omnes. therefore in Greece musicians-nom flourished learned=and it all-nom ‘[The Greeks thought that the highest culture resided in the music of strings and voices...] Accordingly musicians flourished in Greece: everybody learned music.’  (Cic. Tusc. 1.4) (22) Hae permanserunt aquae dies complures. these-nom lasted floods-nom days several ‘The floods lasted for several days.’

(Caes. Civ. 1.50.1)

Sentences containing the verb desum ‘to fail’ are expected to focus on the subject (question ‘what?’). However, this verb is mainly (13 occ. out of 14) found in final position, and its subject, typically an inanimate, contextually independent entity, comes before it. (23)

Ne militibus quidem ut defessis neque equitibus ut paucis not soldiers-dat even as tired-dat nor cavalry-dat as few et labore confectis studium ad pugnandum virtusque deerat. and by toil exhausted zeal-nom for fighting courage-nom=and failed ‘And neither the infantry though tired out, nor the cavalry, though few and exhausted by their efforts, lacked enthusiasm or courage for the fight.’  (Caes. Civ. 2.41.3)

6.6

Conclusions

The examination of monovalent verbs does not weaken the main observations that have been made before. However, as I have shown, the question concerning monovalent verbs is more complex. It would be too rough a simplification to propose a placement rule such as “bound subjects are placed leftwards and unbound ones rightwards in the sentence”. Nevertheless, we can say that the {verb > first argument} order is mainly found in sentences that lack a Topic and that sentences exhibiting subjects placed rightwards typically answer the underlying question ‘who, what?’ Multiple and further expanded subjects show the same tendency.

 Constituent order in Classical Latin prose

As in other languages, placement of an unbound subject rightwards can be found in Latin, without constituting a general rule. Latin also uses the pre-verbal or initial position for such constituents. Furthermore, initial verbs can have different functions: they can be contextually dependent as well as independent. In the latter case, verbs may either carry Focus or be pragmatically unmarked, especially in presentative sentences.

7. The verb sum ‘to be’ The verb sum appears in various types of sentences, mainly existential, locative, and predicative ones. Latin scholars are well aware that it is not always easy to determine at first glance which sum they are dealing with, because there is – as will be demonstrated in this section – absolutely no difference in placement between the copula sum and the existential sum in Latin. Furthermore, as noun phrases are relatively often discontinuous, it is not easy to distinguish quickly between attributive and predicative adjectives either. The liberty of placement of the verb sum seems to be remarkable in comparison with, for example, the Slavic languages, where we can usually determine its value on the basis of the position it occupies. Compare the Czech examples indicated below. In predicative sentences, the subject typically has a definite reading and may remain implicit as in (1). Initial placement of the verb ‘to be’ with an expressed subject would be unusual, and ungrammatical, except for existential sentences such as (2), with indefinite reading of kniha ‘a book’. At least, I do not know of any current examples of the predicative (non-enclitic) copula ‘to be’ standing in initial position in sentences with an expressed subject. The pattern exemplified in (3) is typical for interrogative sentences; it is unacceptable in a declarative one. In other words, the initial verb ‘to be’ with an expressed subject is identified either as existential or as belonging to an interrogative sentence. The copula ‘to be’ with an expressed subject is not sentence-initial in Czech. (1) Predicative (Ta) kniha je zajímavá. (this) book-nom is interesting ‘The (this) book is interesting.’ Je zajímavá. ‘It is interesting.’ (2) Existential Je zajímavá kniha (o...) is interesting book-nom (about...) ‘There is an interesting book (about...)’ (3) Interrogative Je (ta) kniha zajímavá? is (this) book-nom interesting ‘Is the (this) book interesting?’



Chapter 3.╇ Declarative sentences 

In this section, not all possible usages of sum ‘to be’ will be presented;59 I will concentrate on its use with predicative adjectives (Section 7.1) and nouns (Section 7.2), and on existential and locative sentences (Section 7.3). My aim is to describe these sentences and to find criteria for distinguishing between the copula sum and the existential or locative sum.

7.1

The predicative adjective

Marouzeau (1910), in his study devoted to the verb sum, as well as in his synthesis, (1953:€34) states that, on the one hand, “the mobility of the copula is enormous; one can find it in every sentence-position”; on the other hand, “the placement of the copula is determined with respect to the predicative adjective or noun”. He also states (Marouzeau 1938:€ 9, 23, and 1953:€ 35) that the predicative normally precedes the verb sum,60 without, however, making a distinction between predicative adjectives, nouns, pronouns, and prepositional phrases. Adams (1994a: 84) also considers the ordering of the type navigatio perdifficilis fuit “the voyage was very difficult” regular. Other orderings imply an idea of contrast (navigatio fuit perdifficilis and perdifficilis fuit navigatio); the initial verb (fvit navigatio perdifficilis) has, according to him, the value of verum Focus (‘it was actually, indeed’ or ‘it was, unlike expectations’). From my corpus 1, I collected 225 occurrences of the copula verb sum combined with an adjective. In order to analyse the placement of these constituents, I examined the positioning of the adjective with respect to the verb sum in continuous (adjective sum and sum adjective) as well as discontinuous (...) sequences. The data in Table 21 concern all instances, with or without an expressed subject.61 Table 21.╇ The verb sum and a predicative adjective: continuous and discontinuous sequences (corpus 1) Pattern adjective sum adjective... sum sum adjective sum... adjective Total

Total

Percentage

112 â•⁄ 31 â•⁄ 41 â•⁄ 41 225

â•⁄ 50% â•⁄ 14% â•⁄ 18% â•⁄ 18% 100%

59. I disregard here constructions of identification, such as Hae sunt meae imagines (Sall. Jug. 85.30) ‘these are my statues’. 60. Besides, I disagree with Marouzeau’s (1938:€9) statement that the copula is “highlighted” when the predicative follows it. 61. Most occurrences are from Cicero (66%). There are some differences between authors: Caesar uses the ordering {adjective > sum} as frequently as {sum > adjective} (10 occurrences of both); Sallust mainly uses the ordering {adjective > sum} (37 occurrences out of 46).

 Constituent order in Classical Latin prose

Table 22.╇ The verb sum, first argument and predicative adjective (corpus 1) Pattern A1 > adjective > sum A1 > sum > adjective adjective > sum > A1 adjective > A1 > sum sum > adjective > A1 sum > A1 > adjective Total

Total

Percentage

â•⁄ 58 â•⁄ 33 â•⁄ 23 â•⁄ 20 â•⁄â•⁄ 9 â•⁄ 10 153

â•⁄ 37% â•⁄ 22% â•⁄ 15% â•⁄ 13% â•⁄â•⁄ 6% â•⁄â•⁄ 7% 100%

The direct sequence {adjective sum} is the most frequent one (50%). Postposition of the adjective with respect to the verb is less frequent (18%). Nevertheless, sum and the adjective need not stand next to each other; the figures obtained for {adjective... sum} (14%) and for {sum... adjective} (18%) are not negligible. I now look at sentences with an expressed subject. The patterns found are indicated in Table 22; other elements may intervene between the first argument (A1), the predicative adjective, and/or the verb sum. As we see, all possibilities of placement are attested. The ordering {first argument > adjective > sum} is the most common (37%).62 Other patterns are found quite often; only these with initial sum are on the whole less frequent. A more detailed examination shows several placement tendencies concerning the verb sum and the predicative adjective that I summarize below. 1. The most frequent pattern {first argument > adjective > sum}, exemplified in (4), corresponds to the following description: the initial subject is contextually dependent, Focus is on the adjective, and the copula sum in sentence-final position has no special pragmatic function. However, I would not qualify this pattern as the “normal” order. I would say that contextually dependent subjects placed initially and adjectives with Focus function standing rightwards in the sentence represent the “prototypical” pragmatic situation for this syntactic type. At the same time, we cannot view placement of the predicative adjective before (or after) the copula as a syntactic constraint (pace Marouzeau 1953:€34) because the predicative adjective need not occur next to it, as we will see from the examples hereafter. (4) Domus illa mea prope tota vacua est. house-nom that my-nom almost whole empty is ‘That house of mine is almost entirely empty.’

(Cic. Dom. 116)

2. A focal predicative adjective appears in initial position when it is emphatic, as in (5) with talis ‘such’ (Cassius is contextually given). Such a placement is usual for 62. Here again, Sallust does not show much variation and uses this pattern in 78% of instances.



Chapter 3.╇ Declarative sentences 

intensive words (such as talis and tantus), but is also found for other adjectives expressing subjective evaluation, for example stulta ‘silly’ in (6), which comes after the conclusive particle itaque. The constituent consolatio is contextually bound, because the idea of consolation brought by the Ides of March is a part of shared knowledge between Cicero and Atticus.63 Focal predicative adjectives may also go in another position, as gratissimae et bellae stand after a contrastive Topic (patri) in (7). (5) Talis ea tempestate fama de Cassio erat. such-nom at that time reputation-nom of Cassius was ‘Such at that period was the reputation of Cassius.’

(Sal. Jug. 32.5)



(6) [Quod scribis te nescire quid nostris faciendum sit, iam pridem me illa ἀπορία sollicitat.] Itaque stulta iam Iduum Martiarum est consolatio. so foolish-nom now of Ides of March is consolation-nom ‘[You say you don’t know what our friends ought to do; that perplexity has been worrying me for some time.] So there is no sense any longer in consoling ourselves with the Ides of March.’ (Cic. Att. 15.4.2) (7) [De Quinto filio tibi adsentior;] patri quidem certe father-dat part certainly gratissimae et bellae tuae litterae fuerunt. very pleasant and nice-nom your letter-nom was ‘(I agree with you about Quintus junior;) your letter certainly gave much pleasure to his father and was admirably done.’ (Cic. Att. 15.1.4) 3. Other pragmatic situations can be found, even if rarely. For example, the predicative adjective in (8) does not convey salient information; the subject (Mauri) carries Focus function and is placed before the verb. Sallust here gives a geographical description. (8)

Cetera loca usque ad Mauretaniam Numidae tenent, other regions-acc up to Mauretania Numidians-nom hold proxumi Hispania Mauri sunt. nearest-nom to Spain Moors-nom are ‘The other regions up to Mauretania are held by the Numidians; nearest to Spain are the Moors.’ (Sal. Jug. 19.4)

4. It is noteworthy that the copula sum can stand in initial position in Latin.64 This ordering is found in two, closely related, special situations. On the one hand, initial 63. Cf. for example: itaque me Idus Martiae non tam consolantur quam antea. (Cic. Att. 14.22.2) ‘Accordingly, the Ides of March do not console me so much as they did.’ 64. My corpus gives 26 occurrences of such a placement, including sentences without an expressed subject.

 Constituent order in Classical Latin prose

sum appears in sentences providing an explanation, and is accompanied by the particle enim ‘for’, as illustrated in (9). Such an initial verb is not focal; Focus in (9) is carried by odiosa, whereas hiberna navigatio is contextually dependent. The initial sum is also found in other sentences belonging to the secondary, background level, mainly those that contain information of a descriptive type as in (10); Brutus is a contextually dependent constituent. However, viewing sum as a marker of “discontinuity” of the discourse (cf. Luraghi 1995:€ 370) seems incorrect to me. Sum does not have such a function; what it does have is a semantic value referring to a state. Verbs with such a semantic value do not push the narrative further, unlike verbs of action. In sentences providing an explanation or information of a descriptive type, the verb is placed in the initial position instead of the contextually bound subject that does not function as Topic. It might have been selected as Topic, but it is not; it plays no role afterwards. Another example of the same type is given in (11): Caesar is talking about a town of the Salones. Another interpretation of est – existential (‘there is a town’) or identificatory (‘it is a town’) – can be excluded due to the definite reading of oppidum ‘this town’. It is worth adding that the sentences (9) – (11) do not provide a description of these entities; they only offer explanatory observations. (9)

[De meo itinere variae sententiae... Dubitemus igitur. ] Est enim hiberna navigatio odiosa. is for wintry voyage-nom detestable-nom ‘[Views about my journey vary... So let us leave it undecided.] For a winter voyage is detestable.’ (Cic. Att. 15.25)

(10) Erat multo inferior numero navium Brutus. was much fewer in number of ships Brutus-nom ‘Brutus was greatly outnumbered in ships.’ (11)

7.2

(Caes. Civ. 1.57.1)

[... oppidum oppugnare instituit.] Est autem oppidum et loci natura et colle munitum. is but town-nom both of place by nature and by hill protected-nom ‘[...and started to attack the town.] However, the town was protected by the nature of its site and by a hill.’ (Caes. Civ. 3.9.2)

Predicative nouns (or noun phrases)

I proceed now to an examination of sentences with a predicative noun (or noun phrase). Data concerning sentences with explicit subjects (A1) are presented in Table 23.



Chapter 3.╇ Declarative sentences 

Table 23.╇ The verb sum, first argument and predicative noun (corpus 1) Pattern A1 > noun > sum A1 > sum > noun noun > sum > A1 noun > A1 > sum sum > A1 > attribut Total

Total

Percentage

20 6 5 1 3 35

57% 17% 14% 3% 9% 100%

The data in this table mostly come from Cicero’s works (30 occurrences out of 35), especially his philosophical treatises and speeches. The most frequent ordering is {first argument > predicative noun > sum} (57%); other orderings are possible and quite well represented. Two typical examples are given in (12) and (13), which demonstrate at the same time a problem with the interpretation of predicative sentences in Latin. (12) At aegritudo perturbatio est animi. but distress-nom disorder-nom is of soul ‘But distress is a disorder of the soul.’

(Cic. Tusc. 3.15)

(13) Nam corpus quidem quasi vas est aut aliquod animi receptaculum. in fact body-nom part as if vessel is or some of soul container ‘The body is in fact a kind of vessel or container for the soul.’(Cic. Tusc. 1.52) Latin uses no formal means – such as an article, or special case marking – for distinguishing between the subject (aegritudo) and the predicative noun (perturbatio); in consequence, both are theoretically eligible for both syntactic functions.65 It is not the constituent order but only the context that permits us to analyse aegritudo ‘sorrow’ as the subject; it occurs in a passage devoted to grief, and represents a contextually dependent constituent with Topic function. Perturbatio ‘disorder’ is a contextually unbound, predicative noun with Focus function. Similarly in (13), corpus is the contextually dependent Topic and the contextually unbound predicatives (vas and receptaculum) function as Focus. Here, the distinction between the subject and the predicative noun is facilitated by the word quasi expressing an approximation, the indefinite pronoun aliquod, and by the disjunctive aut connecting the two complex constituents. Difficulties with the interpretation of sentences such as (12) are due to the fact that they contain two nouns referring to entities of the same semantic type, especially if they are both abstract nouns. I give one more example of this type in (14), with gloria ‘glory’ and laus ‘praise’; the first one is contextually dependent (the subject), the second one is contextually independent and further developed (the predicative noun). This example also shows an initial placement of the copula. 65. Apparently, this did not represent a problem for native Latin speakers.

 Constituent order in Classical Latin prose

(14)

[credo enim vos... gloriam concupivisse.] Est autem gloria laus recte factorum is but glory-nom praise-nom rightly of deeds magnorumque in rem publicam meritorum. of great=and towards State of services ‘[for I believe that you... have been eager to acquire... glory.] Now glory is praise won by honourable deeds, and great services towards the State.’(Cic. Phil. 1.29)66

In other cases, semantic properties of the nouns involved in predicative constructions play an important role: nouns referring to function and status (princeps ‘chief ’, censor ‘censor’, civis ‘citizen’ and similar) as well as generic nouns (res ‘thing’, species ‘sort’, causa ‘reason’, and so on) are more likely eligible for the predicative function than nouns bearing a higher degree of individuation, for example nouns referring to persons or singular objects. Two examples are given with predicative causa in (15) and dux atque imperator in (16); animus is contextually independent. In both examples, saliency is carried by the subjects. In other words, the predicative noun cannot be identified by the proximity with the copula (pace Marouzeau 1953:€35): no such contiguity is required. (15)

Causa fuit ambulatio et monumentum et ista pretext-nom was promenade-nom and monument-nom and this Tanagraea oppressa libertate Libertas. Tanagran-nom being oppressed liberty-abl Liberty-nom ‘He used as a pretext a promenade, a monument, and, after having oppressed the liberty, that Tanagran lady Liberty.’ (Cic. Dom. 116)

(16) Sed dux atque imperator vitae mortalium but leader-nom and commander-nom of life of mortals animus est. Qui… mind-nom is it ‘But it is the mind which is the leader and commander of the life of mortals. It...’  (Sal. Jug. 1.3) The above observations can also be applied to other predicative constructions. Take a verb such as appello ‘to call’ requiring a direct object and a predicative noun in the accusative. Also here, the same criteria can be utilised: contextual dependency and the semantic value of the nouns involved. Two examples are given below. In a passage devoted to disorders, perturbationes is contextually dependent and interpreted as direct object; morbos, therefore, is the predicative noun (17). L. Murena in (18) is a contextually dependent constituent but referring to a highly individuated entity – this makes it 66. It is worth adding that this passage is quoted by Isidore of Seville (Etym. 2.30.2) in the following way: Gloria est laus… meritorum, with the subject placed initially. See A. Boulanger (1966:€70), Cicéron, Discours, tome 19, Paris.



Chapter 3.╇ Declarative sentences 

eligible to be the direct object; saltatorem expressing a qualification is the predicative noun and bears Focus function. (17)

Omnis autem perturbationes animi morbos all-acc but movements-acc of mind sickness-acc philosophi appellant negantque... philosophers-nom call deny=and ‘Now philosophers apply the word “sickness” to all such emotional movements and they say that no...’ (Cic. Tusc. 3.9)

(18) Saltatorem appellat L. Murenam Cato. dancer-acc calls L. Murena-acc Cato-nom ‘Cato calls Lucius Murena a dancer.’

(Cic. Mur. 13)

To sum up, the predicative noun, the subject and the verb sum can be found in all possible combinations. The most frequent pattern {first argument > predicative > sum} is explained by the fact that subjects are often contextually dependent and predicative nouns often convey salient information. However, other orderings are possible. The initial placement of the copula sum, especially in explicative sentences, is not a sign of a “truth” value of the verb sum.

7.3

Existential sum

Lyons (1967:€ 390)€ distinguishes between existential sentences (19a), exhibiting the verb sum ‘to exist’ and a noun phrase, and locative sentences containing in addition a localising expression.67 This second category can be divided into two types with different pragmatic values (Lyons 1968:€495; cf. Clark 1978:€87, and Cabrillana 2001:€276). Both concern the localisation of an entity in space, but the first type (19b) indicates what is localised (a book), whereas the other one (19c) indicates the place (on the table) of a definite entity. (19) a. There are white bears. (= White bears exist.) b. There is a book on the table. c. The book is on the table. In English, the constituent order is not very flexible, just as in French and in German, and a special existential formula is used for (19a) and (19b) (there is or there are) that permits speakers – with the help of a dummy subject in French (il y a un livre sur la table) – to place the indefinite word (a book) rightwards in the sentence. This is, indeed, its privileged position (Dik 1997, I: 316). In the Slavic languages (see ibid.€and Siewierska 1988:€78), which have no articles, the placement rightwards in the sentence is an indicator of the newness of the constituent. For example in Czech, kniha in (19b’) 67. Lyons also considers possessive sentences as belonging to this category. I excluded possessive constructions from my corpus.

 Constituent order in Classical Latin prose

has an indefinite reading (‘a book’). By contrast, kniha in (19c’) placed leftwards is contextually given and has a definite reading (‘the book’). The pragmatic values of these sentences are not the same; whereas (b’) answers the implicit question ‘what is on the table?’, (c’) informs us about ‘where?’ is the book. (19) b.′ Na stole je kniha. on table is book-nom c. ′ Kniha je na stole. book-nom is on table The function of existential (19a) and locative sentences of the (19b) type is to introduce a new entity into the discourse (Hannay 1985a: 10 and 1991:€144 f.; cf. Lambrecht 1988). Their properties in Latin were described by Pinkster (1991:€ 78) and Rosén (1998).68 Their subject, animate or inanimate, is typically indefinite. In the case of proper names, the person is expected to be unknown to the addressee, and this can be explicitly signalled by a specific word such as quidam ‘a certain’ (20). In the majority of cases, the indefinite subject appears rightwards in the sentence, but such a placement is not obligatory (Rosén 1998:€732, and Spevak 2004:€392),69 as is shown in (21); palus ‘a marsh’ has an indefinite reading. Initial placement of the verb sum in presentative sentences such as (20) is due to its semantic properties and the descriptive character of the sentences containing it (cf. Bolkestein 1996b: 12; cf. Luraghi 1995:€367 f.). Initial placement of a contextually independent entity as in (17) – this is a remarkable difference with respect to the Slavic languages (Spevak 2005a: 257) – can be explained as a strategy of putting Focus first, especially when a new section starts. Note that the noun phrase (palus non magna) is discontinuous (cf. Devine and Stephens 2006:€211). (20) Est autem C. Herennius quidam, tribunus plebis... Is... is but G. Herennius certain-nom tribune of people he-nom ‘However, there is a certain plebeian tribune named Gaius Herennius... He...’  (Cic. Att. 1.18.4) (21) Palus erat non magna inter nostrum atque hostium marsh-nom was not big-nom between our and of enemies exercitum. Hanc... army this-acc ‘There was a small marsh between our army and that of the enemy. This march...’  (Caes. Gal. 2.9.1)

68. Cf. also Orlandini (1995:€15 f.) on quidam. 69. As Rosén (1998:€732) rightly observes, the constituent order cannot be taken as a criterion for the identification of presentative sentences. They do not constantly exhibit an initial verb, pace Marouzeau (1910:€107) and also Devine and Stephens (2006:€208f.).€



Chapter 3.╇ Declarative sentences 

7.3.1 Existential sentences The frequency of two main patterns of existential sentences is presented in Table 24. Only positive sentences were considered.70 Table 24.╇ Existential sentences (corpus 1) Pattern sum > A1 A1 > sum sum inserted inside a noun phrase Total

Occurrences

Percentage

19 â•⁄ 7 â•⁄ 1 27

â•⁄ 70% â•⁄ 26% â•⁄â•⁄ 4% 100%

The pattern {sum > first argument} (with possible intervention of other elements) is observed in 70% of cases; the verb sum frequently figures in sentence-initial position (14 occurrences out of 19). The ordering {first argument > sum} represents one quarter of all instances. Existential sentences are a kind of presentative sentence, the constituents of which are not anchored in the current context. From this point of view, the verb sum ‘to exist’ is also new but not salient, unlike the subject that carries Focus. The rightward placement of subjects in existential sentences is explained by the fact that they are contextually independent, often further developed, and convey new information (Pinkster 1991:€78, and Spevak 2004). An example is given in (22), with discontinuity of navigia luculenta ‘fine vessels’. The ordering {first argument > sum} can be found, especially for inanimate entities (Rosén 1998:€732). In other words, in the case of highly individuated animate entities, {sum > first argument} is preferred; inanimate entities enter into both patterns. (22)

[Nam et ipse et Domitius bona plane habet dicrota] suntque navigia praeterea luculenta Sesti, Buciliani, ceterorum. are=and vessels-nom besides fine-nom of Sestius of Bucilianus of others ‘[For both he and Domitius have some good double-banked crafts;] there are also some fine vessels belonging to Sestius, Bucilianus, and others....’  (Cic. Att. 16.4.4)

(23)

[Illud satis opportune:] duo sinus fuerunt quos tramitti oporteret, Paestanus two bays-nom were which be crossed must of Paestum et Vibonensis. and of Vibo

70. In negative sentences, existential as well as locative, the verb is usually in the sentence-final position, preceded by non or another negative word. When the entity is negated, negative words often stand initially.

 Constituent order in Classical Latin prose

‘[That was rather lucky,] for there were two bays to be crossed, that of Paestum and that of Vibo.’ (Cic. Att. 16.6.1) When the first argument is represented by a noun accompanied by a modifier, we can sometimes hesitate over whether the sentence is an existential or a predicative one. As was stated in the introductory part concerning the verb sum, there is no formal differentiation between these constructions, and it is only recourse to the contextual dependency which can help us to decide. In (22) and (23) we have contextually independent entities, whereas, in (20), locus ‘the place’ is contextually given, deducible from sub montem ‘below the hill’. This permits us to conclude that the sentence in (24) is not an existential one and that praeruptus is not an attributive but a predicative adjective conveying salient information (‘how was the place?’). (24)

[et sub montem in quo erat oppidum positum Ilerda succedunt...] Praeruptus locus erat, precipitous-nom place-nom was [utraque ex parte derectus ac tantum in latitudinem patebat, ut tres instructae cohortes eum locum explerent.] ‘[and coming up below the hill on which the town of Ilerda stood...] The place was precipitous [and sheer on both sides, and of such a width that three cohorts in formation filled it.]’ (Caes. Civ. 1.45.4)

It is possible to find an existential verb sum carrying Focus.71 In such cases, we have to do with a verum Focus in the sense of ‘it really exists’, standing in sentence or clauseinitial position. A good example is given in (25) where such an interpretation is confirmed by the presence of the particle profecto ‘assuredly’. (25)

[Qui vero probari potest ut sibi mederi animus non possit...?] Est profecto animi medicina, philosophia; cuius auxilium… is assuredly of mind medicine-nom philosophy-nom whose aid-nom ‘[And then who can prove that the mind cannot be healed...?] Assuredly, a medical science for the mind does exist: it is philosophy, whose aid...’(Cic. Tusc. 3.6)

7.3.2 Locative sentences Locative sentences contain an existential sum and a locative complement (L). Patterns collected from the corpus 1 are indicated in Table 25.

71. Cf. also Marouzeau’s (1953:€47) nice example with repetition: Est, est illa vis profecto. (Cic. Mil. 84) ‘There is, there is indeed, such a heavenly power.’ On sum with truth value, see also Adams (1994a: 69 ff.)



Chapter 3.╇ Declarative sentences 

Table 25.╇ Locative sentences (corpus 1) Pattern sum > L > A1 L > sum > A1 A1 > sum > L sum > L A1 > L > sum L > sum sum > A1 > L L > A1 > sum Total

Occurrences

Percentage

14 12 10 â•⁄ 3 â•⁄ 4 â•⁄ 2 â•⁄ 3 â•⁄ 8 56

â•⁄ 26% â•⁄ 21% â•⁄ 23% â•⁄ 11% â•⁄â•⁄ 5% â•⁄ 14% 100%

The presence of a locative complement increases the number of possible patterns and no single one of them is dominant. The subject often (26 + 21% of cases) comes after the verb sum and the locative complement. The locative complement can be found rightwards in the sentence (23 + 5%). The verb sum can appear in sentence-final position (14 occurrences), but it more frequently occupies the sentence-initial position, either absolute initial or after a connector (17 occurrences).72 Locative sentences appear with two main pragmatic values: they answer either the question ‘who, what?’ or the question ‘where?’ (i) When one says who or what is somewhere, the subject is usually contextually unbound. The following examples show sentences with initial verbs; the subjects, both further expanded, occupy different positions: oppidum ‘a city’ (26) comes last whereas vallis ‘a valley’ (27) appears after the verb and before the locative complement. Both sentences lack a Topic, although duas acies in (27) is contextually given and might have been selected as such. Erat haud longe ab eo itinere... oppidum Numidarum was not far from the route town-nom of Numidians nomine Vaga, forum rerum venalium totius regni by name Vaga market of merchandise of whole kingdom maxime celebratum... Huc most frequented there ‘Not far from the route... there was a Numidian town by the name of Vaga, the most frequented market for merchandise in the whole kingdom... Here...’  (Sal. Jug. 47.1)

(26)

72. My data differ considerably from those collected by Cabrillana (2001:€285). According to her, the verb is placed in sentence-final position in 86% of the cases in Livy.

 Constituent order in Classical Latin prose

(27)

Erat vallis inter duas acies..., non ita magna, at was valley-nom between two lines neg so big-nom but difficili et arduo ascensu. Hanc... with difficult and strenuous way up this-acc ‘Between the two lines there was... a valley, not particularly deep, but with sides that were difficult and strenuous to climb. This valley...’ (Caes. Civ. 2.34.1)

(28)

Ea tempestate in exercitu nostro fuere complures novi at that time in army our were numerous-nom new-nom atque nobiles... qui Iugurthae... animum accendebant. and noble-nom who of Jugurtha mind-acc incited ‘At that time in our army there were numerous new men and nobles...; they fixed Jugurtha’s... mind...’ (Sal. Jug. 8.1)

However, the verb sum need not stand initially; it can be preceded by another constituent, such as a time or locative complement, as in (28). Another example of a locative sentence with a topical anchor in the immediately preceding context is given in (29). In such cases, the verb comes after the Topic constituent (hoc itinere). (29)

[Abest derecto itinere ab Vtica paulo amplius passus mille.] Sed hoc itinere est fons quo mare succedit but in this way is spring-nom in which sea comes up longius lateque is locus restagnat. from distance broadly=and this place-nom is boggy ‘[It is rather more than a mile from Utica in a direct line.] But there is a spring in the way, to which the sea comes up from a fair distance; the ground is boggy over a wide area.’ (Caes. Civ. 2.24.4)

(ii) The other main pragmatic value of locative sentences concerns the place where an entity is found; these sentences answer the question ‘where?’ The first argument is contextually bound; for example, is in (30) or Pompeius in (31), functions as Topic and stands in sentence-initial position. The localities, Ravennae and in Candavia, are contextually unbound and have Focus function. Note the different positions of the time complement; eo tempore ‘at that time’ stands both before and after the verb erat; such a variation seems pragmatically unimportant to me. (30) (Caesar) Is eo tempore erat Ravennae exspectabatque... he at that time was at Ravenna was waiting=and ‘He was at that time at Ravenna and was awaiting...’ (Caes. Civ. 1.5.5) (31)

erat eo tempore in Candavia Pompeius Pompey-nom was at that time in Candavia [iterque ex Macedonia in hiberna Apolloniam Dyrrachiumque habebat.] ‘Pompey was at that time in Candavia, [on his way from Macedonia to winter camps in Apollonia and Dyrrachium.]’ (Caes. Civ. 3.11.2)



Chapter 3.╇ Declarative sentences 

However, the instances quoted in this section do not allow us to say that the constituent placed rightwards is necessarily the focal one. In (32) I give an example taken from a letter by Cicero written in Puteoli. In theory, it could answer the question ‘who?’ as well as ‘where?’ In the given context, preceding as well as following (note the anaphoric continuation of Hirtius by ei), I interpret it as answering the question ‘who?’ Hirtius is Focus of its sentence without being placed in sentence-final position. Similarly in (33), it is not Brundisii, in sentence-final position, but Antonius which is Focus of the sentence. This town is contextually dependent and the surprising news is that Antonius appears to be there; and in addition, the subsequent context concerns him. (32)

Sed casu, cum legerem tuas litteras, Hirtius erat but by chance when I read your letter-acc Hirtius-nom was apud me in Puteolano. Ei legi et egi. by me in Puteoli him-dat I read and explained ‘But by chance, when I was reading your letter, Hirtius was at my house near Puteoli. I read it to him and took the matter up.’ (Cic. Att. 15.1.2)

(33) [Brundisium...] Erat eo tempore Antonius Brundisii; was at that time Antonius-nom at Brundisium [is virtute militum confisus...] ‘[Brundisium...] At that time Antonius was at Brundisium; [trusting in the courage of his soldiers...]’ (Caes. Civ. 3.24.1) To conclude, in the existential and locative sentences in my corpus 1, the subject has an early position in the sentence in 70% and 47% of cases, respectively. In existential sentences, the subject conveys new information. In locative sentences, Focus is either on the subject or on the locative complement, and its placement can only be identified with the help of the contextual status of these constituents. Rightwards placement of the subject is therefore usual, but not systematic nor obligatory. From this point of view, it is incorrect to call the sequence {verb > first argument} in existential and locative sentences an inverted order, for such an ordering is not surprising. On the other hand, it is not exceptional to find contextually unbound subjects in initial position. This corresponds to the Focus-first strategy of presenting something new. As for the verb sum in initial position, such a placement is frequent in existential sentences (52%), and less frequent in locative ones (30%). This difference is due to the fact that locative sentences, unlike existential sentences, may have a Topic. However, an initial existential verb does not carry Focus and, in my view, neither does it carry emphasis, except for several cases of verum Focus (‘there is really’).

chapter 4

Interrogative sentences 1. Introduction In many languages, the sentence type is relevant to constituent order: interrogative and imperative sentences may differ in constituent ordering from declarative sentences (see Pinkster 1995:€217). There are only a few studies on constituent order in interrogative sentences in Latin.1 Panhuis (1982:€61–68), using Firbas’s (1992) concept of communicative dynamism, proposed the following pattern: rheme proper > theme proper > theme > rheme According to him, interrogative sentences present the most informative element (rheme proper) in initial position, and the verb is usually not sentence-final. Babič’s (1992:€61 and passim) conclusions are mostly the same as Panhuis’s. He argues that the question-word stands almost always in sentence-initial position.2 However, he shows that a contextually given constituent may appear at the beginning of an interrogative sentence, for example viginti minas placed before the word that is questioned, dabin, hosting the interrogative particle -ne in (1): (1) [Roga me viginti minas...] Viginti minas dabin? – Dabuntur. twenty minae-acc will you give=q they will be given ‘[Ask me to pay you eighty minae...] Will you give me twenty minae? – They will be given.’ (Pl. Ps. 1077) Panhuis’s and Babič’s studies, however, are based on Plautus’s comedies, which are interactive texts formed by dialogues and intended for oral presentation with comical effects. Prosodic and metrical factors may also play a role, as both scholars indicate. The aim of this chapter is to examine interrogative sentences in Classical Latin prose,3 especially the placement of question-words and particles. It is organised as 1. Devine and Stephens (2006) pay no special attention to interrogative sentences, although they have occasional observations. 2. This is common knowledge; see Kühner and Stegmann (1914, II: 501). 3. In my corpus of Classical Latin prose (corpus 1), word-questions are slightly more frequent (54% of all interrogative sentences) than sentence-questions (44%); disjunctive questions are not very well represented (2%) In this corpus, interrogative sentences mostly appear in Cicero’s

 Constituent order in Classical Latin prose

follows. Word-questions are examined in Section 2. Section 3 is devoted to sentencequestions, introduced by interrogative particles (3.1) and without particles (3.2). Section 4 concerns disjunctive questions.

2. Word-questions When using a word-question, the speaker or author aims to obtain a piece of information that he does not know (Dik 1997, I: 328). In (2a), the question concerns the place where John is going; the question-word, where, signals what piece of information is missing and represents the Focus of the question.4 In (2b), the constituent to the market is new information and functions as Focus of its sentence; John is going is contextually given. Answers may be reduced to Focus constituents, as illustrated in (2c). (2) a. Where is John going? b. John is going to the market. c. To the market. In many languages, question-words are typically confined to a special position that is initial, pre-verbal or some other position (Dik 1997, II: 264 f.). In Latin, as well as in English and in Polish (Siewierska and Uhlířová 1998:€128), question-words are typically sentence-initial. Word-questions may concern any sentence constituent: arguments, satellites, verbs or their modifiers. Question-words may be used alone (quis ‘who’ ubi, quo ‘where’), or in a phrase with attributive function (qui homo); in the latter case, the noun phrase is usually discontinuous, as in (3) and (4). (3) Quas tu mihi, inquit, intercessiones? Quas religiones? what-acc you me-dat he says vetoes-acc what sanctities-acc “What vetoes,” he says, “what sanctities are you telling me of?”(Cic. Phil. 1.25) (4) (mens hominis) In quo igitur loco est? in which then place is ‘(the mind of man) Where is then its place?’

(Cic. Tusc. 1.29)

In about 8% of the cases, the question-word does not stand in sentence-initial position in my corpus. It is preceded by a contextually bound constituent, such as various types

speeches, correspondence and philosophical treatise (Tusculan Disputations); historical narratives are poor in questions. In this corpus, questions are usually not used in an interactive way, and lack answers. In Cicero’s speeches, questions are often rhetorical with an implicit answer; in his correspondence, they are authentic but we do not know the reaction of the addressee. Only Tusculan Disputations offer pieces of dialogues. 4. However, Firbas (1992:€98) does not consider the interrogative word as rheme proper.



Chapter 4.╇ Interrogative sentences 

of Topic, especially contrastive (5), Sentence Topics (6) or setting elements.5 Example (7) shows a Sub-Topic, adventus atque introitus, deducible from the preceding context; the question-word qui, postposed to its noun phrase, comes before the verb. (5) Atque ille tamen ad collegium rettulit, tu ad quem rettulisti? and he yet to College submitted you to whom submitted ‘Yet he submitted his dedication to the College; and to whom did you submit yours?’ (Cic. Dom. 132) (6) Esto, collegium non adhibuisti. Quid? De collegio quis well College-acc neg you called in what from College who tandem adfuit? at least was present ‘Very well, you did not call in the College. And of the College, who was present?’  (Cic. Dom. 117) (7)

[in patriam revertisse] Adventus meus atque introitus in urbem qui fuit? arrival my-nom and entry-nom in city how-nom was ‘[returned to my country] What was the character of my arrival and entry into the city?’ (Cic. Dom. 75)

Similar observations also hold true for cur ‘why’: 9% of the instances present non-initial placement. Cur is mainly preceded by a contrastive Topic, as in (8). Such ordering is often found after a rhetorical quid ‘what’.6 (8) Quid? Sacra Clodiae gentis cur intereunt? what rites-nom of Clodia family why perish ‘What? Why should all the religious traditions of the Clodii perish?(Cic. Dom. 34) As Siewierska (1988:€124) points out, it is possible to find multiple placements of three and even four successive question-words in first position in Polish, as is shown in (10). In such multiple questions, several items are questioned in the same sentence. (9) Kto co komu dał? who-nom what-acc who-dat gave ‘Who gave what to whom?

5.

I. e. by the same type of constituents as in the case of subordinators (see p. 13 ff.).

6. Cf. Example (6), quoted above, and Devine and Stephens (2006:€238) on nonne.

 Constituent order in Classical Latin prose

One can encounter this phenomenon in Latin also, although examples are few:7 (10) Quis despondit? Quis dedit?] Cui quando nupsit? who-dat when she married ‘[Who betrothed her? Who gave her away?] When and to whom was she married?’ (Ter. Ad. 670) To sum up, question-words are typically sentence-initial and noun phrases containing a question-word are usually discontinuous. However, priority may be given to Topic constituents. With the exception of questions containing no other constituents but a question-word, for example: quid? ‘what?’ (and several cases of ellipsis), a questionword does not go in sentence final position: it is never the last word of a sentence.8

3. Sentence-questions Questions that expect the answer yes or no concern the truth of the sentence content; for example: (11) a. Does John travel to London? b. Yes. or No. In Latin, sentence-questions can be marked by the interrogatives particle num, an, nonne, the enclitic -ne (Hoff 1985, and Bodelot 1987:€60 f.). In my corpus of Classical Latin prose (corpus 1), 24% of yes-no questions are marked by the enclitic -ne, 14% by the particle an, 10% by nonne, 10% by num and 2% by ecquis; questions without a particle represent 40% of the occurrences.

7. Cf. Devine and Stephens (2006:€89), who give examples of multiple question-words in indirect questions. 8. Unlike for example in colloquial French: Il t’a dit quoi? Cf. also Siewierska and Uhlířová (1998:€128 f.) on variations of placement in some Slavic languages; for example a question-word in pre-verbal position in Russian: Tebja kak zvať? ‘What is your name?’ and in sentence-final position in Czech: A s prací mám být hotov dokdy? ‘When shall I finish the work?’ My Latin example of ellipsis is the following one: Studiose enim plerique praesertim in hac partem facta principum imitantur, ut L. Luculli, summi viri, virtutem quis? (Cic. Off. 1.140) ‘For many people imitate zealously the actions of great men; for example, who imitates the virtue of Lucius Lucullus, excellent man?’ Quis contrasts with plerique; the verb imitatur that could follow quis is unexpressed. Cf. also inimicum Antonio quando aut cur? (Cic. Att. 15.22) ‘as to being hostile to Anthony, when and why?’



Chapter 4.╇ Interrogative sentences 

3.1

Sentence-questions with an interrogative particle

3.1.1 The particle -ne The interrogative particle -ne usually accompanies the salient word: Focus of the question. Sentence-questions concern the whole content, it is true, but one word can be treated as salient (Bodelot 1987:€57). For example in (12), the enclitic -ne is tacked onto das, the Focus of the question; the same verb reappears in the affirmative answer, which confirms this analysis. (12)

Hoc dasne aut manere animos post mortem aut this you accept=q either remain souls-acc after death or morte ipsa interire? – Do vero. at death itself perish I accept indeed ‘Do you admit that either souls survive after death, or they perish at death itself? – Yes, I admit that.’ (Cic. Tusc. 1.25)

However, there are instances in which -ne does not mark the word to which it is attached itself as Focus, but the pragmatic unit as a whole (cf. Hoff 1979:€78 f.). For example in (13), -ne accompanies tamen, in (14), iam, words that starts the focal part of the interrogative sentence. The conditional clause introduced by si in (13) functions as the setting, nausea in (14) has Topic function. It is worth adding, however, that tamen and iam, respectively, which host -ne, can be used as a minimal positive answer to such questions.9 (13) Si per vim tulisti, tamenne lex est? if by force you carried still=q law-nom is ‘If you employed force to carry your measure, is it, in spite of this, a law?’  (Cic. Dom. 53) (14) Nausea iamne plane abiit? sickness-nom already=q quite passed ‘Has your stomach trouble quite gone?’

(Cic. Att. 14.10.2)

The word hosting -ne frequently occupies the initial position, which makes up 89% of the instances in Cicero’s works (corpus BTL).10 When this is not the case, another constituent is given priority, especially a Topic constituent: a contrastive Topic (15) or a Sub-Topic (16). They are syntactically integrated into their sentences, and thus do not represent Theme constituents. (15)€ Tibi omnisne animi commotio videtur insania? – you-dat all-nom=q of mind agitation-nom seems insanity-nom 9. On tamen used as an answer, see Spevak (2006d: 243); on iam, cf. Iamne operuit ostium? – Iam. (Ter. Ph. 816) ‘Has she shut the door yet? – Yes, she has.’ 10. Cf. Hoff (1984:€21). According to him, -ne accompanies the first word of the sentence in 76% of cases; in 37% of cases, the first word is a verb.

 Constituent order in Classical Latin prose



Non mihi quidem soli... maioribus quoque nostris neg me-dat part only-dat ancestors-dat also our-dat hoc ita visum... this so seemed ‘Do you think every emotion amounts to insanity? – I am not the only one to hold this opinion... our ancestors held the same view.’ (Cic. Tusc. 3.8)

(16) De reliquis rei publicae malis licetne dicere? about remaining of State evils is allowed=q speak Mihi vero licet... me-dat indeed is allowed ‘Am I permitted to speak of the remaining ills of the State? I am permitted...’  (Cic. Phil. 1.14) 3.1.2 The particles num and nonne The particles num (suggesting a negative answer) and nonne (expecting a positive answer) behave in a similar way: they are usually placed at the beginning of interrogative sentences, but unlike -ne, they do not mark a single constituent in particular. Two typical examples are given in (17) and (18). (17)

Num ille furor tribuni plebis... fraudi Metello q that frenzy-nom of tribune of people injury-dat Metellus-dat fuit...? – Certe non fuit. was certainly neg was ‘Did that frenzy of a plebeian tribune... do any injury to Metellus...? – Certainly not.’ (Cic. Dom. 123)

(18) Nonne vobis Numa Pompilius videtur loqui? q you-dat Numa Pompilius-nom seem to speak ‘Does not Numa Pompilius appear to be speaking to you?’ (Cic. Dom. 127) Num and nonne, just like the word hosting -ne, can be preceded by a Topic constituent. This is frequently found with nonne,11 for example in (19) where it follows in dedicatione. In (20), totum prope caelum summarizes what has already been said: Cicero is writing about heroes living in heaven and considered as gods. Furthermore, an element bearing emphasis or contrast can appear directly after nonne (cf. Example (21) below and Devine and Stephens 2006:€ 239); this is the case of humano in (20) that contrasts with caelum. (19) ‘(dedicatio) Quid? In dedicatione nonne et quis dedicet what in dedication q and who dedicates

11. Nonne is preceded by a Topic constituent in 36% of cases (12 occurrences out of 33), num only in 13% (4 occurrences out of 31).



Chapter 4.╇ Interrogative sentences 

et quid et quo modo quaeritur? and what and how is inquired ‘(dedication) What? In the case of a dedication do not people inquire who says such and such a thing, and what he says, and how?’ (Cic. Dom. 127) (20)

Totum prope caelum, ne pluris persequar, nonne whole-nom nearly sky-nom neg further I pursue is not humano genere completum est? by human race filled up ‘Is not nearly the whole sky (to give no further examples) filled up with the human race?’ (Cic. Tusc. 1.28)

Whereas Topics in (19) and (20) are contextually dependent, I also noted a few instances with entities preceding num that are contextually unbound; for example in (21). Such a constituent is “set” as Topic and forms a point of departure for the sentence. In (22), we have a contextually bound entity (Cyrenaeus Theodorus) that functions as Focus and is referred to with the relative connective cui in the subsequent sentence. The verb miramur can be considered as contextually dependent because in this passage Cicero gives examples of remarkable persons, worthy of admiration. (21)

Quid? Municipia, colonias, praefecturas num what boroughs-acc colonies-acc prefectures-acc q aliter iudicare censetis? otherwise judge you think [Omnes mortales una mente consentiunt...] ‘What? Do you think that the boroughs, the colonies, the prefectures, determine otherwise? [All men agree with one mind...]’ (Cic. Phil. 4.7)

(22)

Cyrenaeum Theodorum, philosophum non ignobilem, of Cyrene-acc Theodorus-acc philosopher-acc neg unknown-acc nonne miramur? q we admire [Cui cum Lysimachus rex crucem minaretur... inquit...] ‘Do we not admire Theodorus of Cyrene, no mean philosopher? [When King Lysimachus threatened him with crucifixion... he said...]’ (Cic. Tusc. 1.102)

3.1.3 The particle an Unlike num and nonne, the particle an always stands in initial position in non-disjunctive interrogative sentences (cf. Section 4 on disjunctive questions). I give only one example for illustration: (23) An tu haec non credis? – Minime vero. q you this neg believe by no means certainly ‘But you do not believe it? – Certainly not.’

(Cic. Tusc. 1.10)

 Constituent order in Classical Latin prose

3.2

Sentence-questions without interrogative particles

Sentence-questions without any interrogative word show three main orderings. Firstly, when the verb itself is involved in the question, and thus represents the Focus, it is found in initial position. This interpretation is confirmed by the negative answer minime in (24). The constituent order in such cases is different from that of declarative sentences, because it exhibits an initial verb. (24)

Infirmas igitur tu acta C. Caesaris, viri invalidate then you acts-acc of G. Caesar of man fortissimi? – Minime. very valiant by no means ‘Do you then invalidate the acts of Gaius Caesar, that most admirable citizen? – By no means.’ (Cic. Dom. 39)

Secondly, contrastive elements are placed in sentence-initial position. An example is given in (25) with a double contrast between the Topics timor and ratio ac sapientia, and between potuit and non poterit, verbs functioning as Focus. In such cases, the constituent order does not seem to differ from declarative sentences. (25)

Timor igitur ab his aegritudinem potuit repellere, fear-nom then from these distress-acc could drive away ratio ac sapientia vera non poterit? reason-nom and wisdom-nom true-nom neg could ‘Therefore fear had the power to drive away their distress; would not reason and true wisdom have the same power? (Cic. Tusc. 3.66)

Thirdly, the sentence starts with an emphatic word such as tantus, or a demonstrative pronoun, as for example in (26). This rhetorical question corresponds to the following affirmation: hanc… non comprobabimus ‘we will not sanction this’. Also here, the constituent order resembles the one used in declarative sentences. (26)

Hanc vos, pontifices, tam variam, tam novam in omni this-acc you priests-voc so various-acc so new-acc in every genere voluntatem, impudentiam, audaciam, sort will-acc impudence-acc audacity-acc cupiditatem comprobabitis? covetousness-acc approve ‘Will you, gentlemen, sanction this universal and unprecedented tyranny of every sort, this impudence, audacity, and covetousness?’ (Cic. Dom. 116)



Chapter 4.╇ Interrogative sentences 

4. Disjunctive questions Disjunctive questions question the truth of the content of the sentence and present a choice between two (or more) alternatives12 (Dik 1997, II: 261). The speaker may choose the correct alternative (27b) or (implicitly) negate them both and supply the correct answer (27c). (27) a. Does John travel to London or to Paris? b. Yes, (he travels) to Paris. c. No, (he travels) to New York! In Latin, disjunctive questions are marked by the pairs utrum… an…, or -ne… an...; the first element may lack the particle. Example (28) shows a disjunctive interrogative sentence with two alternatives, a bonis rebus and a malis, that function as Focus (the verb discessisset is contextually given). The addressee rejects both by selecting a miseris as the correct answer. Vtrum stands in initial position, and the particle an introduces the alternative a malis. (28)

Vtrum igitur, si tum (Pompeius) esset extinctus, a bonis whether so if then had died from good rebus an a malis discessisset? – Certe a miseris. things or from bad had departed certainly from wretched ‘Well, if he had died at that time, would he have departed from the good things or the bad? – Certainly from the wretched things.’ (Cic. Tusc. 1.86)

Vtrum can be preceded by a Topic constituent: in (29), the contrastive Topic (me) is followed by huic maledicto, which is a contextually bound constituent; utrum introduces the focal part of the question (non respondere). (29)

Me huic maledicto utrum non respondere an me-acc this-dat slander-dat whether neg reply or demisse respondere decuit? shyly reply was suitable ‘Ought I when accused in this manner, not to have replied at all, or to have replied with a bowed head?’ (Cic. Dom. 93)

Finally, it is worth mentioning a special type of questions with necne ‘or not’, that negates the alternative. Its place is at the end of the sentence. Example (30) shows an initial verb functioning as Focus (cf. Example 24).

12. For three, see for example: Vtrum causa novi consili capiendi non fuit, an meae partes in ea causa non praecipuae fuerunt, an alio potius confugiendum fuit? (Cic. Dom. 10) ‘Was the occasion not such as to justify our embarking upon a new policy? Was my role upon that occasion not that of a protagonist? Or ought we to have looked rather to another quarter for safety?’

 Constituent order in Classical Latin prose

(30) Sunt haec tua verba necne? are these your words-nom or not ‘Are these your words or not?’

(Cic. Tusc. 3.41)

5. Conclusions Latin word-questions exhibit a particular constituent order that is due to the focal character of question words. These are mostly placed in initial position, but they may be preceded by Topic constituents: especially contrastive Topics and setting elements. Instances of question-words standing at the absolute end of sentences are (as far as I know) not attested except in cases of ellipsis. Sentence-questions containing interrogative particles (–ne, num and nonne) can also come after a Topic constituent, in particular when they are preceded by rhetorical quid? ‘what?’ The particle an regularly stands in initial position. Sentence-questions without interrogative particles exhibit three main placement strategies: the initial position is occupied by the verb, a contrastive, or an emphatic word. Questions with the verb placed initially have a special constituent order; questions exhibiting contrastive or emphatic words in initial position do not differ in constituent ordering from declarative sentences.

chapter 5

Imperative sentences 1. Introduction The common opinion is that Latin imperative sentences distinguish themselves by having the verb constrained to the initial position. This statement is found in Latin grammars1 but also in Devine and Stephens’s (2006:€149) recent book on Latin word order. Marouzeau (1938:€38 and 51 f.) states that imperative sentences have verbs in initial position; a placement other than initial is, according to him, exceptional. However, he admits that priority can be given to some other constituents, especially personal pronouns. Panhuis (1982:€68–72) does not subscribe to this traditional view. When analysing the verbs do ‘to give’ and reddo ‘to give back’ used in Plautus, he claims that imperative sentences are organized very much like declarative sentences and obey pragmatic rules of placement. Therefore the verb in the imperative is not limited to a particular position; for example, in (1), the imperative both precedes and follows its argument, fenus. (1)

[fenus reddundum est mihi...] Cedo fenus, redde fenus, fenus reddite. give interest-acc pay interest-acc interest pay-pl ‘[you have to pay me back the interest...] Give the interest, pay the interest, the interest, pay it back!’ (Pl. Mos. 603)

According to him (p. 69), the imperatives (cedo, redde and reddite) function as theme (in the terminology of Functional Sentence Perspective) whereas fenus is rheme. However, in my view, such an analysis is not correct. I would interpret not the interest (fenus), which is contextually dependent, but the verbs in the imperative as Focus. The sentence is not an implicit answer to the question ‘what?’; it orders somebody to do something. A similar situation appears in another passage in Plautus (As. 664–692), analysed by Panhuis (1982:€37 f.). The verb do is used here with personal pronouns (mihi, nobis), dependent on the situation, and with second arguments, dependent on the context: argentum ‘money’, istas viginti minas ‘these twenty minas’, istuc ‘this’, and istuc argentum ‘this money’. As Risselada (2004) shows, the imperative sentences, which I give in (2) – (5), are not used as orders, but as supplications: two speakers, the freeman 1. See Kühner and Stegmann (1914, II: 598 f.),€Szantyr (1972:€403), and Touratier (1994:€712). In addition, the sentence-initial placement of verbs in the imperative mood is also claimed for Indo-European; see Richter (1903:€66).

 Constituent order in Classical Latin prose

Argyrippus and his friend Philenium, are asking two slaves to deliver them the money that belongs to them:2 (2) Da..., Leonida, argentum mihi. give Leonida-voc money-acc me-dat ‘Leonida..., do give me the money.’

(Pl. As. 664–665)

(3) Da mihi istas viginti minas. give me-dat these twenty minae-acc ‘Do give me those twenty minae.’

(Pl. As. 685)

(4) O Libane, mi patrone, mihi trade istuc... o Libanus-voc my master-voc me-dat deliver that-acc ‘O Libanus, my master, do give me that.’

(Pl. As. 689)

(5) Mi Libane..., amabo..., da istuc argentum nobis. my Libanus-voc please give this money-acc us-dat ‘My Libanus..., please, do give us that money....’ 

(Pl. As. 692)

In these examples, the verb is almost always initial (with exception of (4)). This can be explained by the Focus function it fulfils: the request is not what to give nor to whom but to do so, to deliver – finally – the money. This interpretation is confirmed by the question in (6) where the verb datis is accompanied by -ne and functions as Focus of the question. (6) [quoniam ut est lubitum nos delusistis,] datisne argentum? you give=q money-acc ‘[Well now – since you’ve both made fun of us just as you liked,] are you going to give the money?’ (Pl. As. 711) Additionally, the Examples (2) – (5) show that the order of the second and third arguments is variable. However, I would not consider it as a result of communicative dynamism in the sense that prominence is given to mihi in sentence-final position in (2) and to istas viginti minas in (3).3 In my view, different orders are allowed in Latin. From a more general point of view, imperative sentences may or may not present a special ordering of constituents (cf. Siewierska 1988:€93).4 In other words, they may exhibit the same ordering as declarative sentences.5 However, in Latin declarative 2. Freemen asking their slaves for something produces a comic effect. 3.

Mihi in (3) can be interpreted as contrastive, as Panhuis (1982:€39) suggests.

4. Cf. also Siewierska and Uhlířová (1998:€130) on imperatives in Slavic that do not display any peculiarities of ordering. 5. In Panhuis’s (1982:€68) corpus of Plautus, the verb is initial in 57% of cases, medial in 11% and final in 32% of cases.



Chapter 5.╇ Imperative sentences 

sentences, the verb is often sentence-final, which is not the case in imperative sentences, as Pinkster (1995:€217) observed. Should we assume that Latin imperative verbs have to occur in sentence-initial position? Before starting my examination, I will first ask what kinds of imperatives are found in Classical Latin prose. A typology of imperative sentences and the proportion with which they occur in a concrete corpus is the topic of Section 2. Section 3 is devoted to periphrastic imperatives; Section 4 to imperatives formed from verbs of saying, thinking, and knowing. Imperatives of other verbs (verbs of action, intransitive and trivalent verbs) are discussed in Sections 5–7. Two final Sections, 8 and 9, deal with future imperatives and the placement of pronouns.

2. A typology of imperative sentences In her study on Latin imperatives, Risselada (1993) shows that imperative sentences have different illocutionary and perlocutionary values. The imperatives in (7) and (8) are designed to modify the behaviour of the addressee (this is their perlocutionary value) and represent “directive” imperatives. The function of dic in (9) is different; it serves to accompany another speech act, a question (Risselada 1989:€370). Such imperatives, which Risselada (1993:€ 258) calls “metadirective”, create explicit perlocutionary effects linked with the speech act that they accompany. Cave in (10) is nothing but a lexical expression referring to a very general action (‘do not’), which is specified afterwards by a complement clause (ignoscas). Cave in combination with a subjunctive verb can be considered as a periphrasis. I give this brief sketch of directive, metadirective and periphrastic imperatives in order to argue that one cannot expect the same placement in each of them because their functions are different. In particular, it is not surprising to find them, especially metadirective (9) and periphrastic (10) imperatives, placed before what they introduce. (7) Recta perge in exsilium. straight go to exile ‘Go straight into exile.’ (8) ‘Catonem’ tuum mihi mitte; [cupio enim legere.] Cato-acc your-acc me-dat send ‘Send me your “Cato”, [I want to read it.]’ (9) Dic: quid addidit? say what he added ‘Tell me: What has he added?’ (10) Cave ignoscas! take care you pardon ‘Take care you do not pardon him.’

(Cic. Cat. 1.23)

(Cic. Fam. 7.24.2)

(Cic. Ver. 1.143)

(Cic. Lig. 16)

 Constituent order in Classical Latin prose

Table 1.╇ The frequency of imperative sentences in Cicero (LASLA) Type directive imperatives verbs of saying, thinking, perceiving periphrases

occurrences

percentage

314 371 278

32% 39% 29%

To examine the frequency of these types of imperative sentences, I will use the corpus of Cicero’s works available on Lasla6 and consider verbs in the imperative that appear at least five times.7 For practical reasons, I will consider metadirective imperatives together with various other verbs of communication and perception. Table 1 summarizes my results. This table permits us to conclude that directive imperatives (such as da8 ‘give’, conserva ‘conserve’, accipe ‘receive’) occur in only 32% of cases, whereas imperatives formed from verbs of saying, knowing, thinking, and perceiving (such as dic ‘say’, puta ‘consider’, scito ‘know’) make up 39%, and periphrases (such as fac ‘do’, noli ‘do not want to’) 29% of all occurrences.9 In other words, verbs of communication and periphrases are strongly represented in the attested corpus. In addition to this, present imperatives are in the majority; future imperatives make up 11% of all occurrences.

3. Periphrastic imperatives Imperatives used in periphrases, for example cave ‘beware’, noli ‘do not’, fac ‘let’ desine ‘stop’, vide ‘see’10 show the same tendencies of placement. They are mostly found in sentence-initial position (11) or, in complex sentences, in the clause-initial position

6. In LASLA, the following works are available: all Cicero’s speeches and On Friendship, On Duties, and Cato the Elder. 7. In total 963 occurrences out of 1.241 instances in all, i. e. 78%. Prohibitions with ne + subjunctive were not taken into account. 8. In this chapter, I only mention singular forms, but plural ones were considered too. For future imperatives, see below, Section 8, p. 218 f.. 9. Even if I assumed that the verbs in imperative attested less than five times – and thus absent from my counts – were directive imperatives, they would not become the majority. The figures would be the following: directive imperatives 48%, verbs of saying, knowing, thinking, perceiving 30%, periphrases 22%. 10. The imperative vide is not restricted to periphrases. It also appears with indirect questions, with accusative complements, and other complements.



Chapter 5.╇ Imperative sentences 

(12). Their usual position is before the complement clause or the completive infinitive they introduce. (11) Vide, inquit, ne veteranos offendas. see he says neg veterans-acc you offend ‘See, he says, that you do not offend the veterans.’

(Cic. Phil. 11.37)

(12) Quando te proficisci istinc putes, fac ut sciam. when you-acc set out from there you think let that I know ‘Let me know when you think of leaving Rome.’ (Cic. Att. 2.6.2) If expressed, the pronouns tu or vos in the nominative precede the imperative because they mark contrast with another person, as in (13): tu and ego function there as contrastive Topics. An additional example with a priority Topic constituent is given in (14); Quintum marks a change of Topic. A subject pronoun appears after the imperative when the speaker is talking to a group but addresses his words to only one person (Pinkster 1987:€370), for example in (15). (13) [Ego autem quo modo sim adfectus ex Lepta et Trebatio poteris cognoscere.] Tu fac ut tuam et Tulliae valetudinem cures. you do that your-acc and of Tullia health-acc you care for ‘[How it affects me, you can learn from Lepta and Trebatius.] Be sure you take care of your own and Tullia’s health.’ (Cic. Fam. 14.17) (14) Quintum, fratrem meum, fac diligas. Quintus-acc brother my-acc do that you love ‘Be sure to be good to my brother Quintus.’ (15)

(Cic. Att. 3.13.2)

Noli, inquam, tu, Q. Titini, Cn. Fannio fratri do not I say you Q. Titinius-voc Gn. Fannius-dat brother-dat tuo credere. your-dat believe ‘Do not believe – you I mean, Quintus Titinius – do not believe Gnaeus Fannius, your own brother.’ (Cic. Ver. 1.128)

Whereas the imperative fac is always followed by its complement clause, imperatives noli, cave et desine are in exceptional circumstances preceded by it (in 8% of cases). Such an ordering is explained by the presence of contextually given constituents or by a contrast, as in (16), between two completive infinitives. (16) [Tu me, ut facis, opera, consilio, gratia iuva.] Consolari iam desine, obiurgare vero noli. comfort now stop scold but do not ‘[On your side, aid me, as you are doing, with effort, advice, and influence.] Give up trying to comfort me, but don’t scold.’ (Cic. Att. 3.11.2)

 Constituent order in Classical Latin prose

4. Imperatives of verbs of saying, thinking, knowing, and perceiving Verbs of saying, thinking, knowing, and perceiving share some properties in the sense that they contribute to the interactivity of the discourse and to its progression. This section will be devoted to metadirective imperatives and communicative expressions that introduce questions, attract the addressee’s attention, introduce new content, remind the addressee of an important point, or introduce an exclamation.

4.1

Metadirective imperatives

Metadirective imperatives are expressions that accompany a speech act (Risselada 1993:€ 259). They may appear with assertions: crede mihi ‘believe me’ reinforces the truth of content, and scito ‘know’ inserts content into the knowledge of the addressee. Questions may be announced by dic ‘say’ or responde ‘answer’; audi ‘listen’ serves to attract the attention of the addressee and vide ‘look’ invites him to share the emotions of the speaker. Such imperatives have remarkable properties: some of them are parenthetical and contribute to the expression of illocutionary force (Risselada 1989:€370), and some of them admit evaluative words that are typical of exclamations (Vairel-Carron 1975:€77, and Pinkster 1995:€263). Imperative sentences such as dic mihi lack a Topic (17); this is why the imperative comes first. Furthermore, as dic ‘say’ and similar words are intended to elicit an answer, they mostly precede the direct – syntactically independent – question they announce. (17) Dic mihi, C. Antonius voluitne fieri septemvir? tell me-dat G. Antonius-nom wanted=q become septemvir-nom ‘Tell me, did Gaius Antonius want to be on the Board of Seven?’(Cic. Att. 15.19.2) Crede reinforces an assertion,11 and in Cicero’s works (see BTL) it appears almost always in combination with mihi in the following order: mihi crede/credite.12 In principle, this expression is inserted into an assertion and appears after the word whose truth is to be underlined; for example rubeo in (18). (18) Rubeo, mihi crede, sed iam scripseram; delere nolui. I blush me-dat believe but already I had written erase I didn’t want ‘I am blushing, believe me; but I had written the words, and did not want to erase them.’ (Cic. Att. 15.4.3)

11. Parenthetical use is not the only use of crede; it may govern a complement clause or complements in the accusative or the dative. 12. The ordering crede mihi only represents 4% of the occurrences.



Chapter 5.╇ Imperative sentences 

4.2

Communicative expressions

Unlike the parenthetical expressions examined in the preceding section, communicative expressions have a syntactic function, because they govern a complement. They invite the addressee to produce a speech act: to answer a question (responde ‘reply’), to give explanations (doce ‘explain’), to pay attention to new information (adde ‘add’), or to consider an important point (puta ‘consider’, recordare ‘remember’). Imperatives that are intended to elicit an answer govern indirect questions, infinitive clauses, or simply a direct object in the accusative. These complements, not the imperative verb, convey a new piece of information, as in (19). Further expanded arguments mostly follow the imperative (20). Cataphoric pronouns usually stand sentenceinitially (21). However, when a complement of the imperative has Topic function, it comes first, as illustrated in (22). Neglegentiam meam comes first in (23) because it has a subjective character and bears emphasis. (19)

Quo die?... Dic, Naevi, diem. on what day tell Naevius-voc day-acc [– A. d. V Kal. intercalares.] ‘On what day?... Give the date, Naevius. [– On February 20.]’(Cic. Quinct. 79)

(20)

At videte hominis intolerabilem audaciam cum but see-pl of man intolerable-acc audacity-acc with proiecta quadam et effrenata cupiditate. linked some and unbridled covetousness ‘But observe how the man’s intolerable effrontery is joined to a spirit of overreaching and unbridled covetousness.’ (Cic. Dom. 115)

(21) [Desine id me docere.] Hoc doce, doleam necne doleam, nihil interesse. this prove I suffer or not suffer no way matter ‘[Stop proving that for me.] Prove that it makes no difference whether I am in pain or not.’ (Cic. Tusc. 2.29) (22) Atque istam ipsam quaestionem dicite qui obsignarit. and this-acc itself-acc report-acc say-pl who has signed ‘And tell me who has signed this report.’ (Cic. Clu. 185) (23) Nunc neglegentiam meam cognosce:... now carelessness-acc my-acc notice ‘Now notice my carelessness:...’

(Cic. Att. 16.6.4)

Imperatives that call for attention govern complement clauses, complements in the accusative, or are used in an absolute manner (see below, Example 27). Imperatives such as audi ‘listen’ and attende ‘pay attention’ often (in 67% of the cases) precede their second argument. This holds true especially for complements in the accusative that

 Constituent order in Classical Latin prose

convey new information. However, there are variations such as (24) and (25); the latter instance can be explained by the emphasis drawn to reliqua. (24) Nunc audi reliqua. now listen rest-acc ‘Now listen to the rest.’ (25) Nunc reliqua attendite. now rest-acc listen-pl ‘Now listen to the rest.’

(Cic. Att. 7.1.2)

(Cic. Phil. 13.30)

Emphatic words (such as maximus or tantus) and evaluative expressions, such as for example stuporem in (26), as well as expressions accompanied by focusing particles quoque or etiam, are usually placed initially. In the case of complex sentences starting with an adverbial clause, the imperative usually goes in sentence-final position (27). (26) Sed stuporem hominis vel dicam pecudis attendite. but stupidity-acc of man or I would say of beast listen-pl ‘But regard the stupidity of this man, or – I should say, of the blockhead.’  (Cic. Phil. 2.30) (27)

Tametsi... istius improbitatem perspicuam esse confido, although of this wickedness-acc transparent-acc be I believe tamen paulum etiam attendite. nevertheless little still listen-pl ‘Although I feel sure... that man’s wickedness has long been transparent to you, I ask your attention a little further.’ (Cic. Ver. 2.108)

Imperatives such as adde ‘add’ or recordare ‘remember’ serve to announce a new piece of information or to recall an important one. They are mostly combined with complement clauses and typically stand before them (in 89% of the cases; see Example 28). Such an ordering is due to the fact that these sentences mostly lack a Topic and that the verb precedes the complement clause conveying new information. Similar observations can be formulated about accusative complements, as is shown in (29); the constituents sermones, suspiciones, and invidiam are contextually independent and the imperative adde is repeated. On the other hand, the presence of a Topic, for example in ipso Cn. Pompeio, results in the ordering exemplified in (30) with a sentence-final verb. (28)

Concede igitur ut tibi anteponatur in accusando allow then that you-dat be given preference in accusation provincia. province-nom ‘Allow therefore the province to have preference over you in the privilege of prosecuting.’ (Cic. Div. Caec. 54)



Chapter 5.╇ Imperative sentences 

(29) Adde istuc sermones hominum, adde suspiciones, adde invidiam. add here talk-acc of men add suspicions-acc add odium-acc ‘Add to that the talk of men, add their suspicions, add the odium.’( Cic. Phil. 11.23) (30)

In ipso Cn. Pompeio... quam multa sint nova in himself Gn. Pompey how many-nom are new-nom summa Q. Catuli voluntate constituta recordamini. with entire of Q. Catulus approval established-nom recall-pl ‘Call to mind how many precedents have been established in the case of Gnaeus Pompey himself... with the entire approval of Quintus Catulus.’ (Cic. Man. 60)

5. Verbs of action This section is devoted to various verbs of action (recita ‘read’, defende ‘defend’, etc.) that are used as directive imperatives. The relative position of the verb (V) and its complement in the accusative (A2) is indicated in Table 2. From a statistical point of view, the ordering {verb > second argument} is the most frequent one (75%). For the analysis of the results, I will distinguish three main types of pragmatic values of the sentence, depending on the constituent that conveys new information: (i) an argument, (ii) the verb, and (iii) another constituent. (i) Second arguments placed rightwards in the sentence bear Focus function. They may be contextually independent, as in (31) and (32), or contextually dependent but presented as salient (33). (31)

Accusa senatum; accusa equestrem ordinem...; accusa accuse Senate-acc accuse equestrian order-acc accuse omnes ordines, omnes cives... all classes-acc all citizens-acc ‘Accuse the Senate; accuse the equestrian order...; accuse all classes, and all the citizens...’ (Cic. Phil. 2.19)

(32) Quaere argumenta. look for arguments-acc ‘Look for arguments.’

(Cic. Arch. 11)

Table€2.╇ Verbs of action and their (pro)nominal complement (BTL) Pattern

Occurrences

Percentage

A2 > V V > A2 Total

â•⁄ 29 â•⁄ 87 116

â•⁄ 25% â•⁄ 75% 100%

 Constituent order in Classical Latin prose

(33) [Inveni duos solos libellos...] Recita mihi quaeso read me-dat please hunc primum libellum, deinde illum alterum. this first-acc document-acc then that second-acc ‘[I found only two small books...] Kindly read us this document first, and then that one.’ (Cic. Ver. 2.183) When an imperative sentence contains a Topic, this goes in the initial position, as is shown in (34). The initial position is also occupied by cataphoric pronouns that announce the content expressed in a subordinated clause, as in (35). (34)

A quo periculo defendite, iudices, civem fortem from this peril defend-pl judges-voc citizen-acc brave-acc atque innocentem. and innocent-acc ‘From this peril, gentlemen, defend a brave and innocent citizen.’(Cic. Font. 49)

(35) Hoc igitur tibi propone: amplitudinem animi... this therefore you-dat set before largeness-acc of soul unam esse omnium rem pulcherrimam. one-acc be of all thing-acc most beautiful-acc ‘Set this thought before you: that largeness of soul... is the fairest thing of all.’  (Cic. Tusc. 2.64) However, the author can start his sentence with a Focus constituent. Such an ordering is chosen when these constituents are given emphasis or subjective consideration. For example in (36), nothing will support the accusation, not even testium dicta ‘the statements of witnesses’. (36)

[Nihil, nihil, inquam, aliud, iudices, reperietis...] Testium dicta recita, doce in illorum iudiciis of witnesses statements-acc read show in their judgements quid... sit obiectum. what was objected ‘[Nothing, gentlemen; I repeat, you will find nothing...] Read the statements of the witnesses, and tell me... what was objected to them.’ (Cic. Clu. 62)

(ii) The verb in the imperative may convey new information. In such cases, the second argument is typically contextually dependent. Example (37) shows this pragmatic value nicely: the repeated retinete functions as Focus and is followed by its second argument, the contextually dependent hominem, referring to Verres. This sentence does not answer the underlying question who?, but it says what to do with this man. Other imperatives bearing the same Focus function, parcite and conservate, present ellipse of the second argument. Similarly, zero coordination is used in (38), which connects two focal imperatives; the sentence starts with a contrastive pronoun (vos) expressing the addressee.



Chapter 5.╇ Imperative sentences 

(37) Retinete, retinete hominem in civitate, iudices, parcite et conservate. keep-pl keep-pl man-acc in city judges-voc spare-pl and preserve-pl ‘Keep, keep that man in the city, gentlemen, spare him and preserve him.’  (Cic. Ver. 2.76) (38)

Vos autem, Adherbal et Hiempsal, colite, observate you but Adherbal and Hiempsal-voc honour-pl respect-pl talem hunc virum. such this man-acc ‘As for you, Adherbal and Hiempsal, pay attention and respect to such a man as this.’ (Sal. Jug. 10.8)

When the sentence contains a Topic, the verb with Focus function comes after it. The examples in (39) and (40), both€with an imperative tollite and hanc, offer an interesting comparison. In the first case, the sentence informs us what to suppress (hanc anticipates causam and conveys new information), whereas the second one says what to do with the cruelty in question; here hanc is treated as Sentence Topic. (39) Tollite hanc: nullam tam pravae sententiae take away-pl this-acc no-acc such of perverse proposal causam reperietis. reason-acc you will find ‘Take away this reason, and you will find none for so perverse a proposal.’  Cic. Phil. 14.3) (40) (crudelitas) Hanc tollite ex civitate, iudices, this-acc banish-pl from city judges-voc hanc pati nolite diutius in hac re publica versari. this-acc allow do not-pl any longer in this republic remain ‘Banish this cruelty from the State, gentlemen; do not allow it to stalk abroad any longer in this republic.’ (Cic. S. Rosc. 154) Topics may also be contrastive, as regno Numidiae and me in (41). (41)

Regno Numidiae, quod vostrum est, uti lubet, kingdom-dat of Numidia which yours is as you please consulite; me manibus inpiis eripite. treat-pl me-acc from hands impious rescue-pl ‘Pay what heed you please to the kingdom of Numidia, which is yours; but rescue me from heinous hands.’ (Sal. Jug. 24.10)

(iii) In imperative sentences with a transitive bivalent verb another constituent may represent Focus. An example is given in (42) with clarius. The sentence answers the question ‘how?’ and the Focus constituent is found in final position.

 Constituent order in Classical Latin prose

(42) Quid? Recita istam restipulationem clarius. what read this counter-stipulation-acc clearly ‘What? Read the counter-stipulation clearly.’

(Cic. Q. Rosc. 37)

6. Intransitive verbs In the Ciceronian corpus, I noted only four occurrences of intransitive verbs with directional complements. In these very few cases, the directional complement follows its verb, as in (43). (43) Perge in Siciliam. go to Sicily ‘Now go on into Sicily.’

(Cic. Agr. 2.48)

7. Trivalent verbs Trivalent verbs – other than these belonging to the groups discussed in Sections 4.1 and 4.2 p. 210 ff. – in the imperative, do ‘to give’ and reddo ‘to give back’, are quite well represented in my corpus. After eliminating instances of ellipses of the third argument (A3), there are 28 full occurrences of these verbs. Table 3 shows orderings observed. From these data we can see that all possibilities of placement are attested. The verb mostly precedes its arguments (17 occurrences, 9 of which are in sentence-initial position). However, it is difficult to determine the relative position of the second and the third argument: they are often pronominal and mainly occur before the argument expressed by a noun or a noun phrase. Table€3.╇ The imperatives da/date and redde/reddite in Cicero (BTL) Pattern V > A2 > A3 V > A3 > A2 A2 > V > A3 A2 > A 3 > V A3 > V > A 2 A3 > A 2 > V Total

Occurrences

Pronominal

â•⁄ 7 10 â•⁄ 2 â•⁄ 4 â•⁄ 4 â•⁄ 1 28

5 A2 3 A2; 5 A3 2 A3 1 A2; 1 A3 2 A3 –



Chapter 5.╇ Imperative sentences 

From a pragmatic point of view, the verb generally does not bear the Focus function in the case of trivalent verbs; the only example I found for this group of verbs is given in (44). Both obligatory arguments are contextually bound (homini refers to Pompey). (44) Qua re da te homini; [complectetur, mihi crede.] so give you-acc to man-dat ‘So devote yourself to this man; [he will receive you, believe me, with open arms.]’  (Cic. Fam. 2.8.2) Thus, it is (i) the second argument or (ii) the third argument that usually conveys salient information. (i) When the second argument functions as Focus, it occupies a late position in the sentence. In this case, the third argument is often represented by a personal pronoun (mihi, nobis…), as for example in (45), or is contextually dependent, as rei publicae in (46). Such sentences answer the underlying question ‘who?’, ‘what?’ (45) Da mihi testimonium mutuum. give me-dat evidence-acc mutual-acc ‘Give evidence for me.’

(Cic. Flac. 10)

(46) Reddite rei publicae consulem. give back-pl republic-dat consul-acc ‘Give the country back its consul.’

(Cic. Mur. 90)

Contrastive constituents are found in sentence-initial position. A good example is given in (47): filiam contrasts with the parts of body (membra, manus, digitos and nervos) mentioned in the immediately preceding context. In this passage, Cicero says that Decianus tortured a young woman, and her father asks him to deliver her at least as she is, if she cannot be delivered whole. Both the second and third arguments are contextually given but filiam (note the repetition) is salient. (47) [Membra quae debilitavit lapidibus... manus quas contudit, digitos quos confregit... restituere non potest;] filiam, filiam, inquam, aerumnoso patri, Deciane, redde. daughter-acc I say sorrowing-dat father-dat Decianus-voc restore ‘[He cannot restore the limbs which he has maimed with stones... the hands which he has crushed, the fingers which he has broken...;] the daughter, the daughter, I say, Decianus, give her back to her sorrowing father.’(Cic. Flac. 73) (ii) The third arguments convey new information less frequently. In my corpus, there are only four examples and they all exhibit the third argument in sentence-final position. Example (48) contains an enumeration and answers the question ‘to whom?’ (48) Da igitur, quaeso, negotium Pharnaci, Antaeo, give therefore please commission-acc Pharnaces-dat Antaeus-dat

 Constituent order in Classical Latin prose

Salvio, ut id nomen ex omnibus libris tollatur. Salvius-dat that this name-nom from all copies be deleted ‘So, please, commission Pharnaces, Antaeus, and Salvius to delete that name in all the copies.’ (Cic. Att. 13.44.3) However, it is sometimes not easy to decide which constituent is salient in the case of trivalent verbs, in particular when both arguments are contextually dependent. For example, in (49) both sodalis filius and bona have already been mentioned in the preceding context, but bona is further expanded afterwards (si non quae…). For this reason, I would interpret bona as Focus and the sentence as answering the question ‘what?’ and not ‘to whom?’ Similarly in (50), both hanc licentiam (‘this licence’, i. e. to proscribe someone) and tribuno plebis (‘a plebeian tribune’ in general)€are contextually bound. Here the discontinuity of hanc... licentiam signals that this is the most salient information. (49)

Homo avarissime... redde bona sodalis filio, si man avaricious-voc restore property-acc of friend son-dat if non quae abstulisti, at quae confessus es! neg what you stole at least what you admitted ‘Avaricious man..., restore the property of your friend’s son; if not all that you stole, at least all that you admitted to stealing.’ (Cic. Ver. 1.94)

(50) Date hanc tribuno plebis licentiam et intuemini... give-pl this-acc tribune-dat of people licence-acc and consider-pl ‘Grant this licence to a plebeian tribune, and consider...’ (Cic. Dom. 47) Imperative sentences with da/date may contain contrastive constituents. An example is given in (51) with a double contrast between two contextually dependent constituents (istum and mihi), and two Focus constituents (adversario meo and neminem): (51)

Quaeso, inquit, praetor, adversario meo da please he says praetor-voc opponent-dat my-dat give istum patronum, deinde mihi neminem dederis. this-acc as counsel-acc then me-dat no one-acc you will give ‘Pray, praetor, he says, assign that gentleman as a counsel to my opponent, and then I will not ask you to assign any counsel to me.’ (Cic. De orat. 2.280)

8. Future imperatives Future imperatives represent about 11% of all imperatives (see above, p. 208). It is rare to find in Cicero periphrastic constructions with a future imperative as well as future imperatives formed from verbs of communication. In general, future imperatives refer to an action that will be realised not immediately but at a later moment in the future



Chapter 5.╇ Imperative sentences 

(Vairel-Carron 1975:€224, and Risselada 1993:€122). This characteristic could explain the fact that they appear in complex sentences starting with an adverbial clause and that they take the final position just like verbs in declarative sentences (Pinkster 1995:€260). A typical example is given in (52). (52)

Simulac tute coeperis tua iudicia rescindere, as soon as you-nom will start your verdicts-acc overturn mihi suscensere desinito. me-dat be angry stop ‘As soon as you yourself start overturning the verdicts you have made, stop criticizing me.’ (Cic. Ver. 5.19)

Future imperatives formed from bivalent and trivalent verbs are better represented in my corpus. In complex sentences, they privilege the sentence-final position as in (53). In a non-complex sentence, the imperative is not necessarily final, as for example in (54) where habeto precedes its AcI clause. It is worth noting that in quotations of laws,13 which represent sentences with entirely new content, the verb in the future imperative stands in sentence-final position, and the subject, placed sentence-initially, is given the role of Topic (55). (53) Si rem nullam habebis, quod in buccam venerit scribito. if news no-acc you will have what to lips will come write ‘If you have no news, write whatever rises to your lips.’ (Cic. Att. 1.12.4) (54)

Quae nec possunt scribi nec scribenda sunt; tantum these neg can be written neg to be written are only habeto civem egregium esse Pompeium. take for sure citizen-acc excellent-acc be Pompey-acc ‘This could not and should not be committed to writing. Only take this as certain, that Pompey is an excellent citizen.’ (Cic. Fam. 2.8.2)

(55)

Meretrix coronam auream ne habeto; si habuerit, prostitute-nom crown-acc gold-acc neg have if she will have publica esto. public-nom be ‘A prostitute shall not wear a crown of gold; if she does, it shall be confiscated.’  (Cic. Inv. 2.118)

9. Imperatives and oblique personal pronouns Second and third arguments represented by a pronoun that refers to the speaker or to the addressee often appear immediately after the imperative (cf. above, Example 45, da mihi 13. On sentence-final position of verbs in legal texts, cf. Panhuis (1982:€109).

 Constituent order in Classical Latin prose

testimonium). Is this placement obligatory, or may personal pronouns occupy another position, and thus be separated from the imperative? These are question that I will answer now. The Ciceronian corpus (BTL) offers examples showing that another word, in particular an adverb such as nunc (56) or prius (57) may intervene between the imperative and the pronoun. Sequences with hoc are not fixed either, as is shown in combinations with anaphoric (58) and cataphoric hoc (59): (56) Sequere nunc me in campum. follow now me-acc to Campus ‘Now follow me in the Campus.’

(Cic. Att. 4.15.7)

(57) Reddite prius nobis Brutum. restore-pl first us-dat Brutus-acc ‘First restore to us Brutus.’

(Cic. Phil. 11.24)

(58) Da mihi hoc. give me-dat this-acc ‘Allow me this.’

(Cic. Div. Caec. 23)

(59) Credite hoc mihi, iudices: nulla umquam civitas... believe-pl this-acc me-dat judges no ever city-nom ‘Believe me in this, gentlemen: no city ever...’ (Cic. Ver. 4.133) Furthermore, it is possible to insert a parenthesis such as inquis ‘you say’ or quaeso ‘please’ between the imperative and the personal pronoun, as illustrated in (60) and (61). (60) Crede, inquis, mihi. believe you say me-dat ‘Believe me, you say.’

(Cic. Att. 11.6.2)

(61) Quam ob rem da, quaeso, te huic etiam generi litterarum. therefore give please you-acc this-dat also field-dat of literature ‘Therefore, please, devote yourself to this field of literature also.’(Cic. Ac. 1.12) The pronouns are also found before the imperative. Example (62) shows a non-contrastive nobis that fills the position of the third argument; in this sentence tu contrasts with ego and Focus is on the discontinuous hanc… veniam that announces a complement clause.14 (62) Ego vero... Sed tu hanc nobis da veniam, I indeed but you this-acc us-dat give favour-acc Scaevola, et perfice ut... Scaevola-voc and persuade to ‘(Nay I...) But do us this favour yourself, Scaevola, and persuade...’  (Cic. De orat. 1.163) 14. In Cicero, there are not many occurrences of this type where a pronoun precedes the imperative. However, in other authors, especially poets, such arrangements are more frequent, cf. munus hoc mihi maximi da, Colonia, risus (Catul. 17.7) ‘give to me, Colonia, the gift of greatest laughter’.



Chapter 5.╇ Imperative sentences 

To sum up, personal pronouns often stand after the imperative. However, this is not a constraint but a consequence of the fact that the pronoun does not have a special pragmatic function. It is worth remembering that oblique personal pronouns referring to the speaker and to the addressee are dependent on the situation and, therefore, accessible to the addressee (cf. Panhuis 1982:€45). On the other hand, when oblique personal pronouns are used with contrast or emphasis, they take an important position, especially the initial one, as in (63) and (64). Pronouns of the second person singular in the last example function as reflexives. (63)

Sed tamen perspice quo ista vergant mihique aut but still watch where these tend me-dat=and either scribe aut, quod multo malim, adfer ipse. write or what much I would prefer bring yourself ‘However, watch the way things tend in Rome and write to me, or else, as I should much prefer, bring your news in person.’ (Cic. Att. 16.6.2)

(64) Tecum loquere, te adhibe in consilium, te with refl speak refl-acc bring in deliberation refl-acc audi, tibi obtempera. listen refl-dat obey ‘Be your own confidant and counsellor; listen to yourself, obey yourself.’  (Cic. Fam. 2.7.2)

10. Conclusions The examination of imperative sentences shows that imperatives are not confined to the sentence-initial position. As in declarative sentences, priority may be given to Topic constituents. However, unlike declarative sentences, salient constituents frequently occupy the sentence-final position, in other words, they follow the verb in the imperative. This confirms two main observations: (i) Panhuis’s (1982:€68 ff.) statement that the imperative as a mood does not have a fixed sentence-initial position, and (ii) Pinkster’s (1995:€217) observation that imperative sentences – except for complex sentences with future imperatives – generally do not have the verb in the final position, as is often the case in declarative sentences. A distinction between imperatives forming periphrastic constructions and various communicative formulas on the one hand, and directive imperatives on the other, permits us to understand better the placement of the verbs in the imperative. Indeed, the first group of imperative sentences (periphrases and communicative formulas) often lacks a Topic, and the imperative, placed in the sentence-initial position, is followed by its complements, mainly complement clauses. Sentences containing directive imperatives can have all kinds of Topic constituents in sentence-initial position. Such an ordering is also found when the imperative itself functions as Focus of the sentence.

 Constituent order in Classical Latin prose

Pronouns in the nominative are generally contrastive and occupy the sentence-initial position. Oblique personal pronouns, when not contrastive, avoid this place; however, they do not have a fixed position in the sentence (for example, after the imperative). Future imperatives are infrequent, and they are only formed with a limited number of verbs. Two main tendencies of their placement can be observed: (i) future imperatives placed initially and expanded by a complement clause; (ii) future imperatives used in complex sentences starting with an adverbial clause; the imperative is usually found in sentence-final position.

chapter 6

Noun phrases 1. Introduction Latin noun phrases consist of a head noun and one or more modifiers, such as an adjective, a demonstrative, a possessive, a numeral, a genitive complement, a prepositional phrase, or other constructions.1 The types of modifiers mentioned are optional in the sense that they may in principle2 be added to any noun. An example is in (1). A complement used with bivalent nouns is to be considered as obligatory,3 for example victoriae governed by nuntius ‘messenger’ in (2). (1) ille pastor Idaeus that shepherd Idaean-nom ‘that Idaean shepherd’ (2) nuntius victoriae messenger-nom victory-gen ‘messenger of the victory’

(Cic. Att. 1.18.3)

(Liv. 23.11.7)

This chapter is about the ordering of head nouns and their modifier(s). As pointed out by Lehmann (1991:€224), in Latin noun phrases, the components are relatively loosely connected to each other. This characteristic explains, on the one hand, why modifiers are not mechanically placed before or after the noun, and on the other hand, why modifiers can be quite easily positioned at a distance from their head noun. If we compare studies that have been made on the placement of modifiers in Latin with, for example, the situation in another language with “free” constituent order, such as Polish, we observe a striking difference. In Polish, the tendencies of adjective placement have been quite well explained and any special ordering of modifiers is described in terms of contrast or emphasis (Kłósek 1984). Latin scholars usually deplore the fact that statistical data do not permit them to draw clear conclusions about the relative position of modifiers and, additionally, that there are important variations not only 1. I will not deal here with relative clauses, participial clauses, gerunds and gerundival constructions. For genitives in a gerundive, see Bolkestein (1998c). 2. If no semantic restrictions apply. 3. On bivalent nouns and their complements, see Pinkster (1995:€92 ff.), and on the semantically different status of nouns such as potestas ‘power, opportunity’, spes ‘hope’ or nuntius ‘messenger’, see Bolkestein (1998c: 187). However, the distinction between optional and obligatory complements will not be considered in this chapter.

 Constituent order in Classical Latin prose

between authors, but also in works of one and the same author.4 These difficulties with the description of Latin noun phrases are not only due to the lack of Latin native speakers; they also signal the fact that Latin noun phrases are very flexible, even more so than noun phrases in Polish. The aim of this chapter is to examine Latin noun phrases with particular attention to pragmatics. The chapter is organised as follows. Adjective modifiers (their placement, fixed expressions, juxtaposed and coordinated adjectives) are examined in Section 2. Determiners are divided into four sections: demonstratives (Section 3), indefinites (Section 4), numerical and non-numerical quantifiers (Section 5), and possessives (Section 6). Section 7 is devoted to combinations of modifiers, with both determiners and adjectives. Genitive complements are examined in Section 8, which also includes instances of hyperbata involving modifiers of the noun phrase. Discontinuity brought about by constituents that are not part of the noun phrase is the topic of Section 9.

2. Adjective modifiers 2.1

Placement of adjectives

Marouzeau’s theory of adjective placement is well-known. According to him (Marouzeau 1922:€17 and 75), adjectives placed before their head noun are supposed to have “qualifying” value whereas “determinative” value is ascribed to adjectives standing after their head noun. The distinction between qualifying and determinative adjectives is not without problems but in general, qualifying adjectives are those that can be used as predicatives (populus est ferus ‘the people are savage’) and allow comparative and superlative degrees (Fugier 1983:€ 239). Determinative adjectives lack these properties (*populus est Romanus *‘the people are Roman’). Marouzeau’s theory was criticised by de Neubourg (1977 and 1978), and more recently by Lisón Huguet (2001).5 “Marouzeau drew his inspiration from French”, says de Neubourg (1978:€ 359); “in Old French, all adjectives were systematically in prenominal position”. The major objection is that the principles formulated by Marouzeau do not apply in a systematic way. Certain (but not all) qualifying adjectives do prefer pre-nominal placement, for example magnus ‘big’ and parvus ‘small’,6 but determinative adjectives are placed after the noun only in 50% of cases (Lisón Huguet 2001:€78, and Hoff 2003). Thus, not only is it difficult to establish a clear-cut distinction between 4.

See in particular Lisón Huguet (2001:€74). For differences in Cicero’s works, see Menk (1925:€16).

5. See also Pinkster (1995:€107 ff.). The theory about anteposed qualifying and postposed determinative adjectives poses a problem in Ancient Greek as well. H. Dik (1997:€61) argues that there is one default ordering: postposition of adjectives with respect to the noun. 6. See Marouzeau (1922:€75 f.) and Devine and Stephens (2006:€417, 474 and 468) on magnus, maximus and parvus, that are placed pre-nominally in 85 – 95% of cases.



Chapter 6.╇ Noun phrases 

qualifying and determinative adjectives, but their placement as described by Marouzeau is also not systematic. There is another point that should be considered: qualifying adjectives such as magnus express a subjective evaluation and may bear emphasis, which is responsible for their very frequent pre-nominal placement. Furthermore, with these adjectives discontinuity is quite regular, as has already been observed by Marouzeau himself (1922:€ 216), quoting the following example with discontinuous magnae res and non-discontinuous magnis copiis: (3) Magnae saepe res non ita magnis copiis great-nom often things-nom neg so great-abl forces-abl sunt gestae. have been accomplished ‘Great things have often been accomplished with not so very great forces.’  (Nep. Pel. 2.3) In my approach, I will examine principles governing the placement of adjectives in the light of pragmatics. As a point of departure, I will take observations made by de Jong (1983) and Pinkster (1995:€239)7 that adjectives are normally postposed to their head noun; they are preposed when they carry emphasis or contrast. This basic assumption is to be refined, as my analysis further on in the chapter will show, because the placement of adjectives correlates with pragmatic functions. The pre-nominal placement is favoured not only by contrastive or emphatic adjectives but also by adjectives belonging to a contextually dependent noun phrase (with or without Topic function). Adjectives pragmatically marked as Focus are found post-nominally but can also go in the pre-nominal position. Adjectives that are pragmatically unmarked are mainly found in post-nominal position. For example, the position of adjectives like novus ‘new’ and vetus ‘old’ is not due to semantic but to pragmatic reasons.8 Post-nominal vetus in (4) forms part of the new information; in (5), pre-nominal vetus is in explicit contrast with quolibet. The noun phrase containing post-nominal novus in (6) figures in a whathappens sentence and is contextually independent. In (7), novi consules functions as Topic and presents already given information. (4) Corpus eius leniter ex oleo vetere... permulceatur. body-nom his lightly with oil old should be rubbed ‘his body should be lightly rubbed... with old oil.’€

7.

(Cels. 3.23.2)

Cf. also de Sutter (1986:€160 ff.).

8. Pace Marouzeau (1922:€93), and Devine and Stephens (2006:€446 f.) who claim that there is a semantic difference between post-nominal novus ‘unused’ and pre-nominal novus ‘new’. Similarly, post-nominal vetus is supposed to mean ‘old’, pre-nominal ‘former’.

 Constituent order in Classical Latin prose

(5)

perfricandae quoque eae partes manibus unctis to be rubbed also these parts-nom hands-abl anointed-abl ex vetere oleo sunt with old oil are [... ex quolibet oleo] ‘the said parts are also to be rubbed by hands anointed with old olive oil [... with any kind of oil]’ (Cels. 3.11.2)

Extemplo igitur consules novi L. Aemilius Mamercinus at once therefore consuls new-nom L. Aem. Mamercinus-nom et C. Plautius... conparare inter se provincias iussi... and G. Plautius-nom to divide between refl provinces-acc were ordered ‘Without delay, the new consuls Lucius Aemilius Mamercinus and Gaius Plautius... were accordingly ordered... to divide commands between them...’  (Liv. 8.20.3)

(6)

(7)

2.2

[consules creati sunt T. Veturius, Sp. Postumius] Novi deinde consules... ingressi hostium fines... new-nom then consuls-nom entered of enemies territory-acc [the consuls elected were Titus Veturius and Spurius Postumius] The new consuls then... entered the enemy’s territory...’ (Liv. 8.17.1)

Distinctive function of adjectives

An adjective (this also holds true for determiners) can either function together with its head noun, in which case the underlying question is ‘who, what?’ and concerns the whole unit, or has itself a distinctive function associated with the question ‘how?’ or ‘which (one)?’ The distinctive function of an adjective (or a determiner) means that the property it refers to is significant with respect to other similar entities (cf. Marouzeau 1953:€3). The distinctive function may, but may not, imply contrast. Two examples are given for illustration: the property of being gold is not particularly relevant in (8) whereas the properties expressed by adjectives are distinctive in (9);9 the author says which oxen should be sold: not all of them but only those who are vetulos. (8)

Mihi quidem videtur Brutus noster iam vel coronam me-dat part seems Brutus our-nom now even crown-acc auream per forum ferre posse. gold-acc in Forum wear can ‘It looks to me as though our Brutus can now walk through the Forum with a crown of gold upon his head.’ (Cic. Att. 14.16.2)

9. Example (9) from Marouzeau (1953:€4).



Chapter 6.╇ Noun phrases 

(9)

Boves vetulos, plaustrum vetus..., servum senem, oxen worn-out-acc wagon old-acc slave old-acc servum morbosum, vendat. slave sickly-acc he should sell ‘Sell worn-out oxen, an old wagon..., an old slave, a sickly slave.’(Cato Agr. 2.5)

The distinctive function of adjectives manifests itself in the ordering. This explains, for example, variations such as the pre-nominal and post-nominal placement of adjectives derived from a proper name, as in (10) and (11).10 (10)

Vndecimo anno Punici belli consulatum inierunt M. Marcellus... in eleventh year of Punic war consulate-acc entered M. Marcellus-nom et T. Quinctius Crispinus. and T. Quinctius Crispinus-nom ‘In the eleventh year of the Punic war the consuls entering office were Marcus Marcellus... and Titus Quinctius Crispinus.’ (Liv. 27.22.1)

(11)

Belli Punici secundi quod contra maiores nostros of war Punic second which against ancestors our Hannibal gessit causa fuit Sagunti oppugnatio. Hannibal-nom waged reason-nom was of Saguntum blockade-nom ‘The blockade of Saguntum was the cause of the second Punic war, which Hannibal waged against our ancestors.’ (Cic. Phil. 5.27)

In (10), the order {adjective > noun} is used for dating events. Livy does not say during which war Marcellus and Quinctius started their consulate, but simply “when”; the phrase Punici belli is a part of a temporal setting. In (11), Cicero informs us about the cause of “which” war. In this case, the post-nominal adjective secundi ‘of the second’ has a distinctive value with respect to primi ‘of the first (Punic war)’. In this chapter, I will not discuss typological considerations of placement of adjectives in Latin at all.11 Nevertheless, unlike constituents of sentences that allow a number of possible arrangements, each with their own pragmatic value, I assume that there is a pragmatically neutral order for noun phrases. Nouns – and noun phrases – may function as names of entities: an entity can bear a “title” (i. e. this is XY, this entity is named XY). Examples can be taken from encyclopaedic descriptions: cedrus magna ‘big cedrus’ (Plin. Nat. 24.17), herba lanaria ‘wool-plant’ (Plin. Nat. 24.168), or from geographic or other local names. In such cases, adjectives are usually in post-nominal position, for example: Carthago Nova ‘New Carthage’, Turris Alba ‘White Tower’, or Cloaca Maxima ‘Great Sewer’. Determiners such as numerals stand before the noun: Septem Aquae ‘Seven Waters’. These are lexical units, however, which do not present a postnominal adjective or a prenominal determiner in a systematic way. In Latin, it is 10. On variations of this type, see Lisón Huguet (2001:€79), and Spevak (2006c). 11. See mainly Adams (1976a).

 Constituent order in Classical Latin prose

possible to use a special arrangement for these units. A good example is given in (12): Livy is talking about foundation of a new city, Alba Longa. When he explains the motivation of the name (cf. porrectae... urbis ‘city stretching along’), he says Longa Alba; the pre-nominal position can thus be considered contrastive. Further on, he uses the “current” form, Alba Longa. Longa Alba in (12) has Focus function; so does the discontinuous phrase saxum Tarpeium, denoting the Capitoline hill, with a pre-nominal adjective in (13). In (14), there is an explicit contrast between Old and New Carthage, and therefore, the pre-nominal position of Nova is chosen. (12) (Ascanius) novam ipse aliam sub Albano monte new-acc himself other-acc under Alban mount condidit, quae ab situ porrectae in dorso urbis founded which from position of stretched along the side of town Longa Alba appellata. Inter Lavinium conditum et Albam Longa Alba-nom was named between Lavinium founded and Alba Longam coloniam deductam triginta ferme interfuere anni. Longa colony led out thirty-nom about intervened years-nom ‘(Ascanius) he himself founded a new city below at the foot of the Alban Mount, which from its position, stretching along the side of the hill, was named Alba Longa. An interval of thirty years elapsed between the foundation of Lavinium and the colonisation of Alba Longa.’ (Liv. 1.3.3–4) (13) [hic mons ante Tarpeius dictus...] etiam nunc eius rupes even now its cliff-nom Tarpeium appellatur saxum. Tarpeian-nom is called rock-nom ‘[this hill was previously called the Tarpeian...] even now its cliff is called the Tarpeian Rock.’ (Var. L. 5.41.2) (14)

2.3

[magnam Carthaginem...] Masaesulii... in regionem Hispaniae maxime qua Masaesulians-nom in region of Spain principally where sita Nova Carthago est spectant. placed New Carthage-nom is they face ‘(big Carthage...) The Masaesulians directly face the region of Spain in which lies New Carthage.’ (Liv. 28.17.5)

Fixed formulas

Lisón Huguet (2001:€65) rightly separates “freely” formed noun phrases, such as navis magna ‘big ship’, from fixed expressions. In juridical, administrative, military and



Chapter 6.╇ Noun phrases 

religious formulas, adjectives are always post-nominal,12 for example: pontifex maximus ‘supreme pontiff ’, praetor urbanus ‘praetor of the city’, aes alienum ‘debt’, res publica ‘State’, patres conscripti ‘senators’, res gestae ‘achievements, exploits’, di immortales ‘immortal gods’, populus Romanus ‘Roman people’. These expressions present a fixed order and they are usually not subject to hyperbaton (Fruyt 1990:€181 ff.). Instances with an inverted order and/or discontinuity such as in (15), showing a contrast between alieno and se, are exceptional (although they can be found). (15) (P. Decius) pro alieno se aere devovit for sb else’s refl money he devoted ‘(P. Decius) devoted himself for the debts of another’

(Cic. Phil. 11.13)

However, the question of fixed formulas is complicated by the fact that besides the extremely rigid expressions, such as those I presented above, there are other lexical units with more flexibility.13 This is the case with navis oneraria ‘transport ship’, ora maritima ‘sea coast’ and similar, which allow not only an inverted order, but also discontinuity, as for example in (16) and (17). It is obvious that there are different degrees of “fixedness” of lexical units, and I leave their typology for further research. (16) Naves interim Caesaris onerariae errabundae ships-nom meanwhile of Caesar merchant-nom lost-nom male vagabantur... wrongly were wandering ‘In the meantime Caesar’s merchant ships, lost, were sailing fruitlessly about...’  (B. Afr. 21.3) (17) [Iisdem mandatum ut...] maritima omnis Thraciae maritime-nom whole-nom of Thrace ora a Philippo et Macedonibus liberaretur. coast-nom from Philip and Macedonians should be freed ‘[They were also instructed...] and that the whole Thracian sea cost should be freed from Philip and the Macedonians.’ (Liv. 39.33.4)

2.4

Juxtaposed attributive adjectives

2.4.1 Hierarchical principle of adjective ordering A noun can be modified by more than one adjective. If several adjectives occur in a noun phrase or a prepositional phrase, they are either juxtaposed or coordinated (see Section 2.5, p. 237 ff.). In this section, I will examine noun phrases containing two 12. Apart from the possessive expression erilis filius ‘master’s son’ (see also Adams 1976a: 89). For fixed formulas containing a genitive, see p. 266. 13. Pace Lisón Huguet (2001:€69 f.).

 Constituent order in Classical Latin prose

or more juxtaposed adjectives. In many languages, the ordering of multiple adjectives co-occurring in a noun phrase is governed by semantic principles of hierarchy (Hetzron 1978, and Siewierska 1988:€ 63, among others). A scale of hierarchy proposed by Hetzron, which I will present below, is based on the semantic principle that “an adjective expressing an objective, indisputable property is placed close to the noun, whereas an adjective expressing subjective qualification is usually more distant”.14 In the English Example (18a) below, the adjective expressing subjective evaluation (beautiful) is the most distant from the noun (ball); it is followed by another evaluative adjective (big) that expresses the dimension; after it comes the objective expression of colour (red). The same order is observed in Polish (18b) (Hetzron 1978:€170): (18) a. a beautiful big red ball b. piękna duża czerwona piłka Both English and Polish are considered by Hetzron as languages with the order {adjectives > noun}; the adjectives form a block and precede the noun.15 Languages with the order {noun > adjectives} where the adjective block follows the noun (for example Arabic) use the mirror order: ball red big beautiful, so that here too the evaluative adjective is the most remote one from the noun.16 French is one of the languages with the order {adjective(s) > noun > adjective(s)} where evaluative adjectives are placed before the noun and objective adjectives after the noun (18c):€ (18) c. un joli gros ballon rouge In previous studies on adjective placement in Latin, one question remained unnoticed: if Marouzeau’s distinction between “qualifying” (pre-nominal) and “determinative” (post-nominal) adjectives applies, what happens in noun phrases with more than one modifier? One should expect that a qualifying adjective appears before the noun and a determinative adjective after it (19a). Such an arrangement is usual in Polish in a

14. This principle can also be explained in terms of levels of hierarchy; see Risselada (1984:€202), and Section 2.5.1 below, p. 237. 15. Hetzron (1978:€173) also argues that in some cases the order of adjectives is interchangeable, in particular adjectives expressing dimension and evaluation. Polish allows piękna duża (‘beautiful big’) as well as duża piękna (‘big beautiful’). In English, one can say beautiful old paintings as well as old beautiful paintings; the first expression, more common, refers to old paintings that are beautiful; the second one, to beautiful paintings that are old (see Quirk 1985:€1340, and Pinkster 1995:€110). 16. As Sproat and Shih argue (1991:€587), there are languages (for example Celtic) that use the same ordering as English but in post-nominal position: ball beautiful big red; other languages, such as Japanese, have free placement of adjectives. See also Devine and Stephens (2006:€477).



Chapter 6.╇ Noun phrases 

pragmatically neutral context (19b).17 Consider for example the combination of adjectives meaning ‘big’ and ‘(of) transport’ with the noun ‘ship’: (19) a. magna navis oneraria b. duży statek transportowy18 big ship transport However, the ordering in (19a) is quite rare in Latin (cf. Risselada 1984:€223);19 more frequently we find blocks of adjectives as exemplified in (20) and (21), in both prenominal and post-nominal position. (20) navem onerariam maximam ship-acc transport-acc very big-acc ‘a transport ship of the largest size’ (21) parva oneraria nave small-abl transport-abl ship-abl ‘a small transport ship’

(Cic. Ver. 2.13 and 4.150)

(Liv. 35.51.4)

The applicability of Hetzron’s hierarchical principle of adjective ordering to Latin was discussed by Risselada (1984:€ 208) and de Sutter (1986:€ 156).20 Figure 1 displays Hetzron’s scale (cf. Risselada 1984:€214) with Latin examples, assuming that adjectives follow their head noun en bloc. When the block is prenominal, a mirror order should be understood. Figure 1╇ (Hetzron 1978:€178) noun > “denominal” adjective21 (onerarius) > substance (ligneus) > origin (Punicus) > colour (ruber) > physical defect (caecus) > shape (rotundus) > age (vetus) > social property (felix) > utilitarian qualifications (carus) > physical property (crudus) > dimension (longus) > evaluation (bonus) > affectivity (splendidus) Adjectives denoting an inherent property of an entity (such as onerarius ‘of transport’) are supposed to be close to the noun. They are followed by adjectives expressing substance (ligneus ‘wooden’), origin (Punicus ‘Punic’), colour (ruber ‘red’), 17. Hetzron (1978), when classifying Polish as an {adjective > adjective > noun} language, unfortunately neglects Polish “denominal” adjectives that normally stand in post-nominal position (Kłósek 1984:€33), for example język ojczysty ‘mother tongue’, or psychologia eksperymentalna ‘experimental psychology’. See also the following note. 18. I thank Anna Siewierska for confirming this point to me. Another example would be: podarty bilet tramwajowy ‘a torn tram ticket’. For post-nominal placement of “denominal” adjectives in Polish, see Siewierska and Uhlířová (1998: 134). 19. Cf. Example (34) below where veterem is contrastive. 20. Devine and Stephens (2006:€476) use a simplified scale. 21. I prefer the term “denominal” adjective to Hetzron’s “purpose/destination”; see below for discussion.

 Constituent order in Classical Latin prose

physical defect (caecus ‘blind’), shape (rotundus ‘round’), and age (vetus ‘old’). These properties represent observable and objective judgements that a better informed person may correct. The other qualifications are the result of subjective judgements that one cannot falsify; one can only disagree with them. These are social properties of human beings (felix ‘happy, prosperous’), properties of inanimate entities (carus ‘expensive’), physical property (crudus ‘raw’), dimension (longus ‘long’), evaluation (bonus ‘good’) and affective expressions (splendidus ‘splendid’). The above-presented scale represents a theoretical model; noun phrases containing more than three adjectives are too complex and avoided in Latin (Pinkster 1995:€111) as well as in modern languages. Instead of Hetzron’s term “purpose/destination” adjectives (for example ironing board in English), I prefer the term “denominal” for this category of adjectives, which has already been described by Quirk (new edition 1985:€432 and 1339 ff.). It includes non-gradable adjectives, mostly derived from a noun, adjectives that do not admit intensifiers, degrees of comparison and predicative use.22 This group, very frequent in Latin, includes not only adjectives expressing purpose or destination, for example adjectives denoting what is related to wine (vinarius), transport (onerarius) or corn (frumentarius), but also adjectives that concern an agent or a cause: militaris ‘military’, civilis ‘civil’, consularis ‘consular’, virilis (related to man), agrestis (related to the field) and others. These adjectives express the typical feature of an entity.23 In order to check whether the above-presented hierarchical principle applies to the ordering of adjectives in Latin, I use a corpus24 containing 130 noun phrases with at least two adjectives expressing physical appearance (pulcher ‘beautiful’), dimension (magnus ‘big’), age (novus ‘new’, vetus ‘old’), form (rotundus ‘round’, oblongus ‘elongated’), colour (niger ‘black’, ruber ‘red’, purpureus ‘purple’), or matter (ligneus ‘wooden’, ferreus ‘iron’, marmoreus ‘marble’, argenteus ‘silver’), and denominal adjectives (onerarius ‘of transport’, frumentarius ‘related to corn’, maritimus ‘maritime’, etc.). The principle of hierarchical ordering can be illustrated with the help of the following example 22. This is precisely the category of “denominal” adjectives that privileges postposition in Polish, see note 17 above. 23. When applying the Hetzron’s scale to Latin, Risselada (1984:€ 218 f.) suggested dividing Hetzron’s category of “purpose/destination” into three categories: agent, purpose/destination, and typical characteristics. In my view, these semantic labels have a common denominator: they all express a distinctive feature of an entity that semantically varies with their point of application. For example, frumentarius in navis frumentaria ‘ship destined to transport corn’ has a different point of application (purpose/destination of a ship) from civilis ‘civil’ in bellum civile ‘civil war’ (agent of a war). On the other hand, adjectives derived from a proper name express an agent and are closer to the noun, for example: lex Sempronia frumentaria (Brut. 222) ‘the corn law of G. Sempronius Gracchus’. 24. Occurrences were collected with the help of the BTL (1999), and only concern prose. Classical authors do not offer a sufficient number of examples and therefore writers from other periods of Latin are included.



Chapter 6.╇ Noun phrases 

(see Figure 2): vasa made from silver, argentea (matter), are qualified as a whole by caelata€(physical property); the group is quantified by multa. One can also call this a difference in scope: multa in (20) has a broader scope than caelata, and argentea has in its scope only the noun vasa. Latin allows coordination of the quantifier multus with another modifier, in this case a noun phrase in the genitive (magni ponderis). (22) vasa argentea caelata multa magnique ponderis vessels silver embossed many-nom great-gen=and weight-gen ‘many embossed silver vessels of great weight’ (Liv. 37.46.3) Figure 2.╇ Scope of adjectives [[[vasa argentea] caelata] multa magnique ponderis] 2.4.2 Data Table 1 presents the placement of adjectives (A) and the head noun (N). By ‘nonrespected order’ are meant cases where the adjective ordering does not follow the hierarchical principle indicated above in Figure 1, for example when an evaluative adjective stands closer to the noun than a “denominal” adjective (such as navis magna oneraria).€ We can see at first glance that the order {noun > adjective > adjective} is statistically the most frequent (48%) in comparison with the {adjective > adjective > noun} (29%) and {adjective > noun > adjective} (21%) orders. If we consider only data concerning adjectives in a block, i. e. the orders {noun > adjective > adjective} and {adjective > adjective > noun}, Hetzron’s hierarchical principle is respected in 82% of the cases. This result accords with Risselada’s (1984:€224) findings. We can state that adjectives expressing inherent, objective properties mostly stand close to the noun, whereas evaluative adjectives are placed farther away. On the one hand, this means that unlike the ordering of constituents at the sentence level, the ordering of adjectives is fairly predictable. On the other hand, the deviation of 18% from the semantic principle is relatively high. Table 1.╇ Juxtaposed adjectives in noun phrases (BTL) Noun phrase

Respected order occurrences

N>A>A A>A>N A>N>A>A A>N>A Total

â•⁄ 50 â•⁄ 28 â•⁄â•⁄ 2 â•⁄ 27 107

82%

Non-respected order occurrences 13 10 â•⁄ 0 – 23

18%

Total

Percentage

â•⁄ 63 â•⁄ 38 â•⁄â•⁄ 2 â•⁄ 27 130

48% 29% 2% 21% 100%

 Constituent order in Classical Latin prose

2.4.3 Hierarchical and non-hierarchical ordering of adjectives We have seen that multiple adjectives mostly, but not exclusively, go as a block in the post-nominal position. This holds true for noun phrases used in a neutral context, i. e. without emphasis or contrast (23), as well as for noun phrases with Focus function (24). There are also instances of contrast, for example contrast between Tarentina and Signina in (25). This example shows a contrastive adjective (Tarentina) placed in the middle, and not, as might be expected, at the right end of the noun phrase. (23) ut signa marmorea nuda curantur like statues marble naked-nom are treated ‘in the same way as naked marble statues’

(Vitr. 7.9.3)

(24) [erant] cenationes laqueatae tabulis eburneis versatilibus dining rooms panelled-nom tablets-abl of ivory-abl turning-abl ‘[there were] dining rooms with fretted ceilings of ivory, whose panels could turn’ (Suet. Nero 31.2)  (25)

In grande vas conicienda sunt... pira Tarentina viridia, in large vessel to be thrown are pears Tarentine green-nom vel Signina. or Signine-nom ‘You have to throw in a large vessel... green Tarentine or Signine pears.’(Cels. 4.26)

Noun phrases with multiple adjectives in the pre-nominal position are usually pragmatically pertinent; either they convey new information (26), or they are contextually bound, as vetus Aminneum vinum is in (27). Pre-nominal placement also appears for emphatic or contrastive adjectives, as for example veteri placed at the left end of the noun phrase in (28); regia classis is contextually given. (26)

Conlocari iussit hominem in aureo lecto strato to be placed he ordered man-acc in golden couch covered pulcherrimo textili stragulo... very beautiful-abl woven covering-abl ‘He ordered the man to be placed on a golden couch which was covered with beautiful woven tapestries...’ (Cic. Tusc. 5.61)

(27) (radix) Ea cum vetere Aminneo vino decocta exprimitur. it-nom with old Aminnean wine boiled-nom is pressed ‘It (root) is boiled with old Aminnean wine and pressed.’ (Col. 9.13.8) (28) [...regia classis...] cum coniungi veteri regiae classi... cuperet when to join old-dat royal-dat fleet-dat he wanted ‘[... fleet of the king] when he wanted to join the old fleet of the king...’  (Liv. 37.24.11) Table 1 shows that adjectives are not placed in the hierarchical order in 18% of the cases. Semantic principles governing the ordering of adjectives give way to pragmatic



Chapter 6.╇ Noun phrases 

needs in two situations.25 Firstly, the non-hierarchical order appears when an adjective is contrastive. It stands at the left or right end of the noun phrase, as illustrated in (29) where Tuscis contrasts with Campani. A contrastive adjective can also be placed in the middle of the noun phrase, as magna is in (30); cf. also Example (25) quoted above. (29)

sicut Campani Capuam Tvscis veteribus just as Campanians-nom Capua-acc Etruscan-dat old-dat cultoribus ademptam habitants-dat taken-acc ‘just as the Campanians held Capua, wrested away from its former Etruscan habitants’ (Liv. 28.28.6)

(30)

C. Gracchi frumentaria magna largitio exhauriebat of G. Gracchus of corn-nom big-nom largesse-nom was exhausting igitur aerarium, modica M. Octauii rei publicae therefore treasury-acc moderate-nom of M. Octavius for republic tolerabilis... practicable-nom ‘Thus the extensive largesse of grain, initiated by G. Gracchus, was to exhaust the exchequer whereas the moderate one, inaugurated by M. Octavius, was practicable for the State.’ (Cic. Off. 2.72)

Secondly, the semantic order is not respected in non-contrastive situations also, especially when noun phrases function as Focus; see the examples in (31) and (32). The only factor responsible for the “permuted order” of the adjectives seems to be the pragmatic function of these noun phrases: they are salient, Focus constituents. However, I also found a few instances where the noun phrase does not have a special pragmatic function (33). (31) Eadem sanant ulcera vetera nigra... the same-nom heal ulcers-acc old black-acc ‘These plants... also heal chronic black ulcers.’

(Plin. Nat. 27.62)

(32) Contra haec Pompeius naves magnas onerarias... adornabat. against this Pompey ships-acc big transport-acc fitted out ‘In answer, Pompey fitted out large cargo ships...’ (Caes. Civ. 1.26.1) (33) signum Apollinis pulcherrimum cuius in femore statue-nom of Apollo very beautiful-nom of which on thigh 25. Non-hierarchical order of adjectives is mainly expected in contrastive contexts and enumerations, as illustrated with the following theoretical examples: when one is talking about Corinthian vessels differ in material, vasa Corinthia argentea, vasa Corinthia aurea ‘Corinthian vessels of silver, Corinthian vessels of gold’ would be expected; when about silver vessels of different origine, supellex aerea Corinthia, supellex aerea Deliaca ‘furniture of brass made at Corinth, at Delos’ would signal such a contrast.

 Constituent order in Classical Latin prose

litteris minutis argenteis nomen Myronis erat inscriptum letters-abl small silver-abl name-nom of Myron was inscribed ‘a beautiful statue of Apollo, on whose thigh was inscribed the name of Myron in small silver letters...’ (Cic. Ver. 4.93) 2.4.4 Other orderings It would be erroneous to say that in complex noun phrases exhibiting the order {adjective > noun > adjective}, the evaluative adjective always stands before the noun and the non-evaluative one comes after it. The pre-nominal placement of adjectives is explained by pragmatic reasons: adjectives that appear in pre-nominal position either carry contrast or are contextually bound (with or without Topic function). For example, veterem in (34) enters into explicit contrast with novam and is placed before the noun viam. In (35), only the adjective privata contrasting with publice stands in prenominal position, and the other adjectives are placed post-nominally. Contrastive adjective can also be found in post-nominal position, as in (36); nigras contrasts with candidis, and Arabicas comes before the noun because it is contextually bound. (34)

Q. Fabium Labeonem... derexisse finem Philippo veterem Q. Fabius Labeo-acc fixed as limit for Philip old-acc viam regiam...; Philippum novam postea deflexisse viam... road royal-acc Philip-acc new-acc later directed road-acc ‘that Quintus Fabius Labeo... fixed as the boundary for Philip the ancient royal road... Philip had later made a new road...’ (Liv. 39.27.10)

(35)

tibi apud eosdem privata navis oneraria you-dat by those same private-nom ship transport-nom maxima publice est aedificata enormous-nom publicly was built ‘for you they (Mamertines) built, at their own public expense, an enormous cargo ship for you private use.’ (Cic. Ver. 5.136)

(36) [Arabicam... onychem...] Arabicas onychas nigras inveniri candidis zonis. Arabian-acc onyx-acc black-acc be found white-abl bands-abl ‘[Arabian onyx...] There are black Arabian onyx with white bands.’  (Plin. Nat. 37.90) To sum up: in neutral contexts, Latin noun phrases have their adjectives placed as blocks in post-nominal position and ordered in accordance with Hetzron’s hierarchical principle. Other arrangements correlate with pragmatic functions (Topic, Focus) and features (contrast, emphasis). Noun phrases bearing Focus may have several orderings: postposition, anteposition or permutation (i. e. non-hierarchical order) of adjectives. Noun phrases functioning as Topic, as well as other contextually given noun phrases, show anteposition of adjectives. Contrastive adjectives are anteposed,



Chapter 6.╇ Noun phrases 

but also postposed or placed medially. In other words, pragmatic functions and features can affect the internal order of components of Latin noun phrases, and pragmatically pertinent noun phrases can exhibit various arrangements.

2.5

Coordination of adjectives

In this section, I will pay attention to the coordination of adjectives: attributive (Section 2.5.1) and predicative (Section 2.5.2). Whereas attributive adjectives allow, in well defined conditions, both coordination and juxtaposition, predicative adjectives cannot be juxtaposed but only coordinated. 2.5.1 Attributive adjectives In principle, multiple attributive adjectives belonging to the same hierarchical level are coordinated, whereas adjectives belonging to different levels are juxtaposed (Pinkster 1972:€112, and Risselada 1984). For example two adjectives denoting colour are coordinated (ruber nigerque ‘red and black’), whereas an adjective denoting colour is not coordinated with an adjective denoting material (*niger ferreusque ‘black and iron’) but is juxtaposed to it. Latin belongs to the languages that allow coordination of adjectives denoting quantity, evaluation and dimension.26 This can be explained as prominence given to the first word. The coordinated group usually stands before the head noun, as illustrated in (37). Example (38) shows that coordination is not required in such cases but optional. (37) exponit... plurima et pulcherrima vasa argentea he displays numerous and very beautiful vessels silver-acc ‘he sets out the numerous lovely silver vessels’ (Cic. Ver. 4.62) (38) plurima signa pulcherrima... deportasse te numerous statues very beautiful-acc brought away you-acc negare non potes deny neg you can ‘you cannot deny that you brought away a large number of beautiful statues...’  (Cic. Ver. 1.61) Coordinated adjectives of all types, including “denominal”, usually precede their head noun, as is illustrated in (39).27

26. See Pinkster (1995:€108, and 1972:€112), Risselada (1984:€214), and Kühner and Stegmann (1914, I: 241). For coordination such as ad prima ac dubia signa veris (Liv. 21.58.2) ‘on the first doubtful signs of spring’, see Risselada (1984:€213), and cf. Kühner and Stegmann (1914, II: 25). 27. Cf. statistics given by Lisón Huguet (2001:€139) concerning groups of adjectives (not only coordinated pairs but also blocks of adjectives): 84.5% of these adjectives are in pre-nominal position in Cicero, 82.8% in Livy and 79.2% in Seneca.

 Constituent order in Classical Latin prose

(39)

cum magna terrestri navalique praeda with big-abl from land-abl from ships-abl=and booty-abl Lilybaeum rediit. to Lilybaeum he returned ‘he returned to Lilybaeum with a great quantity of booty taken from the land and from the ships’ (Liv. 27.29.8)

When several adjectives are coordinated, there is no hierarchical order applying, as in (40), concerning qualities of the best type of a he-goat. Their ordering can sometimes be explained by formal reasons: a short adjective is followed by a longer one in prenominal position and vice versa in post-nominal position, as in (41). (40)

Caper... optimus habetur amplissimi corporis... he-goat-nom best-nom is considered with large-gen body-gen nigro densoque et nitido atque longissimo pilo. with black thick=and and glossy and very long hair-abl ‘The points of the best type of he-goat... are a large frame... and black, thick, glossy and very long hair.’

(Col. 7.6)

(41) ingenio hominis pulcherrimo et maximo... genius-abl of man very noble-abl and very great-abl ‘(persuaded by) the genius of this man, so noble and great’ (Plin. Ep. 8.12.4) 2.5.2 Predicative adjectives Whereas juxtaposed adjectives can occur in a noun phrase or prepositional phrase, adjectives with predicative function can only be coordinated and they are, therefore, permutable; their order does not depend on the hierarchical principle of ordering. One example is given in (42). (42)

quod in fluminis modum longum et angustum et because in manner of river long-nom and narrow and vadosum ostium portus sit full of shoals-nom mouth-nom of harbour is ‘because the mouth of the harbour was like a river, long, narrow and full of shoals’  (Liv. 37.14.6)

The predicative use of adjectives is not only found with the verb sum, as in (42), but also with other verbs, for example habeo. In (43), the adjectives denote properties of the leaves. This sentence answers the implicit question ‘how are its leaves?’, and not ‘what does it have?’ Folia is Topic and the adjectives coming after habet convey new information. (43) (persollata) Folia habet maiora etiam cucurbitis et leaves-acc it has larger-acc even than gourd and



Chapter 6.╇ Noun phrases 

hirsutiora nigrioraque et crassiora, radicem albam et grandem. more hairy darker=and and thicker-acc root-acc white and large-acc ‘(persollata) Its leaves are larger, more hairy, darker and thicker even than those of a gourd; its root is white and large.’ (Plin. Nat. 25.113)

3. Demonstrative determiners I now turn to determiners, starting with the anaphoric pronoun is and the demonstratives hic and ille. According to Marouzeau (1922:€149) and Lisón Huguet (2001:€114 ff.), they most often occupy the pre-nominal position.28 For my part, I will examine noun and prepositional phrases containing these determiners in Cicero (Att. 1–4), Caesar (Gal. 1–5), and Sallust, and will pay special attention to the pragmatic functions they fulfil. In the following four sections, I will examine noun and prepositional phrases containing only one (demonstrative) determiner (and possibly an adjective);29 juxtaposed determiners of various types are discussed in Section 7. Table 2 presents the results for the ordering of an anaphoric or demonstrative determiner (D) with respect to its noun (N). Ellipses indicate intervention of other elements between the noun and the determiner.30 Table 2.╇ Noun phrases containing demonstrative modifiers (corpus 2) Noun phrase DN D... N ND Total

is

hic

occ.

%

occ.

%

426 â•⁄â•⁄ 8 â•⁄â•⁄ 2 436

â•⁄ 98 % â•⁄â•⁄ 2 % â•⁄â•⁄ 0 % 100 %

216 â•⁄ 34 â•⁄â•⁄ 8 258

â•⁄ 84 % â•⁄ 13 % â•⁄â•⁄ 3 % 100 %

ille occ. â•⁄ 67 â•⁄ 12 â•⁄ 26 105

% â•⁄ 64 % â•⁄ 11 % â•⁄ 25 % 100 %

28. However, from Marouzeau (1953:€17), one might conclude that the ‘normal’, ‘unmarked’ position of demonstratives is the post-nominal one. 29. In Sections 3 – 6, the same corpus and the same method are used. 30. By other intervening elements, here and in Tables 3 – 6, are meant: hyperbaton within the domain of the noun phrase (for example: ea veterum tribunorum plebis rogatio ‘this bill of the old plebeian tribunes’ Cic. Att. 3.23.2) or the prepositional phrase (his de causis ‘for those reasons’, Caes. Gal.€4.17.1) and hyperbaton involving alien elements (has dabam litteras ‘I wrote this letter’, Cic. Att. 3.21.1). Enclitics and postpositive particles were not counted as intervening elements because the discontinuity they produce is obligatory.

 Constituent order in Classical Latin prose

These figures show31 that is, which has a considerable number of occurrences, strongly prefers immediately pre-nominal position (98%); hic is also mostly pre-nominal (84%) with discontinuity allowed (13%); ille appears in pre-nominal (64%) as well as in post-nominal position (25%). It is also interesting to observe that noun phrases containing is are rarely separated by other elements; hyperbaton mainly occurs in noun phrases containing hic or ille. I have noticed no case of discontinuity with demonstratives in the final position (N... D). The fact that the demonstratives behave differently is in correlation with the functions they fulfil. They stand in pre-nominal position when the noun phrases they belong to function as Topic (1) or when they pick up a referent known from the preceding context (2).32 In other words, these noun phrases represent contextually bound constituents. (1) [unam legionem...] His militibus Q. Titurium Sabinum these soldiers-dat Q. Titurius Sabinus-acc et L. Aurunculeium Cottam legatos praeesse iussit. and L. Aurunculeius Cotta-acc legates-acc command he orders ‘[one legion...] He ordered his legates Quintus Titurius Sabinus and Lucius Aurunculeius Cotta to take command of these soldiers.’ (Caes. Gal. 5.24.5) (2) intromissis militibus nostris sectionem eius oppidi being sent in soldiers our-abl spoil-acc of this town universam Caesar vendidit. whole-acc Caesar-nom sold ‘Caesar sent our soldiers inside, and sold everything in the town in one lot.’  (Caes. Gal. 2.33.6) Furthermore, is and hic appear in various resumptive expressions such as temporal or local settings (eo tempore ‘at the time’), and other phrases (eius modi ‘of that nature’, hoc modo ‘in this manner’, etc.).33 These phrases are usually not discontinuous34 and the pronoun always stands before the noun. The rare postposition of is and hic (see Table 2) occurs when they have a cataphoric function (Marouzeau 1953:€18), either in correlation with a relative pronoun (3) or without one.

31. Something similar can be concluded from the figures given by Lisón Huguet (2001:€115). 32. This example is about a town (Caesar does not give its name) where the Aduatuci brought together all their possessions (see Caes. Gal.€2.29.1 ff.). 33. Such expressions represents 31% of all instances of is and 25% of hic in the three authors. Cicero and Sallust mainly use expressions with is, Caesar with hic (cf. p. 75). 34. An exception is formed by his rebus in participial clauses in Caesar that occasionally allow discontinuity.



Chapter 6.╇ Noun phrases 

(3) liberabam agrum eum qui... publicus fuisset. I released land-acc this-acc which public-nom was ‘I proposed to release the land that was public...’

(Cic. Att. 1.19.4)

Ille does not usually appear in resumptive expressions in my corpus. It refers to a known or retrievable entity belonging to shared knowledge.35 Pre-nominal ille figures in noun phrases with Focus function, as in (4), or in noun phrases without a special pragmatic function, as in (5): (4) Qua re incumbamus, o noster Tite, ad illa praeclara studia. therefore let us throw o our Titus-voc to those famous studies ‘So, Titus mine, let me throw myself into my studies.’ (Cic. Att. 2.16.3) (5) Res illo die non peroratur, iudicium dimittitur. pleading-nom that day neg is completed court-nom is adjourned ‘That day the pleadings were not completed before the court adjourned.’  (Cic. Ver. 2.70) Numerous examples with post-nominal ille allow a pragmatic explanation. In (6), we have a noun phrase functioning as contrastive Topic,36 and as (a part of) Focus in (7). A cataphoric post-nominal ille is illustrated in (8). (6) Vettius ille, ille noster index, Caesari Vettius-nom that-nom that our informer-nom to Caesar pollicitus est sese curaturum... promised refl to find means ‘That man Vettius, my old informer, promised to Caesar to find means...’  (Cic. Att. 2.24.2) (7) Huic ioculatorem senem illum, ut noras, interesse this-dat humorist-acc old man that-acc as you know participate sane nolui. indeed I didn’t want ‘I did not want the old gentleman, as you know he was, to participate.’  (Cic. Att. 4.16.3) (8) Versabatur mihi tempus illud ante oculos cum ille appeared me-dat time-nom that-nom before eyes when he

35. On the functions of ille, see Pinkster (2005a). 36. For ille in postposition, especially with proper names, cf. Lisón Huguet (2001:€117), who also gives further bibliographical references. There are also several instances of ille inserted into a proper name; for example: Marci illius Auri perpropinquus (Cic. Clu. 23) ‘a close relative of this Marcus Aurius’.

 Constituent order in Classical Latin prose

aut lictores dimitteret... either lictors-acc would dismiss ‘I pictured the moment when he would either have to dismiss his lictors.’  (Cic. Att. 3.9.1)

4. Indefinite modifiers Indefinite pronouns are lexical means for expressing the indefinite character of an entity, concrete as well as abstract. According to previous studies, indefinite pronouns occupy the pre-nominal position (Marouzeau 1953:€19, and Lisón Huguet 2001:€120). I will concentrate on only two indefinite pronouns, aliquis ‘some’ and quidam ‘a certain’.37 Table 3 presents the ordering of indefinite pronouns (I) and their nouns (N); ellipses indicate discontinuity. Pre-nominal placement is statistically dominant in my corpus also; on the other hand, the remaining numbers of post-nominal indefinites (17% of aliquis and 33% of quidam) are not negligible. Discontinuity mainly concerns aliquis. Typical examples of pre-nominal placement are given in (9) and (10); they show indefinite pronouns with a concrete and an abstract entity respectively. Whereas in quandam planitiem is Focus of its sentence, ex quodam rumore has no particular pragmatic function. (9)

pervenit in quandam planitiem locis paulo he arrived at certain-acc plain-acc places-abl little superioribus circumventam elevated-abl surrounded-acc ‘he arrived at a plain, surrounded by a rather more elevated area...’(Sal. Jug. 68.2)

Table 3.╇ Noun phrases containing indefinite pronouns (corpus 2) Noun phrase IN I... N NI N... I Total

aliquis occ.

%

17 â•⁄ 6 â•⁄ 5 â•⁄ 1 29

â•⁄ 59 % â•⁄ 21 % â•⁄ 17 % â•⁄â•⁄ 3 % 100 %

quidam occ. 29 â•⁄ 1 15 â•⁄ 1 46

% â•⁄ 63 % â•⁄â•⁄ 2 % â•⁄ 33 % â•⁄â•⁄ 2 % 100 %

37. For a semantic description of aliquis and quidam, see Orlandini (1995:€15 f., and 1983:€233).



Chapter 6.╇ Noun phrases 

(10)

Nos hic te ad mensem Ianuarium exspectamus ex we here you-acc by month January expect from quodam rumore an ex litteris tuis ad alios missis. certain rumour or from letters yours-abl to others sent-abl ‘I am expecting you here by January, from a current rumour or it may be from letters of yours to other people.’ (Cic. Att. 1.3.2)

However, indefinites belonging to a noun phrase with Focus function also occur in postposition (11). (11) Quo maior est suspicio malitiae alicuius, cum id... by this greater-nom is suspicion-nom of wickedness some when it ‘That makes one all the more suspicious of some malicious intention, when it...’  (Cic. Att. 3.23.4) The second indefinite pronoun examined, quidam, offers remarkable combinations with proper names. Such instances do not always involve indefiniteness properly speaking, as is the case for common nouns such as quidam miles ‘a (non-specified) soldier’. With a proper name, quidam signals that this person is supposed to be unknown to the addressee (Rosén 1998:€ 728). Such noun phrases are typically Focus (Future Topics) of their sentence and quidam stands in post-nominal position.38 A well-known example is the presentative sentence in (12);39 another one, taken from Caesar, is given in (13); note the discontinuity of the noun phrase produced by ex Hispania indicating the person’s origin. (12)

Est autem C. Herennius quidam, tribunus plebis, is but G. Herennius certain-nom tribune-nom of people quem tu fortasse ne nosti quidem. whom-acc you perhaps not know even ‘However, there is a certain plebeian tribune named Gaius Herennius, whom you, perhaps, do not even know.’ (Cic. Att. 1.18.4)

(13) Mittitur ad eos conloquendi causa... Q. Iunius is sent to them negotiating for sake Q. Iunius-nom ex Hispania quidam, qui... from Spain certain-nom who ‘Quintus Junius, a certain person from Spain, who... is sent to negotiate with them.’  (Caes. Gal. 5.27.1)

38. I noticed only one instance of anteposition of quidam in my corpus: Post eum diem quidam L. Tarquinius ad senatum adductus erat, quem... (Sal.€Cat. 48.3) ‘On the following day, a certain L. Tarquinius was brought before the Senate, who... ’. 39. Quoted and commented on by Bolkestein (1996b: 10 f.).

 Constituent order in Classical Latin prose

5. Quantifiers 5.1

Non-numerical quantifiers

In this section, numerical and non-numerical quantifiers are examined separately. I will start with the latter. Both Marouzeau (1953:€19 f.) and Lisón Huguet (2001:€112) state that non-numerical quantifiers such as multus ‘much’, paucus ‘little’, omnis ‘all, every’, totus ‘whole’ mainly occupy the pre-nominal position. Marouzeau (1953:€22) explains this behaviour as a result of their ‘intensive character’. I will consider here only three quantifiers (Q), omnis, multus and nullus. Their placement is summarized in Table 4. Whereas my global figures more or less correspond to Marouzeau’s (1953:€19 f.) and Lisón Huguet’s (2001:€112) findings in the sense that the pre-nominal position is statistically dominant, there are two other points to be signalled. The first one is that there are a considerable number of discontinuous phrases, especially with multus and with nullus. They occur in all authors, even in Sallust, who is otherwise spare with hyperbaton. The second one is that omnis and nullus, but not multus, are frequently placed in post-nominal position. The universal quantifier omnis is often found in pre-nominal position, as in (14); the noun phrase it belongs to functions as Focus. Omnis goes in post-nominal position especially when emphasis is given to this quantifier; for example in (15) miserias omnes has a particular, subjective value for the author. In historical narrative, omnis may be postposed in order to signal ‘not a part but an entity as a whole’, as in (16).40 (14) Adhibentur omnes pontifices qui erant senatores. are called in all-nom pontiffs-nom who were senators-nom ‘All the Pontiffs who were Senators were called in.’ (Cic. Att. 4.2.4) Table 4.╇ Noun phrases containing non-numerical quantifiers (corpus 2) Noun phrase QN Q... N NQ N... Q Total

omnis

multus

occ.

%

occ.

%

295 â•⁄ 45 â•⁄ 46 â•⁄â•⁄ 2 388

â•⁄ 76 % â•⁄ 12 % â•⁄ 12 % â•⁄â•⁄ 0 % 100 %

â•⁄ 76 â•⁄ 19 â•⁄â•⁄ 4 â•⁄â•⁄ 1 100

â•⁄ 76 % â•⁄ 19 % â•⁄â•⁄ 4 % â•⁄â•⁄ 1 % 100 %

nullus occ. 31 33 13 â•⁄ 5 82

% â•⁄ 38 % â•⁄ 40 % â•⁄ 16 % â•⁄â•⁄ 6 % 100 %

40. This placement of omnis could be linked with the placement of numerical quantifiers (see Section 5.2). Omnis also occurs in postposition at the end of enumerations (Marouzeau 1953:€22).



Chapter 6.╇ Noun phrases 

(15) Non faciam ut enumerem miserias omnes in quas incidi. neg I will do that I enumerate miseries-acc all-acc in which I fell ‘I will not attempt to enumerate all the miseries into which I have fallen.’  (Cic. Att. 3.7.2) (16) equitatum omnem, ad numerum quattuor milium..., praemittit cavalry-acc all-acc in number four thousand he sends forward ‘he sends forward all his cavalry ahead – 4,000 in number...’(Caes. Gal. 1.15.1) Multus privileges the pre-nominal position (17); the post-nominal position is due to emphasis given to multus (18). (17) multa castella et oppida... capit incenditque numerous fortresses and towns-acc he takes burns=and ‘he captured and burned many strongholds and towns...’

(Sal. Jug. 54.6)

(18) cum auro et argento multo Romam legatos mittit with gold and silver much to Rome legates-acc he sends ‘he sends legates to Rome with a quantity of gold and silver’

(Sal. Jug. 13.6)

Nullus is a negative numeral that often implies emphasis. This explains not only its prenominal placement but also the large number of hyperbata. Noun phrases containing nullus function as Focus, as in (19). The noun phrase in (20) with a post-nominal determiner also has Focus function, but nullus is in implicit contrast with aliqua (medicina is contextually given). (19)

In Italia nullus exercitus, Cn. Pompeius in extremis terris in Italy no-nom army-nom Gn. Pompey-nom in distant lands bellum gerebat. war-acc was waging ‘There was no army in Italy, Gnaeus Pompey was waging war in the most distant lands.’ (Sal. Cat. 16.5)

(20) aperteque loquantur et iam clare gemant, tamen openly=and they speak and now aloud groan yet medicina nulla adferatur remedy-nom no-nom is applied ‘and people now speak openly and groan aloud, yet no remedy is applied’  (Cic. Att. 2.20.3) The results of the examination of omnis, multus and nullus, combined with the previous findings of Marouzeau and Lisón Huguet, can be extended to other non-numerical quantifiers such as paucus ‘little’, tantus ‘so great’, plures ‘several’, ullus ‘any’, etc. They all privilege the pre-nominal position, and several of them are fairly frequently separated from their head noun.

 Constituent order in Classical Latin prose

5.2

Numerical quantifiers

In Marouzeau’s view (1953:€23), information given by ordinal numerals is of the “discriminative” type, and he therefore claims that they are normally post-nominal. As for cardinal numerals, it is not very clear what Marouzeau (1953:€24) considers their “normal placement”. Table 5 displays data concerning numerical quantifiers (Q) collected from my corpus 2. The data obtained for cardinal and ordinal numerals show that the pre-nominal position is statistically dominant, pace Marouzeau. However, the post-nominal position is also quite common (29%). Similar results were reported by Lisón Huguet (2001:€108). Furthermore, as we can see from Table 5, there is no notable difference between cardinals and ordinals, unlike Marouzeau’s statement (1953:€24). A similar observation was also made by Lisón Huguet (2001:€132), who additionally examined distributive and multiplicative numerals. For my part, I will describe the placement of numerals from the point of view of pragmatics, and signal some problems with their description. Numerals are used with countable animate or inanimate entities. They are used in expressions of time, when one counts hours, days or years; in expressions of dimension or value; or in other expressions indicating the quantity of persons, horses, legions, letters, etc. Numerals can have a distinctive value; in such cases, the associated question is ‘how many?’ This is particularly evident from various enumerations such as those of a death toll in which the numeral is usually post-nominal, as for example in (21). Besides enumerations, this post-nominal placement is also expected when the numeral has distinctive numerical value (22); however, pre-nominal placement may be used here as well (23). In the first case, the whole noun phrase carries Focus, whereas in the second one, only quattuor is salient; the noun pagus here is contextually bound. In (24), there is an implicit contrast with exemplum unum; this explains why duo stands in post-nominal position. Table 5.╇ Noun phrases containing numerical quantifiers (corpus 2) Noun phrase

cardinal num. occ. %

QN Q... N NQ N... Q Total

â•⁄ 83 â•⁄â•⁄ 9 â•⁄ 42 â•⁄ 10 144

â•⁄ 58 % â•⁄â•⁄ 6 % â•⁄ 29 % â•⁄â•⁄ 7 % 100 %

ordinal num. occ. % â•⁄ 66 â•⁄ 10 â•⁄ 34 â•⁄â•⁄ 6 116

â•⁄ 57 % â•⁄â•⁄ 9 % â•⁄ 29 % â•⁄â•⁄ 5 % 100 %



Chapter 6.╇ Noun phrases 

(21)

Duobus his unius diei proeliis Caesar desideravit in two these-abl of one day battles-abl Caesar-nom lost milites DCCCCLX et notos equites Romanos... tribunos soldiers 960-acc and known equestrians Roman-acc tribunes militum quinque et centuriones XXXII... signaque military five-acc and centurions 32-acc standards-nom=and sunt militaria amissa XXXII. were military-nom lost 32 ‘In these two battles fought on a single day Caesar lost 960 legionary soldiers and some well-known Roman equestrians... 5 military tribunes and 32 centurions... and 32 military standards were lost.’ (Caes. Civ. 3.71.1)

(22) Gallia est omnis divisa in partes tres, quarum unam... Gaul-nom is whole-nom divided in parts three of which one-acc ‘Gaul is as whole divided into three parts, one of which...’ (Caes. Gal. 1.1.1) (23) (pagus) Nam omnis civitas Helvetia in quattuor pagos divisa est. for whole State Helvetia-nom in four districts divided is ‘(district) For the whole State of the Helvetii is divided into four districts.’  (Caes. Gal. 1.12.4) (24)

[Cur igitur cuiquam misi prius?] Vrgebar ab eo ad quem misi et non habebam exempla duo. I was pressed by him to whom I sent and neg I had copies two-acc ‘[Why then did I send it to anybody else first?] The person to whom I sent it was pressing me and I did not have two copies.’ (Cic. Att. 4.5.1)

However, we know from contemporary languages that not all noun phrases containing numerals necessarily have a numerical value. This is the case for various expressions that originally indicated a quantity but later behave as a unit, for example a place in the Roman countryside called Tres tabernae ‘Three taverns’, or the lake Septem Aquae ‘Seven Waters’.41 Similarly, septima legio ‘the Seventh legion’ can be used not only as an indication of the rank (with the underlying question quanta?)42 – where legio is omissible – but also as a simple expression with the underlying question ‘who’? – in which case the head noun is not omissible. This distinction may have consequences for the ordering of the modifier and permits us to explain Caesar’s expressions such as cum legionibus tribus (25) that appear as well as cum tribus legionibus (26).43 In such cases, the post-nominal numeral has distinctive value (de Sutter 1986:€169; cf. Marouzeau 1953:€26): cum legionibus tribus answers the question ‘with how many legions?’, whereas cum tribus legionibus answers the question ‘with whom?’ 41. Mentioned by Cic. Att. 2.1 and 4.15.5. The place name will be maintained even if the ‘three taverns’ are demolished or another is constructed. 42. Cf. Legione ex consuetudine una frumentatum missa, quae appellabatur septima... (Caes. Gal.€4.32.1) ‘One legion, called the Seventh, had been sent as usual to collect corn...’ 43. For such variations, see Marouzeau (1922:€189).

 Constituent order in Classical Latin prose

(25) Titum Labienum legatum cum legionibus tribus subsequi iussit. Titus Labienus-acc legate-acc with legions three to follow he ordered ‘He ordered his legate Titus Labienus to follow in support with three legions.’  (Caes. Gal. 2.11.3) (26) Ipse... cum tribus legionibus eum locum petit, quo... himself with three legions that place-acc he seeks where ‘He himself... with three legions seeks that place to which...’(Caes. Gal. 7.60.4) Furthermore, the post-nominal placement that correlates with distinctive value (question: ‘how many?’ or: ‘which one?’) is typically chosen when the noun phrase conveys new information (27). By contrast, when an entity has already been mentioned in the preceding context, it is not the post-nominal but the pre-nominal placement that appears, as is illustrated in (28). (27) cum in dextro cornu legio duodecima et non although at right flank legion twelfth-nom and neg magno ab ea intervallo septima constitisset at great-abl from it distance-abl seventh-nom took position ‘although the Twelfth legion was on the right flank, and the Seventh not far off ’  (Caes. Gal. 2.23.4) (28) [milites legionis decimae...] Non inridicule quidam neg unwittily certain-nom ex militibus decimae legionis dixit:... from soldiers of tenth legion said ‘[soldiers of the Tenth legion...] One of the soldiers of the Tenth legion said, not without a touch of humour:...’ (Caes. Gal. 1.42.6)

5.3

Temporal expressions

It is beyond the scope of this book to examine temporal expressions in detail and I will do so only briefly. For their description, I will adopt a distinction between constituents functioning as setting and constituents functioning as a specification of the action expressed by the verb, as suggested by Firbas (1992:€49 f.). Compare the examples in (29a) and (29b): (29) a. Yesterday I met an old friend.44 b. This happened yesterday. The first temporal expression (29a) indicates when an event took place, and represents its temporal setting. The question underlying this sentence is ‘what happened at that time?’ Temporal settings such as yesterday in (29a) serve as a point of departure for the 44. In English, it is possible to say I met an old friend yesterday as well without any noticeable pragmatic difference.



Chapter 6.╇ Noun phrases 

sentence and represent one kind of Topic. The second temporal expression (29b) has a different function: it specifies the action expressed by the verb and indicates when the action occurred. In this case, the underlying question is ‘when did it happen?’ When functioning as a specification of the verb, the temporal expression can represent Focus, or at least be a part of it. Two Latin examples are given in (30) and (31): (30) Postera die circiter hora tertia pervenit in quandam planitiem... on next day around at hour third he arrives in certain plain ‘Around the third hour on the next day he arrived at a plain...’ (Sal. Jug. 68.2) (31)

E Pompeiano navi advectus sum in Luculli nostri from Pompei boat-abl I arrived to of Lucullus our hospitium VI Id. hora fere tertia. house on 10th at hour about third ‘I arrived by boat from Pompeii to enjoy our friend Lucullus’s hospitality on the 10th at about nine o’clock in the morning.’ (Cic. Att. 14.20.1)

Hora tertia is a part of a more complex temporal setting in (30); Focus of this sentence is on quandam planitiem (‘where did he arrive?’), and not on the temporal indication. The sentence exemplified in (31) answers the question ‘when?’ and hora fere tertia conveys, together with the date and the directional complement, new information.45 Although I will not go into detail concerning the position occupied by temporal expressions, this fundamental distinction between setting and specification seems to have consequences for the positioning of numerals. Before showing this, some more general aspects related to expressions of time in Latin should be mentioned. The division of day and night time is expressed in terms of hours (hora) and, in military texts, by watches (vigilia), combined with ordinals. As Marouzeau (1953:€ 28) rightly observed, hora is normally followed by the ordinal (cf. Examples 30 and 31), whereas vigilia is normally preceded by it (for example, prima vigilia ‘at the first watch’ in Sal. Jug. 106.4). The almost total absence of variation in the ordering can be explained by the fact that these expressions refer to relatively fixed sections of the day and the night and are therefore strongly lexicalised. On the other hand, except for fixed dating formulas such as ante diem quartum Kalendas ‘three days before Kalends’, where the ordinal is always post-nominal, a unit such as a day (dies), in combination with an ordinal, is used in a relative way. That is, its reference point is variable; for example, die septimo ‘on the seventh day’ does not refer to a specific day of the month but indicates a distance between two events. In such cases, my corpus shows a difference between setting expressions such as (32), with a pre-nominal ordinal, indicating what happened on that day, and expressions that specify when the action expressed by the verb, as in (33), was accomplished (‘when did they arrive?’), and have a post-nominal ordinal.46 45. For the ordering of complex temporal expressions, see Spevak (forthcoming, b). 46. I would explain Marouzeau’s examples (1953:€24) with septimo die (Caes. Gal.€1.41.5) and die septimo (Caes. Gal.€1.10.5) in the same way.

 Constituent order in Classical Latin prose

(32) Tertio die Caesar vallo castra communit. on third day Caesar-nom rampart-abl camp-acc fortified ‘On the third day Caesar fortified the camp with a rampart.’(Caes. Civ. 1.42.4) (33)

Nullam partem noctis itinere intermisso in fines no-acc part-acc of night march being interrupted to territory Lingonum die quarto pervenerunt. of Lingones on day fourth they arrived ‘After a march discontinued for no part of the night, they reached the territory of the Lingones on the fourth day.’ (Caes. Gal. 1.26.5)

Similar tendencies of placement are found for other time indications, such as with mensis ‘month’ or annus ‘year’. Furthermore, an ordinal can be found with emphatic value, as tertium is in the following example (with discontinuity): (34) Qui miseros publicanos... tertivm iam mensem vexat... This-nom unhappy-acc publicani-acc third-acc now month-acc worries He has been for the last three months worrying those unhappy publicani...’  (Cic. Att. 1.18.7)

6. Possessive modifiers Latin has a set of possessive pronouns for the first and the second person, and a reflexive pronoun suus for the third person. According to Marouzeau (1922:€133), possessive pronouns normally stand in the post-nominal position whereas pre-nominal possessives have special relevance. Nevertheless, Marouzeau admits (1953:€16) that there is much variation among authors, and both placements, i. e. pre-nominal and post-nominal, are found. Data obtained for possessive pronouns (P) meus and suus in my corpus are summarized in Table 6. Table 6.╇ Noun phrases containing possessives (corpus 2) Noun phrase PN P... N NP N... P Total

meus occ.

%

â•⁄ 82 â•⁄â•⁄ 4 104 â•⁄â•⁄ 3 193

â•⁄ 42 % â•⁄â•⁄ 2 % â•⁄ 54 % â•⁄â•⁄ 2 % 100 %

suus occ. 125 â•⁄ 17 153 â•⁄â•⁄ 0 295

% â•⁄ 42 % â•⁄â•⁄ 6 % â•⁄ 52 % 0 100 %



Chapter 6.╇ Noun phrases 

These figures perfectly confirm those given by Lisón Huguet (2001:€124 f.). According to him, roughly 50% of possessive pronouns appear in anteposition and 50% in postposition, with variations between authors.47 When examining possessive pronouns, it is important to consider the conditions of their use (Lehmann 2005 and Fruyt 2007). In Latin, possessive pronouns are not expressed with several relational nouns that denote inalienable entities such as parts of the body, kinship, clothing, or personal belongings. They are also absent when the possessor is the subject of the sentence (35), which is a kind of co-reference.48 In all other cases, a possessive pronoun is used. Meus in (36) is an instance of contrast with Marcus Antonius. Emphasis on the pronoun is illustrated in (37) with a pre-nominal placement. (35) Nam quid ego de consulatu loquar? for why I about consulship should speak ‘For why should I speak of my consulship?’

(Cic. Pis. 3)

(36)

Non placet M. Antonio consulatus meus. neg pleases M. Antonius-dat consulship my-nom [At placuit P. Servilio.] ‘Marcus Antonius does not approve of my consulship. [But Publius Servilius approved of it.]’ (Cic. Phil. 2.12)

(37)

Quis autem meum consulatum praeter te ac P. Clodium who but my consulship-acc except you and P. Clodius-acc qui vituperaret inventus est? who would criticize has been found ‘But who has been found to criticize my consulship except you and Publius Clodius?’ (Cic. Phil. 2.11)

Historical narratives from the classical period offer numerous instances of the noun copiae ‘troops’. In accordance with the conditions of the use of possessive pronouns, the possessor should be specified when troops do not belong to the subject of the sentence. However, suus is often expressed in Caesar, even if it is not necessary, and stands both in post-nominal (38) and pre-nominal (39) position. Marouzeau (1953:€15) suggests that in the first case the post-nominal use of suus is “normal”, whereas suus in (39) is pre-nominal because it contrasts with hostibus. However, I would not interpret them in the same way. In both cases, the reflexive could have remained unexpressed, because the troops belong to Caesar and it is evident that he led off his own forces and not somebody else’s. Caesar’s use of suus, both pre-nominal and post-nominal, is a

47. In my corpus (figures not included in Table 6), Caesar has more pre-nominal possessives (57%) than Sallust (23–25%); the proportion is roughly the same in Cicero. 48. There are other constructions for expressing possession, especially dative ones (Bolkestein 1983), for example: sese omnes flentes Caesari ad pedes proiecerunt (Caes. Gal.€1.31.29) ‘they all threw themselves in tears at Caesar’s feet’.

 Constituent order in Classical Latin prose

kind of emphasis: he never omits suus when speaking about himself, but he omits it, for example, for his enemy Afranius in (40). (38)

[... quattuor reliquas legiones in castra maiora reduxit.] Proximo die instituto suo Caesar ex castris utrisque next day as usual Caesar-nom from camp both copias suas eduxit. forces his-acc led out ‘[... the four remaining legions he led back the main camp.] Next day Caesar led his forces out of both camps as usual.’ (Caes. Gal. 1.50.1)

(39)

[dicit montem, quem a Labieno occupari voluerit, ab hostibus teneri...] Caesar suas copias in proximum collem subducit, Caesar-nom his forces-acc to nearby hill leads off aciem instruit. battle-line-acc arranges ‘[he said that the summit which Caesar had expected Labienus to be occupying was in enemy hands...] Caesar led off his forces to a nearby hill and arranged them for battle.’ (Caes. Gal. 1.22.3)

(40) Potestate facta Afranius copias educit. opportunity being made Afranius-nom forces-acc leads out ‘On being offered this opportunity, Afranius leads his forces out.’( Caes. Civ. 1.41.2) To sum up, the question of the placement of possessive pronouns is complicated by their nature; their use depends on various factors such as co-reference and the alienability or inalienability of entities. The consequence is that possessive pronouns are often found with emphasis and in contrastive contexts. However, contrastive or emphatic pronouns may be placed either pre- or post-nominally, as is shown in (41), where meo is in implicit contrast with ‘others’, and in (42), where mea contrasts with Iugurthae. In (43), the noun phrase with a pronoun in post-nominal position (meae) represents contextually given information and could therefore have been omitted. (41)

Invident honori meo: ergo invideant labori, they begrudge honour my-dat so let begrudge toil-dat innocentiae, periculis etiam meis. blamelessness-dat dangers-dat even my-dat ‘They begrudge my honorific office: so let them begrudge my toil, my blamelessness, even my dangers.’ (Sal. Jug. 85.18)

(42) Non mea culpa saepe ad vos oratum mitto, patres neg my fault-abl often to you to beg I send conscript



Chapter 6.╇ Noun phrases 

conscripti, sed vis Iugurthae subigit... fathers-voc but violence-nom of Jugurtha compels ‘It is not my fault that I often send advocates to speak to you, conscript fathers, but the violence of Jugurtha compels it.’ (Sal. Jug. 24.2) (43)

Toπoθεσíαν quam postulas Miseni et Puteolorum description-acc that you ask of Misenum and of Puteoli includam orationi meae. I will put speech my-dat ‘I shall put in my speech the topographical description of Misenum and Puteoli, which you ask for.’ (Cic. Att. 1.13.5)

There are several other remarks to be made concerning the placement of possessives. Firstly, when one expression of possession is coordinated with another, the noun phrase usually has the coordinated pair placed before the noun, as in (44): (44) Ita cuncti suae hostiumque vitae iuxta pepercerant. so all-nom their-dat of enemies=and life-dat equally had spared ‘So, they had no more spared their own lives than those of their enemy.’  (Sal. Cat. 61.5) Secondly, possessives referring to relatives and familiars, such as frater meus ‘my brother’ or amicus meus ‘my friend’, are regularly placed after the head noun, provided that the conditions for expressing the possessive are fulfilled.49 (45) Amalthea mea te exspectat et indiget tui. Amalthea my-nom you-acc awaits and needs you ‘My Amalthea awaits you and needs you.’

(Cic. Att. 2.1.11)

Thirdly, possessive expressions in the genitive eius ‘his, her’ and eorum ‘their’, referring to a possessor of the third person that is not the subject of the sentence (in which case suus would be used), belong to the paradigm of possessive pronouns, and they occupy, in principle, the same position. For this reason, I suggest that it is reasonable to consider them together with possessive pronouns rather than with genitive complements.50 From this point of view, Example (47) is not an instance of discontinuity in the phrase multis amicis; compare it with (48). However, unlike possessives such as meus, eius and eorum cannot be contrasted with other potential possessors; in contrastive contexts, huius or illius are used.

49. Marouzeau (1953:€ 16) only talks about maiores nostri ‘our ancestors’, but from Cicero’s correspondence and from Sallust’s works it is clear that all relatives (pater ‘father’, filius ‘son’, etc.) are concerned. 50. For eius and eorum, see also p 272.

 Constituent order in Classical Latin prose

(46)

Tribulis enim tuus est et Sextus, pater of tribe-nom indeed your-nom he is and Sextus-nom father-nom eius, nummos vobis dividere solebat. his money-acc you-dat distribute used to ‘For he belongs to your tribe, and his father Sextus used to disburse your gratuities.’ (Cic. Att. 1.18.4)

(47)

Litterae complures a multis eius amicis ac necessariis letters numerous-nom from many his friends and acquaintances adferuntur. are brought ‘A number of letters reached him from a number of his friends and acquaintances.’  (Cic. Ver. 2.97)

(48)

Paucae tamen bestiarum... inter velitum ordines a few-nom yet of beasts among of light-armed ranks cum multis suis volneribus ingentem stragem edebant. with many their wounds great laughter-acc caused ‘A few of the beasts, however... caused great losses among the ranks of the light-armed, though suffering many wounds themselves.’ (Liv. 30.33.14)

7. Juxtaposed determiners 7.1

General observations

Combinations of modifiers have not yet been examined in an adequate way. De Sutter (1986:€156), whose observations are based on Cato, suggests that multiple determiners appear in pre-nominal position in the following order: {demonstrative > quantifier > noun}. When examining the hierarchical structure of noun phrases, Pinkster (1995:€107 f.) proposes a more developed sequence: {demonstrative > indefinite > quantifier > identifier (alius, idem) > noun}. This order would be in accordance with observations made for modern languages (cf. Rijkhoff 2002:€247). Lisón Huguet (2001:€114 ff.) confines himself to statistical data and lists combinations of modifiers that occur in his corpus. Devine and Stephens (2006:€517) do not present a very helpful schema, mainly because of the not very clear distinction between strong and weak modifiers; Marouzeau (1922 and 1953) was not really interested in this problem. I will now examine whether modifiers are organised according to the above presented semantic order, and investigate what the variations from this order imply when they occur. Unlike juxtaposed adjectives that, if permuted, do not involve semantic nuances, variations in the order of determiners might have semantic consequences for



Chapter 6.╇ Noun phrases 

the noun phrase. We can assume that, in principle, the most remote modifier has other modifiers (and adjectives, if any) and the noun in its scope. For example:51 (49) [duo [prima genera]] ‘[two [[first sorts]]’ or [[genera prima] duo] However, we cannot conclude that the cardinal necessarily precedes the ordinal (pace Lisón Huguet 2001:€144 ff.), for somebody might want to say: (50) [prima [duo capita]] ‘the first two paragraphs’ Possible combinations of determiners forming a block (in pre-nominal or, rarely, postnominal position with mirror order), collected with the help of BTL, are indicated in Table 7. Determiners are divided into five groups and include: Numerical quantifiers: cardinals duo ‘two’, tres ‘three’, quattuor ‘four’, quinque ‘five’ and the ordinal primus ‘first’; Non-numerical quantifiers: multus ‘many’, omnis ‘all’, totus ‘whole’, nullus ‘no’; Indefinites: quidam ‘certain’, aliquis ‘certain’, and alius ‘other’; Possessives: meus ‘my’, suus ‘his’; Anaphor is and demonstratives hic, ille.

Table 7 is organised as follows: the determiners of the first row (across) are combined with determiners indicated in columns (down); for example, numerical quantifier plus numerical quantifier (duo prima), then numerical quantifier plus non-numerical quantifier (*duo multi), plus indefinite (duo aliqua), plus possessive (duo sui), plus anaphoric or demonstrative pronoun (duo hi). All combinations concern noun phrases, i. e. determiners are always followed by a noun. Attested combinations are indicated as +, nonattested combinations as -; the question mark? means that there are very few examples or that the combination was found in a pronominal phrase without noun (for example, ii multi). Plus signs between brackets mean that the combination does not concern the whole group but only a part of it; for example, it is found for alius but not for quidam. Table 7.╇ Combinations of modifiers in noun phrases: modifier + modifier + noun (BTL)

Num. quantifier Non-n. quantifier Indefinite Possessive Demonstrative

Numerical quantifier

Nonnumerical quantifier

Indefinite

+ – (+) (+) +

– – (+) + (+)

+ (+) + ? –

51. Both instances are discussed below, see Section 7.2, p. 257.

Possessive Demonstrative

+ + ? – +

(+) (+) – + –

 Constituent order in Classical Latin prose

The categories examined are not homogeneous; consequently, it is worth mentioning some further points that do not appear in the table. Cardinal (duo) and ordinal (primus) numerical quantifiers may be combined together, with possessives, and with alius and quidam, but not with aliquis. They do not occur with non-numerical quantifiers multus and nullus, but combinations with omnis (tres omnes) are possible. As for demonstratives, only combinations with hic (duo haec) are found. The non-numerical quantifiers multus and nullus do not appear with numerical quantifiers, with the exception of omnis (omnes tres). They are not mutually combined. Multus, omnis and nullus are found with alius but not with other indefinites. They combine with possessives. Omnis may be combined with the anaphor is and the demonstratives; multus can only combine with hic (multis his); nullus does not co-occur with demonstrative pronouns. Among indefinite modifiers, only alius can be combined with a numeral (alia duo) and non-numerals (alia multa and alia omnia). Indefinite modifiers can be mutually combined, which signals that their status is different: alia aliqua, alia quaedam and vice versa aliqua alia and quidam alii. This can be explained by the fact that alius is an identifier. Surprisingly, I noted practically no combinations of alius, quidam and aliquis with possessives. Indefinites do not combine with demonstratives. Except for meus primus, there are practically no instances of possessives followed by numerals, which would be expected. Possessives are combined with non-numerical quantifiers but not with nullus. There are isolated combinations of possessives with indefinites. The demonstrative pronouns hic and ille occur in combination with possessives. The anaphor is and the demonstratives hic and ille appear with numerical quantifiers. They can combine with omnis, but not with multus and nullus. There are also combinations with possessives. They do not combine with indefinites, and they do not combine with each other. The question that should be answered now is to what extent the order of determiners is significant. Examination of the data shows that with several combinations the order is free; in some other cases, semantics are involved, and in the remaining ones, pragmatic factors are decisive. Table 8 summarizes my findings, which I will discuss in detail below: semantic differences (SEM.) involved in the ordering of numerical and non-numerical quantifiers and of indefinite modifiers in Section 7.2, the placement of demonstratives reflecting different pragmatic values (PRAG.) in Section 7.3, and placement of possessives as a case of free variation (VAR.) without semantics involved in Section 7.4. In these sections, comments concern blocks of determiners in prenominal position, which is their most frequent position; when determiners are placed post-nominally, the mirror order should be understood.



Chapter 6.╇ Noun phrases 

Table 8.╇ Combinations of modifiers in noun phrases: modifier + modifier + noun (BTL) Numerical Non-numerical Indefinite quantifier quantifier Num. quantifier Non-n. quantifier Indefinite Possessive Demonstrative

SEM. – (SEM.) (VAR.) PRAG.

– – (SEM.) VAR. (PRAG.)

SEM. (SEM.) SEM. (VAR.) –

Possessive Demonstrative VAR. VAR. (VAR.) – PRAG.

(PRAG.) (PRAG.) – PRAG. –

( ) = not all determiners of the group are concerned

7.2

Quantifiers and indefinite determiners

As stated above, the order of the combined quantifiers and indefinites implies a semantic difference. Compare the examples in (51) and (52), with different orderings of the cardinal duo and the ordinal prima. (51) [Crassus ostendit] duo prima genera quaestionum esse in quibus two first sorts-acc of questions be in which eloquentia versaretur, unum infinitum, eloquence-nom is concerned one-acc abstract-acc alterum certum. other-acc concrete-acc ‘[Crassus showed] that there are two main divisions of the questions wherewith eloquence is concerned, the one abstract, the other concrete.’  (Cic. De orat. 2.41) (52)

[...faciendum mihi putavi ut tuis litteris brevi responderem.] Sed prima duo capita epistulae tuae tacita but first two paragraphs-nom of letter your in silence-nom mihi quodam modo relinquenda sunt. me-dat in certain way to be passed over are ‘[...I felt bound to make a brief reply to your letter.] But the first two paragraphs of your letter I am obliged in a way to pass over in silence.’ (Cic. Fam. 3.8.2)

In (51), Cicero is talking about eloquence; one can practise it by using two main sorts of questions (duo prima genera), which Cicero intends to focus on; he passes over other potential ways of subdividing the topic. Example (52) is about a letter from Atticus that has several paragraphs. Cicero will successively discuss them all with the exception of the first two. By prima duo capita ‘first two paragraphs’ Cicero means the two paragraphs at the very beginning of the letter.

 Constituent order in Classical Latin prose

The different ordering of a numeral (duo) and an indefinite (alia)52 is also semantically relevant, as in (53) and (54): (53)

[Cicadae... quarum duo genera: minores... volatura...] Quidam duo alia genera faciunt earum: some-nom two other kinds-acc distinguish of them surculariam...; frumentariam. surcularia-acc frumentaria-acc ‘[Cicadas are divided into two kinds: the smaller kind... the flying kind...] Some authors divide these animals into two other kinds, the surcularia... and the frumentaria.’ (Plin. Nat. 11.92–94)

(54)

[Est inter candidas et quae ceraunia vocatur...] Sotacus et alia duo genera fecit cerauniae, Sotacus-nom also other two kinds-acc mentions ceraunia-gen nigrae rubentisque. black-gen red-gen=and ‘[Among the white stones, there is also one known as ceraunia...] Sotacus mentions also another two varieties of ceraunia, one black and the other red.’  (Plin. Nat. 37.134–135)

Pliny the Elder is presenting a classification of cicadas divided into two kinds. Then he mentions that other authors also distinguish two, but these are other, different, kinds (duo alia genera). Alia duo genera ‘another two varieties’ in (54) occurs in the following context: Pliny is describing a stone named ceraunia that is white. Another author, Sotacus, recognises this white ceraunia and adds another two kinds of ceraunia (black and red) that Pliny does not describe himself. The orderings multus alius and alius multus also imply a semantic difference: ‘many other testimonies’ in (55) means that there were several pieces of evidence, and one of them is mentioned; ‘other numerous victims’ in (56) means that there were numerous similar victims. (55)

Inter multa alia testimonia ad causam pertinentia haec among many other evidences to case relevant this-nom quoque vox... valuit cur Hostilia damnaretur. also word-nom counted why Hostilia-nom should be condemned ‘There were many other pieces of evidence relevant to the case but it was this comment... that secured Hostilia’s conviction.’ (Liv. 40.37.7)

52. However, a combination of duo and the indefinite quidam ‘certain’ only appears in the order duo quidam, for example: duo quidam Arcades familiares (Cic. Div. 1.57) ‘two friends from Arcadia’.



Chapter 6.╇ Noun phrases 

(56)

Censorem, aedilem curulem, tres tribunos plebis censor-acc curule aedile-acc three tribunes-acc plebeian mortuos ferunt, pro portione et ex multitudine alia died-acc they state proportionately and from large number other multa funera fuisse. many deaths-acc were ‘It is stated that a censor, a curule aedile, and three plebeian tribunes died, with a correspondingly large number from the rest of the population.’  (Liv. 7.1.8)

Unlike other non-numerical quantifiers, the universal quantifier omnis may be combined with numerals. It usually precedes the numeral, for example omnes tres legiones (Bell. Alex. 35.2) ‘all three legions’. The negative quantifier nullus does not allow variations in placement: it figures before the modifier that is in its scope. This also holds true for ullus, a variant of nullus in a negative sentence. Variable orders concerning quantifiers and indefinites are summarized in Table 9.53

7.3

The anaphor is and the demonstratives hic and ille

Table 7 has shown that the anaphor is and the demonstratives hic and ille can be combined with numerals, with omnis, and with possessives; they do not appear with indefinites. They can stand remote from the noun as well as close to it. However, variable placement of is, hic and ille involves not a semantic difference, but a pragmatic one. Table 9.╇ Combinations of determiners duo prima genera ╇ ‘two first (= main) sorts (of)’ duo alia genera ╇ ‘two other classes’ multa alia testimonia ╇ ‘much other testimony’ alia quaedam crimina ╇ ‘certain other crimes’ alia aliqua arte ╇ ‘to some other art’ nulla alia suspicio ╇ ‘no other suspicion’

prima duo capita ╇ ‘first two paragraphs’ alia duo genera ╇ ‘another two classes’ alia multa funera ╇ ‘other numerous victims’ quidam alii dei peregrini ╇ ‘certain other alien gods’ aliqua alia causa ╇ ‘some other cause’ –

53. Expressions in first three rows were quoted above as Examples (51) – (56). Here are remaining references: alia quaedam crimina (Cic. Inv. 2.58), quidam alii dei (Cic. Leg. 2.37), alia aliqua arte (Cic. De orat. 2.36), aliqua alia causa (Sen. Nat. 3.15.5), nulla alia suspicio (Cic. Ver. 47). I have found no example for the sequence alius nullus, apart from an occurrence in a very late author.

 Constituent order in Classical Latin prose

The position of the pronoun depends on its function. The noun phrase eas omnes copias – copiae are deducible from the preceding sentence – in (57) has an anaphoric determiner54 whereas eam in omnem eam materiam (58) is cataphoric: (57)

[Omnes Galliae civitates... contra se castra habuisse;] eas omnes copias a se uno proelio pulsas ac these all forces-acc by refl in one battle routed-acc and superatas esse. overcome-acc be ‘[that all the states of the Gaul... had set up their camp against him;] that all their forces had been routed and overcome by him in a single battle.’  (Caes. Gal. 1.44.3) (58) omnem eam materiam, quae erat caesa, conversam ad hostem conlocabat all that timber-acc which was cut turned-acc to enemy he placed ‘he placed all that timber which was cut down on the side facing the enemy’  (Caes. Gal. 3.29.1) A similar strategy of placement may be observed in combinations with numerals: the ordering hi duo ‘these two’ is chosen when the modifier is anaphoric or resumptive (59) and duo hi when it is cataphoric (for example tres eas partes quas ‘the three parts, which...’ in Cic. Inv. 1.8). Example (60) is an instance of emphasis on his. (59)

[duasque ibi legiones conscribit et tres... ex hibernis educit...] et cum his quinque legionibus ire contendit. and with these five legions go he hastens ‘[and there he enlists two legions, and leads out from winter-quarters three...] and with these five legions he marches rapidly...’ (Caes. Gal. 1.10.3)

(60) Si locare voluisses, duobus his muneribus liberasses. if hire out you had wanted two these duties-abl you would have cleared ‘If you had cared to hire them out you would have cleared your outlays in these last two shows.’ (Cic. Att. 4.4a.2)

7.4

Possessive modifiers

Possessives usually appear close to the noun (Lisón Huguet 2001:€145); for example: hic meus reditus ‘this return of mine’ (Cic. Dom. 100). However, possessives – which are often emphatic – can be given prominence as in (61), and stand at the start or the end of the noun phrase. It is worth adding that emphatic or contrastive possessive pronouns occur with hic and ille, but not with is.

54. It is also possible to find an emphatic omnis placed before a resumptive modifier, for example: omnes eas civitates (Caes. Gal.€2.34.1) ‘all these nations’.



Chapter 6.╇ Noun phrases 

(61)

Meus ille laudatus consulatus mihi te, liberos, my that lauded consulship-nom me-dat you children-acc patriam, fortunas; tibi velim ne quid country-acc possessions-acc you-dat I would like neg anything eripuerit praeter unum me. has robbed except only me ‘That much-lauded consulship of mine has robbed me of you and my children and my country, and my possessions; I only hope that it has robbed you of nothing but myself.’ (Cic. Q. fr. 1.3.1)

Unlike other modifiers, possessives may also be placed post-nominally, as illustrated in (62): (62) Omnibus hic labor meus propositus est quicumque... all-dat this toil my-nom offered is whoever ‘These my endeavours are placed at the disposal of all who...’ (Cic. Clu. 149) Both orders are also found for combinations with omnis (omnis meus... and meus omnis... ‘all my...’). When omnis is placed after the possessive as in (64), it usually bears emphasis. Variations such as multa mea (‘many my’) and mea multa55 are possible as well, without semantic or pragmatic differences. In very complex noun phrases, as in (65), multus can be emphatic and placed at the start of the noun phrase. However, nullus always precedes the possessive (nullus meus ‘no my’). (63) coniectus in fraudem omnia mea praesidia neglexi thrown in fraud all my defences-acc I neglected ‘trapped as I was, I neglected all my defences’ (Cic. Att. 3.15.7) (64) Id est maximum et miserrimum mearum omnivm miseriarum. that is greatest and saddest my-gen all-gen miseries-gen ‘That is the greatest and saddest of all my sorrows.’ (Cic. Att. 3.7.3) (65) mvltis meis et firmis praesidiis obsessus many my and trusty guards-abl surrounded ‘surrounded by my many and trusty guards’

(Cic. Cat. 1.6)

Numerals in combination with possessives are not very well attested. They appear prenominally in the order {numeral > possessive}, as in (66), and also in the other order, as in (67) with a post-nominal block. There is also an arrangement with a pre-nominal numeral and a post-nominal possessive. In all these cases, no semantic nuances are involved. (66) ut... tribus suis patrimoniis deleniret that three his patrimonies-abl he would mollify ‘to use his three patrimonies as a means of mollifying them’

(Cic. Mil. 95)

55. See, for example: multis suis vulneribus (Liv. 30.33.14) ‘suffering many wounds’ and tua multa... officia (Cic. Fam. 6.5.4) ‘your many kindnesses’.

 Constituent order in Classical Latin prose

(67) Legi orationes duas tuas, quarum altera... I read speeches two your-acc of which one-nom ‘I have read your two speeches; one of them...

(Cic. ad Brut. 2.3.4)

In the infrequent combinations with indefinites the possessive may be placed either before the noun or before the indefinite. There seems to be no semantic difference in these orders, and both allow an emphatic interpretation, which can be ascribed to the value of the possessive itself. Compare (68) and (69): (68)

[De mea autem benevolentia erga te, etsi potes ex eodem Phania cognoscere,] tamen videntur etiam aliquae meae partes. yet seem also some my contributions-nom ‘[Now as regards my goodwill towards you, though you can learn it from Phania,] nonetheless, there seems to be some contribution of mine.’ (Cic. Fam. 3.1.1)

(69)

quod laudem adulescentis propinqui existimo etiam because praise-acc of young kinsman I think also ad meum aliquem fructum redundare to my some interest react ‘because I think that the praises of a young kinsman react in some degree to my personal interest.’ (Cic. Lig. 8)

As a recapitulation, I give the variations in the order of anaphoric, demonstrative, and possessive pronouns in Table 10.56 Table 10.╇ Combinations of determiners eas omnes copias ╇ ‘all their forces’

omnem eam materiam quae... ╇ ‘all that timber which...’

his quinque legionibus ╇ ‘with these five legions’

tres eas partes quas... ╇ ‘the three parts which...’

hic meus reditus ╇ ‘this return of mine’

meum illum casum ╇ ‘that disgrace of mine’

omne suum studium ╇ ‘all his energy’

suas omnes opes ╇ ‘all his fortunes’

tribus suis patrimoniis ╇ ‘his three patrimonies’

orationes duas tuas ╇ ‘your two speeches’

aliquae meae partes ╇ ‘some contributions of mine’

meum aliquem fructum ╇ ‘somewhat to my personal interest’

56. Several expressions were quoted above as Examples (57) – (60) and (66) – (69). Here are the remaining references: meum illum casum (Cic. Sest. 53), omne suum studium (Cic. Q. fr. 1.1.29), suas omnes opes (Cic. Quinct. 98).



Chapter 6.╇ Noun phrases 

7.5

Multiple determiners

It is possible to combine more than two modifiers,57 but such combinations are quite rare. Devine and Stephens (2006:€517) give an example of this phenomenon (hae is a textual variant of haec; the latter is accepted by most editors): (70) [omnes [hae [tres partes]]] ‘all these three parts (of exoneration)’

(Rhet. Her. 2.16.24)

However, the attested combinations of multiple determiners usually contain the universal quantifier omnis and a possessive, both of which may involve emphasis. It is unnecessary that an emphatic omnis should occupy a remote position; it can be adjacent to the noun as in (71). I would also interpret omnem in (72) as emphatic. In any case, permutation of these modifiers does not involve semantic nuances. (71) (domus) Ac mea quidem his tribus omnibvs and my-nom part these three all-abl iudiciis... liberata est. decisions-abl was liberated ‘My house... has been exonerated by all these three decisions.’ (Cic. Har. 30) (72)

7.6

Ad honestatem enim illum omnem conatum suum to honour indeed that whole enterprise his-acc referebat, non ad voluptatem. he referred neg to pleasure ‘For honour and not pleasure was the aim of that great enterprise.’(Cic. Fin. 2.56)

Multiple determiners and adjectives

One can more easily find complex noun phrases containing two determiners and (an) adjective(s). In general, in such phrases, determiners are remote from the head noun, whereas adjectives stand next to it. This reflects the general principle of scope, described by Rijkhoff (2002:€346): determiners such as demonstratives have the widest scope and appear first (or last) in the sequence; adjectives only have nouns in their scope and are adjacent to them; quantifiers never appear between an adjective and a noun. This principle also applies to Latin. Furthermore, in complex Latin noun phrases, the adjective or the possessive may go in the post-nominal position, as illustrated in (73). (73) ut... Sabazius et quidam alii dei peregrini iudicati that Sabazius-nom and certain other gods alien-nom judged

57. On combinations of determiners, cf. also Fugier and Corbin (1977).

 Constituent order in Classical Latin prose

e civitate eiciantur. from State are banished ‘by representing Sabazius and certain other alien gods as brought to trial and banished from the State’ (Cic. Leg. 2.37) Such complex noun phrases frequently involve contrast or emphasis, as in (74) and (75). The first example displays a contrast between nostra and Messi, and presents prenominal determiners and a post-nominal adjective. The second one illustrates prenominal placement of the whole bloc; the adjective familiarem contrasts with alienas (manus). Consider also the emphasis on the adjectives ambitiosae and fucosae that are entwined with the modifiers illae and nostrae in (76). (74) Illa nostra lex consularis nunc modesta videtur, that our law consular-nom now modest-nom seems haec Messi non ferenda. this-nom of Messius neg tolerable ‘Our consular law now looks quite modest; Messius’s is felt to be intolerable.’  (Cic. Att. 4.1.7) (75)

Nolebam illum nostrum familiarem sermonem in alienas I didn’t want that our familiar talk-acc in strangers’ manus devenire. hands get into ‘I did not want our familiar talk to get into strangers’ hands.’ (Cic. Att. 1.9.1)

(76)

Nam illae ambitiosae nostrae fucosaeque amicitiae for these ambitious our pretentious friendships-nom sunt in quodam splendore forensi, fructum are in certain splendour public benefit-acc domesticum non habent. domestic-acc neg they have ‘For my friendships with great and artificial attractions make a fine show in public, but at home they are barren things.’ (Cic. Att. 1.18.1)

Special arrangements mainly concern noun phrases functioning as Focus. For example the emphatic adjective ponderosam is placed at the left end of the noun phrase in (77). In (78), there is a permuted – with respect to the semantic principle mentioned above – order of modifiers (instead of illum nostrum splendorem forensem). (77)

tamen isti puero... da ponderosam aliquam epistulam yet this boy-dat give massive-acc some letter-acc plenam omnium non modo actorum sed etiam full-acc of all not only news-gen but also



Chapter 6.╇ Noun phrases 



opinionum tuarum opinions-gen yours ‘yet do give this boy... a massive letter, full of news and also of your own comment’  (Cic. Att. 2.11.1)

(78)

Splendorem nostrum illum forensem et in senatu splendour-acc our that public-acc and in Senate auctoritatem et apud viros bonos gratiam, magis authority-acc and by men good esteem-acc more quam optaramus consecuti sumus. than we had hoped we attained ‘I have attained, in larger measure than I had dreamed possible, my brilliant standing at the bar, my authority in the Senate, and my influence among the honest men.’ (Cic. Att. 4.1.3)

It is difficult to determine what combinations of modifiers are avoided in Latin, especially because the Latin texts that are available to us are relatively poor in combinations of numerals, demonstratives, and possessives.58 Rijkhoff (2002:€348), in his comparative study concerning a number of modern languages, signals two orders which are avoided: {demonstrative > adjective > noun > numeral} and {adjective > noun > numeral > demonstrative}. I have not found these orders in Latin (*hi fortes viri duo and *fortes viri duo hi), and I would not expect them either. Additionally, I have not found orderings such as *meum librum illum, with a pre-nominal placement of the possessive and a post-nominal placement of the demonstrative.

8. Genitive complements The placement of genitive complements in Latin is a very intriguing matter, especially because statistics show no clear preference for their placement: genitives stand either before or after their head noun, each in approximately 50% of cases.59 In several studies, a number of typological considerations have been discussed (especially by Adams 1976a: 73–83),60 but without clear conclusions. It is noteworthy that in Latin fixed

58. The expression and non-expression of demonstratives and possessives certainly plays a role; see Sections 3 and 6 above, p. 239 ff and 250 ff. 59. See Lisón Huguet (2001:€175) and Polo (2004:€235 ff.) for Petronius. 60. For him, the order {genitive > noun} is characteristic of the archaic period, whereas the order {noun > genitive} is commoner in the Classical and later periods.

 Constituent order in Classical Latin prose

expressions, for example military or juridical (Adams 1976a: 75, Lisón Huguet 2001:€164 ff.), both a pre- and a post-nominal placement of genitive complements is found: Pre-nominal genitive

Post-nominal genitive

senatus consultum ╇ ‘decree of the Senate’ iuris consultus ‘lawyer’ plebis scitum ╇ ‘decree of the commons’

magister equitum ╇ ‘Master of the Horse’ pater familias ‘father of family’ tribunus plebis ‘plebeian tribune’

The category of post-nominal genitives in the right column is interpreted by Adams (1976a: 75)61 as being parallel to expressions containing “determinative” adjectives such as patres conscripti ‘the conscript fathers’ or aediles curules ‘curule aediles’ (see p. 229). The expressions indicated in the left column, to which we can add aquae ductus ‘aqueduct’ and terrae motus ‘earthquake’ (Marouzeau 1953:€31), are considered to be similar to compound nouns including a pre-nominal adjective. It is also worth adding the patronymic expression, genitive + filius, meaning ‘son of…’ (for example, Marcus, Luci filius ‘Marcus, son of Lucius’)62 that allows no variation in the placement of the genitive. Despite numerous studies on genitive complements, not only those following the traditional distinctions between possessive, objective, subjective and other genitives (Polo 2004:€238 ff.), but also those adopting another approach that distinguishes between genitives with a collective reference, patronymic expressions, partitive and pronominal constructions (Lisón Huguet 2001:€175 ff.),63 the question of their “normal” order remains without a clear answer. I will not try to answer the question of whether Latin genitive complements are normally pre-nominal or post-nominal;64 instead, I will concentrate on pragmatic aspects of noun phrases expanded by a genitive and various arrangements of such phrases. Before doing this, several points are worth establishing. Firstly, genitives are nouns or noun phrases, and therefore they exhibit more autonomy than adjectives or determiners; consequently, they also have greater

61. Cf. Marouzeau (1922:€124 and 1953:€28); he tried to establish a link between “determinative”, i. e. post-nominal adjectives and genitive complements. 62. See Adams (1976a: 75), and Lisón Huguet (2001:€178). 63. In addition, Devine and Stephens (2006:€314 ff.) present a very detailed study of genitives depending on various governing nouns, where we can observe a great variety of constructions. 64. The question is whether we should consider that there is one, “normal” placement of genitive complements at all. According to Dryer (2005:€350), there are 7% of languages with both {genitive > noun} and {noun > genitive} orders with neither order dominant. Latin may belong to this category.



Chapter 6.╇ Noun phrases 

mobility. In (1), a genitive complement governs a relative clause; obviously, with adjectival modifiers or determiners, this is impossible.65 (1) Conclamant legionis XIII, quae aderat, milites... shout out legion 13th-gen which was present soldiers-nom ‘The soldiers of the Thirteenth legion, which was there..., shouted that...’  (Caes. Civ. 1.7.8) Secondly, genitives vary in what they refer to: they can be proper names, concrete or abstract, singular or plural entities. The question is whether we should expect the same behaviour for all this variety of constructions. Thirdly, it would be incorrect to limit the question of the placement of genitives to sequences {noun > genitive} or {genitive > noun}. Indeed, a head noun can be expanded by more than one modifier and it can govern not a single genitive but a genitive phrase. As with other modifiers discussed in this chapter, it also holds true for genitive complements that they can either fulfil a pragmatic function (Topic, Focus), or carry pragmatic features (emphasis, contrast), or not. They may – or may not – have a distinctive value. A noun phrase containing a genitive complement with a distinctive value answers the question ‘which one’? (cf. Example 2, quoted below), whereas noun phrases containing a genitive complement without such a distinctive value function as a unit and answer the question ‘who, what?’ (cf. Example 3). The examination of genitive placement will start with noun phrases formed by a single noun and a single genitive complement; complex instances will follow later.

8.1

Simple instances

From a pragmatic point of view, several tendencies for the placement of genitive complements can be established. Genitives with a distinctive function usually stand after the head noun as in (2); the distinctive function often goes together with the Focus function, as is the case for castra Sexti ‘Sextus’s camp’. Genitives referring to already given information stand in pre-nominal position. Thus in (3), reference is made to a well established Topic (Scipio). (2) Restat ut in castra Sexti aut, si forte, Bruti nos remains that to camp Sextus-gen or maybe Brutus-gen refl conferamus. we betake ‘The alternative is to betake ourselves to Sextus’s camp, or maybe to Brutus’s.’  (Cic. Att. 14.13.2)

65. Prepositional phrases that modify a noun also have a certain autonomy. However, I will not pay special attention to them in this study.

 Constituent order in Classical Latin prose

(3) Non mediocri cura Scipionis animum pepulit. neg slight anxiety-abl Scipio-gen mind-acc he aroused ‘He aroused no slight anxiety in the mind of Scipio.’

(Liv. 30.14.1)

A single noun that governs a genitive phrase is usually not incorporated into it. The genitive phrase is placed post-nominally when the information it provides is new, or pre-nominally when the genitive expresses contextually bound information. Genitive phrases containing an anaphoric element that links them with the preceding context belong to the second category, as in (4). In (5), decimae legionis appearing before its head noun (cohortatione) is contextually bound, because it was mentioned before. The determiner decimae stands before its head noun legionis (see above, p. 247 f.) for the same pragmatic reason: (4)

Sed huiusce rei totius vitium, quod... praesenti tibi but this thing whole-gen fault-acc that present-dat you-dat commodius exponam. more conveniently I will explain ‘But it will be better to explain the problem of the whole thing, that..., when I talk to you.’ (Cic. Att. 1.17.3)

(5)

[Caesar... ad legionem decimam devenit, milites... est cohortatus...] Caesar ab decimae legionis cohortatione ad dextrum cornu profectus... Caesar from 10th legion-gen encouragement to right flank went ‘[Caesar... came to the Tenth legion, and encouraged the soldiers...] After his words of encouragement to the Tenth, Caesar made his way to the right flank.’  (Caes. Gal. 2.25.1) When a genitive complement is pre-nominal in a prepositional phrase, it is incorporated between the preposition and the head noun, which frame the whole unit, as in (6). (6) In istius lenonis turpissimi domo... Capitoli ornamenta in this profligate foul-gen home Capitol-gen adornments-nom ponentur? will be put ‘In the home of that foul profligate, shall the adornments of the Capitol be set...?’  (Cic. Ver. 4.71)

8.2

Complex instances

In this section various cases of complex noun phrases will be examined; first of all, the sequence {noun + modifier + genitive}, which has been analyzed by Hoff (1995). According to him (p. 248), in Classical Latin prose, the genitive stands outside the sequence noun + adjective in 60% of cases; it is incorporated into the noun phrase in



Chapter 6.╇ Noun phrases 

40% (cf. Adams 1976a: 80). The most frequent orderings are (Hoff ibid.): {modifier > noun > genitive} (45%) and {modifier > genitive > noun} (28%). However, a closer examination shows that the order of these complements is not mechanical but fits in with the general placement tendencies, presented in this chapter, and is therefore not determined syntactically (pace Hoff 1995:€252). Partem ultimam pontis in (7) is a contextually unbound constituent; its complements in post-nominal position (ultimam, pontis) specify the noun partem. By contrast, genitive complements providing contextual links (the connecting relative quorum, the anaphor eorum, etc.) are placed initially; other contextually bound genitive complements behave in the same way, as for example the genitive Britanniae in (8). (7)

partem ultimam pontis, quae ripas Vbiorum part-acc farthest-acc bridge-gen which banks-acc Ubii-gen contingebat, in longitudinem pedum ducentorum rescindit touched in length of feet 200 he breaks up ‘He breaks up a 200-foot length of the bridge at its farthest end, which touched the bank in the territory of the Ubii’ (Caes. Gal. 6.29.2)

(8) (Britanni) Britanniae pars interior ab iis incolitur, quos... Britain-gen part inland-nom by these is inhabited who ‘(Britons) The inland regions of Britain are inhabited by people whom...’  (Caes. Gal. 5.12.1) I will add two more examples. The genitive Sequanorum in (9) has a distinctive value – several tribes were mentioned in the preceding context – and stands post-nominally, at the right-hand end of the phrase, whereas the genitive Haeduorum, without any distinctive value, is contextually dependent and goes in the pre-nominal position (10): (9)

Ariovistum... ad occupandum Vesontionem, quod est Ariovistus-acc to seize Vesontio-acc which is oppidum maximum Sequanorum, contendere town-nom main-nom Sequani-gen hasten ‘that Ariovistus was hastening... to seize on Vesontio, which is the main town of the Sequani’ (Caes. Gal. 1.38.1)

(10)

compluris annos portoria reliquaque omnia for several years river tolls-acc other-acc=and all-acc Haeduorum vectigalia parvo pretio redempta habere Aedui-gen taxes-acc low price-abl purchased-acc have

 Constituent order in Classical Latin prose

‘that for several years he had purchased, at a low price, the collection rights on river tolls and all the Aedui’s other taxes’ (Caes. Gal. 1.18.3) I come now to more complex noun phrases containing a head noun governing both an attributive adjective and a genitive noun with a modifier. Such instances were also examined by Hoff (2003). According to his data (2003:€ 208), the most frequent arrangements are the following: modifier > noun > genitive noun > genitive modifier (24%) modifier > noun > genitive modifier > genitive noun (18%) modifier > genitive modifier > genitive noun > noun (16%) genitive modifier > genitive noun > modifier > noun (15%). The genitive phrase can come after the head noun and its modifier; it can also precede it or be inserted inside it. Pragmatic values of concrete instances will show under which conditions these patterns are chosen. The ordering {head noun > genitive modifier} with the noun followed by the genitive phrase appears when the complex constituent has Focus function, as in (11).66 As we will see in the following section, it is also possible to insert the genitive constituent between the adjective and the head noun. (11)

Quod quidem si facis, magnum fructum studiorum this indeed if you do great-acc benefit-acc studies-gen optimorum capis, in quibus... excellent-gen you take in which ‘If this is what you are doing, you are reaping the rich reward of the excellent studies in which...’ (Cic. Fam. 6.10b.1)

Topic constituents or constituents expressing contextually given information show a pre-nominal placement of modifiers. A good example is given in (12): the noun phrase summarizes what has been said in the preceding context. (12)

Multae res ad hoc consilium Gallos hortabantur: superiorum many things-nom to this decision Gauls-acc encouraged previous dierum Sabini cunctatio, perfugae confirmatio... days-gen Sabinus-gen hesitancy-nom deserter-gen corroboration-nom ‘Many factors encouraged the Gauls to adopt this strategy: Sabinus’s previous hesitancy, the deserter’s corroboration...’ (Caes. Gal. 3.18.6)

66. Concerning this example and also (13) below, cf. Rijkhoff (2002:€261f.). He argues that in an SOV language, a sequence such as {adjective > noun > genitive (phrase)} should be impossible: “in no verb-final language does the adjective precede and the genitive follow the noun”. I do not believe that Latin is typologically a genuine SOV language; in any case, from examples quoted in (11) and (13), it is evident that such a rule does not apply in Latin.



Chapter 6.╇ Noun phrases 

The first word of the complex noun phrase may carry contrast or emphasis; for example, in (13), domesticis is in contrast with alienis; rei frumentariae is contextually bound. In (14) populi Romani bears emphasis and stands even before tanta ‘such’, which is usually emphatic and placed first. (13) [ut potius in suis quam in alienis finibus decertarent] et domesticis copiis rei frumentariae uterentur and domestic-abl resources-abl supply-gen corn-gen they would use ‘[that they could fight better in their own territory than in that belonging to others,] and could use their own supplies of corn’ (Caes. Gal. 2.10.4) (14) ut populi Romani tanta mercede nihil sapere disceres that people Roman-gen such price-abl not be reasonable you would learn ‘for this enormous price paid by Roman people, you might learn to be foolish’  (Cic. Phil. 2.43) Furthermore, there are noun phrases with more than one genitive complement. A well known example is in (15), quoted among others by Devine and Stephens (2006:€316),67 containing both a subjective and an objective genitive. From a pragmatic point of view, the genitive Helvetiorum is contextually dependent, and it comes first for this reason. (15)

[Helvetiorum...] pro veteribus Helvetiorum iniuriis populi Romani for old Helvetii-gen outrages people Roman-gen ‘(of the Helvetii...) for outrages long ago inflicted on the Roman people by the Helvetii’ (Caes. Gal. 1.30.1)

However, in cases such as (16), the ordering is due not to pragmatics but to syntax. The noun centurio governs the genitive phrase primi pili that governs tertiae legionis. The order of these genitive phrases is usually not permuted.68 (16)

Centurio primi pili tertiae legionis per centurion-nom front rank-gen third legion-gen through munitionem... locum hostibus introeundi dedit. fortification occasion-acc enemies-dat to enter gave ‘A front-rank centurion of the Third legion allowed the enemy to enter the fortification.’ (Sal. Jug. 38.6)

67. For other examples, see ibid. 68. Cf. the following complex phrase: ut coactus sit praecipuum illud opusculum suum Annalium Ennii Elenchorum... vendere (Suet. Gram. 8.3) ‘to be forced to sell that admirable little work of his “Criticisms of the Annals of Ennius”. Annalium Ennii precedes its head Elenchorum.

 Constituent order in Classical Latin prose

8.3

Hyperbaton within the domain of the noun phrase

One of the ways to order a noun phrase that contains both a modifier and a genitive (or genitive phrase) is the order: {modifier > genitive (phrase) > noun}. In other words, the genitive (phrase) is incorporated into or framed by the governing phrase. This can be called a case of “internal hyperbaton”: a noun phrase is split up by its own modifier. Generally speaking, as the complexity of a noun phrase increases, the number of its possible arrangements increases also. However, frame-structured noun phrases are optional; they represent only one of the solutions for organising a complex noun phrase; we have seen in the preceding section that genitive complements can also stand outside the governing noun phrase. In my view, instances such as splendorem nostrum illum forensem (Cic. Att. 4.1.3, quoted above, p. 265, as Example 78) ‘my brilliant standing at the bar’ do not represent cases of internal hyperbaton (pace Menk 1925:€15). It is more likely to be an instance of permutation of components within a noun phrase, used instead of illum nostrum splendorem forensem. As suggested above (see p. 253), I also do not consider as discontinuous noun phrases containing the non-contrastive possessive genitives eius and eorum, that is, a non-reflexive expression of possession for the third person singular and plural. Nonetheless, a noun phrase may incorporate possessive genitives formed from pronouns other than is or from nouns, as in (17), where a possessive genitive phrase maiorum suorum is coordinated with the reflexive possessive suis. This last example is also a good one to show a very complex Latin noun phrase containing adjectives, a genitive phrase and a prepositional phrase; the whole is framed by multis and meritis. (17)

Halaesini pro multis ac magnis suis maiorumque Halaesini-nom for many-abl and great their-abl ancestors-gen=and suorum in rem publicam nostram meritis atque their-gen towards republic our-acc services-abl and beneficiis suo iure... leges ab senatu nostro petiverunt. benefits-abl justly laws-acc from Senate our asked ‘The citizens of Halaesa, in recognition of many valuable services and benefits rendered to Rome by themselves and by their ancestors... justly asked our Senate to legislate for them.’ (Cic. Ver. 2.122)

According to Hoff (2003:€208), noun phrases incorporate their complex genitive complements in 28% of the cases in Classical Latin prose; genitive phrases which are not incorporated but placed after or before the governing noun phrase are more frequent.



Chapter 6.╇ Noun phrases 

The frame-structured noun phrases mainly show the following order: {adjective > genitive phrase > noun}. An instructive example is given in (18); the framing can be viewed as a signal that the noun phrase is treated as a single unit, parallel to istis faucibus and istis lateribus. (18)

Tu istis faucibus, istis lateribus, ista gladiatoria you that gullet-abl that chest-abl that gladiatorial-abl totius corporis firmitate tantum vini... exhauseras... whole body-gen strength-abl so much vine had drunk up ‘You with that gullet of yours, that chest, that strong gladiator’s body, you drunk up such a quantity of wine...’ (Cic. Phil. 2.63)

The ordering starting with the head {noun > genitive phrase > adjective}, is also possible but not very frequent (Gettert 1999:€180 and 196). In (19), the adjective perexiguam is placed with emphasis at the right end of the noun phrase. This adjective is not predicative; the clause answers the question ‘what did I ask for?’ Cum ego diem in Siciliam inquirendi perexigvam postulavissem, when I time-acc in Sicily for inquiring very short-acc I had asked [invenit iste, qui sibi in Achaiam biduo breviorem diem potularet.] ‘When I had demanded a very short time to prosecute my inquiries in Sicily, [he found a man to ask for two days less to make investigations in Achaia.]’  (Cic. Ver. 1.6)

(19)

It has already been observed that a noun phrase is rarely incorporated inside its genitive phrase.69 Despite Hoff ’s (2003:€208) detailed analysis of possible sequences and his statement that certain combinations are not attested, I would argue, together with Kühner and Stegmann (1914, II: 611) that theoretically all possibilities of placement for magnum fructum studiorum optimorum (quoted above as Example 11) ‘the rich reward of the excellent studies’ are allowed. One only needs a pragmatic context for using them. I do not believe that syntactic restrictions apply to possible combinations and I would not be surprised to find one of the orders labelled “avoided” by Hoff (2003:€ 215). The following example shows a genitive phrase (reliquorum... tyrannorum) incorporating its governing noun phrase. I see nothing ‘deviant’ in this arrangement because all elements are contextually dependent except one: reliquorum, which stands initially and is contrastive with respect to huius.70 (20) nec vero huius tyranni solum... sed reliqvorvm neg but this tyrant-gen only but other-gen 69. According to Bolkestein (2001:€247), they represent 1.3% of the instances she examined. Hoff (2003) argues that sequences starting with a genitive adjective are attested (the example in (20) is quoted by him), but sequences starting with the genitive noun are avoided. 70. Cf. also humanarum quaedam simulacra virtutum (Cic. Fin. 2.110) ‘some semblance of the human virtues’ with emphasis on the first word.

 Constituent order in Classical Latin prose

similes exitus tyrannorum similar-nom deaths-nom tyrants-gen ‘not only the death of this tyrant... but also the similar fate of all other despots’  (Cic. Off. 2.23) In Classical Latin prose, it is possible to find other arrangements such as in (21). Maximum, placed initially, bears emphasis. Although populus Romanus universus has already been mentioned in the preceding context, we can assume a certain emphasis also on universi, because it is standing at the right-hand end of the noun phrase. (21)

maximvm vero populi Romani iudicium universi strongest-nom but people Roman-gen judgment-nom whole-gen consessu gladiatorio declaratum est. at assembly gladiatorial was declared ‘But the strongest expression of the judgment of the whole Roman people was plainly given by an audience at gladiatorial games.’ (Cic. Sest. 124)

To sum up, from all examples quoted in this section we can conclude that this form of internal hyperbaton is essentially a means that permits a speaker or writer to delimit the boundaries of a noun phrase and to stress the fact that this phrase is to be taken as a unit. This internal hyperbaton does not seem to be per se a device used for pragmatic purposes. Nevertheless, the ends of such units, especially on the left, are apt to receive emphatic or contrastive elements.

9. Hyperbaton produced by alien elements Latin noun phrases allow separation by words or groups of words that do not belong to the noun phrase. This phenomenon is called hyperbaton or discontinuity of noun phrases. It goes against a general iconic principle called the principle of domain integrity,71 which states that “what belongs together semantically is expressed together in the actual linguistic expression”. Rijkhoff (2002:€250) characterises it in the following way: “constituents prefer to remain within the boundaries of their proper domain”, and “constituents of a domain prefer not to be interrupted by other constituents”. The aim of this section is to evaluate the frequency of hyperbaton in Classical Latin prose and to describe its pragmatic motivations.72

71. See Rijkhoff (2002:€250), Dik (1997, I: 402), and Bolkestein (1998c: 186 f.) for Latin. 72. Hyperbaton cannot be viewed as a solely stylistic device; see p. 23 ff.



Chapter 6.╇ Noun phrases 

9.1

Frequency of hyperbaton of noun phrases

A typology of hyperbata in Latin, as well as some general aspects, was presented in Chapter 1, Section 7, p. 24 ff;73 the preceding section was devoted to internal hyperbaton. This section will deal with optional hyperbata produced by one or more alien elements, that is, elements that do not belong to the noun phrase. These are adverbs, pronouns, verbs or their combinations. First of all, it is important to evaluate the frequency of these hyperbata. Three extracts from Caesar (Civ. 1.2–8), Cicero (Tusc. 1.1– 23) and Sallust (Jug. 1–13) will be considered here, each containing approximately 2,200 words. Only discontinuity in complex noun phrases is considered; prepositional phrases were disregarded. The figures obtained are indicated in Table 11. Table 11.╇ Hyperbaton of complex noun phrases Author

Caesar Cicero Sallust Total

Number of noun phrases in the texts (100%)

Internal hyperbaton Numbers

Percent.

119 â•⁄ 67 115 301

18 â•⁄ 7 â•⁄ 5 30

15% 10% â•⁄ 4% 10%

Hyperbaton caused by alien elements Numbers Percent. 15 16 â•⁄ 5 36

13% 24% â•⁄ 4% 12%

Table 11 shows that 10% of all noun phrases examined (301 occ., 100%) involve internal hyperbaton and 12% of all noun phrases involve hyperbaton by alien elements; the remaining 78% of noun phrases are non-discontinuous. My figures are even higher than the evaluation of 10% provided by Lehmann (1991:€ 224). Additionally, as this table shows clearly, and as was previously well-known (Pinkster 2005b), there is a great deal of variation between authors: whereas hyperbata, especially the type involving alien elements, are very frequent in Cicero, they are rare in Sallust.

9.2

Hyperbaton produced by alien elements

From a pragmatic point of view, hyperbaton is mainly described as a means of pragmatic highlighting – see especially works of de Jong (1986), Bolkestein (1998c and 2001), and Pinkster (2005b)74 devoted to discontinuity produced by alien element or elements. However, this form of hyperbaton produced by alien elements does not concern only Focus constituents and emphatic words such as magnus or tantus, as might appear to be the case at the first glance. 73. The obligatory hyperbaton produced by enclitics and postpositive particles will not be considered here any more. 74. Cf. also Panhuis (1982:€72–80) and Herman (1985).

 Constituent order in Classical Latin prose

I will first briefly summarize previous findings. From his examination of magnus, de Jong (1986) concludes that discontinuous noun phrases bear Focus; he shows that the order {adjective > other elements > noun} is the most frequent. For him, the first word, the adjective, is practically always emphatic, whereas the words intervening in the middle are not concerned with “focality” (22). (22)

Cuius adventu ab hostibus cognito / magnae his arrival by enemies being learnt large-nom ex finitimis civitatibus copiae convenerunt. from neighbouring states forces-nom gathered ‘When the enemy learned of his coming, a large force from the neighbouring states gathered.’ (Caes. Gal. 7.57.2)

According to Bolkestein (2001:€251), discontinuous phrases containing an anaphoric element always have the order {anaphoric pronoun > other elements > noun}, as in€(23). It is obvious that discontinuity here concerns a Sentence Topic: (23)

His ille rebus ita convaluit by these he things so gained in power [ut nunc in uno civi spes ad resistendum sit.] ‘By these steps he has become so strong [that hope of resistance depends on one man.]’ (Cic. Att. 7.3.4)

Hyperbaton can also proceed in the other direction, starting with the noun: {noun > other elements > adjective}; but this is found less frequently, as is stated by de Jong (1986:€331) and Bolkestein (2001:€254). In another article, Bolkestein (1998b: 9)€showed interesting cases of this type of hyperbaton with intervening parenthetical sentences; salient words such as humanus in (24) are found after a parenthesis (vere ut dicam): (24)

Est quidam Graecus qui cum isto vivit, homo, vere ut is certain Greek-nom who with this lived man-nom truly to dicam – sic enim cognovi – , humanus, sed... say thus indeed I found cultured-nom but ‘There is a certain Greek who lives with him, a man, to tell the truth (I speak as I have found him) of good manners, at least...’ (Cic. Pis. 68)

Bolkestein (2001) was also interested in constraints on Latin hyperbaton. She observed (p. 256) that discontinuity is mainly caused by obligatory constituents of a sentence, not by satellites. Furthermore, there is never any discontinuity between a preposition and the rest of the prepositional phrase.75 To this general survey, I will add several important points. The first one concerns the position of the modifier and the head noun. They may stand not only at the beginning or at the end of the sentence, as in (25), but also take another position. As Latin 75. Cf. the arrangements listed by Devine and Stephens (2006:€571 ff.).



Chapter 6.╇ Noun phrases 

sentences are sensitive to colon division, the first word of a discontinuous noun phrase is often found sentence-internally at the beginning of a colon – see Example (26) with tantus (/ marks the colon subdivision). This ordering is found in particular with emphatic words such as multus, tantus or nullus. (25) Multa mihi dant solacia. many-acc me-dat they give consolations-acc ‘They give me many consolations.’ (26)

(Cic. Att. 4.18.2)

His rebus gestis / omni Gallia pacata / tanta huius these things being done whole Gaul pacified such-nom of this belli ad barbaros opinio perlata est, uti... war to barbarians fame-nom spread was that ‘Once this action was over and the whole of Gaul was pacified, so impressive a fame of this war spread among the barbarians, that...’ (Caes. Gal. 2.35.1)

From a pragmatic point of view, the discontinuous constituents in both the examples quoted above function as Focus. This is also the case in (27), where the quantifier duas in pre-verbal position is salient. This example illustrates the less frequent ordering {noun > other elements > modifier}; this can be explained by the fact that the modifier is developed by a relative clause (quarum alteram...). The discontinuous phrase omnem equitatum in (28) forms part of a complex piece of information. The first element omnem comes after the Topic Caesar and starts the focal part of the sentence. (27) Orationes autem me duas postulas; quarum alteram... speeches-acc but me-acc two-acc you ask of which one-acc ‘But you ask me for two speeches; one of them... ’ (Cic. Att. 2.7.1) (28)

Qua re nuntiata Caesar omnem ex castris equitatum this thing being reported Caesar all-acc from camp cavalry-acc suis auxilio misit. his-dat aid-dat sent ‘Which being reported, Caesar sent the entire cavalry to his men’s assistance from the camp.’ (Caes. Gal. 4.37.2)

Not only Focus, but also Topic constituents may be discontinuous: they are shown in (29) and (30). In these examples, the Focus constituents are vincit76 and quorsum respectively. Whereas vincit comes after haec and stands inside the discontinuous noun phrase, quorsum is not involved in the hyperbaton. (29)

[Disputatur in consilio... Plerique censebant... Alii, quod...] Haec vincit in consilio sententia this-nom won in council decision-nom [et prima luce postridie constituunt proficisci.]

76. However, I cannot exclude another interpretation according to which Focus is on haec sententia.

 Constituent order in Classical Latin prose

‘[A council of war was held to deliberate... The greater number were for... Others argued against...] This view prevailed in the council, [and they decided to start out at first light next day.]’ (Caes. Civ. 1.67.6) (30) Quorsum haec omnis spectat oratio? where this-nom all-nom regards speech-nom ‘What is the object of my whole speech?’

(Cic. Phil. 7.26)

It is noteworthy that in Latin discontinuity concerns not only noun phrases with one modifier, but also noun phrases containing multiple modifiers, as in (31). The noun iudiciis is separated from de his singulis. Another example is given in (32). This noun phrase is, together with silentio tuetur, the most informative part of the sentence, framed by togulam77 and suam, which bears the emphasis. (31)

Petam a vobis ut me dum de his singulis disputo I will beg from you that me while about these one by one I deal iudiciis attente audiatis. trials-abl with attention you listen ‘I will beg of you to listen to me with attention, while I deal with these trials one by one.’ (Cic. Clu. 89)

(32)

Qui poterat, familiaris noster [– sic est enim, volo te hoc scire – ] who could friend our Pompeius, togulam illam pictam silentio Pompey-nom toga-acc that embroidered-acc by silence tuetur svam. protects his-acc ‘The man who might have been one, my friend Pompey [– for so he is, let me tell you – ], lives up to that lovely embroidered toga of his by holding his tongue.’  (Cic. Att. 1.18.6) Not only modifiers, but also genitive complements may be separated from their head noun, as pointed out by Bolkestein (1998c: 193). In (33), Focus is on the genitive complement caedis impiae that takes a late position in the sentence. In (34), mei peccati, coming first, summarizes what has been said in the preceding context and functions as Topic; Focus is on the action. Such instances represent violations of the principle of head proximity (cf. Rijkhoff 2002:€251), the tendency to keep modifiers and heads as close together as possible. (33)

Auctores fuere contra ius caedis impiae instigators-nom were against law murder impious-gen legatorum nostrorum. legates our-gen

77. Pompey, by special law, was authorized to wear the toga picta ‘embroidered toga’ of a triumphator at the games. Cicero uses the contemptuous diminutive, togula.



Chapter 6.╇ Noun phrases 

‘They instigated the impious murder of our envoys in violation of the law of nations.’ (Liv. 5.4.13) (34) Itaque mei peccati luo poenas. so my mistake-gen I pay penalty-acc ‘Well, I am paying for my mistake.’

(Cic. Att. 3.9.1)

In the first chapter (p. 25) we have seen that in Latin prose it is possible (but rare) to find two hyperbata in the same sentence. Furthermore, a double discontinuity of one noun phrase is allowed, for example tuas Epiroticas litteras in (35); etiam might have come at the absolute beginning of the sentence. This noun phrase functions as Focus and saliency seems to be given to Epiroticas, which is occupying the middle position. (35) Tuas etiam Epiroticas exspecto litteras, ut... your-acc also from Epirus-acc I expect letter-acc so that ‘I am expecting a letter from you from Epirus, so that...’ (Cic. Att. 5.20.9) For hyperbaton of question words in interrogative sentences, see p. 196. The discontinuous noun phrases involving alien elements presented above are pragmatically motivated. On the other hand, in some cases hyperbaton seems to be used for rhythmical, prosodic reasons in order to produce a clausula at the end of a sentence or a clause (Adams 1971:€3, and cf. Menk 1925:€12, among others), or at least a “better” prosodic sequence. The verb is typically the only intervening constituent. An example is given in (36), with discontinuity of fortunarum suarum that, together, would present a sequence of two genitive plural endings. Nonetheless, the hyperbaton in (36) is also pragmatically grounded because it concerns a Focus constituent: the genitive phrase fortunarum suarum is the most salient constituent. (36) Eodem magnam partem fortunarum traiecit suarum. there great-acc part-acc fortunes-gen he transported his-gen ‘He transported there a great part of his fortunes.’ (Nep. Att. 2.3) So far we have looked for pragmatic reasons for the discontinuity of the noun phrases; the intervening word(s) are only instrumental. In some cases, when a single constituent is involved, we can envisage that two semantically related words are brought close together and contrasted, as suam quisque in (37). Apart from such situations, elements intervening in discontinuous noun phrases play no role at all. (37)

[Eo mortuo ad neminem unum summa imperii redit] sed separatim suam quisque classem ad arbitrium suum administrabat. but separately his-acc each fleet-acc to wish his operated ‘[After his death, no one inherited the overall command,] but each admiral operated his own fleet as he thought best.’ (Caes. Civ. 3.18.2)

 Constituent order in Classical Latin prose

10. Conclusions The examination of Latin noun phrases confirms de Jong’s (1983) and Pinkster’s (1995:€111) statements that in a neutral context, in the absence of contrast and emphasis, adjectives are placed after their head noun; determiners usually stand before it. Furthermore, as I sought to show, there are special arrangements possible for constituents fulfilling a pragmatic function. Noun phrases functioning as Focus normally present post-nominal adjective(s), but they also admit pre-nominal adjectives without contrast or emphasis. Similarly, determiners may appear in post-nominal position. Noun phrases with Topic function often have pre-nominal adjectives and determiners. Noun phrases containing multiple juxtaposed adjectives show post-nominal (48%), pre-nominal (29%) or combined pre- and post-nominal placement of adjectives. Juxtaposed adjectives are ordered according to the hierarchical principle following which adjectives expressing inherent, objective properties stand close to the noun whereas evaluative adjectives take more distance. However it is possible to permute the internal order of adjectives in around 18% of cases, especially when the noun phrase functions as Focus. Noun phrases containing coordinated adjectives exhibit pre-nominal placement of these elements. Noun phrases containing juxtaposed determiners mostly show pre-nominal placement of them in blocks. The ordering of determiners is not fixed and allows variations. Variations of the placement of quantifiers and indefinites involve semantic differences; the placement of the anaphor is and the demonstratives hic and ille depends on the functions they fulfil; variations in the placement of possessives with respect to other determiners are free. Genitive complements have more syntactic autonomy than adjectives and determiners. In general, when a genitive (phrase) depends on a complex noun phrase, it may be incorporated into it and the governing noun phrase forms a frame. As for their positioning, genitive complements conveying contextually given information are expected in pre-nominal position, whereas genitives expressing new information are mainly found after the noun. Hyperbaton or discontinuity of noun phrases is optional. Internal hyperbaton, i. e. noun phrases separated by their own complements (genitive and prepositional phrases) occurs with about 10% of all noun phrases. Hyperbaton produced by alien elements makes up 12% of all noun phrases. Internal hyperbaton produced by a complement belonging to the noun phrase does not seem to be a pragmatic device per se. It permits the speaker or author to mark a noun phrase as a unit. Hyperbaton involving alien elements is used for pragmatic purposes. It is an optional expressive means that concerns Focus and Topic constituents. The detached part of a noun phrase usually bears emphasis or contrast. To sum up the most important properties, the Latin noun phrase has a relatively low degree of integration of its elements. Noun phrases with a pragmatic function



Chapter 6.╇ Noun phrases 

Topic or Focus allow special arrangements of the internal order of their components. Contextually dependent constituents (with or without Topic function) often exhibit adjectives in pre-nominal position. Focus constituents may have adjectives in either pre-nominal or post-nominal position, and also permuted adjectives or postposed determiners. Furthermore, pragmatically pertinent constituents with Topic or Focus function allow discontinuity.

Conclusion This study has aimed at describing pragmatic, semantic, and syntactic aspects of constituent order in Classical Latin prose (Cicero, Caesar, and Sallust). It examines in detail three main sentence types (declarative, imperative and interrogative sentences), as well as noun phrases. Question tests are used in a systematic way in order to determine the pragmatic value of a sentence. As a theoretical framework, I adopted, on the one hand, the concepts of Topic and Focus described by Functional Grammar, and, on the other hand, the concept of contextual dependency, developed by Functional Sentence Perspective. These two pragmatic approaches were combined because they both provide concepts that are helpful for our understanding of Latin constituent order. Firstly, Functional Sentence Perspective works with the concept of contextual dependency that seems to be crucial for the interpretation of Latin sentences. Secondly, the distinction between four main pragmatic functions (Focus, Topic, Theme, and Tail) used by Functional grammar seems to be convenient for a description of a language known only from its literary tradition. In particular, the concept of various types of Topic permits us to have a better orientation in the study of Latin texts. On the other hand, I would not subscribe to the principle of communicative dynamism as systematically applying in Latin (pace Panhuis 1982). My major objection is that Latin sentences may, but may not, exhibit such an organisation, going from the least to the most informative element, because the final position (or pre-verbal position when the final position is occupied by the verb) is not an indicator of the saliency of a constituent. The salient constituent may stand in sentence-final position but may be placed in an internal position as well. The distinction between two main pragmatic functions in Functional Grammar, Topic (what the sentence is about) and Focus (the most salient information), is useful for our understanding of Latin constituent order. It is indeed important to determine what a sentence is about and what information it conveys. Whereas Topic constituents are found in the sentence-initial position, Focus constituents are sometimes difficult to determine. Identification of Focus constituents is complicated not only by the “mobility” of Focus constituents, but also by the complexity of the information conveyed in a prose text. The concept of dependency, as described by Firbas, distinguishes between elements that are known from the previous context or from the situation – they are contextually or situationally dependent – and elements that are absent from the previous context or from the situation – these are contextually or situationally independent. Such a distinction is very important for our understanding of Latin sentences, because the relationship of a constituent with the previous context is often the only means that permits us to interpret it as “definite” or “indefinite”. Contextual dependency is helpful

 Constituent order in Classical Latin prose

for the identification of Focus constituents because contextually independent entities are strong candidates for Focus function. It is indispensable for the interpretation of existential, locative and predicative sentences. The concept of several types of Topic, especially Sentence Topic (Hannay) and Discourse Topic, is of particular interest for the organisation of Latin sentences. The Sentence Topic is an entity chosen as a point of departure – this term goes back to Weil (1844:€25) – of a sentence and put in initial position. Not only an entity, but also a temporal or a local setting can be selected as the “foundation of a message”. When an entity is chosen as Sentence Topic, the speaker or author will say something about it; when a sentence starts with a temporal or local setting, the speaker or author will inform us about what happened then or there. Sentence Topic is to be distinguished from Discourse Topic, which refers to a person well established in the discourse. Discourse Topics belong to a higher level and come after Sentence Topics or setting elements. As I hope to have shown, the concept of Sentence Topic permits us to explain the placement of anaphoric pronouns in a sentence. Anaphoric pronouns placed in initial position function as Sentence Topic, i. e. as the point of departure of the sentence, and they usually pick up a salient element of the preceding sentence. Anaphoric pronouns occupying an internal position do not have a special pragmatic function. In addition, new discourse participants – with whatever syntactic function – that play a role afterwards cannot be zero pronominalised in the subsequent sentence; they must be referred to by means of anaphora. In other words, they are topicalised, selected as Sentence Topic in the subsequent sentence. In comparison with previous studies, I examined not only the variability of Latin constituent order but also its constraints. There are several words that must occupy, in an obligatory way, the initial position in Classical Latin prose: these are coordinators and the connective particles nam ‘for’ and itaque ‘so’. Three other particles systematically avoid the initial position: enim ‘indeed’, autem ‘on the other hand’, and vero ‘certainly’ are enclitic and regularly go in second sentence-position. The enclitics -que ‘and’, -ve ‘or’, and the interrogative particle -ne accompany their host and are also excluded from the initial position. Postpositive focusing particles such as quoque, quidem behave in the same way. This placement has consequences for noun phrases and prepositional phrases: enclitics and postpositive particles go after the first word of such a phrase. Other Latin particles such as igitur, as well as subordinators and interrogative words, usually occupy the sentence-initial or clause-initial position, but not in an obligatory way: priority may be given to another constituent. Apart from prepositions, which have their own rules of placement, other Latin words – verbs, nouns, pronouns, etc. – do not obey formal constraints of placement and may occupy any position in a sentence. However, pragmatically marked constituents are admitted in the prominent initial position; constituents that are not pragmatically pertinent avoid this position. There are other striking conclusions that differ from previous studies. Firstly, consideration of semantic properties of objects such as animacy, agentivity and individuation permits us to conclude that animate, agentive and individuated entities manifest



Conclusion 

more mobility than inanimate entities. Inanimate entities often function as a pragmatic unit with the verb, and this explains the high number of instances of the verb preceded by its object in the sentence-final position found in historical narrative. Secondly, a verb in the imperative need not be placed in sentence-initial position. When an imperative sentence contains a Topic, this is placed first; when an imperative lacks a Topic – which is typically the case in several periphrastic constructions and communicative expressions – the verb goes in initial position. Thirdly, verbs placed in initial position are not necessarily salient. In particular, initial verbs in presentative sentences and sentences providing an explanation (typically connected by nam or enim) do not have a special pragmatic function; another constituent bears Focus function in such cases. Their initial placement is a result of the fact that these sentences lack a Topic. Fourthly, noun phrases functioning as Topic and Focus – i. e. pragmatically pertinent – allow special ordering of their components. Nouns with Focus function normally present one or more post-nominal adjectives, but they also admit, without contrast or emphasis, pre-nominal adjectives, as well as determiners in post-nominal position. Noun phrases with Topic function often have pre-nominal adjectives. Noun phrases with Topic and Focus function allow discontinuity. In addition, the placement of genitive complements also depends on whether they fulfil a pragmatic function or not, and on their contextual status. There is one more conclusion to be drawn: Latin constituent order obeys pragmatic rules of placement. This is not an original conclusion because several Latin scholars, including Panhuis and Pinkster, have already claimed it. However, this claim implies one point that must finally be explicitly formulated: if we claim that variability of Latin constituent order is pragmatically motivated, it means that variability is not a matter of stylistics. This does not exclude an additional use of special stylistic arrangements and stylistic figures in elaborated Latin texts. Latin constituent order is basically grounded in pragmatics, and stylistic ornamentation may be superposed to it, but variability in Latin constituent order cannot be regarded as stylistic ornamentation. One might object that my study concerns Classical Latin prose, and mostly deals with literary works written by excellent Latin authors. The reply is simple: texts without literary elaboration do not become, for that reason, texts without variable order. The tendencies of placement are the same, to take one example among others, in Egeria’s Itinerarium.1

1.

Cf. Spevak (2005a).

References Editions of Latin texts and translations Latin texts are quoted from the CD-Rom Bibliotheca Teubneriana Latina (=BTL) 19991, Cetedoc, Catholic University of Louvain-la-Neuve, Brepols. Translations were taken from the LOEB Classical Library series; for Caesar and Sallust, I also used translation by John Mackenzie Carter, Julius Caesar, The Civil War, Oxford: OUP, 1997, and by Anthony John Woodman, Sallust, Catiline’s War, The Jugurthine War, Histories, London: Penguin Classics, 2007.

Corpora used Corpus 1: Cicero, Tusculan Disputations (Tusculanae disputationes), books 1 and 3; On His House (De domo sua); Philippics (Philippicae), 1 and 4; Letters to Atticus (Epistulae ad Atticum), 13.50 – 16; Caesar, The Civil War (De bello civili), 1 – 3.30; Sallust, The Jugurthine War (De bello Iugurthino). Corpus 2: Cicero, Letters to Atticus (Epistulae ad Atticum), books 1 – 4; Caesar, The Gallic War (De bello Gallico), books 1 – 5; Sallust, The Conspiracy of Catiline (De coniuratione Catilinae) and The Jugurthine War (De bello Iugurthino). Corpus BTL: Bibliotheca Teubneriana Latina, 19911. Corpus LASLA: database of the Laboratoire d’Analyse Statistique des Langues Anciennes (http:// www.cipl.ulg.ac.be/lsl.htm).

Bibliographic references Adams, James Noel. 1971. A type of hyperbaton in Latin prose. ‘Proceedings’ of the Cambridge ‘Philological Society ‘17: 1–16. Adams, James Noel. 1976a. A typological approach to Latin word order. Indogermanische Forschungen 81: 70–100. Adams, James Noel. 1976b. The Text and language of a Vulgar Latin chronicle. Anonymus Valesianus II. London: University of London, Institute of Classical Studies. Adams, James Noel. 1994a. Wackernagel’s Law and the Placement of the Copula esse in Classical Latin. [Proceedings of the Cambridge Philological Society, Supplementary Vol. 18] Cambridge: The Cambridge Philological Society.

 Constituent order in Classical Latin prose Adams, James Noel. 1994b. Wackernagel’s law and the position of unstressed personal pronouns in Classical Latin. Transactions of the Philological Society 92(2): 103–178. Adams, James Noel. 1999. Nominative personal pronouns and some patterns of speech in republican and Augustan poetry. Proceedings of the British Academy 93: 97–133. Allen, W. Sidney. 1973. Accent and Rhythm. Prosodic features of Latin and Greek. A Study in Theory and reconstruction. Cambridge: CUP. Amacker, René. 1998. Ordre des mots et subordination. La traiectio chez Varron. In Estudios de lingüística latina. Actas del IX coloquio internacional de lingüística latina, Universidad autónoma de Madrid, 14–18 de abril de 1997, Benjamín García-Hernández (ed.), 139–154. Madrid: Ed. Clasicás. Amacker, René. 2001. Indices de thématisation segmentée non initiale chez Cicéron. In De lingua Latina novae quaestiones: Actes du Xe colloque international de linguistique latine, ParisSèvres, 19–23€avril 1999, Claude Moussy (ed.), 185–200. Leuven: Peeters. Babič, Matjaž. 1992. Word order in Plautus. interrogative and imperative sentences. živa antika, Antiquité vivante 41(2): 56–63. Bauer, Brigitte L.M. 2009. Word order. In New Perspectives on Historical Latin Syntax, Philip Baldi & Pierluigi Cuzzolin (eds), 241–316. Berlin: Mouton de Gruyter. Behaghel, Otto. 1909. Bewiehungen zwischen Umfang und Reihenfolge von Satzgliedern. Indogermanische Forschungen 25: 110–142. Behaghel, Otto. 1932. Deutsche Syntax. Eine geschichtliche Darstellung, Vol. 4, Wortstellung, Periodenbau. Heidelberg: Winters. Berger, Ernst. 19424. Stylistique latine, Paris: C. Klincksieck. Bodelot, Colette. 1987. L’interrogation indirecte en latin. Syntaxe, valeur illocutoire, formes. Paris: Société pour l’information grammaticale. Bolkestein, A. Machtelt. 1981. Embedded predications, displacement and pseudo-argument formation in Latin. In Predication and Expression in Functional grammar, A. Machtelt Bolkestein & Henk A. Combé (eds), 63–112. London: Academic press. Bolkestein, A. Machtelt. 1983. Genitive and dative possessors in Latin. In Advances in Functional Grammar, Simon C. Dik (ed), 55–91. Dordrecht: Foris. Bolkestein, A. Machtelt. 1989. Parameters in the expression of embedded predication in Latin. In Subordination and Other Topics in Latin. Proceedings of the Third Colloquium on Latin Linguistics (Bologna, 1–5€April 1985) [Studies in Language Companion Series 17], Gualtiero Calboli (ed), 3–35. Amsterdam: John Benjamins. Bolkestein, A. Machtelt. 1995. Functions of verb-subject order in Latin. Sprachtypologie und Universalienforschung 48(1–2): 32–43. Bolkestein, A. Machtelt. 1996a. Is ‘qui’ ‘et is’? On the so-called free relative connection in Latin. In Aspects of Latin. Papers from the Seventh International Colloquium on Latin Linguistics, Jerusalem, April 1993, Hannah Rosén (ed.), 553–566. Innsbruck: Institut für Sprachwissenschaft. Bolkestein, A. Machtelt. 1996b. Free but not arbitrary. ‘Emotive’ word order in Latin? In On Latin. Linguistic and Literary Studies in Honour of Harm Pinkster, Rodie Risselada, Jan R. de Jong & A. Machtelt Bolkestein (eds), 7–23. Amsterdam: J.C. Gieben. Bolkestein, A. Machtelt. 1998a. What to do with topic and focus? Evaluating pragmatic information. In Functional Grammar and Verbal Interaction [Studies in Language Companion Series 44], Mike Hannay & A. Machtelt Bolkestein (eds), 193–214 Amsterdam: John Benjamins.



References  Bolkestein, A. Machtelt. 1998b. Between brackets. Some properties of parenthetical clauses in Latin. An investigation of the language of Cicero’s letters. In Latin in Use. Amsterdam Studies in the Pragmatics of Latin, Rodie Risselada (ed.), 1–17. Amsterdam: J.C. Gieben. Bolkestein, A. Machtelt. 1998c. Word order variation in complex noun phrases in Classical Latin. In Estudios de lingüística latina. Actas del IX coloquio internacional de lingüística latina, Universidad autónoma de Madrid, 14–18 de abril de 1997, Benjamín García-Hernández (ed.), 185–202. Madrid: Ed. Clasicás. Bolkestein, A. Machtelt. 2000. Discourse organization and anaphora in Latin. In Textual Parameters in Older Languages, Susan C. Herring, Pieter van Reenen & Lene Schøsler (eds), 107– 137. Amsterdam: John Benjamins. Bolkestein, A. Machtelt. 2001. Random scrambling? Constraints on discontinuity in Latin noun phrases. In De lingua Latina novae quaestiones: Actes du Xe colloque international de linguistique latine, Paris-Sèvres, 19–23€avril 1999, Claude Moussy (ed.), 245–258. Leuven: Peeters. Bolkestein, A. Machtelt. 2002. Anaphoric absolute ablatives in Caesar. In Donum grammaticum. Studies in Latin and Celtic Linguistics in Honour of Hannah Rosén, Lea Sawicki & Donna Shalev (eds), 51–63. Leuven: Peeters. Bolkestein, A. Machtelt & Grift, Michel van de. 1994.€Participant tracking in Latin discourse. In Linguistic Studies on Latin. Selected Papers from the 6th International Colloquium on Latin Linguistics (Budapest, 23–27€ March 1991) [Studies in Language Companion Series 28], Jószef Herman (ed.), 283–302. Amsterdam: John Benjamins. Bortolussi, Bernard. 2007. Phénomènes d’ambiguïté syntaxique dans la proposition infinitive. In L’↜ambiguïté en Grèce et à Rome. Approche linguistique, Claude Moussy & Anna Orlandini (eds), 81–91. Paris: Presses de l’Université Paris-Sorbonne. Cabrillana, Concepción. 1999. On the integration of theme constituents in the predication in Latin. Euphrosyne 27: 417–427. Cabrillana, Concepción. 2001. Locative structures in construction with sum. In De lingua Latina novae quaestiones. Actes du Xe colloque international de linguistique latine (Paris-Sèvres, 19–23€avril 1999), Claude Moussy (ed), 275–290. Leuven: Peeters. Chausserie-Laprée, Jean-Pierre. 1969. L’↜expression narrative chez les historiens latins. Histoire d’un style. Paris: E. de Boccard. Clark, Eve V. 1978. Locationals: Existential, locative, and possessive constructions. In Universals of Human Language, Vol. 4, Syntax, Joseph H. Greenberg (ed.), 83–126. Stanford CA: University press. Daneš, František. 1968. Some thoughts on the semantic structure of the sentence. Lingua 21: 55–69. Daneš, František. 1974. Functional sentence perspective and the organization of the text. In Papers on Functional Sentence Perspective, František Daneš (ed.),106–128. Prague/The Hague: Academia/Mouton. Devine, Andrew M. & Stephens, Laurence D. 2006. Latin Word Order. Structured Meaning and Information. Oxford: OUP. Dik, Helma. 1995. Word Order in Ancient Greek. A Pragmatic Account of Word Order Variation in Herodotus. Amsterdam: J.C. Gieben. Dik, Helma. 1997. Interpreting adjective position in Herodotus. In Grammar as Interpretation. Greek Literature in its Linguistic Context, Egbert J. Bakker (ed.), 55–76, Leiden: Brill. Dik, Helma. 2003. On unemphatic ‘emphatic’ pronouns in Greek. Nominative pronouns in Plato and Sophocles. Mnemosyne 56(5): 535–550. Dik, Helma. 2007. Word Order in Greek Tragic Dialogue. Oxford: OUP.

 Constituent order in Classical Latin prose Dik, Simon C. 1997. The Theory of Functional Grammar. 2 vols., 2nd revised edn. ed. by Kees Hengeveld. Berlin: Mouton de Gruyter. Dover, Kenneth James. 1960. Greek Word Order. Cambridge: CUP. Dryer, Matthew S. 2005. Order of genitive and noun. In The World Atlas of Language Structures, Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (eds), 350–353. Oxford: OUP. Firbas, Jan. 1992. Functional Sentence Perspective in Written and Spoken communication, Cambridge: CUP. Fraenkel, Eduard. 19642 [1933]. Kolon und Satz. Beobachtungen zur Gliederung des antiken Satz. In Kleine Beiträge zur klassischen Philologie 2, 93–130. Roma: Edizioni di storia e letteratura. Fraenkel, Eduard. 1965. Noch einmal Kolon und Satz. Munich: Verlag der Bayerischen Akademie der Wissenschaften. Fruyt, Michèle. 1990. La formation des mots par agglutination en latin. Bulletin de la Société de Linguistique 85: 173–209. Fruyt, Michèle. 2007, Le réfléchi latin. Comment expliquer ses emplois. In L’↜énoncé réfléchi, André Rousseau, Didier Bottineau & Daniel Roulland (eds), 249–271. Rennes: University Press. Fugier, Huguette. 1983. Le syntagme nominal en latin classique. In Aufstieg und Niedergang der römischen Welt, Wolfgang Haase (ed.), 2, 29(1), 212–269. Berlin: Walter de Gruyter. Fugier, Huguette & Corbin, Jean-Marie. 1977. Coordination et classes fonctionnelles dans le système nominal latin. Bulletin de la Société de Linguistique 72: 245–273. Gasparov, Boris M. 1978. Ustnaja reč’ kak semiotičeskij ob’jekt [Spoken language as a semiotic object]. In Semantika nominacii i semiotika ustnoj reči [Semantics of nouns and semiotics of spoken language], Acta et Commentationes Universitatis Tartuensis 442, 63–112, Tartu. Gettert, Hans. 1999. Konstituenz und lateinische Syntax. Aachen: Shaker. Giannecchini, Giulio. 1982. Osservazioni su tendenze nominali e ordine delle parole, con particolare riguardo al latino. In Materiali e discussioni per l’analisi dei testi classici 8: 79–122. Givón, T. 1983. Topic continuity in discourse. An introduction. In Topic Continuity in Discourse. A Quantitative Cross-language Study [Typological Studies in Language 3], T. Givón (ed.), 5–41. Amsterdam: John Benjamins. Greenberg, Joseph H. 19662. Some universals of grammar with particular reference to the order of€ meaningful elements. In Universals of Language, Joseph H. Greenberg (ed.), 73–113. Cambridge MA: MIT. Habinek, Thomas. N. 1986. The Colometry of Latin Prose. Berkeley CA: University of California Press. Hannay, Michael. 1985a. English Existentials in Functional Grammar. Dordrecht: Foris. Hannay, Michael. 1985b. Inferability, discourse boundeness, and Sub-topics. In Syntax and Pragmatics in Functional Grammar, A. Machtelt Bolkestein, Casper de Groot & J. Lachlan Mackenzie (eds), 49–63, Dordrecht: Foris. Hannay, Michael. 1991. Word order variation in English. Journal of Pragmatics 16(1): 131–155. Hannay, Michael & Martínez Caro, Elena. 2008. Last things first? A FDG approach to clause-final focus constituents in English and Spanish. In Languages and Cultures in Contrast and Comparison [Pragmatics & Beyond Series 175], María de los Ángeles Gómez González, J. Lachlan Mackenzie & Elsa M. González Álvarez (eds), 33–68. Amsterdam: John Benjamins. Happ, Heinz. 1976. Grundfragen einer Dependenz-Grammatik des Lateinischen. Göttingen: Vandenhoeck & Ruprecht.



References 

Herman, Jószef. 1985. Déclinaison latine et évolution du syntagme nominal. In Syntaxe et latin. Actes du IIe congrès international de linguistique latine (Aix-en-Provence, 28–31€mars 1983), Christian Touratier (ed.), 345–360. Aix-en-Provence: University Press. Hetzron, Robert. 1978. On the relative order of adjectives. In Language Universals. Papers from the Conference Held at Gummersbach near Cologne, Germany, October 3–8, 1976, Hansjakob Seiler (ed.), 165–184. Tubingen: Narr. Hoff, François. 1979. Le système linguistique de l’interrogation en latin classique. PhD dissertation, University of Strasbourg. Hoff, François. 1984. Le système linguistique de l’interrogation en latin classique. In L’Interrogation: Actes du colloque tenu les 19 et 20€décembre 1983, Paul Valentin (ed.), 17–35. Paris: University Press. Hoff, François. 1995. L’ordre des mots chez César. Les groupements adjectif-nom-génitif rares. Lalies 15: 245–257. Hoff, François. 1996. L’ordre des mots chez César: Les séquences interdites. Le cas du perfectum passif. In Acten des VIII. internationalen Kolloquiums zur lateinischen Linguistik, Alfred Bammesberger & Friedrich Heberlein (eds), 372–382. Heidelberg: Winter. Hoff, François. 2003. Magnum fructum studiorum optimorum. L’ordre des mots dans les syntagmes nominaux complexes en latin classique. Lalies 23: 205–223. Hoffmann, Maria E. 1989. A typology of Latin theme constituents. In Actes du Ve colloque de linguistique latine, Louvain-la-Neuve/Borzée, 31€mars – 4€avril 1989, Marius Lavency & Dominique Longrée (eds), 185–196. Louvain-La-Neuve: Institut de linguistique de Louvain. Hoffmann, Maria E. 1991. Word order patterns of excipere in the sense of ‘to follow after’. In New Studies in Latin Linguistics. Selected Papers from the 4th International Colloquium on Latin Linguistics, Cambridge, April 1987 [Studies in Language Companion Series 21], Robert Coleman (ed.), 367–380, Amsterdam: John Benjamins. Hopper, Paul J. & Thompson, Sandra A. 1980. Transitivity in grammar and discourse. Language 56(2): 251–299. Janse, Mark. 1997. Review of Adams, Wackernagel’s Law and the position of unstressed personal pronouns in Classical Latin (1994). Kratylos 42: 105–115. Johnson, Ruth A.L. 1991. The Direct Object Pronoun as a Marker of Transitivity in Latin. PhD dissertation, University of California. Jones, Frederick. 1991. Subject, topic, given and salient: Sentence beginnings in Latin, Proceedings of the Cambridge Philological Society 37: 81–105. Jong, Jan R. de. 1983. Word order within Latin noun phrases. In Latin Linguistics and Linguistic Theory [Studies in Language Companion Series 12], Harm Pinkster (ed.), 131–144. Amsterdam: John Benjamins. Jong, Jan R. de. 1986. Hyperbaton en informatiestructuur. Lampas 19: 323–331. Jong, Jan R. de. 1989. The position of the Latin subject. In Subordination and Other Topics in Latin. Proceedings of the Third Colloquium on Latin Linguistics (Bologna, 1–5€April 1985) [Studies in Language Companion Series 17], Gualtiero Calboli (ed.), 521–540. Amsterdam: John Benjamins. Jong, Jan R. de. 1996. The borderline between deixis and anaphora in Latin. In Aspects of Latin. Papers from the Seventh International Colloquium on Latin Linguistics, Jerusalem, April 1993, Hannah Rosén (ed.), 499–509. Innsbruck: Institut für Sprachwissenschaft. Jonge, Casper. C. de. 2008. Between Grammar and Rhetoric. Dionysius of Halicarnassus on Language, Linguistics, and Literature. Leiden: Brill.

 Constituent order in Classical Latin prose Kłósek, Izabela. 1984. Place de l’adjectif épithète dans le groupe nominal en français et en polonais. Poznan: University Press. Knoth, Tine W. 2006. Prima positio. Et functionelt studie i saetningens første position i latinsk prosa. PhD dissertation, University of Copenhagen. König, Ekkehard. 1991. The Meaning of Focus Particles: A Comparative Perspective. London: Routledge. Kroon, Caroline H.M. 1995. Discourse Particles in Latin. A Study of nam, enim, autem, vero and at. Amsterdam: J.C. Gieben. Kroon, Caroline H.M. 2005. The relationship between grammar and discourse: Evidence from the Latin particle quidem. In Papers on Grammar IX, 2, Latina Lingua. Proceedings of the Twelfth International Colloquium on Latin Linguistics (Bologna, 9–14€June 2003), Gualtiero Calboli (ed.), 577–590. Roma: Herder. Kühner, Raphael & Stegmann, Carl. 1914. Ausführliche Grammatik der lateinischen Sprache, 2, Satzlehre, 2 vols. Hannover: Hahn. Lambrecht, Knud. 1988. Presentational cleft constructions in spoken French. In Clause Combining in Grammar and Discourse [Typological Studies in Language 18], John Haiman & Sandra A. Thompson (eds), 135–179. Amsterdam: John Benjamins. Lambrecht, Knud. 1994. Information Structure and Sentence Form. Topic, Focus, and Mental Representations of Discourse Referents. Cambridge: CUP. Lebreton, Jules. 1901. Études sur la langue et la grammaire de Cicéron. Paris: Hachette. Lehmann, Christian. 1991. The Latin nominal group in typological perspective. In New Studies in Latin Linguistics. Selected Papers from the 4th International Colloquium on Latin Linguistics, Cambridge, April 1987 [Studies in Language Companion Series 21], Robert Coleman (ed.), 203–232. Amsterdam: John Benjamins. Lehmann, Christian. 2005. Sur l’évolution du pronom possessif. In Latin et langues romanes. études de linguistique offertes à József Herman à l’occasion de son 80ème anniversaire, Sandor Kiss, Luca Mondin & Giampaolo Salvi (eds), 37–46. Tubingen: Niemeyer. Lisón Huguet, Nicolás. 2001. El orden de palabras en los grupos nominales en latín. Zaragoza: University Press. Löfstedt, Bengt. 1966. Die Konstruktion c’est lui qui l’a fait im Lateinischen. Indogermanische Forschungen 71: 253–277. Longrée, Dominique. 2002. Sur la concurrence entre ‘relatifs’ et ‘démonstratifs de liaison’. In Theory and Description in Latin Linguistics. Selected Papers from the XIth International Colloquium on Latin Linguistics, Amsterdam, June 24–29, 2001, A. Machtelt Bolkestein, Caroline H. M. Kroon, Harm Pinkster et al. (eds), 205–217. Amsterdam: J.C. Gieben. Longrée, Dominique. 2004. Une approche statistique de la concurrence entre démonstratifs chez les historiens latins (César, Salluste, Tacite). In Anaphore, cataphore et corrélation en latin. Actes de la journée d’étude de linguistique latine, Université Blaise Pascal-Clermont-Ferrand II, 7€janvier 2003, Colette Bodelot (ed.), 157–178. Clermont-Ferrand: University Press. Luraghi, Silvia. 1995. The pragmatics of verb initial sentences in some ancient Indo-European languages. In Word Order in Discourse. Selection of Papers Presented at the 20th Linguistic Symposium Held at the University of Wisconsin-Milwaukee, April 12–14, 1991 [Typological Studies in Language 30], Pamela Downing & Michael Noonan (eds), 355–386. Amsterdam: John Benjamins. Luraghi, Silvia. 1997. Omission of the direct object in Latin. Indogermanische Forschungen 102: 239–257.



References  Lyons, John. 1967. A note on possessive, existential and locative sentences. Foundation of Language 3: 390–396. Lyons, John. 1968. Existence, location, possession and transitivity. Logic, Methodology and Philosophy of Science 3: 495–504. Marouzeau, Jules. 1910. La phrase à verbe ‘être’ en latin. Paris: Geuthner. Marouzeau, Jules. 1922. L’ordre des mots dans la phrase latine, Vol. I. Les groupes nominaux. Paris: Champion. Marouzeau, Jules. 1938. L’ordre des mots dans la phrase latine, Vol. II. Le verbe. Paris: Les Belles Lettres. Marouzeau, Jules. 1947. Place de la préposition. Revue des études Latines 25: 298–327. Marouzeau, Jules. 1949. L’ordre des mots dans la phrase latine, Vol. III. Les articulations de l’énoncé. Paris: Les Belles Lettres. Marouzeau, Jules. 1953. L’ordre des mots en latin. Volume complémentaire. Paris: Les Belles Lettres. Maslova, Elena & Bernini, Giuliano. 2006. Sentence topics in the languages of Europe and beyond. In Pragmatic Organization of Discourse in the Languages of Europe, Giuliano Bernini & Marcia L. Schwartz (eds), 69–120. Berlin: Mouton de Gruyter. Meillet, Antoine. 1903. Introduction à l’étude comparative des langues indo-européennes. Paris: Hachette. Menk, Edgar A. 1925. The Position of the Possessive Pronoun in Cicero’s Orations. Grand Forks IA: University of Iowa. Miller, Jim. 2006. Focus in the languages of Europe. In Pragmatic Organization of Discourse in the Languages of Europe, Giuliano Bernini & Marcia L. Schwartz (eds), 121–214. Berlin: Mouton de Gruyter. Neubourg, Leo de. 1977. Sur le caractère analogique de la place de l’adjectif en latin. Orbis 26(2): 395–403. Neubourg, Leo de. 1978. Arguments supplémentaires en faveur de l’analogie dans l’ordre des mots. Orbis 27(2): 352–372. Odelman, Eva. 1972. Études sur quelques reflets du style administratif chez César. Stockholm: University Press. Oniga, Renato. 1988. I composti nominali latini. Una morphologia generativa. Bologna: Patròn. Orlandini, Anna. 1983. Une analyse sémantique et pragmatique des pronoms indefinis en latin. In Latin Linguistics and Linguistic Theory [Studies in Language Companion Series 12], Harm Pinkster (ed.), 229–240. Amsterdam: John Benjamins. Orlandini, Anna. 1995. Il riferimento del nome. Un’↜analisi semantico-pragmatica del nome latino. Bologna: CLUEB. Orlandini, Anna. 2001. Négation et argumentation en latin. Louvain: Peeters. Orlandini, Anna. 2005. Fonctions adverbiales dans des structures corrélatives en latin. In Structures parallèles et corrélatives en grec et en latin. Actes du colloque de linguistique grecque et latine, Bordeaux, 26–27€septembre 2002, Paolo de Carvalho & Frédéric Lambert (eds), 159– 180, Saint-étienne: University Press. Ostafin, David M. 1986. Studies in Latin Word Order. A Transformational Approach. PhD dissertation, University of Connecticut. Panhuis, Dirk G.J. 1980. Gapping in Latin. The Classical Journal 75: 229–240. Panhuis, Dirk G.J. 1981. Word order, genre, adstratum: The place of the verb in Caesar’s topographical excursus. Glotta 59: 295–308. Panhuis, Dirk G.J. 1982. The Communicative Perspective in the Sentence. A Study of Latin Word Order [Studies in Language Companion Series 11], Amsterdam: John Benjamins.

 Constituent order in Classical Latin prose Panhuis, Dirk G.J. 1984a. Is Latin an SOV language? A diachronic perspective. Indogermanische Forschungen 89: 140–159. Panhuis, Dirk G.J. 1984b. Topic shift and other discourse functions of passives in Latin narratives. Glotta 62: 232–240 Pennell Ross, Deborah. 1996. Anaphors and antecedents in narrative text. In Aspects of Latin. Papers from the Seventh International Colloquium on Latin Linguistics, Jerusalem, April 1993, Hannah Rosén (ed.), 511–523. Innsbruck: Institut für Sprachwissenschaft. Perrot, Jean. 1978. Fonctions syntaxiques, énonciation, information. Bulletin de la Société de Linguistique 73: 85–101. Pinkster, Harm. 1972. On Latin Adverbs. Amsterdam: North-Holland. Pinkster, Harm. 1985. The discourse function of the passive. In Syntax and Pragmatics in Functional Grammar, A. Machtelt Bolkestein, Casper de Groot & J. Lachlan Mackenzie (eds), 107–118. Dordrecht: Foris. Pinkster, Harm. 1987. The use of subject pronouns in Latin. In études de linguistique générale et de linguistique latine offertes en hommage à Guy Serbat, 369–379. Paris: Société pour l’information grammaticale. Pinkster, Harm. 1990. La coordination. L’information grammaticale 46: 8–13. Pinkster, Harm. 1991. Evidence for SVO in Latin? In Latin and the Romance Languages in the Early Middle Ages, Roger Wright (ed.), 69–82. London: Routledge. Pinkster, Harm. 1992. Notes on the Syntax of Celsus. Mnemosyne 45: 513–524. Pinkster, Harm. 1995. Sintaxis y semántica del latín, revised and enhanced edn. of Latin Syntax and Semantics, 1990. Madrid: Ed. Clasicás. Pinkster, Harm. 2005a. The use of is and ille in Seneca Rhetor. In Latin et langues romanes. études de linguistique offertes à József Herman à l’occasion de son 80ème anniversaire, Sandor Kiss, Luca Mondin & Giampaolo Salvi (eds), 57–64. Tubingen: Niemeyer. Pinkster, Harm. 2005b. Changing patterns of discontinuity in Latin, paper presented at the 13th International Colloquium on Latin Linguistics, Brussels, 4–9€ April, 2005. <www.harmpinkster.nl>, section ‘handouts’. Pinkster, Harm. 2005c. The language of Pliny the Elder. In Aspects of the Language of Latin Prose, Tobias Reinhardt, Michael Lapidge & James N. Adams (eds), 239–256, Proceedings of the British Academy 129. Oxford: OUP. Polo, Chiara. 2004. Word Order between Morphology and Syntax. Padova: Unipress. Quirk, Randolph. 1985. A Comprehensive Grammar of the English Language. London: Longman. Renský, Miroslav. 1966. English verbo-nominal phrases. Some structural and stylistic aspects. Travaux linguistiques de Prague 1: 289–299. Richter, Elise. 1903. Zur Entwicklung der romanischen Wortstellung aus der lateinischen. Halle: Niemeyer. Rijkhoff, Jan. 2002. The Noun Phrase. Oxford: OUP. Risselada, Rodie. 1984. Coordination and juxtaposition of adjectives in the Latin NP. Glotta 62: 202–231. Risselada, Rodie. 1989. Latin illocutionary parentheticals. In Actes du Ve colloque de linguistique latine, Louvain-la-Neuve/Borzée, 31€ mars – 4€ avril 1989, Marius Lavency & Dominique Longrée (eds), 367–378. Louvain-La-Neuve: Institut de linguistique de Louvain. Risselada, Rodie. 1991. Passive, perspective and textual cohesion. In Selected Papers from the 4th International Colloquium on Latin Linguistics, Cambridge, April 1987 [Studies in Language Companion Series 21], Robert Coleman (ed.), 401–414. Amsterdam: John Benjamins.



References 

Risselada, Rodie. 1993. Imperatives and Other Directive Expressions in Latin. A Study in the Pragmatics of a Dead Language. Amsterdam: J.C. Gieben. Risselada, Rodie. 2004. Actes de langage et rapports sociaux: Les directifs dans Plaute, Asinaria 649–732. In Registros lingüísticos en las lenguas clásicas, Antonio López Eire & Agustín Ramos Guerreira (eds), 269–290. Salamanca: University Press. Rosén, Hannah. 1981. Studies in the Syntax of the Verbal Noun in Early Latin. Munich: Fink. Rosén, Hannah. 1990. La coordination assymétrique comme critère de fonction syntaxique en latin. L’information grammaticale 46: 34–37. Rosén, Hannah. 1998. Latin presentative sentences. In Estudios de lingüística latina. Actas del IX coloquio internacional de lingüística latina, Universidad autónoma de Madrid, 14–18 de abril de 1997, Benjamín García-Hernández (ed.), 723–742. Madrid: Ed. Clasicás. Rosén, Hannah. 1999. Latine loqui: Trends and Directions in the Crystallization of Classical Latin. Munich: Fink. Rosén, Hannah. 2009. Coherence, sentence modification, and sentence-part modification – The contribution of particles. In New Perspectives on Historical Latin Syntax, Philip Baldi & Pierluigi Cuzzolin (eds), 317–441. Berlin: Mouton de Gruyter. Rosén, Hannah. Forthcoming. Latin epitaxis in historical and typological view. Paper presented at the 13th International Colloquium on Latin Linguistics, Brussels, 4–9€April 2005 Rosengren, Inger. 1997. The thetic/categorical distinction revisited once more. Linguistics 35: 439–479. Salvi, Giampaolo. 2004. La formazione della struttura di frase romanza. Ordine delle parole e clitici dal latino alle lingue romanze antiche. Tubingen: Niemeyer. Sasse, Hans-Jürgen. 1995. Theticity and VS order. a case study. Sprachtypologie und Universalienforschung 48(1–2): 3–31. Schneider, Nicolaus. 1912. De verbi in lingua Latina collocatione, Munster. Sgall, Petr, Hajičová, Eva & Panevová, Jarmila. 1986. The Meaning of the Sentence: Its Semantic and Pragmatic Aspects. Prague/Dordrecht: Academia/Reidel. Siewierska, Anna. 1984. Phrasal discontinuity in Polish. Australian Journal of Linguistics 4(1): 57–71. Siewierska, Anna. 1988. Word Order Rules, London: Croom Helm. Siewierska, Anna. (ed.). 1998a. Constituent Order in the Languages of Europe. Berlin: Mouton de Gruyter. Siewierska, Anna. 1998b. Polish main clause constituent order and FG pragmatic functions. In Functional Grammar and Verbal Interaction [Studies in Language Companion Series 44], Mike Hannay & A. Machtelt Bolkestein (eds), 243–266. Amsterdam: John Benjamins. Siewierska, Anna & Uhlířová, Ludmila. 1998. An overview of word order in Slavic languages. In Constituent Order in the Languages of Europe, Anna Siewierska (ed.), 105–149. Berlin: Mouton de Gruyter. Siewierska, Anna. 2004. Person. Cambridge: CUP. Solodow, Joseph B. 1978. The Latin particle quidem. University Park PA: The American Philological Association. Somers, Maartje H. 1994. Theme and topic. The relation between discourse and constituent fronting in Latin. In Linguistic Studies on Latin. Selected Papers from the 6th International Colloquium on Latin Linguistics (Budapest, 23–27€March 1991) [Studies in Language Companion Series 28], Jószef Herman (ed.), 151–163. Amsterdam: John Benjamins. Sörés, Anna. 2004. La notion d’‘ordre de base’ dans la typologie des langues et son application à la langue hongroise. Bulletin de la Société de Linguistique 99: 281–310.

 Constituent order in Classical Latin prose Spevak, Olga. 2004. Verb-subject order in Latin: The case of existential and locative sentences. Classica et Mediaevalia 55: 381–396. Spevak, Olga. 2005a. Itinerarium Egeriae. L’↜ordre des constituants obligatoires. Mnemosyne 58(2): 235–261. Spevak, Olga. 2005b.€A propos de ‘uerbum primo loco’. Essai de synthèse. In Papers on Grammar IX, 2, Latina Lingua. Proceedings of the Twelfth International Colloquium on Latin Linguistics (Bologna, 9–14€June 2003), Gualtiero Calboli (ed.), 731–740. Roma: Herder. Spevak, Olga. 2006a. L’↜ordre des constituants en latin: Aspects pragmatiques, syntaxiques et sémantiques. Habilitation, University of Paris IV-Sorbonne. <www.olgaspevak.nl>, section ‘L’ordre des mots’. Spevak, Olga. 2006b. Les enclitiques en latin. Indogermanische Forschungen 111: 249–274.€ Spevak, Olga. 2006c. Recent studies on word order in Latin. A review article, N. Lisón Huguet, El orden de palabras en los grupos nominales en latín. Universidad de Zaragoza, 2001; C. Polo, Word Order between Morphology and Syntax. Padova, 2004; G. Salvi, La formazione della struttura di frase romanza. Ordine delle parole e clitici dal latino alle lingue romanze antiche. Tubingen, 2004. Mnemosyne 59(3): 433–452. Spevak, Olga. 2006d. Tamen. Essai d’une description syntaxique. Glotta 82: 221–248. Spevak, Olga. 2007a. La place du focus en latin. Quelques observations. In Cohérence textuelle et ordre des syntagmes (13e Colloque international de linguistique latine, Bruxelles, 4–9€avril 2005), Gérald Purnelle (ed.), 191–201. Liège: Faculté de Philosophie et Lettres. Spevak, Olga. 2007b. L’anaphore, la deixis et l’ordre des constituants en latin. Latomus 66(4): 853–870. Spevak, Olga. 2007c. Review of A.M. Devine & L. D. Stephens, Latin Word Order: Structured Meaning and Information, OUP, 2006. Mnemosyne 60(3): 497–501. Spevak, Olga. 2008. The position of focus constituents in Latin. A comparison between Latin and Czech. Journal of Pragmatics 40(1): 114–126. Spevak, Olga. Forthcoming a. The placement of reflexive pronouns se and sibi in Late Latin. Spevak, Olga. Forthcoming b. Le temps et l’espace dans la correspondance de Cicéron et le concept de ‘cadre’ et de ‘spécification’. Sproat, Richard & Shih, Chilin. 1991. The cross-linguistic distribution of adjective ordering restrictions. In Interdisciplinary Approaches to Language. Essays in honor of S.-Y. Kuroda, Carol Georgopoulos & Roberta Ishihara (eds), 565–593. Dordrecht: Kluwer. Sutter, Marc de, 1986. A theory of word order within the Latin noun phrase, based on Cato’s De agri cultura. In Studies in Latin Literature and Roman History 4, Carl Deroux (ed.), 151– 183, Brussels: Latomus. Szantyr, Anton. 19722. Lateinische Syntax und Stilistik mit dem allgemeinen Teil der lateinischen Grammatik. Munich: C.H. Beck. Sznajder, Lyliane. 1998. Verbes transitifs sans objet en latin. In Estudios de lingüística latina. Actas del IX coloquio internacional de lingüística latina, Universidad autónoma de Madrid, 14–18 de abril de 1997, Benjamín García-Hernández (ed.), 791–808. Madrid: Ed. Clasicás. Torrego, M. Esperanza. 2005.€Grammar and pragmatics. The textual use of repente and subito. In Papers on Grammar IX, 2, Latina Lingua. Proceedings of the Twelfth International Colloquium on Latin Linguistics (Bologna, 9–14€ June 2003), Gualtiero Calboli (ed.), 763–773. Roma: Herder. Toth, Alfred. 1994. Thema, Topik und Koda im Lateinischen. Zu einigen syntaktischen, semantischen und pragmatischen Problemen der lateinischen Grammatik. In Papers on Grammar IV, Gualtiero Calboli (ed.), 177–210. Bologna: CLUEB.



References 

Touratier, Christian. 1987. Valence verbale et intransitivation. In études de linguistique générale et de linguistique latine offertes en hommage à Guy Serbat, 405–416. Paris: Société pour l’information grammaticale. Touratier, Christian. 1991. Transitivité et valence ou syntaxe et sémantique. Stemma 1: 105–113. Touratier, Christian. 1994.€Syntaxe latine. Louvain-la-Neuve: Peeters. Vairel-Carron, Hélène. 1975. Exclamation, ordre et défense: Analyse de deux systèmes syntaxiques en latin. Paris: Les Belles Lettres. Vet, Co. 1981. Subject assignment in the impersonal constructions of French. In Predication and Expression in Functional grammar, A. Machtelt Bolkestein & Henk A. Combé (eds), 143– 163. London: Academic press. Vincent, Nigel. 1988. Latin. In The Romance Languages, Martin Harris & Nigel Vincent (eds), 26–78. London: Croom Helm. Vogel, Wilhelm S. 1938. Zur Stellung von esse bei Caesar und Sallust. Würzburg/Ausmühle: K. Triltsch. Watt, William. S. 1980. Enim Tullianum. The Classical Quarterly 30: 120–123. Wehr, Barbara. 1984. Diskurs-Strategien im Romanischen. Ein Beitrag zur romanischen Syntax. Tubingen: Narr. Weil, Henri. 1844. Question de grammaire générale: De l’ordre des mots dans les langues anciennes comparées aux langues modernes. Paris: Crapelet [1991, Paris, Didier; 1978. The Order of Words in the Ancient Languages Compared with Modern Languages. transl. by A. Scaglione, Amsterdam: J. Benjamins].

Index locorum Abbreviations of ancient authors and works follow the Oxford Latin Dictionary (Oxford, Clarendon Press, 1968). B. Afr. 21.3╇ 229 Caes. Civ. 1.5.5╇ 192 Caes. Civ. 1.7.8╇ 267 Caes. Civ. 1.12.2╇ 141 Caes. Civ. 1.13.4╇ 147, 149 Caes. Civ. 1.13.5╇ 121 Caes. Civ. 1.15.3╇ 121 Caes. Civ. 1.15.6╇ 169 Caes. Civ. 1.24.3╇ 176 Caes. Civ. 1.25.1╇ 168 Caes. Civ. 1.26.1╇ 236 Caes. Civ. 1.33.3╇ 148 Caes. Civ. 1.41.2╇ 252 Caes. Civ. 1.41.6╇ 142 Caes. Civ. 1.42.4╇ 250 Caes. Civ. 1.44.4╇ 170 Caes. Civ. 1.45.4╇ 190 Caes. Civ. 1.50.1╇ 179 Caes. Civ. 1.57.1╇ 184 Caes. Civ. 1.57.3╇ 152 Caes. Civ. 1.65.3╇ 128 Caes. Civ. 1.67.6╇ 178, 278 Caes. Civ. 1.71.2╇ 176 Caes. Civ. 1.72.4╇ 148 Caes. Civ. 1.76.2╇ 178 Caes. Civ. 1.83.3╇ 149 Caes. Civ. 1.84.2╇ 148 Caes. Civ. 1.85.1╇ 162 Caes. Civ. 2.3.3╇ 125 Caes. Civ. 2.4.5╇ 169 Caes. Civ. 2.5.1╇ 104 Caes. Civ. 2.10.2╇ 147 Caes. Civ. 2.10.3╇ 106 Caes. Civ. 2.11.2╇ 120 Caes. Civ. 2.21.4╇ 168 Caes. Civ. 2.21.5╇ 167 Caes. Civ. 2.22.2╇ 156 Caes. Civ. 2.23.4╇ 171 Caes. Civ. 2.24.4╇ 192 Caes. Civ. 2.34.1╇ 192 Caes. Civ. 2.34.6╇ 128, 171

Caes. Civ. 2.41.3╇ 179 Caes. Civ. 2.44.3╇ 171 Caes. Civ. 3.1.1╇ 148 Caes. Civ. 3.9.2╇ 184 Caes. Civ. 3.9.8╇ 178 Caes. Civ. 3.11.2╇ 193 Caes. Civ. 3.11.3╇ 144 Caes. Civ. 3.12.1╇ 169 Caes. Civ. 3.12.3╇ 176 Caes. Civ. 3.18.2╇ 280 Caes. Civ. 3.24.1╇ 193 Caes. Civ. 3.29.1╇ 105, 169 Caes. Civ. 3.33.2╇ 156 Caes. Civ. 3.48.1╇ 153 Caes. Civ. 3.53.1╇ 154 Caes. Civ. 3.71.1╇ 247 Caes. Civ. 3.93.7╇ 154 Caes. Civ. 3.99.2╇ 152 Caes. Civ. 3.99.5╇ 154 Caes. Gal. 1.1.1╇ 247 Caes. Gal. 1.7.1╇ 109 Caes. Gal. 1.9.1╇ 173 Caes. Gal. 1.10.3╇ 260 Caes. Gal. 1.12.4╇ 247 Caes. Gal. 1.13.2╇ 132 Caes. Gal. 1.15.1╇ 245 Caes. Gal. 1.18.3╇ 270 Caes. Gal. 1.22.3╇ 252 Caes. Gal. 1.26.5╇ 250 Caes. Gal. 1.30.1╇ 271 Caes. Gal. 1.38.1╇ 270 Caes. Gal. 1.41.1╇ 131 Caes. Gal. 1.42.6╇ 248 Caes. Gal. 1.44.3╇ 260 Caes. Gal. 1.47.1╇ 137 Caes. Gal. 1.50.1╇ 252 Caes. Gal. 2.9.1╇ 189 Caes. Gal. 2.10.4╇ 271 Caes. Gal. 2.11.3╇ 248 Caes. Gal. 2.14.1╇ 129 Caes. Gal. 2.23.4╇ 248

Caes. Gal. 2.25.1╇ 268 Caes. Gal. 2.31.1╇ 138 Caes. Gal. 2.33.6╇ 240 Caes. Gal. 2.35.1╇ 277 Caes. Gal. 3.18.6╇ 271 Caes. Gal. 3.29.1╇ 260 Caes. Gal. 4.37.2╇ 277 Caes. Gal. 5.3.5╇ 137 Caes. Gal. 5.12.1╇ 269 Caes. Gal. 5.24.5╇ 240 Caes. Gal. 5.27.1╇ 244 Caes. Gal. 6.7.6╇ 172 Caes. Gal. 6.9.6╇ 137 Caes. Gal. 6.29.2╇ 269 Caes. Gal. 6.32.1╇ 138 Caes. Gal. 7.41.1╇ 128 Caes. Gal. 7.57.2╇ 276 Caes. Gal. 7.60.4╇ 248 Cato Agr. 2.5╇ 227 Cato Agr. 5.3╇ 108 Cels. 3.11.2╇ 226 Cels. 3.23.2╇ 226 Cels. 4.26╇ 234 Cic. Ac. 1.12╇ 220 Cic. ad Brut. 2.3.4╇ 262 Cic. Agr. 2.48╇ 216 Cic. Arch. 11╇ 214 Cic. Att. 1.3.2╇ 109, 243 Cic. Att. 1.5.6╇ 109 Cic. Att. 1.9.1╇ 264 Cic. Att. 1.12.4╇ 219 Cic. Att. 1.13.5╇ 253 Cic. Att. 1.16.10╇ 172 Cic. Att. 1.17.3╇ 268 Cic. Att. 1.18.1╇ 264 Cic. Att. 1.18.3╇ 223 Cic. Att. 1.18.4╇ 188, 243, 254 Cic. Att. 1.18.6╇ 278 Cic. Att. 1.18.7╇ 250 Cic. Att. 1.19.4╇ 241 Cic. Att. 2.1.11╇ 253

 Constituent Order in Classical Latin Prose Cic. Att. 2.4.3╇ 110 Cic. Att. 2.6.2╇ 209 Cic. Att. 2.7.1╇ 277 Cic. Att. 2.11.1╇ 265 Cic. Att. 2.16.3╇ 241 Cic. Att. 2.20.3╇ 245 Cic. Att. 2.24.2╇ 241 Cic. Att. 3.5╇ 127 Cic. Att. 3.6╇ 113 Cic. Att. 3.7.2╇ 245 Cic. Att. 3.7.3╇ 261 Cic. Att. 3.9.1╇ 242, 279 Cic. Att. 3.11.2╇ 210 Cic. Att. 3.13.2╇ 209 Cic. Att. 3.15.7╇ 261 Cic. Att. 3.20.2╇ 108 Cic. Att. 3.23.4╇ 243 Cic. Att. 4.1.3╇ 265 Cic. Att. 4.1.7╇ 264 Cic. Att. 4.2.4╇ 245 Cic. Att. 4.4a.2╇ 260 Cic. Att. 4.5.1╇ 247 Cic. Att. 4.15.7╇ 220 Cic. Att. 4.16.3╇ 242 Cic. Att. 4.18.2╇ 277 Cic. Att. 5.3.3╇ 130 Cic. Att. 5.20.9╇ 279 Cic. Att. 6.1.7╇ 144 Cic. Att. 6.1.13╇ 129 Cic. Att. 7.1.2╇ 212 Cic. Att. 7.3.4╇ 276 Cic. Att. 10.10.1╇ 140 Cic. Att. 11.6.2╇ 220 Cic. Att. 11.17.1╇ 144 Cic. Att. 12.49.3╇ 139 Cic. Att. 13.42.1╇ 164 Cic. Att. 13.44.3╇ 218 Cic. Att. 13.52.1╇ 110 Cic. Att. 14.1.1╇ 161 Cic. Att. 14.10.2╇ 199 Cic. Att. 14.11.2╇ 114 Cic. Att. 14.13.2╇ 268 Cic. Att. 14.13.6╇ 111 Cic. Att. 14.16.2╇ 227 Cic. Att. 14.18.2╇ 158 Cic. Att. 14.20.1╇ 249 Cic. Att. 15.1.2╇ 110, 193 Cic. Att. 15.1.4╇ 183 Cic. Att. 15.1a.1╇ 167 Cic. Att. 15.4.1╇ 176 Cic. Att. 15.4.2╇ 152, 158, 183 Cic. Att. 15.4.3╇ 211 Cic. Att. 15.4.5╇ 135 Cic. Att. 15.5.1╇ 141

Cic. Att. 15.6.2╇ 163 Cic. Att. 15.11.1╇ 179 Cic. Att. 15.11.4╇ 149 Cic. Att. 15.13.1╇ 140 Cic. Att. 15.13a.1╇ 123 Cic. Att. 15.15.2╇ 160 Cic. Att. 15.17.2╇ 159 Cic. Att. 15.19.2╇ 210 Cic. Att. 15.24╇ 135, 160 Cic. Att. 15.25╇ 158, 184 Cic. Att. 15.29.1╇ 140 Cic. Att. 15.29.2╇ 160, 161, 162 Cic. Att. 16.1.6╇ 161 Cic. Att. 16.2.1╇ 111 Cic. Att. 16.4.2╇ 161 Cic. Att. 16.4.4╇ 190 Cic. Att. 16.6.1╇ 190 Cic. Att. 16.6.2╇ 221 Cic. Att. 16.6.4╇ 211 Cic. Att. 16.10.2╇ 123 Cic. Att. 16.13.1╇ 177 Cic. Att. 16.15.3╇ 164 Cic. Cael. 12╇ 112 Cic. Cat. 1.6╇ 261 Cic. Cat. 1.23╇ 207 Cic. Clu. 4╇ 111 Cic. Clu. 62╇ 214 Cic. Clu. 89╇ 278 Cic. Clu. 149╇ 261 Cic. Clu. 185╇ 211 Cic. De orat. 1.163╇ 221 Cic. De orat. 2.41╇ 257 Cic. De orat. 2.280╇ 218 Cic. Div. Caec. 23╇ 220 Cic. Div. Caec. 54╇ 213 Cic. Dom. 11╇ 153 Cic. Dom. 34╇ 197 Cic. Dom. 39╇ 202 Cic. Dom. 47╇ 218 Cic. Dom. 53╇ 199 Cic. Dom. 65╇ 148 Cic. Dom. 75╇ 197 Cic. Dom. 93╇ 203 Cic. Dom. 102╇ 147 Cic. Dom. 112╇ 144 Cic. Dom. 115╇ 211 Cic. Dom. 116╇ 182, 186, 203 Cic. Dom. 117╇ 197 Cic. Dom. 119╇ 152 Cic. Dom. 123╇ 200 Cic. Dom. 127╇ 164, 200, 201 Cic. Dom. 130╇ 105 Cic. Dom. 132╇ 197 Cic. Fam. 2.7.2╇ 221

Cic. Fam. 2.8.2╇ 217, 219 Cic. Fam. 3.1.1╇ 262 Cic. Fam. 3.8.2╇ 257 Cic. Fam. 6.10b.1╇ 270 Cic. Fam. 7.24.2╇ 207 Cic. Fam. 12.4.2╇ 123 Cic. Fam. 14.1.2╇ 130 Cic. Fam. 14.17╇ 209 Cic. Fam. 15.4.5╇ 130 Cic. Fin. 2.56╇ 263 Cic. Flac. 10╇ 217 Cic. Flac. 73╇ 217 Cic. Font. 49╇ 214 Cic. Har. 30╇ 263 Cic. Inv. 2.118╇ 219 Cic. Leg. 2.37╇ 264 Cic. Lig. 8╇ 263 Cic. Lig. 16╇ 208 Cic. Man. 60╇ 213 Cic. Mil. 46╇ 114 Cic. Mil. 77╇ 56 Cic. Mil. 95╇ 262 Cic. Mur. 13╇ 187 Cic. Mur. 90╇ 217 Cic. Off. 2.23╇ 274 Cic. Off. 2.57╇ 156 Cic. Off. 2.72╇ 235 Cic. Phil. 1.14╇ 200 Cic. Phil. 1.15╇ 105 Cic. Phil. 1.25╇ 163, 196 Cic. Phil. 1.29╇ 186 Cic. Phil. 1.31╇ 164 Cic. Phil. 2.5╇ 156 Cic. Phil. 2.11╇ 251 Cic. Phil. 2.12╇ 251 Cic. Phil. 2.19╇ 214 Cic. Phil. 2.30╇ 212 Cic. Phil. 2.43╇ 271 Cic. Phil. 2.63╇ 273 Cic. Phil. 2.97╇ 153 Cic. Phil. 4.7╇ 201 Cic. Phil. 5.27╇ 227 Cic. Phil. 7.26╇ 278 Cic. Phil. 11.13╇ 229 Cic. Phil. 11.23╇ 213 Cic. Phil. 11.24╇ 220 Cic. Phil. 11.37╇ 209 Cic. Phil. 13.30╇ 212 Cic. Phil. 14.3╇ 215 Cic. Pis. 3╇ 251 Cic. Pis. 68╇ 276 Cic. Q. fr. 1.3.1╇ 261 Cic. Q. fr. 2.1.3╇ 129 Cic. Q. Rosc. 37╇ 216



Index locorum  Cic. Quinct. 79╇ 211 Cic. Red. sen. 30╇ 131 Cic. Rep. 2.22╇ 155 Cic. Sest. 124╇ 274 Cic. S. Rosc. 18╇ 113 Cic. Tusc. 1.4╇ 179 Cic. Tusc. 1.10╇ 202 Cic. Tusc. 1.25╇ 199 Cic. Tusc. 1.28╇ 201 Cic. Tusc. 1.29╇ 196 Cic. Tusc. 1.52╇ 162, 185 Cic. Tusc. 1.76╇ 178 Cic. Tusc. 1.86╇ 155, 203 Cic. Tusc. 1.99╇ 159 Cic. Tusc. 1.102╇ 201 Cic. Tusc. 1.105╇ 125 Cic. Tusc. 2.29╇ 211 Cic. Tusc. 2.64╇ 214 Cic. Tusc. 3.2╇ 143 Cic. Tusc. 3.6╇ 190 Cic. Tusc. 3.8╇ 200 Cic. Tusc. 3.9╇ 187 Cic. Tusc. 3.15╇ 185 Cic. Tusc. 3.41╇ 204 Cic. Tusc. 3.48╇ 124 Cic. Tusc. 3.49╇ 161 Cic. Tusc. 3.66╇ 202 Cic. Tusc. 5.1╇ 130 Cic. Tusc. 5.61╇ 234 Cic. Ver. 1.6╇ 273 Cic. Ver. 1.61╇ 238 Cic. Ver. 1.94╇ 218 Cic. Ver. 1.128╇ 209 Cic. Ver. 1.143╇ 208 Cic. Ver. 2.13╇ 231 Cic. Ver. 2.70╇ 241 Cic. Ver. 2.76╇ 215 Cic. Ver. 2.97╇ 254 Cic. Ver. 2.122╇ 272 Cic. Ver. 2.183╇ 214 Cic. Ver. 4.62╇ 237 Cic. Ver. 4.71╇ 268 Cic. Ver. 4.93╇ 236

Cic. Ver. 4.133╇ 220 Cic. Ver. 5.19╇ 219 Cic. Ver. 5.136╇ 236 Col. 7.6╇ 238 Col. 9.13.8╇ 234 Liv. 1.3.3–4╇ 228 Liv. 1.49.1╇ 116 Liv. 5.4.13╇ 279 Liv. 7.1.8╇ 259 Liv. 8.17.1╇ 226 Liv. 8.20.3╇ 226 Liv. 23.11.7╇ 223 Liv. 27.22.1╇ 227 Liv. 27.29.8╇ 238 Liv. 28.17.5╇ 229 Liv. 28.28.6╇ 235 Liv. 30.14.1╇ 268 Liv. 30.33.14╇ 254 Liv. 35.51.4╇ 231 Liv. 37.14.6╇ 238 Liv. 37.24.11╇ 235 Liv. 37.46.3╇ 233 Liv. 39.27.10╇ 236 Liv. 39.33.4╇ 229 Liv. 40.37.7╇ 259 Nep. Att. 2.3╇ 279 Nep. Pel. 2.3╇ 225 Pl. As. 664–665╇ 206 Pl. As. 685╇ 206 Pl. As. 689╇ 206 Pl. As. 692╇ 206 Pl. As. 711╇ 206 Pl. Curc. 62╇ 5 Pl. Mos. 603╇ 205 Pl. Poen. 1069╇ 109 Pl. Ps. 1077╇ 195 Pl. Ps. 1222–1223╇ 112 Plin. Ep. 8.12.4╇ 238 Plin. Nat. 11.92–94╇ 258 Plin. Nat. 25.113╇ 239 Plin. Nat. 27.62╇ 235 Plin. Nat. 37.90╇ 236 Plin. Nat. 37.134–135╇ 258

Rhet. Her. 2.16.24╇ 263 Sal. Cat. 16.5╇ 245 Sal. Cat. 61.5╇ 253 Sal. Jug. 1.3╇ 186 Sal. Jug. 8.1╇ 192 Sal. Jug. 10.8╇ 215 Sal. Jug. 11.2╇ 177 Sal. Jug. 13.1╇ 117, 124 Sal. Jug. 13.6╇ 245 Sal. Jug. 19.4╇ 183 Sal. Jug. 24.2╇ 253 Sal. Jug. 24.10╇ 215 Sal. Jug. 25.5╇ 135 Sal. Jug. 31.27╇ 122 Sal. Jug. 32.5╇ 183 Sal. Jug. 38.6╇ 272 Sal. Jug. 42.1╇ 122 Sal. Jug. 47.1╇ 192 Sal. Jug. 48.3╇ 178 Sal. Jug. 52.3╇ 175 Sal. Jug. 52.6 – 53.1╇ 170 Sal. Jug. 53.1╇ 120 Sal. Jug. 54.3╇ 122 Sal. Jug. 54.6╇ 245 Sal. Jug. 57.3╇ 106 Sal. Jug. 58.1╇ 121 Sal. Jug. 60.6╇ 121 Sal. Jug. 61.4 – 62.1╇ 120 Sal. Jug. 65.3╇ 117 Sal. Jug. 68.2╇ 243, 249 Sal. Jug. 74.2╇ 178 Sal. Jug. 85.18╇ 253 Sal. Jug. 86.5╇ 149 Sal. Jug. 91.3╇ 167 Sal. Jug. 94.3╇ 167 Sal. Jug. 106.2╇ 175 Sal. Jug. 106.6╇ 124 Suet. Nero 31.1╇ 234 Ter. Ad. 670╇ 198 Var. L. 5.41.2╇ 228 Vitr. 7.9.3╇ 234

Index rerum ablative absolute╇ 10, 67-69, 83, 87 AcI (accusative plus infinitive clause)╇ 55, 117, 157-162, 219 agent╇ 9-10, 116-117, 232 colon (syntactic unit)╇ 10-11, 170-171, 277 communicative dynamism╇ 4-6, 29-31, 132, 195, 206, 283 complex information╇ 38-39, 40, 44-45, 49, 53, 58, 60, 68, 80-82, 85, 121, 164, 169-170, 171, connector╇ 64, 67-68, 74, 135, 164, 191 coordination of adjectives╇ 224, 229, 233, 237-239, 280 of clauses╇ 77, 79, 97, 100, 103 of constituents╇ 17, 103-105, 126, 131 of determiners╇ 253, 272 of Focus constituents╇ 39-40, 76, 91, 141, 154 of prepositional phrases╇ 1819 zero coordination╇ 40, 214 coordinators╇ 13, 15, 17, 19, 52, 100, 171, 284 copula╇ 152, 180-183, 185-187, 306, 311 co-reference╇ 34, 98, 171, 251-252 definite (reading), definiteness╇ 9, 30, 69, 117, 123, 140, 166, 173, 180, 183, 187-188, 282, denominal adjectives╇ 231-233, 237 determinative adjectives╇ 224225, 230, 266-267 discontinuity fixed formulas╇ 17-18, 229 noun (and prepositional) phrases╇ 22, 23-26, 28, 39, 48, 82, 95, 129, 139, 143, 146, 164, 177, 180, 188-189, 218, 220, 225, 228, 240, 242, 243, 244, 250, 253, 272, 280-281, 285

proper names╇ 17-18, 21-22, 25 question-words╇ 196, 198 produced by alien elements╇ 24-25, 274-279 enclitics╇ 17-18, 239 focusing particles╇ 20-22, 51-53, 239 genitive complements╇ 2526, 272-279 discontinuous analytic verb forms╇ 150, 153, 156 Discourse Topic (entity established in the discourse)╇ 6-7, 29, 34, 37, 56, 64-73, 76, 80, 90, 98, 101, 103, 109, 141, 161, 284 distinctive value of adjectives╇ 226-227, 232, 246-248, 267, 269 domain integrity╇ 10, 22-23, 159, 274 enclitics╇ 2, 16-19, 23, 25-26, 239, 275, 284 existential (verb)╇ 172, 180-181, 184, 187-190, 193, 284 final position contextually dependent constituents╇ 31, 113, 206 emphatic constituents╇ 47, 130 focal constituents╇ 6, 29-31, 58, 193, 215, 217, 221 interrogative pronouns and particles╇ 15, 198 subjects╇ 112-113, 123, 176-177 final verbs in declarative sentences╇ 2, 5-6, 97, 150-151, 154, 156, 160, 179, 182, 189, 191 in imperative sentences╇ 206207, 212, 219, 221-222 in interrogative sentences╇ 195 fixed expressions╇ 17-18, 228-229, 249, 266 Focus-first strategy╇ 41-44, 124, 188, 193

focusing particles╇ 8, 20-22, 39, 49-51, 52-53, 284 frame noun phrase╇ 26, 272-273, 278, 280 prepositional phrase╇ 268 verb and its complement╇ 121, 171, 313 Future Topic (new discourse participant)╇ 6-7, 56-60, 62-63, 77, 79, 80, 87, 120, 243 head proximity╇ 22-23, 278 hierarchy╇ 8-10 agent > patient╇ 9, 116-117 given > new╇ 9, 29, 116, 123, 133 ordering of adjectives╇ 229236, 238, 280 ordering of determiners╇ 254 hyperbaton see discontinuity iconic, iconicity╇ 9-10, 274 indefinite (reading), indefiniteness╇ 9, 29-30, 42, 57, 102, 117, 139-140, 167, 173, 180, 185, 187-188, 283 infinitive clause╇ 8, 10, 22, 82, 102, 109-110, 157-162, 211 initial position╇ 13-16 anaphoric pronouns╇ 61-63, 68, 74-77, 80, 82, 86, 91, 116, 134 cataphoric pronouns╇ 211, 214 connecting relative qui╇ 13, 14, 69, 87 connectors╇ 13, 15, 53 contextually dependent constituents╇ 34, 54, 72, 124, 140, 177, 180, 269 contextually independent constituents╇ 146-147, 180, 188, 193 contrastive constituents╇ 14, 46, 93, 95, 109, 123, 202, 217, 273 coordinators╇ 13

 Constituent order in Classical Latin prose emphatic constituents╇ 47, 53, 55, 93, 95, 129, 130, 144, 182, 274 focal constituents╇ 14, 31, 39, 41-42, 44, 124, 161, 182 focusing particles╇ 51, 212 interrogative words and particles╇ 13, 15, 19, 195-196, 198-199, 201, 203, 204 pronouns╇ 32, 93-94, 171, 221-222 setting╇ 70-71 subjects╇ 43, 44, 70, 147, 176177, 182, 193, 219 Topics╇ 28, 43, 46, 58, 64-65, 68, 70, 74, 90, 148, 192, 214, 219, 221 initial verbs in declarative sentences╇ 2, 30-31, 46-47, 54, 73, 129, 139, 149, 151-152, 155-156, 162, 172, 176, 180-181, 184, 188-193 auxiliary verb╇ 150-152, 155, 157 copula╇ 180, 182-185, 187 in imperative sentences╇ 205208, 216, 221-222 in interrogative sentences╇ 202-204 internal position╇ 2, 24, 277 anaphoric pronouns╇ 74-77, 80-82, 86, 91 personal pronouns╇ 32, 95-96 lexical units╇ 17, 227, 229 LIPOC╇ 8, 122, 162 parenthesis╇ 111, 158, 164, 220, 276 participial clause╇ 8, 10, 22, 57, 67, 81, 86, 88, 135, 223, 240 patient╇ 9-10, 116-117 permutation (non-hierarchical order of adjectives)╇ 237, 263, 272 post-nominal placement of adjectives╇ 225, 227, 229, 230-231, 234, 236-237, 280, 285 of determiners╇ 239-252, 261, 263-265, 280

of genitive complements╇ 266-269 postpositive particles╇ 20-21, 23, 26, 52, 239, 275, 284 post-verbal position directional complements╇ 165-167 objects╇ 118-119, 122-123, 133, 134, 143 subjects╇ 72, 148, 173, 177 pragmatic unit╇ 38, 40, 44-45, 81, 120-121, 125-126, 128-131, 138,141, 143-144, 169-170, 199, 285 pragmatically marked, pertinent╇ 28, 81, 225, 234, 237, 281 pragmatically unmarked, insignificant╇ 32, 47, 96, 180, 192, 225 pre-nominal placement of adjectives╇ 224-225, 227228, 230, 234, 236-237, 266, 280-281, 285 of determiners╇ 239-240, 242, 244-246, 248-251, 254-255, 261, 265, 280 of genitive complements╇ 266-270 prepositional phrase╇ 5, 10, 1623, 25-26, 59, 67-68, 107, 10911, 128-129, 136-139, 142,181, 223, 229, 238-239, 267-268, 272, 275-276, 280, 284 presentative sentences╇ 6, 39, 4144, 57-58, 60, 62, 145, 152, 157, 173, 175, 180, 188-189, 243, 285 preverbal position contextually independent constituents╇ 143 directional complements╇ 165-166, 168, 171 focal constituents╇ 6, 29, 31, 58, 91, 277 objects╇ 119-122, 125, 132-135, 137-138, 143-144 subjects╇ 124, 146, 177, 180 proper names adjectives derived from╇ 227, 232

discontinuity╇ 17, 21, 25 with ille╇ 241 with quidam╇ 188, 243 prosody╇ 8, 28, 49, 107, 112113,195, 279 qualifying adjectives╇ 37, 224225, 230 referential chains╇ 56, 61-63, 98, 101 referential properties╇ 13, 67, 74 reminder (of the subject)╇ 73, 149, 177 rheme╇ 4-6, 132, 142, 195-196, 205 Sentence Topic (point of departure of a sentence)╇ 14-15, 56, 65-68, 72, 76-77, 90-91, 94, 116, 122, 148, 162, 197, 215, 276, 284 setting╇ 5, 28, 30, 41, 58, 67, 68-72, 90, 135, 197, 199, 204, 240, 284 spatial╇ 37, 66, 170 temporal╇ 30, 66, 88, 113, 171, 175, 227, 248-249 specification╇ 111, 133, 168, 248-250 Sub-Topic (deducible entity selected as Topic)╇ 6-7, 56, 64-65, 148, 197, 199 stylistics╇ 11, 23, 103, 274, 285 subordinators╇ 13-14, 16, 71, 162, 164, 197, 284 theme (Firbas)╇ 4-5, 29, 132, 142, 195, 205 topicalisation (selection of a constituent as Topic)╇ 61, 68, 75-87, 90-92, 101-102, 105, 114, 165, 284 transitivity╇ 115, 117-118 high degree of transitivity╇ 117-118, 125 low degree of transitivity╇ 125 verum Focus╇ 46-47, 73, 128, 178, 181, 190, 193 weak semantic value (verb)╇ 125127, 147, 178 what-happens sentences╇ 36, 39, 41-44, 48, 147 zero anaphora╇ 7, 33-34, 60-63, 98, 100, 102-105

appendix

Three commented texts In this appendix, I add three integral paragraphs analysed from a pragmatic point of view:1 Sallust’s The Conspiracy of Catiline, Cicero’s speech Against Verres and a letter from Cicero to Atticus.

1. Sallust, The Conspiracy of Catiline (55.1–6) Text: 1Postquam, ut dixi, senatus in Catonis sententiam discessit, consul optumum factu ratus noctem, quae instabat, antecapere, ne quid eo spatio novaretur, tresviros, quae supplicium postulabat, parare iubet. 2Ipse praesidiis dispositis Lentulum in carcerem deducit; idem fit ceteris per praetores. 3Est in carcere locus, quod Tullianum appellatur, ubi paululum ascenderis ad laevam, circiter duodecim pedes humi depressus. 4Eum muniunt undique parietes atque insuper camera lapideis fornicibus iuncta; 5sed incultu, tenebris, odore foeda atque terribilis eius facies est. 6In eum locum postquam demissus est Lentulus, vindices rerum capitalium, quibus praeceptum erat, laqueo gulam fregere. 7Ita ille patricius ex gente clarissuma Corneliorum, qui consulare imperium Romae habuerat, dignum moribus factisque suis exitium [vitae] invenit. 8De Cethego, Statilio, Gabinio, Caepario eodem modo supplicium sumptum est. Translation: ‘1After the Senate (as I have said) had divided in favour of Cato’s proposal, the consul, deeming that the best thing to do was to forestall the impending night to prevent any revolutionary move during the course of it, ordered the triumvirs to prepare what the punishment demanded. 2After the deployment of guards, he personally escorted Lentulus to the gaol; the same was done for the others by the praetors. 3There is in the gaol a place called the Tullianum, on the left when you have gone up a short way, and sunk into the ground about twelve feet. 4It is fortified on all sides by walls, and the roof above is spanned by stone vaulting; 5but neglect, gloom and stench give it a foul and terrifying appearance. 6That was the place to which Lentulus was taken down, whereupon the executioners of capital cases, whose commission it was, broke his neck with a noose. 7So it was that a patrician from the most distinguished clan of the Cornelii, 1. In this book, attention has been paid to subordinate clauses only in an indirect way. Their analysis is included in this appendix.

 Constituent order in Classical Latin prose

who had held consular command at Rome, met the end which his behaviour and deeds deserved. 8On Cethegus, Statilius, Gabinius, and Caeparius punishment was exacted in the same way.’ Previous context: The Senate decided that the conspirators must be executed, as had been proposed by Cato. After a comparison between Cato and Caesar (the proponent of a milder treatment of the conspirators), Sallust returns to the main narrative line. The consul, M. Tullius Cicero, was responsible for the safety of the Republic and it was his duty to act. Analysis: 1 Postquam, ut dixi, senatus in Catonis sententiam discessit, consul optumum factu ratus noctem, quae instabat, antecapere, ne quid eo spatio novaretur, tresviros, quae supplicium postulabat, parare iubet. The section starts with a reminder of the situation, formulated in a temporal clause (postquam... discessit) that represents the temporal setting of the whole complex sentence. The author repeats whose proposal was accepted by the Senate: Cato’s. Catonis in the prepositional phrase (in Catonis sententiam) stands before its head noun because it is contextually given information (see Cat. 53.1). Consul is Discourse Topic of this (and the following) sentence and comes immediately after the temporal clause. The consul acts on the Senate’s decision (iconicity). The section optumum... novaretur expresses the motivation of the consul’s action; it is a new piece of information but not the most important one. Optumum stands first in its participial domain (optumum... antecapere) because it is emphatic; it is followed by the supine factu that it governs. The most informative part of this participial clause is noctem (followed by its relative clause) antecapere; a final clause (ne quid... novaretur) explains why. Then comes the most informative part of the sentence, represented by the pragmatic unit tresviros2... parare iubet. The object clause depending on parare precedes its governing verb iubet. The information is complex here, and the sentence answers the question ‘what does the consul do?’ 2 Ipse

praesidiis dispositis Lentulum in carcerem deducit; idem fit ceteris per praetores. Another action by the consul is expressed (‘what does the consul do?’): taking the prisoner to the prison, Lentulum in carcerem deducit, functions as a pragmatic unit and conveys the most important information. Ipse, referring to the consul, is a contrastive Topic. Lentulum is a contextually given constituent, but belongs here to the complex information.3 After Catiline’s flight from Rome, Lentulus was the remaining “chief ” of conspirators. Praesidiis dispositis expresses accompanying circumstances and has no particular pragmatic function. In the following clause, idem fit summarizes the previous action, i. e. in carcerem deducit. It is followed by ceteris (with a reference deducible 2. Tresviros capitales are meant, minor magistrates who had charge of prisons and executions. 3.

Cf. de P. Lentulo ceterisque statuetis (Sal.€Cat. 52.17).



Three commented texts 

from Lentulum), and by per praetores, that contrasts with ipse and is salient (contrastive Focus) in this clause. From a stylistic point of view, ipse, Lentulum, ceteris and per praetores form a chiastic structure. 3 Est

in carcere locus, quod Tullianum appellatur, ubi paululum ascenderis ad laevam, circiter duodecim pedes humi depressus. The author leaves the main narrative line in order to provide a description of the place. The sentence starts with the existential verb est4 (in the present tense, with reference to the moment of the speech act), followed by a locative complement that picks up the phrase in carcerem expressed in the preceding sentence. The locative complement could have come first, if it summarized a salient element of the previous sentence (which is not the case here).5 Locus, expanded by a relative clause providing us with its name, Tullianum, is Focus. This sentence answers the underlying question ‘what?’ The temporal clause (ubi... laevam) as well as the participial one (circiter... depressus) are included to describe the place. Ad laevam in clause-final position can be taken as the most informative element in the subordinate clause.6 The participial clause (circiter... depressus) provides the depth of the place; the complements precede their participle. The numeral quantifier duodecim is used here without any idea of contrast or any other special informative value, and thus precedes its head noun (pedes). The author does not particularly stress the measurements (as they would be in the case of pedes circiter duodecim). 4 Eum

muniunt undique parietes atque insuper camera lapideis fornicibus iuncta. The Sentence Topic eum picks up a previous Focus constituent (locus). Muniunt is a semantically weak verb expressing a state; the salient constituent is parietes, coordinated by atque with camera. The sentence answers the question ‘what?’ Iuncta, expanding camera, is preceded by its complement, lapideis fornicibus, with a pre-nominal adjective lapideis that would normally be expected in post-nominal position. I would 4. As we have seen in Chapter 3, Section 7.3, p. 188 ff., it is quite usual to find the existential verb sum in sentence-initial position. Unlike commentators, I believe there is no need to explain it either as opening a “pathetische Ekphrasis” (K. Vretska, C. Sallustius Crispus, De Catilinae coniuratione, Heidelberg, 1976:€640), or as an “epic beginning” with reference to Homer (I. Mariotti, G. Sallustio Crispo Coniuratio Catilinae, Bologna, 2007:€657). See my examples in the above quoted section; and cf. Erat in Italia eius exercitus (Cic. Sest. 41) ‘His army was in Italy’ in a nondescriptive passage of Cicero’s. 5. In the preceding sentence in carcerem is a part of complex information; it is not a salient constituent, because the sentence does not answer the question ‘where does he lead him?’ 6. If the transmission of the text is good, ubi ascenderis ad laevam expresses nothing else but the idea “upwards to the left”, and could have been rendered by two adverbials. The formulation ascenderis is remarkable, as if Sallust supposed that his reader, a good Roman citizen, might have entered this prison. At the least, the author wants to draw him inside. Commentaries, however, consider it as unclear (among others Vretska 1976:€640, quoted in the note above), because the Tullianum, the death-chamber, was in the lower (and not the upper) part of the prison.

 Constituent order in Classical Latin prose

not exclude the possibility that the anteposition is used here with the idea of an implicit contrast (lapideis as opposed to other material found in the prison, for example ‘wood’ or to a more common type of fornicibus). 5

Sed incultu, tenebris, odore foeda atque terribilis eius facies est. The roof spanned by stone vaulting could imply an impression of an attractive place, but this is not the case, as is signalled by the connector sed (in obligatory initial position). In this sentence, neither eius facies, deducible from the context, nor the copula est is focal. Focus is on the predicative adjectives foeda (and its three preceding complements) and terribilis. They are coordinated (and not juxtaposed) because they are predicative; furthermore, they express subjective evaluation of the physical appearance. Due to the subjective character of the expressed qualifications, this long constituent is put first in its clause. 6 In eum locum postquam demissus est Lentulus, vindices rerum capitalium, quibus prae-

ceptum erat, laqueo gulam fregere. We return to the narrative line. The Sentence Topic in eum locum, referring back once again to the previous salient information, signals that this sentence will inform us about ‘what happened there?’ The temporal clause, postquam demissus est Lentulus, repeats what has already been said. Lentulus coming last in the subordinate clause is not especially informative; its expression is a simple reminder. If this sentence had followed after deducit (or even after praetores) we could simply have had postquam demissus est, but such an ellipsis is not allowed here. Vindices rerum capitalium is a contextually independent entity (note the relative clause quibus praeceptum erat),7 and functions as both semantic agent and syntactic subject of the action. Its genitive complement stands, as is expected for unbound constituents, in postposition; the denominal adjective (capitalium), expressing an inherent property, follows its head noun (rerum). This sentence focuses on the action: laqueo gulam fregere forms a pragmatic unit. In addition, we also note that gulam has no modifier. When expressing an inalienable entity, body parts usually refer to the subject of the sentence, as do reflexive pronouns. Such an interpretation is excluded here for pragmatic reasons and gulam cannot be understood otherwise than gulam eius (Lentuli). The possession could also be expressed by a dative (ei). 7

Ita ille patricius ex gente clarissuma Corneliorum, qui consulare imperium Romae habuerat, dignum moribus factisque suis exitium [vitae] invenit. The particle ita summarizes the previous information about execution of Lentulus. Ille, referring to Lentulus, is a contrastive Topic; this sentence is about him; the information about the others follows in the next sentence. Ille, being a demonstrative modifier,

7. Seemingly, these persons executed the orders of triumvirs. However, one might consider vindices rerum capitalium as deducible from the shared knowledge of author and readers.



Three commented texts 

is usually expected to come before its head, patricius.8 This noun phrase governs a prepositional phrase with ex that qualifies Lentulus. Ex gente has two postnominal modifiers, positioned as is expected for contextually unbound elements: the adjective clarissuma and the genitive Corneliorum. In the relative clause, also providing a qualification of Lentulus, the most informative constituent is consulare imperium, coming first. Due to the focal character of this noun phrase, the denominal adjective comes before its head. All these qualifications would normally be incompatible with being put to death. The Focus of the main clause is the phrase dignum... exitium (question ‘what happened to him?’) that incorporates the modifiers of dignum: moribus factisque suis. Dignum expresses subjective evaluation and thus comes first in this framed complex noun phrase.9 The verb invenit with weak semantic value in sentence-final position is not focal. 8

De Cethego, Statilio, Gabinio, Caepario eodem modo supplicium sumptum est. This sentence starts with a multiple Sub-Topic: de Cethego, Statilio, Gabinio, Caepario, referring to the remaining conspirators (cf. ceteris mentioned above). It tells us what happened to them. However, supplicium sumptum est, in sentence-final position with a nice alliteration is not Focus. We expect their punishment; what we do not know is whether they were treated in different way. This point permits us to determine eodem modo, which expresses the manner, as the salient element.

8. Patricius could also be understood as an apposition. 9. Editors usually suppress vitae considering it a superfluous interpolation (see, for example, P. McGushin, C. Sallustius Crispus, Bellum Catilinae: A Commentary, Leiden, 1977:€276). I would see no difficulty with this word. In any case it follows exitium and is not incorporated into the framed noun phrase. Furthermore, as McGushin (ibid.) signals, this explanatory genitive vitae is commonly used with exitus, to which exitium is an archaic alternative.

 Constituent order in Classical Latin prose

2. Cicero, AgainstVerres (4.86) Text: 1Atque haec – nihil enim praetermittendum de istius impudentia videtur – agebantur in conventu palam de sella ac de loco superiore. 2Erat hiems summa, tempestas, ut ipsum Sopatrum dicere audistis, perfrigida, imber maximus, cum iste imperat lictoribus ut Sopatrum de porticu, in qua ipse sedebat, praecipitem in forum deiciant nudumque constituant. 3Vix erat hoc plane imperatum cum illum spoliatum stipatumque lictoribus videres. 4Omnes id fore putabant ut miser atque innocens virgis caederetur; fefellit hic homines opinio. 5Virgis iste caederet sine causa socium populi Romani atque amicum? 6Non usque eo est improbus; non omnia sunt in uno vitia; numquam fuit crudelis. 7Leniter hominem clementerque accepit. 8Equestres sunt medio in foro Marcellorum statuae, sicut fere ceteris in oppidis Siciliae; 9ex quibus iste C. Marcelli statuam delegit, cuius officia in illam civitatem totamque provinciam recentissima erant et maxima; 10in ea Sopatrum, hominem cum domi nobilem tum summo magistratu praeditum, divaricari ac deligari iubet. Translation: ‘1And all these things (for I do not think that I ought to omit any particular of his impudence) were done openly in the middle of the assembly, while Verres was sitting on his chair of office, in a lofty situation. 2It was the depth of winter; the weather, as you heard Sopater himself state, was bitterly cold; heavy rain was falling; when that fellow orders the lictors to throw Sopater headlong down from the portico on which he himself was sitting, and to strip him naked. 3The command was scarcely out of his mouth, before you might have seen him stripped and surrounded by the lictors. 4All thought that the unhappy and innocent man was going to be scourged. They were mistaken. 5Do you think that Verres would scourge without any reason an ally and friend of the Roman people? 6He is not so wicked. All vices are not to be found in that man; he was never cruel. 7He treated the man with great gentleness and clemency. 8In the middle of the forum there are some equestrian statues of the Marcelli, as there are in most of the other towns of Sicily; 9out of these he selected the statue of Caius Marcellus, whose services to that city and to the whole province were most recent and most important. 10On that statue he orders Sopater, a man of noble birth in his city, and at that very time invested with the chief magistracy, to be placed astride and bound to it.’ Previous context: Verres ordered the people of the city of Tyndaris to take down a statue of Mercury that he wanted to take away. On their refusal, Verres gives the same order to Sopater, the chief magistrate of the city. When he refuses, Verres treats him violently and leaves Tyndaris. Sopater refers the matter to the Senate. The Senate gives no concrete answer but is perplexed by this situation. When Verres comes back, he summons Sopater who persists in his refusal.



Three commented texts 

Analysis: 1 Atque haec – nihil enim praetermittendum de istius impudentia videtur – agebantur in conventu palam de sella ac de loco superiore. The first sentence provides background information about accompanying circumstances. Atque, connecting it with the previous narrative section, obligatorily occupies the initial position. After it comes haec (Sentence Topic) that summarizes preceding events. The verb, agebantur, is semantically weak, which makes it unlikely to be Focus. The most informative part of this sentence is represented by the locative complements (in conventu... superiore); the question underlying this sentence is ‘where?’ They are arranged with the sense going from more general (in conventu) to more specific (de sella ac de loco superiore). Palam placed after in conventu simply reinforces the idea that the events occurred publicly; it could have preceded it as well. The parenthesis, representing the author’s personal comment, starts with an emphatic word (nihil). The enclitic enim is found necessarily in second sentence-position. Nihil together with praetermittendum represents the most informative part of this parenthesis; de istius impudentia is contextually bound, and the verb videtur in sentence-final position has a weak semantic value. 2

Erat hiems summa, tempestas, ut ipsum Sopatrum dicere audistis, perfrigida, imber maximus, cum iste imperat lictoribus ut Sopatrum de porticu, in qua ipse sedebat, praecipitem in forum deiciant nudumque constituant. The description of the situation continues with an indication of the weather. The complex sentence contains a “cum inversum” clause: the first, the syntactically governing clause, expresses a circumstance whereas the second, the formally governed clause, expresses an action. The first clause focuses on the enumerated entities (hiems... maximus) and answers the question ‘what?’ This sentence lacks a Topic, and it is not surprising to find the existential verb erat in initial position.10 However, there is no reason to consider it as emphatic. The nouns hiems, tempestas and imber, bearing Focus function, are each followed by their modifiers functioning as attributes. The author could have treated the attributive adjectives as emphatic (for example, starting the sentence with summa erat hiems), since they convey subjective evaluation. The phrase tempestas perfrigida is interrupted by a parenthesis, which permits the author to stress the adjective (with its intensive prefix per-).11 The clause with cum pushes the narrative forward by expressing an action: what did Verres do next? The pronoun iste marks a Topic change; the main clause focuses on imperat lictoribus, functioning as a pragmatic unit. Lictoribus is a contextually dependent constituent. The order itself is expressed by a completive clause with ut, and informs us what they should do with 10. Here again, commentators talk about “unexpected descriptive movement” (G. Baldo, M. Tulli Ciceronis In C. Verrem actionis secundae liber quartus, De signis, Firenze, 2004:€436). However, we have seen that the previous sentence has already provided background information. 11. For compounds in per- in Cicero’s speeches against Verres, see Baldo (2004:€52) with further references.

 Constituent order in Classical Latin prose

Sopater. Sopatrum... sedebat is an old piece of information. The reference is made by a full noun, in order to avoid ambiguity, but mainly to maintain the contrast between Verres (iste, ipse)12 and Sopater. New information in this clause is conveyed by two coordinated pragmatic units (praecipitem... constituant). 3

Vix erat hoc plane imperatum cum illum spoliatum stipatumque lictoribus videres. This is another complex sentence with a cum inversum. The summarising hoc gives priority to the emphatic vix that starts the sentence. At the same time, vix is the most informative element of this clause. The passive verb has an analytic form (erat imperatum) the components of which are separated, and it is thus discontinuous; but in the case of analytic verb forms, separation of the participle from the auxiliary is not itself a sign of saliency. As for the cum-clause, its non-focal verb videres stands at the end and governs a participial clause with its subject (illum) coming first. Spoliatum stipatumque (with alliteration) are Focus of this clause: the information conveyed is that Verres’s order was executed; lictoribus (contextually bound) only accompanies the participle stipatum. 4 Omnes id fore putabant ut miser atque innocens virgis caederetur; fefellit hic homines opinio.

Omnes is deducible from the context (cf. in conventu palam) and functions as Sub-Topic. The verb putabant is not focal: focus of the main clause is on the cataphoric id that announces an ut-clause containing contextually given elements (miser atque innocens) and virgis caederetur, which is the most informative part of the clause. The next sentence starts with the Focus: fefellit, placed in the prominent initial position; it is a verum Focus expressing an action that is against expectations. The anaphoric hic thus comes after it. Homines is contextually bound, as well as opinio (with a definite reading), deducible from putabant (which does not have, together with fore, a special pragmatic value). 5 Virgis

iste caederet sine causa socium populi Romani atque amicum? A previously salient piece of information is taken up here: virgis caederet, that represents Focus of the direct question. I cannot see any special emphasis on the initial virgis. Both from a grammatical and a pragmatic point of view, the expression of iste is not necessary, but it accompanies well the ironic tone that begins here and continues afterwards. Sine causa... amicum: it was against the law to punish Roman allies without a decision by the tribunal and these constituents thus belong to the commonly shared knowledge. Socium and amicum both have in common the complement populi Romani, that appears in the middle of them. 6

Non usque eo est improbus; non omnia sunt in uno vitia; numquam fuit crudelis. We already know that Verres was improbus (contextually given information); Focus of this sentence is on the negation non usque eo placed with emphasis in initial position. 12. For this strange alternation of iste and ipse, cf. Baldo, quoted in the previous note (2004:€436). From a pragmatic point of view, iste functions as Topic; ipse does not have this function, but simply reinforces the subject.



Three commented texts 

The two following sentences have the same pragmatic value, and focus on non omnia and numquam. The emphatic omnia is separated from its head noun (vitia). 7

Leniter hominem clementerque accepit. Focus of this sentence is on the adverbs of manner, leniter and clementer, that are separated from each other. Leniter stands with emphasis in initial position; clementer is placed in pre-verbal position. The other constituents, hominem and accepit, are deducible from the context. Before telling us what exactly the treatment was, a short descriptive portion comes first. 8

Equestres sunt medio in foro Marcellorum statuae, sicut fere ceteris in oppidis Siciliae. This sentence, with an existential sunt, answers the question ‘what is in the middle of the forum?’ and focuses on equestres Marcellorum statuae (a contextually unbound constituent). Any interpretation of sunt other than existential is excluded here.13 The Focus function of this noun phrase is responsible for its discontinuity as well as for the anteposition of the adjective and of the genitive with respect to the head noun.14 The adjective placed in initial position is not emphatic; it is only focal. Since medio in foro is contextually bound, the sentence could have started with it, or with the existential verb. The prepositional phrase has the preposition between the adjective and the head noun (medio in foro); the anteposition of the adjective does not have any special value, it seems to me. In the last portion of this sentence ceteris comes first in the prepositional phrase (ceteris in oppidis Siciliae). The genitive modifier is placed as expected, in post-nominal position. I would not exclude the possibility that Cicero arranged both locative complements (medio in foro and ceteris in oppidis) in the same way in order to draw a contrast between them. 9

Ex quibus iste C. Marcelli statuam delegit, cuius officia in illam civitatem totamque provinciam recentissima erant et maxima. The connecting relative ex quibus picks up statuae, and cannot be placed elsewhere. The pronoun iste, with a pejorative connotation, refers back to Verres (Discourse Topic). The verb delegit is contextually independent, but the clause focuses on C. Marcelli statuam, which is salient (‘whose statue?’). The repetition of the head noun (statuam) is unavoidable, but as it is a pragmatically marked constituent the genitive modifier is placed before the governing substantive. The relative clause (cuius... maxima) provides background information of the explanatory type. The part officia... provinciam can be considered as contextually bound: a city does not put up a statue for nothing. The demonstrative modifier illam is placed as expected before its head noun, as well as the quantifier totus (without emphasis). The attributive coordinated adjectives, recentissima et maxima, are salient and separated by the copula erant. 13. Equestris is a denominal adjective that cannot be used as predicative. 14. Commentaries, however, leave this expression without making any remark (cf. Baldo, op. cit., p.€437).

 Constituent order in Classical Latin prose 10

In ea Sopatrum, hominem cum domi nobilem tum summo magistratu praeditum, divaricari ac deligari iubet. The anaphoric expression in ea picks up the preceding Focus constituent (C. Marcelli statuam) and represents Sentence Topic. The underlying question is ‘what is Verres going to do with this statue?’ New information is contained in the action: Focus is on the coordinated infinitives divaricari ac deligari. Due to the previous context, iubet is contextually given. The apposition hominem... praeditum adds qualifications of Sopater and repeats roughly what has already been said, in order to establish some contrast between the status of this man and Verres’s treatment of him.



Three commented texts 

3. Cicero, A letter to Atticus (11.10.1) Text: 1Ad meas incredibilis aegritudines aliquid novi accedit ex iis quae de Quintis ad me adferuntur. 2P. Terentius meus necessarius operas in portu et scriptura Asiae pro magistro dedit. 3Is Quintum filium Ephesi vidit VI Id. Dec. eumque studiose propter amicitiam nostram invitavit; 4cumque ex eo de me percontaretur, eum sibi ita dixisse narrabat, se mihi esse inimicissimum volumenque sibi ostendisse orationis quam apud Caesarem contra me esset habiturus. 5Multa a se dicta contra eius amentiam. 6Multa postea Patris simili scelere secum Quintum patrem locutum; 7cuius furorem ex iis epistulis quas ad te misi perspicere potuisti. 8Haec tibi dolori esse certo scio; 9me quidem excruciant et eo magis quod mihi cum illis ne querendi quidem locum futurum puto. Translation: ‘1Reports reaching me about the Quinti have dropped new bitterness into my overflowing cup. 2My friend P. Terentius had a job as managing director of the customs and pasture rents company in Asia. 3He saw young Quintus at Ephesus on the 8th of December, and was at pains to be hospitable because of his friendship with me. 4He says that when he asked about me Quintus told him that he was my bitter enemy and showed him a roll containing a speech which he was going to make against me in Caesar’s presence. 5Terentius says that he dissuaded him from such a senseless proceeding at great length. 6He adds that later on at Patrae Quintus senior spoke much to him in similarly outrageous fashion. 7The letter which I sent you will have enabled you to judge of his unbalanced state of mind. 8I am sure that all this gives you pain. 9They torture me, all the more because I do not expect even to have the opportunity of remonstrating with them.’ Context: After the defeat of Pompey’s forces at Pharsalus, Cicero comes back to Italy. He is overcome by invectives that his brother Quintus and his nephew Quintus direct against him. The letter of Cicero was written on the 19th January 47 B.C. in Brundisium. Publius Terentius Hispo, who reports Quintus’s actions, was a relative of Cicero; this man was the subject of a letter of recommendation from Cicero to the governor of Bithinia (Fam. 13.65). 1

Ad meas incredibilis aegritudines aliquid novi accedit ex iis quae de Quintis ad me adferuntur. At the start of his letter to Atticus, Cicero is reminding him of his distress: ad meas incredibilis aegritudines is contextually bound information and functions here as Topic. This prepositional phrase has pre-nominal modifiers: such an arrangement is explained by the contextually dependent character of this constituent. However, incredibilis expresses subjective evaluation and we can consider it as placed with emphasis just before the substantive. We are expecting now ‘what was added to his distress?’ After aegritudines comes a contextually unbound part of the sentence: aliquid novi is

 Constituent order in Classical Latin prose

the most salient element. The verb accedit is not Focus; it is only introducing it, conforming to its semantic value of ‘presentation’. Ex iis... adferuntur adds an explanation of what this aliquid consisted of. Cicero was regularly informed about what happened (ad me adferuntur); the new piece of information in this clause is de Quintis. 2

P. Terentius meus necessarius operas in portu et scriptura Asiae pro magistro dedit. The news came in an intermediate way, via Publius Terentius. This sentence conveys completely new content; it is a “what-happens” sentence. Publius Terentius, followed by an explicative apposition meus necessarius,15 is contextually unbound (or at least treated as such), as well as the whole action that follows. The verb stands in sentencefinal position, and its object, operas, coming after the appositions, seems to frame the complements of the verb (in portu... magistro). However, Cicero is not talking about Publius’s action (‘what did P. Terentius do’?) but presenting the circumstance that permitted him to obtain information about the Quinti.16 I would thus interpret P. Terentius as Focus of its sentence. It occupies the sentence-initial position for semantic reasons (being the agent) but also for pragmatic reasons. This person will play an important role afterwards. 3 Is

Quintum filium Ephesi vidit VI Id. Dec. eumque studiose propter amicitiam nostram invitavit. The anaphoric is picks up the salient element of the previous sentence, Publius Terentius, and functions as Sentence Topic: something will be said about him. Then the most informative part of the sentence starts. Cicero has mentioned the Quinti, but Quintum filium informs us about which one of them is concerned. The verb vidit is contextually unbound, but is not the most salient constituent. Ephesi and VI Id. Dec. (more than one month before the letter was written) clarify the action expressed by the verb by providing exact information about the place and the time. With respect to the following context, Quintum filium is Focus of this clause; it is picked up by eum (Sentence Topic) in the subsequent clause. This second clause answers the question ‘what did P. Terentius do with him?’ and focuses on the action invitavit. One might consider the adverb studiose as salient together with the verb. Propter amicitiam nostram has no special pragmatic function. The possessive pronoun nostram follows its head noun and does not bear emphasis. It is worth pointing out that in this sentence and in the following ones, we have two Topics; of these, Terentius is the more central. 4

Cumque ex eo de me percontaretur, eum sibi ita dixisse narrabat, se mihi esse inimicissimum volumenque sibi ostendisse orationis quam apud Caesarem contra me esset habiturus.

15. The presence of this apposition – as well as the description of Terentius’s function – signals that Atticus does not know him (well). 16. A formulation such as: P. Terentius Quintum... vidit would probably be insufficient from the informative point of view. Cicero is explaining who P. Terentius is, and how he happened to meet Quintus.



Three commented texts 

Whereas the passage as a whole is about Terentius (Discourse Topic), this sentence is about Quintus (Sentence Topic ex eo). Since Terentius is Discourse Topic, the subject of the verbs is maintained (percontaretur, narrabat) and refers to him. This complex sentence informs us about what Quintus said. The clause with cum simply presents an accompanying circumstance; the clause eum... narrabat is not very informative either because it contains context-deducible elements. In this clause, the verb narrabat follows the AcI clause which it governs. The ordering eum (referring to Quintus) sibi (referring to Terentius) reflects the hierarchy of agent (subject) > patient (object), just as se mihi does afterwards.17 The most salient element is ita, which announces the informative part of the sentence, se... orationis, syntactically dependent on dixisse.18 These two AcI clauses have their own Focus: the attributes inimicissimum and orationis, which are both placed as the last constituent in their clause. In the second AcI clause, all the constituents from volumen to orationis are contextually independent; but in the case of transitive verbs, the objects with such a contextual status are generally the most informative elements, not the verbs themselves. Here, Focus is in particular on the genitive complement of volumen: that is, on orationis, which is expanded by a relative clause. Note also that the genitive orationis is separated from its head noun volumen; this is explained by the fact that it is a pragmatically pertinent constituent. In the relative clause apud Caesarem, contrasted with contra me, functions as Focus. The verb (esset habiturus) is deducible from orationis. 5

Multa a se dicta contra eius amentiam. This clause is a reaction to Quintus’s words made by Terentius (Discourse Topic) and syntactically depends on narrabat. The constituent multa stands with emphasis in initial position; it also conveys the salient information (Focus). Other constituents are contextually dependent: the verb dicta (esse) is deducible from the context, as well as amentiam. I would also envisage a contrast between a se and contra eius amentiam. 6

Multa postea Patris simili scelere secum Quintum patrem locutum. This clause has the same beginning as the preceding one, with an emphatic multa; that is, however, not salient here. This clause contains contextually unbound elements, namely postea, Patris and Quintum patrem,19 and informs us who else talked to Terentius and under what circumstances20. Focus is on Quintum patrem, which represents the subject of the AcI clause and is placed before its verb, locutum; it is also continued afterwards (cuius). The adverb postea prepares us, so to speak, for the arrival of new 17. It is worth adding that the use of reflexive pronouns is regular; they refer to the subject of the governing clause: sibi refers to Terentius and se to Quintus filius (subject of dixisse). 18. In the second AcI clause, the subject of ostendisse is maintained (i. e. eum referring to Quintum filium). 19. I would not consider Quintum patrem as deducible from Quintum filium, nor from de Quintis. From the context, one cannot tell who Terentius collected the information from. 20. Patrae is an Achean town situated in the Gulf of Corinth.

 Constituent order in Classical Latin prose

content. The constituents Patris, simili scelere and secum do not have a special pragmatic function. Secum referring to Terentius, stands just before Quintum patrem (Focus). I would not consider this placement as a sign of contrast, but as due to the preference for putting constituents referring to people close together. 7 Cuius

furorem ex iis epistulis quas ad te misi perspicere potuisti. The author adds a comment of his own. A connecting relative (cuius) picks up the previous informative constituent (Quintum patrem). A modal verb (potuisti), which is not focal here governs an infinitive (perspicere), with a contextually unbound second argument (furorem). I would consider furorem as Focus of its clause, because it answers the underlying question ‘what could you notice?’ However, I cannot exclude another interpretation according to which furorem belongs to shared knowledge between Cicero and Atticus (Atticus already knows that Quintus the elder – Cicero’s brother – was furious). Consequently, Focus would be on the source, ex iis epistulis, and the underlying question would be ‘from where could you notice that he was furious?’ Certainly, such a sentence would not bring a noteworthy piece of information but would permit the author to close the story about Terentius and to connect with Atticus’s perception of the situation. 8

Haec tibi dolori esse certo scio. The anaphoric haec in sentence-initial position (Sentence Topic) summarizes the whole previous content. It is followed by the personal pronoun tibi, referring to Atticus, which contrasts with me in the subsequent sentence. Focus is on dolori. The verb scio, together with its adverb certo, is contextually independent, but does not convey salient information. The pragmatic value of the sentence is responsible for the disposition of constituents: the AcI clause containing Sentence Topic (haec) as well as Focus (dolori) comes before its governing verb (scio) that is pragmatically unimportant. 9

Me quidem excruciant et eo magis quod mihi cum illis ne querendi quidem locum futurum puto. Me is a contrastive Topic (with respect to Atticus), and the contrast is underlined by the focusing postposed particle quidem. Then comes Focus, the verb excruciant (note the semantic gradation of arguments in dolori esse – excruciant). Et eo magis signals that there will be added content; this is expressed by the clause with quod that focuses on querendi, accompanied by the discontinuous focusing particle ne... quidem. Mihi and cum illis (referring to Quinti) are contextually dependent on the situation and the context, respectively. The remaining part of the clause (ne querendi... puto) contains unbound constituents. Querendi forms a noun phrase with locum but only the gerund is salient, as is signalled by the focusing particle ne... quidem.

Studies in Language Companion Series A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 118 Mühleisen, Susanne: Heterogeneity in Word-Formation Patterns. A corpus-based analysis of suffixation with -ee and its productivity in English. 2010. xiii, 245 pp. 117 Spevak, Olga: Constituent Order in Classical Latin Prose. 2010. xv, 318 pp. 116 Nordström, Jackie: Modality and Subordinators. 2010. xvii, 341 pp. 115 Hasko, Victoria and Renee Perelmutter (eds.): New Approaches to Slavic Verbs of Motion. x, 381 pp. + index. Expected March 2010 114 Roby, David Brian: Aspect and the Categorization of States. The case of ser and estar in Spanish. 2009. xiii, 191 pp. 113 Comrie, Bernard, Ray Fabri, Elizabeth Hume, Manwel Mifsud, Thomas Stolz and Martine Vanhove (eds.): Introducing Maltese Linguistics. Selected papers from the 1st International Conference on Maltese Linguistics, Bremen, 18–20 October, 2007. 2009. xi, 422 pp. 112 Dufter, Andreas and Daniel Jacob (eds.): Focus and Background in Romance Languages. 2009. vii, 362 pp. 111 Polguère, Alain and Igor A. Mel’čuk (eds.): Dependency in Linguistic Description. 2009. xxii, 281 pp. 110 Dimmendaal, Gerrit J. (ed.): Coding Participant Marking. Construction types in twelve African languages. 2009. xvi, 389 pp. 109 Narrog, Heiko: Modality in Japanese. The layered structure of the clause and hierarchies of functional categories. 2009. xxii, 277 pp. 108 Barðdal, Jóhanna and Shobhana L. Chelliah (eds.): The Role of Semantic, Pragmatic, and Discourse Factors in the Development of Case. 2009. xx, 432 pp. 107 Butler, Christopher S. and Javier Martín Arista (eds.): Deconstructing Constructions. 2009. xx, 306 pp. 106 Vanhove, Martine (ed.): From Polysemy to Semantic Change. Towards a typology of lexical semantic associations. 2008. xiii, 404 pp. 105 Van Valin, Jr., Robert D. (ed.): Investigations of the Syntax–Semantics–Pragmatics Interface. 2008. xxiv, 484 pp. 104 Mushin, Ilana and Brett Baker (eds.): Discourse and Grammar in Australian Languages. 2008. x, 239 pp. 103 Josephson, Folke and Ingmar Söhrman (eds.): Interdependence of Diachronic and Synchronic Analyses. 2008. viii, 350 pp. 102 Goddard, Cliff (ed.): Cross-Linguistic Semantics. 2008. xvi, 356 pp. 101 Stolz, Thomas, Sonja Kettler, Cornelia Stroh and Aina Urdze: Split Possession. An areallinguistic study of the alienability correlation and related phenomena in the languages of Europe. 2008. x, 546 pp. 100 Ameka, Felix K. and M.E. Kropp Dakubu (eds.): Aspect and Modality in Kwa Languages. 2008. ix, 335 pp. 99 Høeg Müller, Henrik and Alex Klinge (eds.): Essays on Nominal Determination. From morphology to discourse management. 2008. xviii, 369 pp. 98 Fabricius-Hansen, Cathrine and Wiebke Ramm (eds.): 'Subordination' versus 'Coordination' in Sentence and Text. A cross-linguistic perspective. 2008. vi, 359 pp. 97 Dollinger, Stefan: New-Dialect Formation in Canada. Evidence from the English modal auxiliaries. 2008. xxii, 355 pp. 96 Romeo, Nicoletta: Aspect in Burmese. Meaning and function. 2008. xv, 289 pp. 95 O’Connor, Loretta: Motion, Transfer and Transformation. The grammar of change in Lowland Chontal. 2007. xiv, 251 pp. 94 Miestamo, Matti, Kaius Sinnemäki and Fred Karlsson (eds.): Language Complexity. Typology, contact, change. 2008. xiv, 356 pp. 93 Schalley, Andrea C. and Drew Khlentzos (eds.): Mental States. Volume 2: Language and cognitive structure. 2007. x, 362 pp. 92 Schalley, Andrea C. and Drew Khlentzos (eds.): Mental States. Volume 1: Evolution, function, nature. 2007. xii, 304 pp.

91 Filipović, Luna: Talking about Motion. A crosslinguistic investigation of lexicalization patterns. 2007. x, 182 pp. 90 Muysken, Pieter (ed.): From Linguistic Areas to Areal Linguistics. 2008. vii, 293 pp. 89 Stark, Elisabeth, Elisabeth Leiss and Werner Abraham (eds.): Nominal Determination. Typology, context constraints, and historical emergence. 2007. viii, 370 pp. 88 Ramat, Paolo and Elisa Roma (eds.): Europe and the Mediterranean as Linguistic Areas. Convergencies from a historical and typological perspective. 2007. xxvi, 364 pp. 87 Verhoeven, Elisabeth: Experiential Constructions in Yucatec Maya. A typologically based analysis of a functional domain in a Mayan language. 2007. xiv, 380 pp. 86 Schwarz-Friesel, Monika, Manfred Consten and Mareile Knees (eds.): Anaphors in Text. Cognitive, formal and applied approaches to anaphoric reference. 2007. xvi, 282 pp. 85 Butler, Christopher S., Raquel Hidalgo Downing and Julia Lavid (eds.): Functional Perspectives on Grammar and Discourse. In honour of Angela Downing. 2007. xxx, 481 pp. 84 Wanner, Leo (ed.): Selected Lexical and Grammatical Issues in the Meaning–Text Theory. In honour of Igor Mel'čuk. 2007. xviii, 380 pp. 83 Hannay, Mike and Gerard J. Steen (eds.): Structural-Functional Studies in English Grammar. In honour of Lachlan Mackenzie. 2007. vi, 393 pp. 82 Ziegeler, Debra: Interfaces with English Aspect. Diachronic and empirical studies. 2006. xvi, 325 pp. 81 Peeters, Bert (ed.): Semantic Primes and Universal Grammar. Empirical evidence from the Romance languages. 2006. xvi, 374 pp. 80 Birner, Betty J. and Gregory Ward (eds.): Drawing the Boundaries of Meaning. Neo-Gricean studies in pragmatics and semantics in honor of Laurence R. Horn. 2006. xii, 350 pp. 79 Laffut, An: Three-Participant Constructions in English. A functional-cognitive approach to caused relations. 2006. ix, 268 pp. 78 Yamamoto, Mutsumi: Agency and Impersonality. Their Linguistic and Cultural Manifestations. 2006. x, 152 pp. 77 Kulikov, Leonid, Andrej Malchukov and Peter de Swart (eds.): Case, Valency and Transitivity. 2006. xx, 503 pp. 76 Nevalainen, Terttu, Juhani Klemola and Mikko Laitinen (eds.): Types of Variation. Diachronic, dialectal and typological interfaces. 2006. viii, 378 pp. 75 Hole, Daniel, André Meinunger and Werner Abraham (eds.): Datives and Other Cases. Between argument structure and event structure. 2006. viii, 385 pp. 74 Pietrandrea, Paola: Epistemic Modality. Functional properties and the Italian system. 2005. xii, 232 pp. 73 Xiao, Richard and Tony McEnery: Aspect in Mandarin Chinese. A corpus-based study. 2004. x, 305 pp. 72 Frajzyngier, Zygmunt, Adam Hodges and David S. Rood (eds.): Linguistic Diversity and Language Theories. 2005. xii, 432 pp. 71 Dahl, ÃŒsten: The Growth and Maintenance of Linguistic Complexity. 2004. x, 336 pp. 70 Lefebvre, Claire: Issues in the Study of Pidgin and Creole Languages. 2004. xvi, 358 pp. 69 Tanaka, Lidia: Gender, Language and Culture. A study of Japanese television interview discourse. 2004. xvii, 233 pp. 68 Moder, Carol Lynn and Aida Martinovic-Zic (eds.): Discourse Across Languages and Cultures. 2004. vi, 366 pp. 67 Luraghi, Silvia: On the Meaning of Prepositions and Cases. The expression of semantic roles in Ancient Greek. 2003. xii, 366 pp. 66 Nariyama, Shigeko: Ellipsis and Reference Tracking in Japanese. 2003. xvi, 400 pp. 65 Matsumoto, Kazuko: Intonation Units in Japanese Conversation. Syntactic, informational and functional structures. 2003. xviii, 215 pp. 64 Butler, Christopher S.: Structure and Function – A Guide to Three Major Structural-Functional Theories. Part 2: From clause to discourse and beyond. 2003. xiv, 579 pp. 63 Butler, Christopher S.: Structure and Function – A Guide to Three Major Structural-Functional Theories. Part 1: Approaches to the simplex clause. 2003. xx, 573 pp. 62 Field, Fredric: Linguistic Borrowing in Bilingual Contexts. With a foreword by Bernard Comrie. 2002. xviii, 255 pp.