Discourse Particles in Latin: A Study of Nam, Enim, Autem, Vero and at (Amsterdam Studies in Classical Philology, 4) [Illustrated] 9050634478, 9789050634472

Due to their extremely elusive nature, particles have long been treated in a stepmotherly way, in grammars of individual

113 28 36MB

English Pages 416 [417] Year 1995

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Title Page
Copyright Page
Table of Contents
Preface
1. Introduction
1.1 Outline of the study
1.2 Data
Part I: Theory
2. Relations
2.1 Issues in text relations: conjuncts, relations and relators
2.2 Relations: various classifications
2.3 (Logical-)semantic classifications
2.4 Pragmatic classifications
2.4.1 Semantic-pragmatic relations
2.4.2 Speech act relations
2.4.3 Discourse-pragmatic classifications
2.5 Formal and structural classifications
2.5.1 Clause combining
2.5.2 Text structure
2.6 Evaluation of the usefulness of relation taxonomies
3. Relators: approaches to particles
3. 1 Introduction
3.2 Formal and syntactic accounts of particles
3.3 Semantic approaches to particles
3.4 Discourse and conversation approaches to particles
3.5 Evaluation
4. A framework for the description of connective particles
4.1 The descriptive model: coherence, connectives and levels of discourse
4.1.1 The first parameter: levels of discourse
4.1.2 The second parameter: type of relationship
4.1.2.1 Intratextual and extratextual relationships
4.1.2.2 Structural and semantic-functional relationships
4.2 Comments on the analytical framework
4.2.1 The representational level of discourse
4.2.2 The presentational level of discourse
4.2.2.1 Rhetorical discourse relations
4.2.2.2 Types of rhetorical relationships
4.2.2.3 Organization of the text
4.2.3 The interactional level of discourse
4.2.3.1 Interactional relations
4.2.3.2 Conversational structure
4.2.3.3 'Evaluation' in terms of the communicative situation
4.3 Concluding remarks
5. Heuristics
5.1 Components in a description of particles: basic meaning, discourse function, actual use and side-effects
5.2 Levels of discourse and their linguistic manifestations
5.2.1 Distinguishing 'representational' from 'presentational' connective particles
5.2.2 Distinguishing interactional from other particles: the importance of the concept of discourse type
5.2.2.1 Discourse type: the oppositions monologal-dialogal and monological-dialogical
5.2.2.2 Diaphony
5.3 Overview of linguistic clues
Part II: Description
6. Causal relations and causal connectives: conventional accounts of nam and enim in the literature
6.1 Causal relations
6.2 Previous accounts of nam
6.3 Previous accounts of enim
6.4 Interchangeability of nam and enim
7. Nam
7.1 A discourse-pragmatic description of nam
7.2 'Backward-linking' nam
7.2.1 The rhetorical relation 'evidence'
7.2.2 The rhetorical relation 'justification'
7.2.3 The rhetorical relations 'explanation', 'elaboration' and 'background'
7.2.4 The rhetorical relation 'exemplification'
7.2.5 Additional remarks
7.3 So-called 'forward-linking' nam
7.3.1 Possible instances of 'forward-linking' nam
7.3.2 The relation between 'backward-' and 'forward-linking' nam
7.3.3 Nam as a marker of the organization of the text
7.4 The so-called affirmative particle nam
7.5 Conclusions
8. Enim
8.1 A discourse-pragmatic description of enim
8.2 The 'non-connective' (situating) nature of enim
8.2. 1 Syntactic distributional properties
8.2.2 Discourse type
8.3 The interactional nature of enim
8.3.1 Shared information
8.3.2 Co-occurrence with other particles
8.3.3 Evidence from question-answer pairs
8.3.4 Illocutionary force
8.3.5 Metacommunicative comment
8.4 Actual use: pragmatic motivations for the use of enim
8.4.1 Actual use 1: toning down a challengeable utterance
8.4.2 Actual use 2: irony
8.4.3 Actual use 3: empathy
8.5 Conclusions
8.5.1 Interactional consensus as a unifying concept
8.5.2 Limitations on the interchangeability of nam and enim
9. Adversative relations and adversative connectives: conventional accounts of autem, vero and at
9. 1 Adversative relations
9.2 Previous accounts of autem, vero and at
9.2.1 Previous accounts of autem
9.2.2 Previous accounts of vero
9.2.3 Previous accounts of at
10. Autem
10.1 The discourse function of autem
10.2 The local use of autem: autem as a focusing device
10.2.1 Autem as a marker of parallel focus
10.2.1.1 Parallel constructions with lexical opposites
10.2.1.2 Change of addressee
10.2.1.3 Additional remarks
10.2.2 Autem as a marker of absolute focus
10.2.2.1 Autem as a marker of parallel focus in questions
10.2.2.2 Autem as a marker of absolute focus in questions
10.3 The global use of autem: autem as a marker of the organization of the text
10.3.1 Parallel focus and thematic discontinuity
10.3.2 Autem with discourse topic shifts
10.3.2.1 Autem with given and accessible discourse topics
10.3.2.2 Autem with new discourse topics
10.3.3 Autem with other types of thematic discontinuity
10.3.3.1 Autem with shifted 'settings'
10.3.3.2 Autem as a neutral transition particle
10.3.3.3 Autem with temporary thematic shifts
10.4 Conclusions
10.4.1 A discourse-pragmatic description of autem
10.4.2 Linguistic clues
10.4.3 Pragmatic motivation for the use of autem
10.4.4 Autem and other discourse connectives
11. Vero
11.1 The discourse functions of vero
11.2 Representational vero
11.2.1 Between adverb and particle
11.2.2 Vero as an objective modality marker
11.2.3 Vero in reactive moves
11.2.4 Summary of the representational uses of vero
11.3 Interactional vero
11.3.1 Vero as a subjective modality marker
11. 3.2 Other interactional instances of vero: vero as a conversation particle
11.4 Side-effects
11.4.1 Preliminary remarks: pragmatic motivations and linguistic clues
11.4.2 Side-effect 1: the 'adversative connective' vero
11.4.3 Side-effect 2: vero as an ostensible marker of the organization of the text or as a focus particle; the chiaroscuro effect
11.5 Conclusions
11.5.1 A discourse-pragmatic description of vero
11.5.2 Vera and other discourse connectives
12. At
12.1 Introduction
12.2 The interactional discourse function of at
12.2.1 Discourse type 1: dialogical dialogal (direct dialogue without an overt central reporter)
12.2.2 Discourse type 3: dialogical monologal
12.2.3 Discourse type 5: diaphonic monological discourse
12.2.3.1 Indirectly rendered conversation in oratio obliqua
12.2.3.2 Implied interlocutor with the status of an 'embedded voice'
12.2.3.3 Pseudo-apodotic clauses containing at
12.2.3.4 At in sentences of narration proper
12.3 The presentational discourse function of at
12.4 The so-called pathetic use of at
12.5 Conclusions
12.5.1 A discourse-pragmatic description of at
12.5.2 At and other discourse connectives
13. Conclusions
Bibliography
Index of terms
Index of critically discussed passages
Index of Latin words
Recommend Papers

Discourse Particles in Latin: A Study of Nam, Enim, Autem, Vero and at (Amsterdam Studies in Classical Philology, 4) [Illustrated]
 9050634478, 9789050634472

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

DISCOURSE PARTICLES IN LATIN

Due to their extremely elusive nature, particles have long been treated in a stepmotherly way, in grammars of individual languages as well as in theories of language. This study is representative of the recent upsurge of interest in particles, more particularly in particles with a discourse marking function. By offering a detailed account of a number of Latin discourse particles, the study provides more insight into a virtually neglected area of Latin. At the same time it contributes to the theoretical and methodological foundations of current particle research and, more generally, to the development of linguistic models of discourse. The first part develops, in a step-by-step manner, a general framework for the analysis of discourse particles. To this end a number of valuable insights from pragmatics and discourse analysis are introduced, evaluated, and turned into an adequate set of tools for the actual description of particles. An important role is accorded to linguistic criteria underlying the distinctions made. The second part of the book investigates, on the basis of a large sample of instances from various types of texts, two groups of discourse particles in Latin: the so-called causal coordinators nam and enim (usually described as equivalents of English/or), and the so-called adversative coordinators autem, vero and at (assumed equivalents of English but). By carefully analysing these particles along the lines of the framework developed in the fIrst part, the book offers a new perspective on the functions of these particles in Latin texts and solves a number of problems in their interpretation. On the basis of more unified accounts of the individual particles than those that are traditionally given in Latin grammars, it is furthermore shown that the alleged synonyms nam and enim, and autem, vero and at, have, in fact, clearly distinct functions.

AMSTERDAM STUDIES IN CLASSICAL PHILOLOGY EDITORS

ALBERT RIJKSBARON IRENE LF. DE JONG HARM PINKSTER VOLUME FOUR

PREVIOUSLY PUBLISHED

1. A. Rijksbaron, Grammatical Observations on Euripides' Bacchae. 1991. 2. R. Risselada, Imperatives and other Directive Expressions in Latin. A Study in the Pragmatics ofa Dead Language. 1993. 3. G. Wakker, Conditions and Conditionals. An Investigation of Ancient Greek. 1994.

5. H. Dik, Word Order in Ancient Greek. A Pragmatic Account of Word Order Variation in Herodotus. 1995.

CAROLINE KROON

DISCOURSE PARTICLES IN LATIN A STUDY OF NAM, ENIM, AUTEM, VERO ANDAT

I.e. GIEBEN, PUBLISHER AMSTERDAM 1995

No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the author. © by C.H.M. Kroon, 1995 / Printed in The Netherlands / ISBN 90 50634478

Voor mijn ouders

TABLE OF CONTENTS

Preface 1.

Introduction 1.1 Outline of the study 1.2 Data

xiii

2 3

Part I: Theory 2.

Relations

2.1 Issues in text relations: conjuncts, relations and relators

3.

4.

7 7 9

2.2 Relations: various classifications 2.3 (Logical-)semantic classifications 2.4 Pragmatic classifications 2.4.1 Semantic-pragmatic relations 2.4.2 Speech act relations 2.4.3 Discourse-pragmatic classifications 2.5 Formal and structural classifications 2.5.1 Clause combining 2.5.2 Text structure 2.6 Evaluation of the usefulness of relation taxonomies

20 24 25 26 30

Relators: approaches to particles 3. 1 Introduction 3.2 Formal and syntactic accounts of particles 3.3 Semantic approaches to particles 3.4 Discourse and conversation approaches to particles 3.5 Evaluation

34 34 37 41 45 54

A framework for the description of connective particles

58 58 59 62 63 64 69 69 73 75 78 80

4.1 The descriptive model: coherence, connectives and levels of discourse 4.1.1 The first parameter: levels of discourse 4.1.2 The second parameter: type of relationship 4.1.2.1 Intratextual and extratextual relationships 4.1.2.2 Structural and semantic-functional relationships 4.2 Comments on the analytical framework 4.2.1 The representational level of discourse 4.2.2 The presentational level of discourse 4.2.2.1 Rhetorical discourse relations 4.2.2.2 Types of rhetorical relationships 4.2.2.3 Organization of the text

10

12 12 17

viii

5.

table of contents 4.2.3 The interactional level of discourse 4.2.3.1 Interactional relations 4.2.3.2 Conversational structure 4.2.3.3 'Evaluation' in terms of the communicative situation 4.3 Concluding remarks

89 90 94 94 96

Heuristics 5.1 Components in a description of particles: basic meaning, discourse function, actual use and side-effects 5.2 Levels of discourse and their linguistic manifestations 5.2.1 Distinguishing 'representational' from 'presentational' connective particles 5.2.2 Distinguishing interactional from other particles: the importance of the concept of discourse type 5.2.2.1 Discourse type: the oppositions monologal-dialogal and monological-dialogical 5.2.2.2 Diaphony 5.3 Overview of linguistic clues

97 97 103 103 108 109 111 115

Part II: Description

6.

7.

Causal relations and causal connectives: conventional accounts of nam and enim in the literature 6.1 Causal relations 6.2 Previous accounts of nam 6.3 Previous accounts of enim 6.4 Interchangeability of nam and enim

129 129 132 137 142

144 Nam 144 7.1 A discourse-pragmatic description of nam 145 7.2 'Backward-linking' nam 146 7.2.1 The rhetorical relation 'evidence' 146 7.2.2 The rhetorical relation 'justification' 7.2.3 The rhetorical relations 'explanation', 'elaboration' and 'background' 147 148 7.2.4 The rhetorical relation 'exemplification' 149 7.2.5 Additional remarks 152 7.3 So-called 'forward-linking' nam 152 7.3.1 Possible instances of 'forward-linking' nam 155 7.3.2 The relation between 'backward-' and 'forward-linking' nam 161 7.3.3 Nam as a marker of the organization of the text 163 The so-called affirmative particle nam 7.4 168 7.5 Conclusions

table of contents

ix

8.

Enim 8.1 A discourse-pragmatic description of enim 8.2 The 'non-connective' (situating) nature of enim 8.2. 1 Syntactic distributional properties 8.2.2 Discourse type 8.3 The interactional nature of enim 8.3.1 Shared information 8.3.2 Co-occurrence with other particles 8.3.3 Evidence from question-answer pairs 8.3.4 Illocutionary force 8.3.5 Metacommunicative comment 8.4 Actual use: pragmatic motivations for the use of enim 8.4.1 Actual use 1: toning down a challengeable utterance 8.4.2 Actual use 2: irony 8.4.3 Actual use 3: empathy 8.5 Conclusions 8.5.1 Interactional consensus as a unifying concept 8.5.2 Limitations on the interchangeability of nam and enim

171 171 172 172 175 183 185 186 188 189 195 196 196 198 199 202 203 205

9.

Adversative relations and adversative connectives: conventional accounts of autem, vero and at 9. 1 Adversative relations 9.2 Previous accounts of autem, vero and at 9.2.1 Previous accounts of autem 9.2.2 Previous accounts of vero 9.2.3 Previous accounts of at

210 210 217 218 220 222

10. Autem 226 10.1 The discourse function of autem 226 10.2 The local use of autem: autem as a focusing device 227 10.2.1 Autem as a marker of parallel focus 229 10.2.1.1 Parallel constructions with lexical opposites 231 10.2.1.2 Change of addressee 233 10.2.1.3 Additional remarks 235 10.2.2 Autem as a marker of absolute focus 238 10.2.2.1 Autem as a marker of parallel focus in questions 241 10.2.2.2 Autem as a marker of absolute focus in questions 244 10.3 The global use of autem: autem as a marker of the organization of the text 247 10.3.1 Parallel focus and thematic discontinuity 247 10.3.2 Autem with discourse topic shifts 250 10.3.2.1 Autem with given and accessible discourse topics 250 10.3.2.2 Autem with new discourse topics 261 10.3.3 Autem with other types of thematic discontinuity 263 10.3.3.1 Autem with shifted 'settings' 263 10.3.3.2 Autem as a neutral transition particle 266 10.3.3.3 Autem with temporary thematic shifts 267

x

table of contents 10.4 Conclusions 10.4.1 A discourse-pragmatic description of autem 10.4.2 Linguistic clues 10.4.3 Pragmatic motivation for the use of autem 10.4.4 Autem and other discourse connectives

11. Vera 11.1 11.2

11.3

11.4

11.5

The discourse functions of vero Representational vero 11.2.1 Between adverb and particle 11.2.2 Vero as an objective modality marker 11.2.3 Vero in reactive moves 11.2.4 Summary of the representational uses of vero Interactional vero 11.3.1 Vero as a subjective modality marker 11. 3.2 Other interactional instances of vero: vero as a conversation particle Side-effects 11.4.1 Preliminary remarks: pragmatic motivations and linguistic clues 11.4.2 Side-effect 1: the 'adversative connective' vero 11.4.3 Side-effect 2: vero as an ostensible marker of the organization of the text or as a focus particle; the chiaroscuro effect Conclusions 11.5.1 A discourse-pragmatic description of vero 11.5.2 Vera and other discourse connectives

12. At 12.1 12.2

12.3 12.4 12.5

269 269 272 275 277 281 281 285 285 288 291 298 299 300 304 309 310·

315 319 326

326 329

333 Introduction The interactional discourse function of at 12.2.1 Discourse type 1: dialogical dialogal (direct dialogue without an overt central reporter) 12.2.2 Discourse type 3: dialogical monologal 12.2.3 Discourse type 5: diaphonic monological discourse 12.2.3.1 Indirectly rendered conversation in oratio obliqua 12.2.3.2 Implied interlocutor with the status of an 'embedded voice' 12.2.3.3 Pseudo-apodotic clauses containing at 12.2.3.4 At in sentences of narration proper The presentational discourse function of at The so-called pathetic use of at Conclusions 12.5.1 A discourse-pragmatic description of at 12.5.2 At and other discourse connectives

333

335 336 340 342

343 343

344 348

350 357 362

362 366

table of contents

xi

13. Conclusions

371

Bibliography

377

Index of terms

391

Index of critically discussed passages

397

Index of Latin words

398

PREFACE

The writing of a doctoral thesis is all too often a lonely journey with a highly uncertain outcome. I consider myself particularly fortunate to have been guided on this odyssee by three travelling companions who advised me on the direction and helped me to sail round many difficult rocks. My warmest thanks go, fIrst, to my pilot and navigator Dr Machtelt Bolkestein, who supervised this study from pre-beginning to end. Her faith in the potential of my topic and her never-failing optimism proved invaluable throughout. I have profIted much from her creativity in distilling repeatedly new research questions from my material, which provided me with much more interesting data than I could incorporate in this study. Secondly I want to express my gratitude to Professor Harm Pinkster, who from my fIrst days at the University of Amsterdam coached me and trained me in all the skills of the profession. As a real captain he created the atmosphere of collegiality in our team in which my self-confIdence was able to grow and in which I was able to develop unsuspected talents. I am particularly grateful for his friendship and his encouragement when the going was rough. Last but not least, I thank my fellow seaman and sparring partner Dr Rodie Risselada for her active involvement in the entire project. There is hardly any aspect of this thesis that has not profIted from her sincere interest and penetrating comments. In the fInal stage of the production of the text her editorial help proved indispensable. I thank her most of all for her loyalty and warm friendship over the years and express the heart-felt wish that our fruitful cooperation be continued for at least another two decades. There are two more people who deserve special mention here. Professor Simon Dik fIrst introduced me to the theory of Functional Grammar in 1984, and from that time on followed my academic career with great interest. Although this is not a book on Functional Grammar, it is clear that my views on - and approach to - language have been influenced by this theory to a considerable extent. Dr David Langslow of the University of Oxford generously agreed to correct my English, and did in fact much more than that. Thanks to him, English word order has become less of a mystery to me. In January 1992, with the harbour within reach, I had a small accident with some serious consequences for my health. Thanks to my family, who were

xiv

preface

even more determined than I was to bring this expedition to a happy conclusion, I was able to continue the work on the thesis. I thank all of Nina's grandparents, and especially Sjoerd, for running the house while I was 'away', and for their loving care. During the course of this project I enjoyed the hospitality of many individuals and several institutes. I retain happy memories of my stay at the Linguistics Department of the University of California at Santa Barbara in 1990, made possible by a grant of the Netherlands America Commission for Educational Exchange. I thank Professor Sandra Thompson of the UCSB, who generously advised me and encouraged me during this period. The Netherlands organization for scientific research, NWO, subsidized a visit to the Thesaurus Linguae Latinae in Munich, as well as a three-week stay at the Fondation Hardt in Vandoeuvres, Switzerland. For various reasons both trips proved very rewarding. Many people have helped me during the final stage of the production of the text. Special thanks are due to Ineke Blijleven and Emilie van Opstall, who with great cheerfulness and precision did many a tiresome job on the computer. Practical help was offered also by Dr Mike Hannay, Professor Siem Slings and Dr Elseline Vester. I thank Dr Albert Rijksbaron for reading a prefinal version of the book and preserving me from a number of annoying errors. To my colleague at the Free University Professor Jan den Boeft I offer my apologies for monopolizing the computer and messing up our office for quite some time. I dedicate this thesis to my parents, who contributed more to its realization than they themselves will probably be aware of. It is through their love and support throughout my life that I have come this far. Amsterdam, January 1995

1

INTRODUCTION

From antiquity onwards Latin linguists have described particles such as nam, enim, autem, vero and at as linguistic means that express semantic relations between adjacent main clauses. A reflection of this tradition (which, mutatis mutandis, is to be found also ingrarnmars of modem languages) can be seen in the way in which Latin grammars organize their descriptions of these particles: nam and enim (which are usually regarded as near-equivalents of English 'for') are treated as a rule in sections on causal coordinating particles, autem, vero and at (usually taken as equivalents of English 'but') in sections on adversative coordinating particles. It is generally clear, however, that not all of the actual uses of the particles concerned fit this picture: Latin grammars are usually stuck with a rather embarrassing number of exceptional cases which clearly defy a description in terms of causal or adversative coordination. Popular, though quite unsatisfactory, solutions include the assumption of diachronic developments, or of the synchronic coexistence of different functions of one and the same form (i.e. homonymy), or of an extreme form of polysemy. Modem, more pragmatically-oriented linguistic research, and especially a number of recent discourse studies, have thrown new light on the functions of connective particles in a number of modem languages. Important observations are, for instance, that text relations are not necessarily of the same (i.e. semantic) type (cf. e.g. Halliday and Hasan 1976), that they can obtain between units of any size and nature (i.e. not only between adjacent, grammatical clauses; cf. e.g. Sinclair & Coulthard 1975), and - most recently - that they may apply to various dimensions of discourse (cf. Schiffrin 1987). In the present study I will attempt to demonstrate the usefulness of a discoursepragmatic approach to the study and description of the aforementioned Latin particles, by applying, adjusting and further developing new concepts from the areas of pragmatics and discourse analysis.

2

§ 1.1

1.1 Outline of the study In the first part (Theory, chapters 2-5), I will develop a theoretical framework for the description of 'connective' particles in general and of Latin enim, nam, autem, vera and at in particular, in which I incorporate insights from several fields of linguistic research. Discussions of the state of the art in research on text relations and on (connective and other) particles (chapters 2 and 3, respectively) will be followed, in chapter 4, by the actual outline of this descriptive framework. This framework can be seen as the outcome of the interplay between theory-driven and data-driven investigation, and may claim to apply more widely than merely to the description of Latin connective particles. The last chapter of the theoretical part (chapter 5) is meant to bridge the gap between theory and the actual practice of particle description. It is devoted mainly to heuristic matters and contains, for instance, an overview of linguistic clues for determining the discourse function of individual particles. The theoretical part may interest especially linguists working in the field of text cohesion, particle specialists and linguists concerned with the development of (linguistic) models of discourse coherence. The second part (Description, chapters 6-12) contains systematic, synchronic descriptions of the Latin 'causal' particles nam and enim and the 'adversative' particles autem, vera and at. As such it provides the empirical foundations for the theoretical framework expounded in the first part. The chapters on nam and enim are preceded by a general introduction on the contentious notion of causal relation and the way in which this notion has been used in earlier accounts of these particles (chapter 6). Likewise the chapters on autem, vera and at are preceded by a discussion of the equally vague notion of adversative relation that is used in conventional accounts of the matter (chapter 9). Since there are considerable differences between the items concerned, the discussions of the individual particles do not proceed along identical, predetermined lines: each of the descriptive chapters is arranged in the way that appeared to be the most instructive in view of the particular features of the item under discussion. The descriptive chapters resemble one another, however, in that an important role is accorded to the evidence in linguistic argumentation, and in that comparisons are drawn with (near) synonyms and other Latin particles. This second, descriptive part may be of benefit especially to Latinists, although some of its material may interest general linguists as well. With an eye to the heterogeneity of the groups of readers envisaged, I have included many cross-references from the first part of the study to the second

introduction

3

part, and vice versa. Furthermore, individual chapters will be preceded by a summary of their contents. The book ends in chapter 13 with a brief overview of the main findings of the study.

1.2 Data The research reported in the second part of this study is based on Latin literary texts from about 200 BC to about 200 AD, with special reference to Latin comedy (Plautus and Terence), the complete works of Cicero (letters, orations, philosophical discussions, rhetoric), historiography (notably Caesar, Sallust, Livy, Curtius and Tacitus), the 'novel' Satyricon by Petronius, and Seneca's Letters to Lucilius. The research is not to be characterized as corpus-based research in the sense of aiming to explain all instances in a fixed corpus, a method I regard as not very fruitful for the present undertaking: by studying all instances of a particular particle in a subcorpus of Latin texts, one runs the risk of registering unnecessarily many identical and unproblematic instances, while potentially missing the rare, really interesting ones. I therefore adopt a mixed approach, and will make use, in my discussion, of a sample of representative instances drawn from the corpus described above, supplemented by a motley collection of (usually 'deviant' and therefore very interesting) instances provided by Latin manuals and other studies on the subject. Statistical observations will, accordingly, be supplied on an ad hoc basis only. Where counts are presented, these are usually based on a selection of the Ibycus corpus l . This selection comprises the entire works of Accius, Caecilius, Caesar (incl. the Bellum Africanum, Bellum Alexandrinum and Bellum Hispaniense) , Catullus, Celsus, Cicero, Columella, Curtius, Ennius, Horace, Juvenal, Livy, Lucan, Lucretius, Manilius, Martial, Nepos, Ovid, Pacuvius, Petronius, Phaedrus, Plautus, Pliny the Elder, Pliny the Younger, Propertius, Quintilian, Sallust, Seneca the Elder, Seneca the Younger, Silius ltalicus, Statius, Suetonius: Tacitus, Terence, Tibullus, Valerius Flaccus, Valerius Maximus and Vitruvius. Statistical remarks based on this corpus serve mainly to illustrate and corroborate the argumentation, and are not tested for statistical significance. The Latin examples in the text are quoted from the Oxford Classical Texts editions, or, where the latter are not available, from the Teubner or Bude

I. Regrettably, the Ibycus corpus was not available to me until the research had practically been completed. Counts based on the Ibycus material have for the most part been inserted afterwards.

4

§ 1.2

editions. I confine myself mainly to examples from Latin prose, although every now and then I find it expedient to include parallels from poetry as welf. The abbreviated names of Latin authors and their works follow the practice of the Oxford Latin Dictionary. Most translations of the Latin quotations are taken from the Latin-English Loeb editions, with a few alterations where needed (in particular of the translations of the particles in question, which often turn out to be inadequate or altogether lacking).

2. Lyric poetry is left out of account altogether. Other forms of poetry (such as epic and didactic poetry) are taken into account only in so far as they provide material relevant to the overall picture of a particular particle.

PART I: THEORY

2

RELATIONS

Chapters 2 and 3 are intended primarily to give a structured overview of a number of representative studies which cover the alrrwst unlimited and sometimes poorly defined field of text relations. The systematic subdividing of the field attempted in these chapters is necessary to make clear where and how the present study fits in and what specific contributions it is undertaking to make (and, conversely, how this study can profit from previous research). The present chapter deals especially with those approaches to text relations whose main concern is the development of taxonomies of (logical-semantic or pragmatic-functional) relations in terms of which the relators might eventually be described and subclassified. This approach will be referred to as the 'topdown' approach to the study of connective particles. Representatives of a rrwre 'bottom-up' approach will be discussed in chapter 3.

2.1 Issues in text relations: conjuncts, relations and relators

The issue of relations between text segments may involve several components. First we may distinguish two or more conjuncts, that is, those parts of the text which enter into some relationship with each other. These conjuncts do not necessarily both receive an explicit expression, as one of them may remain implicit!. The second element involved is the relation itself, which is defmable in, for instance, structural, semantic or functional terms. This relation may get, in its turn, an overt linguistic marking in the form of a relator, the presence of which is in general optional, but not (as will become clear in the course of this study) without significance for the ultimate interpretation of the text.

1. Cf. e.g. Levinsohn (1987: 13): "Most utterances relate in some way or other to their context, whether the context is linguistic (previous written sentences or past speeches), or non-linguistic (e.g., the scene before the speaker or his earlier experiences)". See also ch. 4, p. 63.

8

§ 2.1

In the present chapter and in chapter 3 I will give an outline of part of the vast amount of current linguistic research that in one way or another is concerned with the intricate issue of text relations. These studies differ not only as to which component (conjunct, relation or relator) they primarily focus on, but also as to their definitions of the elements involved, depending on their specific aims or theoretical framework. Although many of them are valuable as far as their individual goals are concerned, it turns out that they are only of limited help for the task of the present study. That is to say, in current literature no single set of text relations has been proposed onto which the system of Latin relators can be simply mapped. For ease of survey the material is, in a somewhat artificial manner, split up into two parts. In the present chapter I will discuss approaches which are mainly interested in setting up taxonomies for relations, or which otherwise centre on the issue of relations. The following chapter (ch. 3) deals with research that focuses on the description of the role of relators (conjunctions, particles, discourse markers, and the like) in signalling text relations 2 • This dichotomy reflects to a certain extent the methodological difference between 'top-down' and 'bottom-up' approaches in research on text relations and their linguistic reflection. A top-down approach starts with setting up a system or taxonomy of relations, and works downwards to the linguistic signs (the relators) which are to be described and explained within the overall frame of relations. Bottom-up analyses, on the other hand, focus on the investigation of the linguistic signs (relators), thereby gradually building up a more abstract system of relations. Chapters 2 and 3 end with an evaluation of both approaches in terms of their applicability to and usefulness for the present study. The third component of a text relation, the conjuncts, will not receive a systematic treatment in chapters 2 and 3, the main reason being that most studies - for reasons of simplicity or otherwise - equate conjuncts simply with sentences or clauses. However, conjuncts in the form of speech acts come into my discussion of pragmatic relations in 2.4.2, while 2.5 deals partly with conjuncts that are larger than one single sentence. The issue of the conjuncts will be taken up in more detail in chapter 4, where I will come to speak about 'discourse units'.

2. It is not always clear whether a study takes the relations or the relators as a starting point and centre of interest. A certain overlap between chapters 2 and 3 is therefore unavoidable.

classifications of relations

9

2.2 Relations: various classifications The various classifications of text relations to be discussed in the present section can be distinguished and characterized roughly on the basis of the following set of partly interlocking parameters, of which the last parameter appears to be the most essential one: (i) (ii) (iii)

general versus specific degree of dependence on linguistic signs level of description

The first parameter distinguishes comprehensive studies that aim at an exhaustive overview of possible relations from investigations with the more modest goal of tackling only one subgroup, such as for instance that of the causal or the adversative relation. Examples of the former group are Halliday and Hasan (1976), Longacre (1983), Halliday (1985) and a number of publications within the framework of Rhetorical Structure Theory (see section 2.4.3 below). In addition to these more encompassing taxonomies a lot of work has been done on the detailed description of specific relation types. As examples can be mentioned Altenberg (1984), Anscombre (1984), Harras (1984), Rosengren (1986), Klein (1987) and Vandepitte (1993) for causal relations; Moeschler and De Spengler (1982), Konig and Eisenberg (1984), Konig (1986), Thompson and Mann (1986; 1987), Konig and Van der Auwera (1988) and Spooren (1989) for adversative and, in particular, concessive relations. I will return to some of these more detailed accounts of specific relation types in later chapters, when I come to deal with the so-called causal and adversative connectives. Parameter (ii) covers the fact that taxonomies of text relations also differ as to the extent to which linguistic signs (the relators) are taken into account. This parameter has an important methodological implication in that it differentiates between various (rigid or less rigid) types of top-down approaches. On the one end of the top down scale we find studies that explicitly state that their inventory of relations is universal and independent of the linguistic reflection of the relations in any particular language3 • A more moderate position is taken up by linguists who consider the presence of a specific means of expression in at least one language to be a prerequisite for integration of a relation in their

3. Cf. e.g. Mann and Thompson (1989) when unfolding the principles of Rhetorical Structure Theory. The theory does not claim however that the taxonomy is universally valid.

10

§ 2.2

taxonomy4. Near the other end of the top-down scale we find language specific studies in which classifications of relations are presented alongside the corresponding means of expression5 , and classifications that use the insertion of relators as a diagnostic tool for the identification of a particular text relation6 • The third parameter, level of description, appears to be the most essential one for the characterization of the different taxonomies of text relations. This parameter accounts for the fact that the classifications usually take their criteria from one of the three main levels of linguistic description, commonly referred to as syntax (or: structure), semantics and pragmatics. Syntax can be described as the study of the combination of forms, semantics as the study of meanings, and pragmatics as the study of language in use 7 • In the next few sections I will subsequently discuss classifications with a (logical)-semantic (2.3), pragmatic (2.4) or structural (2.5) basis.

2.3 (Logical-)semantic classifications Since the introduction of logical truth-tables at the end of the 19th century, many taxonomies of text relations have been based on - or have been influenced by - logical-semantic principles of clause combining. Loosely stated this means that a small set of logical symbols or operators, such as &, V, >, ~ and ~, are applied to natural language relations, with the natural language conjunctions functioning as linguistic expressions of these symbols. This approach is characteristic of the linguistic current that is generally known as formal semantics. It is by now, however, well-recognized that natural language does not necessarily obey the laws of logic, and that an approach in which linguistic meaning is formulated in terms of truth-conditions (without admission of, among other things, cultural, social and cognitive factors), causes serious problems for linguistic research8 • 4. E.g. Longacre (1983: ch. 3 passim). 5. This method is best demonstrated by Halliday and Hasan (J 976: ch. 5). 6. Cf. e.g. Hobbs (1985: 31). 7. Pragmatics is not an unequivocal term. In this study I take pragmatics to refer to the study of the relations between the linguistic properties of utterances and their properties as units of (social) interaction. 8. For a discussion of the problems of applying formal semantics to natural language, cf. Van Dijk (1977: 37-40).

(logical-)semantic classifications

11

figure 1: logical-semantic relations 9

relation additive disjunctive causal-conditional temporal-sequential adversative

operator & V

> --?i> ~

Nonetheless, the term 'logical-semantic relations' is found every now and then in the literature to refer more generally to meaning-relations between propositions. The basic logical-semantic categories listed above in this case usually serve as cover terms in a more sophisticated system of semantic subdistinctions lO • The relation causal-conditional, for example, commonly breaks up into direct cause, condition, purpose, reason, result, etc. Usually, the criteria underlying semantic taxonomies of this type remain vague: the distinctions seem merely to reflect a universal set of cognitive categories that are based on general knowledge about possible real world events or facts, and the relations between them. Semantically oriented taxonomies of text relations have found their main application and justification in language-specific studies on the meanings of conjunctions. The semantic approach is furthermore reflected in the way most language-specific grammars organize their chapters on clause combining. In the authoritative Latin grammar by Kuhner-Stegmann, for instance, relationtypes and matching relators are arranged according to the four major distinctions copulative, adversative, disjunctive and causal, an approach which may be traced back as far as the ancient Latin grammarians ll .

9. In some of the taxonomies the disjunctive relation is grouped together with the additive relation. Besides, the adversative relation does not always show up in the list of logical-semantic relations, as it is not generally considered to be a basic logical category: one subgroup of the adversative relation, concession, cannot be described in logical-semantic terms, but seems to need an extra operation (viz. 'frustration' or 'negation', see below and ch. 9); the other subgroup (contrast) is often described as an additive relation. The problematic notion of 'adversativity' will be dealt with later on in ch. 9. 10. Cf. e.g. Crothers (1979). 11. Kiihner-Stegmann's distinctions hold only for coordination ("Beiordnung n). Subordination (nUnterordnung n) gets a separate treatment. For the way relations and matching conjunctions are classified by the ancient Latin grammarians, cf. Baratin (1989: 69-114).

§2.4

12 2.4 Pragmatic classifications

In this section I discuss three types of research which, in varying degrees, integrate pragmatic insights in their studies on text relations: studies on semantic-pragmatic relations, studies on speech act relations and discoursepragmatic approaches to text relations. 2.4.1 semantic-pragmatic relations

One of the first attempts to make the (logical-)semantic approach to text relations a more widely applicable and workable tool is the seminal work by Halliday and Hasan (1976). In this study text relations are discussed as one of the elements within the more general framework called cohesion, a semantic concept referring to the aspects in which a text differs from a collection of unrelated sentences. This work can be said to belong partly to the traditional grammatical approach in that it still starts from a restricted set of more or less universal relations between one proposition and another: additive, adversative, causal and temporal. According to Halliday and Hasan, these basic semantic relationships remain the same, whether the relation is incorporated into linguistic structure (subordination and coordination) or noe 2 : both the structural example (1) and the non-structural example (2) show, for instance, the basic semantic relation of cause13. (1)

Since it's raining, let's stay at home

(2)

It's raining - Then let's stay at home

The main merit of Halliday and Hasan's study lies, however, in the observation that the (logical-)semantic relations hold not only on an external level of reality, that is, between events outside the language itself, but also on an internal linguistic level, i.e. within the process of linguistic interaction 14 • What is meant by internal vs. external relationships can be illustrated with the following two examples:

12. For a discussion of 'structural' relations, see 2.5 below. 13. Cf. Halliday and Hasan (1976: 9). The causal relation in (2) holds regardless of whether the linguistic signal then (or some other overt signal) is present or not. 14. Because Halliday and Hasan are concerned mainly with what they call 'conjunction' (i.e. nonstructural relations), it does not become clear whether the external/internal distinction can be applied also to the structural relation types.

semantic-pragmatic relations

13

(3)

He made a number of telephone-calls and fed the cat. Then he turned on the television.

(4)

So much about pollution. Then there is our drought problem ....

In the former example, then signals a temporal relation on the representational level of the text, that is, between events in an extra-linguistic world. In (4), on the other hand, then indicates the succession of both conjuncts in the communicative setting of the current, unfolding discourse. In other words, then in (3) and (4) signals the same type of relationship, though on different levels of reality. In English, many relators have the capacity to occur in both these functions. Halliday and Hasan's view that an utterance is as much an element of content as it is an element of interpersonal communication, and that, accordingly, every linguistic description (including that of text relations) should at least be able to account for both, can be seen as an important step from a semantically based description of text relations to a more pragmatically based one. This line is continued and worked out in more detail in Halliday (1985). Although Halliday still calls the relations 'logical-semantic', they now get a more or less pragmatic definition in that they are described in terms of the communicative or rhetorical function of the second conjunct with regard to the first, preceding conjunct. The relations are basically of two types, called expansion and projection IS. In the case of an expansion relation the second clause (or clause complex) expands the first clause (or clause complex) by elaborating, extending or enhancing it. Figure 2 below is a gross oversimplification of the more refined inventory of relations that Halliday presents in chapters 7 and 9 of his book. This figure shows how the conventional (logical-)semantic relations 'additive', 'disjunctive', 'causal-conditional', 'temporal', and 'adversative,16 may be distributed over Halliday's functional categories extension and enhancement.

15. The projection type of relationship (which involves transitions to quoted speech) falls outside the scope of the present study, and will therefore not be elaborated on. 16. Adversativity comprises both contrast and concession. In Halliday's framework contrast belongs to the extension relations, concession to the enhancement relations.

14

§ 2.4.1

figure 2: Halliday's taxonomy of text relations ____________ EXPANSION __________

elaboration

extension

enhancement

exemplification clarification

additive disjunctive adversative

causal-conditional temporal adversative

I

I

I

A more or less comparable attempt to design a more pragmatically oriented classification of text relations can be found in Longacre (1983)17. In the third chapter of his Grammar of discourse Longacre distinguishes three interlocking concepts which together determine his taxonomy of text relations. figure 3: Longacre's taxonomy of text relations

basic structures J8

frustrated counterparts

additive disjunctive temporal causal-conditional

(concessive/ adversative)

elaborative deep structures paraphrase illustration deixis attribution

(concessive/ adversative)

Like Halliday, Longacre considers the logical-semantic relations as listed in 2.3 to be basic, on account of the fact that they are assumed to be basic to the formal structure of a text. These basic relations can obtain between conjuncts

17. This work is based on Longacre's earlier study An anatomy of speech notions, Lisse: de Ridder, 1976. 18. Instead of additive, disjunctive, temporal and causal-conditional, Longacre uses the comparable terms conjoining, alternation, temporal and implication. For the sake of clarity I will use in this chapter the former set of terms, unless there is reason to assume that the terms used are not equivalent.

semantic-pragmatic relations

15

of any size 19 • Along with the basic structures the linguistic system can draw upon a set of 'elaborative deep structures', which can be seen as embellishing or rhetorical devices. Languages do express these relations linguistically, but in a less systematic way than the basic relations 20 • The last notional parameter, which intersects the other two, is called 'frustration'. Almost all basic types of relations and a few relations of the elaborative (i.e. pragmatic) type have frustrated counterparts, which are generally known as concessive or adversative relations. Example (5), for instance, is the frustrated counterpart of a temporal (successive) relation: 'leaving for Paris' usually implies 'arriving in Paris'; (6) is a comparable example of a frustrated cause relation. (5)

They left for Paris but didn't arrive

(6)

He was poisoned but he didn't die

My reason for dealing at some length with the taxonomies of Halliday and Longacre is that, although neither of the proposals has been fully worked out and both give the impression of lacking a sound empirical basis, they can be said to make an essential move in the direction of more pragmatically oriented taxonomies of text relations. That is, the integration of 'pragmatic' relations such as exemplification and illustration, and especially Halliday's distinction between external and internal relations, reveal an awareness of the existence of a pragmatic level of description of text relations in addition to a purely semantic one. This insight has proved to be of essential importance for the description of various linguistic phenomena, including the occurrence of connective particles. The reflection of the semantic-pragmatic type of relation taxonomies proposed by Halliday and Longacre can be traced in many later works of linguistic research21 , such as studies on discourse organization, for instance, as we will see later on. It may be noted here that it is not surprising that the important distinction between semantic and pragmatic types of relationships has not always been fully recognized in the literature. The undue concentration on English as the object language of linguistic studies may have been a reason for this: the

19. Cf. Longacre (1979), where he applies this parameter to relations on the level of paragraph structure. 20. The relations paraphrase and illustration are more or less self-explanatory and seem to correspond roughly to Halliday's categories clarification and exemplification. Deixis-relations obtain when conjunct 1 functions as an introduction or identification with regard to conjunct 2. The category attribution concerns quoted speech. 21. See, among others, Lundquist (1980); Martin (1983); Quirk et al. (1985); Longacre (1979).

16

§ 2.4.1

English system of relators does not always provide for separate sets corresponding to this difference, that is, many relators can be used to signal a semantic as well as a pragmatic type of relationship22. This is nicely illustrated by the examples (3) - (4) above, and the following examples (7) - (9). The distinction is however suggested more clearly by the more systematic differences in distribution between for instance French parce que and puisque/car, German weil and da/denn, Ancient Greek w~ and f1rE{/'Yap, and Latin quia/quod on the one hand and quoniam/nam on the other23 . See also chapter 5. (7)

he cannot attend the meeting because he is ill (cause)

(8)

where can I find the best restaurant in town? - because you're the greatest gourmet I know (justification)

(9)

Eric was here, because there is no milk in the fridge (evidence)

In terms of Halliday and Hasan, the sentence cited in (7) displays an external (i.e. semantic) relation of cause; (8) and (9) would both be described as internal (i.e. pragmatic) cause relations. Later studies draw up more refined subcategorizations, in which the internal cause relation is split up further into justification and evidence. Whereas in the justification relation (example 8) the because-clause justifies the utterance of a particular speech act, the evidence relation (example 9) provides an argumentative conclusion based on knowledge24 . Within a semantic-pragmatic classification of relations the dichotomy external cause relation - internal cause relation could thus be replaced by, for instance, a trichotomy, which in French and German seems to correspond more clearly with a difference in means of expression than in English, as is illustrated in figure 4 below. It should be stressed that this micro-taxonomy of 'causal relations' is the only semantic-pragmatic subcategorization of text relations where 'top-down' and 'bottom-up' approaches appear to complement each other fruitfully, thus providing a better-defined (and linguistically based) set of relations than is usually the case. However, it is still an open question whether this taxonomy can be expanded along the same lines for other 'basic' categories of relations (such as for instance additive and adversative?5.

22. Cf. Van Dijk (1979). 23. For French cf. e.g. Ducrot et al. (1975), Anscombre (1984) and Moeschler (1987); for German Dunbar (1985) and Rosengren (1986); for Ancient Greek Rijksbaron (1976); for Latin Fugier (1989), Bolkestein (1991) and Mellet (J 995). 24. Cf. e.g. Davies (J 979) and Sweetser (J 990). 25. Sweetser (1990) alludes to this possibility.

semantic-pragmatic relations

17

figure 4: a cross-linguistic taxonomy of 'causal' relatious level of reality

relation

relator French

relator German

external

cause

parce que

weil

cognition

evidence

puisque

da

interaction

justification

car

denn

internal

Besides, it might be hard to demonstrate the linguistic relevance of distinguishing unequivocally 'internal' (pragmatic/rhetorical/functional) relationships such as exemplification, clarification and illustration, which do not have clear counterparts on the external level of reality, and seem to obtain preferably between more extensive text parts than clauses. It may well be the case that there are no languages that express these relations by means of particular subordinators. 2.4.2 speech act relations A more manifestly pragmatic approach to the issue of text relations than the one sketched above is represented by speech act oriented studies. Here, relations between speech acts (to a certain extent corresponding to Halliday's 'internal relations') are systematically set apart from semantic relations between propositional contents. Consider the following two examples, which resemble the pair in (3) and (4): (10) I was hungry, so I ate some cookies (11) Your little brother seems to be hungry, so let's give him some cookies

In both examples a semantic cause-result relation can be observed between the denoted states of affairs referred to in the connected clauses, that is, between 'being hungry' and 'eating/ giving cookies'. In the latter example one can discern in addition a pragmatic relation between the two constituent speech acts assertion and proposal. To distinguish this pragmatic relation from the semantic relation of result, it could for instance be called conclusion. Characteristic of the treatment of text relations in many speech act oriented studies is that the relations between the speech acts are described in terms of higher order speech acts. On this view the relation between the assertion and

18

§ 2.4.2

the proposal in (11) would give rise to a new, higher level speech act 'conclusion'. In (10), on the other hand, only one speech act can be discerned: both propositions fall within the scope of the basic speech act assertion. As a representative of recent linguistic research dealing with speech act relations I mention Klein's Die konklusive Sprechhandlungen (1987). In this study Klein proposes to extend the apparatus of classical speech act theory so as to include at least the illocutionary acts 26 'ErkUiren Warum', 'Rechtfertigen', 'Begriinden' and 'Folgern', which together constitute the group of conclusive speech acts. How these illocutionary acts relate to the traditional Searlean categories can be demonstrated by a sentence like (12) and its underlying representation (13): (12) Ich verspreche dir: Morgen bekommst du dein Geld. Also: Lass mich in Ruh! (13)

F(F J (p), F2 (q»

In Klein's view F' in (13) represents the higher order speech act 'conclusion' (Folgern). FI and F2 stand for a commissive and a directive illocutionary act, respectively, and p and q are proposition variables. Essential characteristics of the representation in (13) are the illocutionary layering (one higher level speech act dominates two lower level speech acts) and the illocutionary bivalency (the higher order speech act has two illocutionary 'slots'). Comparable though less explicit proposals for an extension of classical

speech act theory are put forward by Rosengren (1987: 'Begriindungen und Folgerungen als kommunikative Handlungen'); and Anscombre (1984: 'La representation de la notion de cause dans la langue'), who refers to the constellation of speech acts as 'acte d'expliquer' or 'acte de consecuter'. Speech act oriented studies of pragmatic relations of the type discussed above 27 share a number of properties that distinguish them from other approaches to text relations. For one thing, they are more explicit about the interrelation of semantic and pragmatic text relations than the earlier-mentioned 'mixed' accounts by Halliday & Hasan (1976), Halliday (1985) and Longacre (1983). Whereas in the taxonomies of Halliday and Longacre

26. Although the notions speech act and illocutionary act are strictly speaking not identical, for the sake of clarity I here use them as mere equivalents. I will come back to these notions in chapter 4. 27. Cf. also Van Eemeren and Grootendorst (1984), who distinguish between illocutionary forces (to a certain extent comparable with speech acts) at sentence level and illocutionary forces at a 'higher' textual level, and discuss the concept of argumentation as an example of higher order speech acts. Note that Van Dijk's term 'macro-speech acts' (1977: 232-247) refers to an altogether different phenomenon and should not be taken into account here.

speech act relations

19

relations of the semantic (representational/external) and pragmatic (functional/internal) type are put more or less on a par with one another, the speech act approaches consciously keep them apart and try to indicate how they may interrelate and co-occur in a text. In other words, in the speech act approach the undifferentiated semantic-pragmatic lists of relations are replaced by a bilevel system in which semantic relations may serve as 'input' to pragmatic relations, giving rise to higher order speech acts. This has been illustrated by example (11), in which both a semantic relation of result and a pragmatic relation of conclusion can be discerned. A second characteristic of speech act oriented studies on text relations is the interest they take in relational devices and mechanisms such as the use of connective particles (cf. parameter (ii) on page 9). In a way, the observation that ostensibly synonymous causal relators in for instance Latin, French and German in fact have different distributional properties can be said to have urged the fundamental distinction between semantic and pragmatic types of relationships: both notions appear to be indispensable for describing German wei! as opposed to da and denn, French parce que as opposed to puis que and car, and Latin quia as opposed to quoniam and nam. Another feature typical of research on speech act relations is a lack of interest in determining the whole range of possible relationships between subsequent speech acts. Not surprisingly, most studies concentrate on the group of cause-like relations explanation, motivation, justification and conclusion. It remains to be seen, however, whether the proposed description in terms of illocutionary layering is applicable to other kinds of speech act relations as well, such as addition or contrasfs. A final property of studies on speech act relations - and their major drawback - is that the object of study is usually the isolated speech act sequence. This means that the wider discourse context is not taken into account, and that there is no systematic analysis of authentic natural discourse. In this respect it differs from the discourse-pragmatic approach to text relations, to be discussed in the next section.

28. The most extensive inventories of speech act relations that I know of can be found in Van Dijk (1981: 265-284) and Ferrara (1985: 146-148), and contain relations like justification, conclusion, explanation, comment, repetition, expansion and preparation.

20

§ 2.4.3

2.4.3. discourse-pragmatic classifications

Discourse-pragmatic approaches to text relations incorporate the achievements of both the semantic-pragmatic and the speech act approach discussed above: they adopt the semantic-pragmatic inventories of relations, and employ them for the description of relations that hold between propositional contents, speech acts or configurations of speech acts. The main assets of discourse-pragmatic studies as compared to the other approaches, is that they are interested not only in the relations between subsequent speech acts and propositions, but also in broader links between more complex units of discourse, and in the overall hierarchical structure of a textz 9 • One of the best-founded proposals for a discourse-pragmatic theory of text relations has been provided by Sandra Thompson, William Mann and a group of close collaborators, under the name of Rhetorical Structure Theory (RST)30. The name Rhetorical Structure Theory stands for: a theory of the structure of texts which is based on rhetorical relations between text parts. The structures are called rhetorical on account of the fact that the underlying relations reflect the writer's purposes and options of organization and presentation of the text, and that they are defined in terms of the effects they are intended to produce. The taxonomy presented by RST is thus stated in terms of discourse organizing principles rather than in terms of semantic principles of clause combining, as is the case in most other research on text relations. The text structure that is disclosed by these rhetorical relations is essentially hierarchical and recursive: by the same set of relations elementary parts of the text are combined into increasingly larger parts, which in the end produce the entire text3 1 • Rhetorical Structure Theory can be regarded as a 'top-down' approach to research on text relations, in view of the fact that it claims to be independent of linguistic realizations and concerned wholly with abstract (cognitive) entities, that is, with the structure and combinations of meanings and intentions underlying the text parts. RST does not, however, rule out the possibility of

29. Although several taxonomies of semantic-pragmatic relations claim that the same sort of relations hold on all levels of the text, usually no connection between text units larger than the clause are actually taken into consideration. 30. Introductions into this theory can be found in: Mann & Thompson (1987; 1988; 1989); Matthiessen & Thompson (1988); Mann, Matthiessen & Thompson (1989). 31. RST admits that in addition to a rhetorical structure other types of structure can be discerned in texts as well, and that some texts (e.g. laws) may even lack a rhetorical structure.

discourse-pragmatic classifications

21

being a useful descriptive tool for a variety of linguistic issues, such as the grammar of clause combining and the meanings of conjunctions32 • A claim to utility in the latter area remains, however, to be substantiated: to date no study has come to my notice in which the RST framework is successfully applied to the description of relators. Figure 5 below gives an overview of the set of rhetorical relations that has been developed within RST in a series of publications from 1983 onwards33 • The relations are grouped according to two basic distinctions. The first distinction (Subject matter vs. Presentational) corresponds more or less to the difference between semantic and pragmatic relations referred to in 2.4.1 34 : in the taxonomy of RST some relations pertain to the Subject matter of the text, while other relations concern the success or felicity of a rhetorical act and are used to facilitate the presentation process of the text. The second distinction covers the fact that a text may be composed of both symmetric (List) and asymmetric (Nucleus-Satellite) relations. In a Nucleus-Satellite relation, which is the more frequent type, one conjunct of the related pair is ancillary to the other, more central conjunct. In a List relation neither member of the pair is ancillary to the other. 'Central' and 'ancillary' are used in a strict functional sense: they stand for central and less central with regard to the communicative goals of the writer35 . Note that in figure 5, which is based on Mann & Thompson (1989), no cross-classifications are identified, that is, the items distinguished do not occur in more than one box. It is not altogether clear whether RST would admit such cross-classifications, but there are some indications that it does. For instance, in contrast to the representation in figure 5, in Matthiessen & Thompson (1988) the rhetorical relation Solutionhood is grouped among the Presentational type of relations and Concession among the Subject matter type. This apparent inconsistency might imply that some of the rhetorical relations belong to both categories. In my opinion, the rhetorical relation Sequence would also qualify for this double status (cf. examples 3 and 4 in 2.4.1).

32. Cf. e.g. Matthiessen & Thompson (1988) for a discourse account of the notion of 'subordinate clause'. 33. The set of relations is based exclusively on a sample of expository written monologue in English. RST admits the possibility of other sets for other genres and languages. 34. Cf. Mann & Thompson (1989). 35. This asymmetry is recognized by many other studies on text relations and text structure. Cf. e.g. Van Dijk (1981); Pike & Pike (1983); Roulet et al. (1985). Judgements about what is central and what is less central in a text are usually not based on linguistic evidence. In chapter 4 I will return to the asymmetry issue in more detail.

22

§ 2.4.3

figure 5: Rhetorical Structure Theory

Subject matter

Presentational

Nucleus-Satellite

List

Elaboration Circumstance Solutionhood Volitional Cause Volitional Result Non-volitional Cause Non-volitional Result Purpose Condition Otherwise Interpretation Evaluation Restatement Summary

Contrast Sequence Joint

Antithesis Concession Motivation Background Enablement Justify Evidence

The cross-classification problem sketched above arises from the fact that RST fails to make sufficiently clear in what respects Subject matter relations differ from Presentational relations, and whether and how they interrelate. Why, for instance, is Evaluation ranked among the Subject matter relations, but Background among the presentational ones? This problem automatically raises the question as to the underlying motivation for setting up this specific (and no other) taxonomy. As the theory stands now, the proposed taxonomy of text relations does not really differ from the 'undifferentiated' semantic-pragmatic taxonomies as discussed in section 2.4.1 (cf. p. 12-17)36. By way of illustration of the above discussion on RST I reproduce here the RST analysis of a short text taken from Mann & Thompson (1989): 36. For an attempt to obviate the drawbacks of the RST-taxonomy, cf. Sanders (1992), who proposes a taxonomy which is based on a restricted number of cognitive primitives and strives after a greater psychological plausibility.

23

discourse-pragmatic classifications

(14) The Transfer of Technology to Underdeveloped Countries 1. The elimination of mass poverty is necessary to supply the motivation for fertility control in such countries. 2. Other countries should assist in this process, 3. not least because they have a moral obligation to do so.

In this text fragment text part 1 serves as a background for 2 and 3 together, which themselves are mutually connected by a non-volitional cause relation. The hierarchical organization of the text in (14) is rendered in RST as follows:

1-3

(15) background

2-3 non-volitional cause

2

-----

3

The essential characteristic of the analysis in (15) is the hierarchical organization of the text: the basic units (clauses) 2 and 3 are connected by a relation of non-volitional cause and together constitute a more complex text part. This complex text part is on a higher level connected to unit 1 by a background relation. The vertical lines in (15) point to the central text parts (Nuclei) (Le. central for the communicative goals of the writer). Rhetorical Structure Theory is one of the most elaborated, though by no means the only proposal for applying the semantic-pragmatic type of relationships to text segments larger than clauses. It acknowledges for instance its debt to previous studies by Grimes (1975) and Hobbs (1985)37, in which, too, large

37. Grimes' list includes alternation, specification, equivalence, attribution, explanation, collection, covariance and adversative. He calls these relations "rhetorical predicates". Hobbs distinguishes occasion, evaluation, background, explanation, expansion, parallel, violated expectation, contrast, elaboration and exemplification. The main difference from RST is that Hobbs's relations (which he

24

§ 2.5

sections of text are grouped according to a small number of organizing relations. In addition we can mention Longacre (1979) as a rudimentary attempt to describe paragraph structure along the lines of a taxonomy of semantic-pragmatic relations. A much more elaborate proposal in this direction is the Geneva model outlined in Roulet et al. (1985), which I discuss in chapter 3. In chapter 4, where I draw the outlines of an analytical framework for the description of connective particles, some of the ideas of RST will return, although in a slightly altered form. The fundamental distinction in RST between symmetric and asymmetric relations appear to be especially useful. I take over also the RST -term 'presentational relations', which, however, I use for a better defined and slightly different set of phenomena.

2.5 Formal and structural classifications To round off this overview of the literature on text relations I turn finally to the longstanding tradition of dealing with text relations from a formal or structural point of view, that is, with relations that are formally incorporated into linguistic structure. In this approach the term 'conjunction' is reserved for the syntagmatic relationship which organizes grammatical units (clauses) hierarchically into larger units (clause complexes )38. Traditional sentence grammars commonly distinguish two types of formal conjunction: subordination and coordination. Lately, in the last two decades, there have been several attempts to study the grammatical organization of clauses into clause combinations of various sizes up to whole texts, a discipline which has come to be known as Text Grammar. Text grammars resemble sentence grammars in that both attempt to describe relationships between grammatical units of text. With regard to the issue of text relations the difference between the two is a matter of scope: sentence grammars restrict their description of relations to combinations of contiguous (sub)clauses (local scope), whereas text grammars also take relations between larger text parts into account (global scope).

calls 'coherence relations') are defined in functional terms of text interpretation and reception, instead of in functional terms of writer's goals (i.e. text production). 38. Note however that in Halliday & Hasan (1976) 'conjunction' is used in a non-structural, semantic sense.

formal and structural classifications

25

From the above description it is clear that studies of the structural aspects of text relations automatically involve the formal determination of the conjuncts, that is, the identification of formal units of structure. Especially in work on structural units larger than the (complex) clause (as in text grammatical research) the formal determination of such units has appeared to cause serious problems. For studies that concentrate their description on the semantic or pragmatic aspects of text relations (as in the field of Discourse Analysis) the formal determination of the conjuncts is less a matter of debate, since in these approaches the conjuncts are essentially defined in cognitive terms, that is, as units of content or of interaction, respectively. 2.5.1 clause combining

In traditional syntax the nature of clause combining has always been described in terms of subordination versus coordination39 • This dichotomy is not uncontroversial, and especially the notion of subordination has been a matter of much dispute in recent linguistic research. It has been pointed out more than once that the concept of subordination, if applied universally, covers a number of widely differing phenomena, and that accordingly it is impossible to give a single, unitary definition of the grammatical category of subordinate clauses. Hence, instead of the bipartition subordination - coordination a richer notional apparatus has been proposed, for instance by Lehmann (1988). According to Lehmann one should assume a continuum of syntactic clause linkage types, with parataxis (minimally tight linkage) and embedding (maximally tight linkage) as end poles and a host of subdistinctions in between. Where a particular clause linkage type is located in this continuum is dependent on the interplay of half a dozen parameters such as the degree of desententialization of the 'subordinate' clause, and the degree of integration of one clause into the other40 •

39. The syntactic notion of clause is usually defined in terms of its constituency, i.e. as a unit of language that consists of at least a predicate and one explicit or implicit argument. Following Halliday (1985) a structural combination of two of such units could be called a clause complex. In more conventional studies the term sentence is used for clauses as well as clause complexes. 40. The ideas were originally advanced by systemic linguists like Halliday, and have been adopted by, among others, Haiman & Thompson (1984) and Matthiessen & Thompson (1988). The tripartition coordination Isubordination Icosubordination as presented in Van Valin (1984) and in Foley & Van Valin (1984) resembles to a certain extent the parataxis Ihypotaxis lembedding distinction.

26

§ 2.5.2

Since the issue of clause combining per se falls strictly speaking outside the scope of my study, I will not elaborate further on it here. For a characterization of the Latin system of clause linkage in accordance with Lehmann's parameters I refer to Lehmann (1989: 177), for a discussion of embedding constructions in Latin to Bolkestein (1989). 2.5.2 text structure

Most linguists regard the combination of clauses into clause complexes as the upper limit of grammatical structure. A broader view is taken by the linguistic current that tries to reveal the grammar of texts by using the same concepts and methods as those applied in traditional sentence grammars for the description of the syntax of clauses. A text grammar thus idealy describes what sorts of segments texts have, and what patterns of constituent categories may underlie the overall structure of a text (e.g. in terms of paradigmatic choices and possible cooccurrence restrictions). Theoretically speaking a text grammar that fulfils these conditions will be able to predict the well-formedness and illformedness of stretches of texts, much in the same way as sentence grammars predict the grammaticality or ungrammaticality of clauses. It is not surprising, however, that text grammatical studies of this type have commonly failed to give an indication of what such macro-syntactic structures and constituents might look like. On the level of the text the syntactic dependencies (if any) appear to be looser and much more variable than in clauses, and the theoretical machinery of sentence grammars by no means suffices to capture the structural organization of texts. This can be explained by the fact that a text is the outcome of a dynamic process of communicative interaction, in which an indefinite number of speaker goals may give rise to an almost indefinite number of linguistic strategies, which in turn may be somehow reflected in the text structure. Text structure is therefore hard to formalize. This is not to say that there are no grammatical features of single clauses that have a textual function, or, in other words, that it is not linguistically relevant to distinguish text units larger than the clause or clause complex. On the contrary, especially in the field of Discourse Analysis a number of linguistic phenomena have been adduced which cannot possibly be accounted for by any grammar that confines itself to the single clause (complex). One of these phenomena forms the central issue of this book, namely the occurrence and function of connective particles in a text.

text structure

27

A good example of how connectives may for instance play a role in segmenting a text into macro-units has been provided by Levinsohn (1987) for New Testament Greek. Levinsohn demonstrates that the difference in distribution of the near-synonyms Oi and Kat (both roughly meaning 'and') in the narrative passages of the Acti 1 can be explained only in terms of a supersentential 'development unit' (DU). The text of the Acts may be viewed as a series of such units. Each DU builds on earlier DUs, and each represents a new development of the story. A DU may be introduced by the connective Oi, but not by Kat, which in turn is reserved for the linking of smaller units within a DU42 • Applied to Acts 5:1-6 this yields the following schema (Levinsohn 1987, p. 84):

liE A certain man called Ananias sold a property Ka{ kept back part of the money Ka{, bringing the rest of the money, laid it at the apostles' feet.

liE Peter condemned Ananias for seeking to deceive the Holy Spirit.

liE Ananias fell down and died Ka{ fear came on all who heard.

liE The young men wrapped up the body Kat, carrying it out, burried it

The above analysis proves that 'macro-structure' and 'macro-structural unit' are linguistically useful concepts, at least as far as the description of the function of connectives is concerned43 •

41. The area of biblical studies has proved to be an inexhaustible source of linguistic observations on text structure. Cf. Grimes (1975: 18-20) for a discussion. 42. For the same observation for classical Greek cf. Ruijgh (1971: 129-135) and Bakker (1993). 43. An analogous linguistic phenomenon which has frequently been mentioned (following Giv6n: 1983) as playing an important role in the segmentation of a text is the concept of continuity/discontinuity. Lexical continuity (continuity of participants, time, place or action) and morphosyntactic continuity (continuity of tense, mood, aspect, voice etc.) may contribute to the division of a text into 'topic units', while discontinuity points to boundaries in the text structure. Especially the use of proforms and zero-anaphora (in covariation with the use of full noun phrases) has been described as a linguistic means for the coding of continuity/ discontinuity, thus providing evidence

28

§ 2.5.2

The above discussion makes clear that the term 'text structure' can be applied to different phenomena, such as to formal structures above the clause, or to functional patterning. Indeed, in the linguistic literature the term 'text structure' is often used in a vague, intuitive way, without specific indication of what is meant by it. Equally confusing and ill-defined are the various terms that are found to refer to the component elements of the assumed textual structures. The following overview is an attempt to disentangle the composite notion of text structure, which is needed to keep the discussion in the ensuing chapters (esp. ch. 4) as lucid as possible. While studies on text structure and text relations often seem to concentrate on only one aspect of text structure, I think it necessary to take into account at least the following three aspects: figure 6: types of structure on the level of the text text structure

constituent units

formal structure thematic structure linguistic action structure

units of form units of content units of communication

As I stated above, 'text structure' is used in the literature first in a strictly formal sense to denote the global-level counterpart of syntactic clause structure. The conjuncts around which these structures are built are essentially units of form, for instance 'paragraph' or 'chapter'. This narrow conception of the notion of text structure (adhered to by the linguistic current known as Text Grammar or Text Linguistics, and built on the principle of constituency relations) has proved to be difficult to maintain, and led Halliday & Hasan to their statement that there are no levels of structure above the clause. The relevance of distinguishing formal categories such as 'paragraph' or 'chapter' (for other than typographical reasons) still remains to be shown44 .

of the need to identify larger units of structure. A range of other linguistic phenomena could be mentioned here as calling for an explanation in terms of text structure: for example, the fact that paragraphs can be referred to anaphorically or cataphorically by means of a proform like it; the existence in some languages of particles and clitics that indicate paragraph introduction or closure, as Longacre (1979) demonstrates for Huichol, Shipibo and Capanahua; switch reference systems; and specific word order patterns at major text structure boundaries. 44. Text Grammars have for instance been proposed by Harweg (1968) within the structuralist tradition, by Pike (1967) in the framework of tagmemic theory, and in a number of studies by Van Dijk, PetOfi and others within a Transformational framework. A more recent Transformational approach to text structure can be found in Rubattel (1988).

text structure

29

The term text structure is used also in relation to configurations of units of content in macro-semantic structures. In order to distinguish this type of text structure from the type of structures studied by text grammarians, one could call the former theflUltic structure. The units involved in thematic structure are based on at least two principles: (i) repeated reference to the same set of entities (usually called referent or topic continuity); and (ii) conventional knowledge of stereotypical sequences of actions or events ('scripts' or 'routines'). The first principle is studied especially within the field of Discourse Analy45 sis , the second within so-called Story Grammars46 . A good illustration of the macro-semantic (= thematic) organization of a text is the Levinsohn analysis of Acts 5: 1-6 discussed above. In order to distinguish terminologically complex units of content from supposed complex units of form, one could, for example, name the former 'thematic chains' and 'episodes', the latter 'paragraphs' and 'chapters'. The third type of structure on the level of the text can be called linguistic action structure. This type of structure accounts for the fact that texts can essentially be viewed as products of linguistic interaction, displaying complex systems of units of communication. In order of increasing complexity these units could be called act, move, exchange, transaction and interaction47 . Linguistic action structure is the primary object of investigation of the field of research called Conversation Analysis48. In chapter 4 I will return to this type of structure in more detail. Today, it seems, more and more linguists are realizing that text structure is a composite concept which should accordingly be approached from a more

45. This line of research has been pursued especially in the wake of the well-known volume on topic continuity edited by Giv6n (1983); see also note 43. 46. Cf. Rumelhart (1975); Van Dijk (1977: ch. 5). Story Grammars usually seek to draw strong parallels between linguistic structure and story structure, but are essentially non-linguistic. For critical remarks cf. Goldberg (1983). 47. The terms exchange, transaction and interaction more or less speak for themselves. The term act is reminiscent of the concept of speech act introduced by Austin (1962) and Searle (1969). Instead of 'move' sometimes the terms 'speech act sequence' or 'illocutionary act complex' are used. The communicative notion 'move' is to be distinguished from the formal notion 'turn' (i.e. tum of speaking): a move may be extended over more than one tum, while a tum may consist of more than one move. 48. A representative and much-cited study in this field is Sinclair and Coulthard (1975) on classroom-interaction. Note that despite its title Towards an analysis of discourse, the study is conducted mainly from a Conversation Analysis point of view.

30

§ 2.5.2

pluralistic point of view49 • From such a viewpoint text structure involves on the one hand semantic and interactional patterns (what might be called coherence structures) and, on the other hand, the ways in which these patterns are formally expressed on the other (cohesion structures). For this 'holistic' approach to text, which covers any study of systematic patterning above the clause, Enkvist (1989) has proposed the name Discourse Linguistics, with the key notion of connexity as the sum total of cohesion plus coherence. It is perhaps needless to say that to date there exists no fully worked out model of text/discourse structure (or, for that matter, a fully worked out text/discourse grammar) in the integrated sense suggested by Enkvist, although it is gradually becoming clearer what such a model should look like50 • A major problem appears to be that discourse, being a dynamic process, is much harder to formalize than the static product that is formed by an isolated, decontextualized clause. Another complicating factor is the incorporation in such a model of genre and text type, which have proved to be of great importance for the way a particular discourse is structured.

2.6 Evaluation of the usefulness of relation taxonomies It has long been common practice to describe and explain languages' systems of connective particles in terms of predetermined taxonomies and theories of text relations. In this chapter I have taken a closer look at traditional and current approaches to text relations to see what they are like and on what criteria they are based, in order to determine to what extent they may be of

use in a study on Latin discourse connectives. It turns out that such a 'top-down' approach to the study of connective particles is not without its problems. For one thing, the field of text relations is covered by a wide variety of studies which differ in aim, scope, the degree to which they take linguistic expressions into account, and, most importantly, the level of linguistic description (semantic, pragmatic or formal/structural)

49. Cf. e.g. De Beaugrande & Dressler (1981), Ellis (1987), Meyer (1987) and the discussion in Enkvist (1989). 50. For an attempt to develop an integrated model of discourse structure in the Systemic tradition I refer to Fawcett (1980). Note that I use 'text' versus 'discourse' to signal the difference between the product of speech or writing that has been detached from its situational context, and the dynamic process of speech and writing in its situational context, respectively (cf. e.g. Enkvist 1989).

evaluation of taxonomies

31

they concentrate on. It appears that none of these taxonomies or theories alone can account fully for the function and distribution of the particles under discussion in the present study: a first, superficial exploration of the Latin material reveals immediately that a single relation type may be expressed in Latin by several connective particles, and, conversely, that a single connective particle may be associated with more than one relation type. This problem is most apparent if one tries to describe the particles involved in terms of the conventional (logical-)semantic taxonomies of text relations (cf. 2.3). For instance, even in a more sophisticated logical-semantic approach the Latin connective items quod, quia, cum, quoniam, quando, enim and nam can all be described as associated, in one way or other, with a relation of the causal-conditional type, subcategory cause (see section 2.3). In traditional grammars these connective particles are accordingly summed up in a catalogue manner under the rubric of causal conjunction, with the subordinationcoordination distinction as the main criterion for further differentiation. This line of approach conceals, however, interesting differences between the individual particles, as I will show in the descriptive chapters in the second part of this book. Semantic-pragmatic taxonomies of text relations (2.4.1) differ from the logical-semantic ones in that they add a parameter for further subcategorization. This parameter accounts for the observation that there are two (or perhaps even three: cf. p. 16-17) 'realities' in which the 'basic' text relations may obtain: the external reality of the reported events, and the internal reality of the communicative interaction (the 'here and now' of the interlocutors). The application of this parameter yields more refined inventories of relations. The 'basic' relation of cause, for instance, may be divided into the external relation cause, and the internal relations justification and evidence, as I argued in 2.4.1. Nevertheless, even when we try to apply the labels of these more refined semantic-pragmatic taxonomies of text relations to the system of 'causal' relators in Latin, the results are quite disappointing. It is admittedly true that the taxonomies concerned help us to differentiate between, for example, quod and quia on the one hand (indicators of external cause relations), and nam and enim on the other (apparent indicators of internal cause relations). On closer inspection, however, it appears that the essential features of nam and, especially, of enim cannot be accounted for adequately in terms of internal cause relations such as justification and evidence. What is more, the difference in function between nam and enim, which one feels intuitively and which reveals

32

§ 2.6

itself in their highly differing distributional properties, remains fully obscured if one tries to describe the particles in terms of common relation taxonomies. As for nam, I will argue (see ch. 7) that its distinguishing feature has to do with the marking of the subsidiary status of the nam-unit with regard to some other unit. On account of this feature nam may occur both in justification conjuncts and in evidence conjuncts, but also in text units which function as exemplification, clarification, background or afterthought with respect to the preceding context. This amounts to saying that the Latin discourse connective nam does not correspond to one single item in a semantic-pragmatic taxonomy of causal relations. In (16) and (17) examples are quoted of nam in an evidence and justification context: (16) times ecastor; ... nam palles male ('good heavens, you're frightened; ... for you are awfully pale', PI. Cas. 982) (I 7)

nos autem (nam id te scire cupere certe scio) publicis consiliis nullis intersumus ('as for me - for I am sure you want to hear about it - I take no part in public deliberations', Cic. Att. 2.23.2)

It appears to be even more difficult to match nam's alleged synonym enim with one single item in a semantic-pragmatic taxonomy of causal relations. This is due partly to the fact that most relation taxonomies (including the discoursepragmatic) confine themselves to consideration of the relations of text units with the textual context only. In chapter 8 I will argue that the relational force of enim should be described rather in terms of the extratextual, communicative situation. The problems encountered in a taxonomy-based description of the 'causal' connectives nam and enim are not exactly isolated. As a matter of fact, as became clear in the course of my investigations, the functions of most discourse connectives in Latin (and presumably in other languages as well) are too complex to be fully captured by the labels of any current, semantically or

pragmatically oriented, taxonomy of text relations 51 : for an adequate description of the function of most discourse connectives we should have recourse in addition to such concepts as text organization and conversation management. Hence, what would be needed in a 'top-down' approach to the study of relators instead of the common relation taxonomies, is a full-fledged multi-

51. Such taxonomies may presumably be applied with more success to the study of grammatical clause combining and the matching relators (mostly subordinators).

evaluation of taxonomies

33

dimensional model of hierarchical discourse organization, in which there is room also for elements of the non-verbal, communicative context. The development of such a model is in full progress, but, as I pointed out in 2.5.2, far from being completed. We have, however, been provided with many interesting insights in the last few years, especially from the fields of Discourse Analysis and Conversation Analysis. As far as the study of connective particles is concerned, we can mention for instance the useful observation that relators may playa role in the structural organization of a text (cf. 2.5.2), and that texts typically consist of central and less central parts (cf. 2.4.3). It is my conviction that 'bottom-up' studies of discourse phenomena such as connective particles may contribute considerably to the realization of a discourse model in the integrated sense referred to above. A few such studies will be discussed in the next chapter.

3

RELATORS: APPROACHES TO PARTICLES

3.1 Introduction

Relations in texts can be expressed or signalled by a number of linguistic means, which together might be called relators. These include all kinds of lexical devices (nouns like conclusion or consequence, verbs like to add or to summarize, prepositional phrases such as in spite of or owing to, adverbs like therefore), but also a phonological-grammatical device such as intonation, and, most characteristically, the use of 'particles'. The present chapter deals with the elusive category of particles, and with a number of representative studies which in one way or another are concerned with particles, especially with connective particles. Ever since the term particle was introduced in ancient rhetorical and grammatical theory it has been used quite loosely and for a number of different phenomenal. For a long time the category was defined in a negative way, as a waste-basket term for those words which cannot be assigned to any of the familiar word classes (parts of speech), that is, for those words which are not a (pro)noun, verb, preposition, adverb, interjection or conjunction. Alternatively the term 'particle' was used in a strictly morphological sense, to refer to all invariable or indeclinable words. Taken most broadly in this sense the term could apply to interjections, conjunctions, adverbs and prepositions. In modern linguistics (say the last 25 years) it has been more usual to define the category of particles on semantic or functional rather than on formal or morphological grounds. The most recent definition (also stated in negative terms) comes from Foolen (1993: 13) and is reminiscent of the well-known

I. For the various uses of the term 'particle' in the Latin grammatical tradition cf. Pinkster (1972: ch. 8 note 2) and especially Schenkeveld (1988). Cf. also Sicking (1986: 126): "De inhoud van de als 'partikels' aangeduide verzameling blijkt van auteur tot auteur te verschillen. Kennelijk hebben we te maken met een restgroep van woorden die men om verschillende - historische, semantische of systematische - redenen niet elders heeft willen behandelen. (... ) De onzekerheid en onduidelijkheid die hier aan de dag treden zijn historisch terug te voeren op het feit dat ook de antieke grammatici - wier classificaties van de taalverschijnselen de beschrijving van het Grieks en het Latijn zo ingrijpend hebben beinvloed - voor een aantal zogeheten partikels geen overtuigende plaats hebben gevonden in het systeem van de taal zoals zij dat zagen". For an extensive discussion of the treatment of particles (notably conjunctions) by the ancient grammarians I refer to Baratin (1989).

introduction

35

distinction between content words and function words: particles are words that do not contribute to the propositional content of a sentence or utterance. Thus defined, 'particles' cover more or less the following form classes: 1. 2. 3.

interjections (well, y'know, I mean) conjunctions (and, but, for) some of the adverbs (certain modal adverbs like English indeed, maybe or German ja, doch, also called 'modal particles'; connective adverbs like so or anyhow, also called 'connectives' or 'connectors'; intensifying adverbs like just or even, also called 'focus particles ')

With regard to the above it should be mentioned that the difference between propositional and non-propositional is not always unproblematic, especially in the case of the modal adverbs. An as-yet-unresolved question in this respect is whether one should differentiate between modal adverbs and modal particles: although from a grammatical viewpoint they display different distributions (as has been shown for Latin by Pinkster 1972, § 8.1), they appear to be functionally equal in that both reveal the language user's attitude to the propositional content of the utterance (rather than themselves being part of the propositional content? I will not pursue this issue further here. For a general discussion I refer to Foolen (1993: 26-32). In my discussion of vero in chapter 11 (section 11.1) I will come back to the problematic distinction between adverbs and particles. In the present study I will employ a rather broad, functional definition of the term particle: particles are those invariable words which have in common that they fit their host unit3 into a wider perspective, which may be the surrounding verbal context and its implications, or the communicative situation in which the text is integrated. In modern linguistics there are, broadly, two currents in particle research, differing as to the type of particles focused on and the approach adopted. On the one hand there is a strong German tradition in research on particles in a 2. In Foolen (1993: 37) it is suggested that modal adverbs pertain primarily to attitudes to the propositional content of the utterance, whereas modal particles primarily indicate attitudes to the illocutionary function of the utterance. No details are given however. 3. I use the term 'host unit' for the particular stretch of text to which the force of the particle involved pertains. The concept of 'unit' or 'host unit' is not necessarily coextensive with the concept of 'clause': it stands rather for a pragmatically (and not formally) determined stretch of text, and may apply to any kind of information-conveying text segment, ranging from single words to a cluster of several sentences.

36

§ 3.1

narrower sense, that is, on modal and focus particles 4 • This tradition started with Weydt (1969) and is characterized by a predominantly semantic approach. Recent studies in this field are for instance Konig (1991a) and Abraham (ed. 1991). The other current in modern particle research is concerned mainly with linguistic expressions that indicate how a discourse unit is integrated into the discourse context. The group of particles investigated includes items from the classes of coordinating conjunctions, connective adverbs and interjections (e.g. and, but, so, y'know, I mean, etc.), and has become known under the name of discourse markers (French marqueurs du discours). As the designation 'discourse marker' already implies, the approach can be characterized as discourse-oriented. The Anglo-American branch of this current (with Schiffrin 1987 as its main representative) is strongly rooted in the ethnomethodological and sociolinguistic tradition. The French research on discourse markers (with the Geneva school around Roulet as its centre, cf. Roulet et al. 1985) has a stronger affinity with the semantic tradition, more specifically with the argumentation theory of Ducrot. The particles under discussion in the present study (notably nam, enim, autem, vero and at) can be regarded as a subset of the larger category of discourse markers, the latter including also words that have a lexical meaning of their own (cf. e.g. French a propos or English I mean). For this reason, and in order to emphasize their general 'relational' function, I will refer to the group of Latin particles involved as discourse particles or discourse connectives rather than as discourse markers 5 • In view of the goals of the present study I will focus in the following sections mainly on accounts of the more 'connective' type of particles. I wish to emphasize, however, that the difference between these two types of particles could be less sharp than the above overview probably suggests: as will become 4. Unifonn definitions of modal and focus particles are still lacking. Modal particles (such as Gennanja or doch) are usually held to indicate the attitude of the language user to the transmitted content, or to modify the illocution type of the sentences they occur in (see note 2). Focus particles (such as Gennan schon and noch, or English only and even) have only part of the proposition in their scope. They are functionally characterized by their property of invoking possible alternatives for the idea or concept focused on. Whereas focus particles occur more or less universally, modal particles seem to be much more language specific: they abound in e.g. Gennan, Dutch, the Slavic languages and the Scandinavian languages, but do not seem to be a relevant category in English and French. Latin is commonly regarded as a language with relatively few modal particles, as compared to e.g. Ancient Greek. Recent studies on focus particles are e.g. Bakker (1988) on Ancient Greek 7r€P and Konig (l99Ja). 5. Another reason for my choice of the tenn 'discourse particle' (or 'discorse connective') instead of 'discourse marker' is that I want to avoid the impression that the description of the particles in the present study is fully in line with Schiffrin's model.

introduction

37

clear from the discussion of the Latin particles in part II, it is not uncommon for particles to display 'connecting' as well as 'modal' features. More or less analogously to the discussion in chapter 2, the survey of the literature in this chapter has been divided into three sections, one on formal or syntactic accounts of particles (3.2), a second on semantic and semanticpragmatic approaches to particles (3.3), and a third on discourse/conversation approaches to particles (3.4). At relevant points some of the classical questions of particle research will be brought up (Do particles have meaning? If so, in what terms should this meaning be described? What is their semiotic status? How are we to account for their polyfunctionality?). In the last section I will evaluate the various approaches in terms of their usefulness for the goals of the present study, that is the study of a number of Latin discourse connectives and the development of an analytical framework for their description. The survey of the literature in this chapter is inevitably incomplete. For more general and complete overviews of the state of the art in particle research I refer to Wolski (1986), Wierzbicka (1986) and especially Foolen (1993).

3.2 Formal and syntactic accounts of particles In the preceding section I mentioned two different formal definitions of 'particles': (i) particles are words that cannot be assigned to any of the conventional parts of speech, and (ii) particles are invariable (or indeclinable) words. These definitions correspond to two formal or syntactic approaches to particles and particle research. The first 'approach' is typically associated with American structuralism and Chomskyan linguistics, whose interest in particles understandably amounts to a virtual neglect of the category: particles are obviously outside the syntactic structure of the sentence and therefore not amenable to linguistic inquiry. The consequence of this position is that particles are regarded as embellishments or meaningless fillers, whose occurrence is not governed by rules and dependent mainly on the idiolect of the language user. It need hardly be said that this position is linguistically quite unsatisfactory6.

6. The view that particles are meaningless fillers is not exactly new. Cf. e.g. Servius' comment (4th century) on Vergil A. 1.331 (sciendum multas particulas ad ornatum pertinere) and A. 8.84 (vacat enim et tantum ad ornatum pertinet). For a comment on the use of enim in the latter

38

§ 3.2

A second, more interesting formal approach to particles consists in attempts to subclassify invariables according to their syntactic behaviour. Characteristic of this approach is the development and application of syntactic tests in order to corroborate intuitive subdistinctions of the group of invariables. The rationale behind these tests is that under identical syntactic conditions invariables may behave differently or identically. Research in this field centres in large part on the issue of the grammatical behaviour of adverbs in a strict sense in contrast with other invariables7 • For Latin I refer on this subject to the study of Pinkster (1972), in which he discusses (among other things) the differences in syntactic distribution of adverbs in a strict sense on the one hand, and interjections, prepositions, connectors (see below), and subordinators on the other. As to the difference between anaphoric (causal) adverbs (e.g. ideo, 'therefore') and consecutive connectors (e.g. igitur, 'so'), which share the semantic role of indicating a cause-result relation between two clauses, Pinkster mentions inter alia the following criteria (ibidem, pp. 156-62):

* * * * *

causal adverbs (in contrast to consecutive connectors) can be used both alone and in a correlative pattern causal adverbs cannot occur in questions introduced by cur? ('why') or quin? ('why not?') causal adverbs can be combined with consecutive connectors but not with other causal adverbs; the same holds mutatis mutandis for consecutive connectors causal adverbs can occur in relative and so-called adverbial clauses which follow the main clause causal adverbs can occur In the second or following member of a coordination pattern which consists of two or more coordinated clauses

In the same vein Rudolph (1979) mentions a number of syntactic criteria for distinguishing between modal adverbs and modal particles in German, for instance:

fragment I refer to chapter 8, p. 183 and 204. Cf. also Kroon (1992). 7. The intuitive notion 'adverbs in a strict sense' covers those invariables which are closely integrated into the syntactic structure of a clause and closely resemble other sentence elements such as the subject or object. Note that a high degree of syntactic integration corresponds with the semantic property of being part of the proposition. 'Adverbs in a strict sense' contrast with 'adverbs in a broad sense' which cover all invariables.

formal and syntactic accounts of particles

*

* *

39

modal adverbs (in contrast to modal particles) can by themselves form the answer to a question modal adverbs can occupy the first position of the clause modal adverbs can be stressed

For English we can mention the account given by Quirk et aI. (1985, ch. 8), in which adverbs (as part of the larger category of adverbials) are divided into four syntactically defined subcategories called adjuncts and disjuncts (i.e. adverbs in a strict sense), subjuncts (i.e. focus particles) and conjuncts (i.e. connectors in the sense of Pinkster, see below). Adjuncts are said to differ from the other three categories on account of a number of grammatical distributional properties, which they share with for instance subjects and objects 8 . The following list (Quirk et aI., p. 504) can be found in one form or other in many other linguistic studies dealing with the syntactic behaviour of adverbs or particles (see also chapter 5):

* * * *

an adjunct can be the focus of a cleft sentence an adjunct can be the basis of contrast in alternative interrogation or negation an adjunct can be focused by focusing elements an adjunct can be elicited by question forms

The other three categories mentioned by Quirk et aI. lack the above properties because they have (i) a "lesser role than the other sentence elements" (this applies to the subjuncts), or (ii) a "superior role than the other sentence elements", being outside the syntactically integrated clause structure (disjuncts and conjuncts)9. Most interesting for the present study are, of course, syntactic accounts of the subcategory of 'connective particles' or, in a more conventional terminology, 'conjunctions'. Traditionally the category of conjunctions is divided into 'coordinating' and 'subordinating' conjunctions, although terminology is often confused here 1o • In Pinkster (1972, ch. 8), however, it is argued that there

8. For the difference between adjuncts and disjuncts cf. e.g. Pinkster (1972; 1990). 9. The distinction drawn by Quirk between disjuncts and conjuncts seems to be based on semantic rather than on syntactic criteria: "[conjuncts] are unlike disjuncts in not typically filling in the semantic roles characteristic of adjuncts" (p. 631). 10. Cf. Pinkster (1972: 153) for an overview of the confused way in which KUhner-Stegmann and Szantyr treat the Latin conjunctions.

40

§ 3.2

are good (syntactic) reasons to divide the coordinating conjunctions into 'coordinators' on the one hand and 'connectors' on the other. Coordinators (e.g. Latin et 'and' or sed 'but') link constituents (or sub-clauses) that fulfil the same syntactic or semantic function in the sentence. Connectors (words like Latin autem 'however' or igitur 'so'; Quirk et al. 1985 use the term 'conjuncts' or 'conjunct adverbs' for roughly the same group of words in English) lack this syntactic property: they occur rather between independent sentences ll . Thus, the category of connectors is - as Pinkster himself admits - defined in a mainly negative way: it contains those invariable words which seem somehow to have connective capacities, but which do not display the syntactic behaviour of the coordinators (nor, for that matter, of the connective adverbs: see above for the differences between Latin ideo and igitur). In the present study I take this syntactically indeterminate group of words as a starting point to see whether we can come to a satisfactory, functional description of the group as a whole and of some of its members. For the sake of clarity I propose to reserve the term 'conjunction' for the structural type of connective particles, that is, for the so-called subordinators and coordinators. Instead of Pinkster's term 'connector' I will use the term 'discourse connective' or simply 'connective'. As to the merits of the syntactic accounts of particles, we can say that they are useful in that they yield the first linguistically based subcategorization of the highly heterogeneous group of invariables. What is more, by the application of a number of syntactic tests they show that some of the invariables (viz. the modal and focus particles, as well as the 'connectors' or 'discourse connectives') should be regarded as grammatical Einze/ganger, whose proper function or meaning cannot be captured by conventional linguistic instruments and terminology. For the actual description of individual particles, however, the syntactic accounts do not have much to offer. Complementary criteria are needed from the areas of semantics and (discourse-) pragmatics. These are' however much harder to formulate than the syntactic ones.

11. Quirk et al. (1985: 752) mention one specific syntactic test to demonstrate the difference between coordinators and connectors ('conjunct adverbs' in their own terminology): connectors (e,g. so, yet) differ from coordinators in that they can be preceded by a coordinator. The syntax of coordination is too complex to discuss here in more detail. For extensive accounts of the syntactic phenomenon of coordination see e.g. Dik (1968) and Quirk et al. (1985: ch. 13).

semantic approaches to particles

41

3.3 Semantic approaches to particles Semantic accounts of particles (i.e. accounts in which one attempts to assess the proper meaning of a particle, e.g in order to set up semantic subclassifications of particles) are undoubtedly the accounts we are most familiar with from dictionaries and reference books: as a rule, dictionaries and reference books provide us with lists of possible meanings and uses of particles, in an attempt Gustified by their particular aim) to describe the entire 'meaning potential' of individual particles. Such an approach seems to be relatively unproblematic (at first sight, at any rate) in the case of connective particles: in chapter 2 (section 2.3) we have seen that connective particles (including what I call '(discourse) connectives') tend to be described in terms of a more or less universal set of abstract semantic relations, such as additive, disjunctive, causal-conditional, temporalsequential and adversative. Connective particles are held to 'express', 'signify' or 'establish' semantic relations between subsequent clauses. Usually this relation-establishing function is tacitly assumed to be equal to the proper 'meaning' or 'semantic value' of a connective particle. Following the Latin grammatical tradition Pinkster for instance (1990: 253) speaks of additive, adversative, disjunctive, causal, consecutive and continuative connectors. Characteristic of recent, more detailed semantic studies of connective particles (more particularly on connectives), is that the semantic classifications of relations in terms of which the particles are described, usually show the influence of modern pragmatic insights 12 • In section 2.4 I referred to this type of classification as the semantic-pragmatic and speech act oriented classifications of text relations, and discussed their advantages as well as their shortcomings for the study and description of discourse connectives. As to the type of 'meaning' assigned to discourse connectives we can say that pragmaticallyoriented approaches assume 'functional' meanings of connective particles alongside the more conventional 'referential' or 'logical' meanings. The

12. Cf. e.g. Halliday and Hasan (1976) and Quirk et al. (1985) for English, and Anscombre & Ducrot (1977) and Ducrot et al. (1980) for French. For Latin this more pragmatically-oriented approach to connective particles can be found in articles by e.g. Fugier (1987; 1989), Bolkestein (1991) and Mellet (1994; 1995). An exception to this tendency is the rather unsuccessful attempt by Warner (1985) to set up a semantic taxonomy of English discourse connectives (such as otherwise, but, so and unless) from a generative semantics point of view, in which the connectives do not have an independent meaning or function, but are defined in terms of the truth conditions of the discourse fragments in which they appear. Descriptions of connective particles within a truthconditional theory of meaning have in general proved to be inadequate.

42

§ 3.3

introduction of pragmatics into the study of connective particles eventually led to a distinction between semantic and pragmatic meanings or uses of connectives. As an example of this approach to connective particles we can mention Van Dijk (1979). In this article semantically-used connective particles are considered to express relations between denoted facts in the extratextual world, whereas pragmatically used connectives are said to express relations between speech acts 13 • For instance, the English connectives and, but, or and yet can be said to have a semantic as well as a pragmatic use. This can be illustrated with the following two examples of but (Van Dijk 1979: 451): (1) (2)

Harry was ill, but he came to the meeting anyway A: Can you tell me the time? B: But, you have a watch yourself!

According to van Dijk the contrastive 'meaning' of but in (2) pertains to linguistic action and not, as in (1) (semantic but), to denoted facts. As to the actual description of 'pragmatic connectives' van Dijk observes furthermore that it cannot be given in terms of conventional meaning, but requires an interpretation in terms of functions with respect to pragmatic contexts (1979: 449-50). More than the connective particles (which from a strictly semantic viewpoint seem to be rather straightforward) the focus particles (e.g. English only, even, also; German schon, eben, noch) have offered themselves as a rewarding subject for semantic studies, especially for the more logically oriented ones. The major concern of these studies is to develop semantic parameters for analysing and describing the complex meaning aspects of focus particles. Konig (1991a) and (1991b) can be mentioned as representative recent studies in the field. The essentially semantic nature of the accounts is especially clear from the way in which Konig defines the group as a whole: "Focus particles may either include or exclude alternatives to the value of their focus (i.e. the part of the utterance that has focus function, CK), as possible values for the variable of the open sentence in their scope" (1991a: 16). Modal particles, in contrast (e.g. German ja, doch, halt), have long been treated in a stepmotherly way in semantic studies of particles, partly because the meaning of these particles is much harder to grasp in conventional seman13. Van Dijk leaves the question unanswered whether we are dealing here with different uses of the same set of connectives, or with different connective meanings which happen to be expressed by the same form. On this issue see below and chapter 5, section 5.1.

semantic approaches to particles

43

tic terms than that of connective and focus particles. However, semantic accounts of modal particles have become more frequent, especially since the study of Weydt (1969) on modal particles in German14 • The studies within this German tradition are concerned in large part with the delimitation of modal particles from other categories, the development of modal particles from other categories, and with cross-linguistic comparisons of modal particles. In their analyses of individual modal particles authors of semantic studies see themselves confronted with two major problems. First there is the issue of the terms in which to state the 'meaning' of the modal particles. It is now clear that modal particles defy a description in terms of referential, lexical, logical or truth-conditional meaning, and that it would perhaps be more fruitful to attempt to describe modal particles (just like the 'pragmatic connectives', see above) in terms of 'functional' or 'pragmatic' meaning: both 'pragmatic connectives' and modal particles seem to apply to the utterance in its capacity as a linguistic act performed by a language user. While 'pragmatic connectives' appear to connect subsequent linguistic acts, modal particles can be held to modify somehow (or to reveal the language user's attitude towards) the illocutionary force of the linguistic act. What is still lacking, however, is an overview of the more specific 'pragmatic functions' that might be associated with the individual particles, as well as a suitable metalanguage 15 • On this point valuable support may be expected from discourse and conversation approaches to particles (see section 3.4 below). A second problem for semantic approaches to modal particles is the extreme poly functionality of this group of words: modal particles are characterized not only by their property of having one or more uses that are commoner for words of other grammatical categories (e.g of adverbs or conjunctions; socalled cross-categorial polyfunctionality), but also by their apparent variety of modal uses (category-internal poly functionality). The conventional way in particle research of dealing with the poly functionality issue has been to assume a great number of unrelated meanings of a particular particle (the so-called meaning maximalist approach or homonymy point of view). In contrast, particle researchers in the German tradition of Weydt and the French tradition of Ducrot, working within a theory of language which assumes that linguistic forms have preferably only one meaning or function, advocate strongly a 14. For an overview of the host of studies on Gennan and Dutch particles in the tradition of Weydt (1969), cf. Foolen (1993: 7-10). 15. For lack of an adequate metalanguage linguists have often taken recourse to metaphorical expressions in particle description. Cf. Kroon (1992).

44

§ 3.3

minimalist approach to particle description (the rrwnosemy point of view)!6. In a strict form this approach involves the attempt to subsume the whole variety of uses of a particle under one (necessarily undifferentiated) meaning. The actual use of a particular particle then depends on the combination of this more or less abstract meaning with the properties of the context and a number of general pragmatic principles (e.g. communicative strategies, cooperative principles, analyses of speaker intention), the nature of which cannot be established within the scope of linguistics proper. More recently there are more and more researchers who regard both extreme positions as quite umealistic and who consequently advise an approach that is neither strictly minimalistic, nor entirely maximalistic. While admitting that forms usually have more than one meaning, they argue at the same time that these meanings are related to one another in a motivated way. The number of meaning distinctions are however kept to a minimum!7. I consider this moderate variant of the minimalist position (also called the polysemy point of view) as the most fruitful approach to particle research. In section 5.1 I will come back to this point and sketch how in practice a moderately minimalistic description of a particle will work out. In this context it is to be noted that the polyfunctionality issue, although raised especially in the literature on modal particles, is equally relevant to most of the 'connectives' as well!8. As for Latin, it is common practice in grammars and reference books to discuss a number of 'additional' meanings or functions of a connective, alongside a more central meaning. Mention is made for instance of an 'adversative use' of the causal connective enim, or of a 'modal use' of the adversative connective autem.

16. For the views of the school of Ducrot cf. e.g. Ducrot et al. (1980). The main drawbacks of maximalist descriptions of particles are: (i) the confusion of 'meaning', 'translation', 'function' and 'use' of a particle (cf. e.g. how KUhner-Stegmann treat the Latin particle enim); (ii) the fact that the function of a particle is systematically confused with the function of its host unit, or with the function of other aspects of the context. Minimalist accounts attempt to avoid this confusion. 17. As to the nature of the unificatory principle(s) underlying the various meanings, opinions differ. Cognitive semanticists (cf. e.g. Sweetser 1990) describe this relationship in terms of metaphorical extensions of an 'original' meaning. Others (e.g. Heringer 1988; Luke 1989) have argued in favour of a Wittgensteinian 'prototype' solution, which involves the assumption of a limited number of prototypical uses or meanings (instead of one unique basic meaning) which are related to each other through a chain of family resemblances. For general overviews of the various methods in particle description (maximalism, minimalism and moderate minimalism) I refer to Heringer (1988) and Foolen (1993). 18. Cf. also the accounts of English and in Posner (1980) and Sweetser (1990).

discourse and conversation approaches to particles

45

3.4 Discourse and conversation approaches to particles

This section deals with the recent current in linguistics that focuses on the role of 'particles' in discourse or, more particularly, in conversation. By their use of terms such as 'discourse markers', 'discourse particles', 'discourse connectives', 'Gesprlichspartikeln', 'phatic connectives' and the like, the studies involved make clear that they are not primarily interested in the formal and strictly semantic aspects of particles. The words investigated are characterized rather in functional terms, as those words which signal how the speaker intends the message to relate to the foregoing or following discourse, or to (some aspect ot) the communicative situation. The class of particles studied is in fact not fully coextensive with particles in the sense described in the preceding sections, since on the one hand some particles (e.g. most of the subordinating conjunctions) do not have discourse uses, and on the other hand certain other (non-particle) types of expression can have a discourse marking function 19 • An important characteristic of discourse and conversation studies on particles (or, for that matter, 'discourse markers') is that they usually do away with the sentence or clause as the relevant unit of text to which the function of the particle pertains. This is in accordance with the following observations: (i)

(ii) (iii)

the significance of discourse markers may extend beyond the sentence (whereas 'interclausal connectives' are held to mark semantic or pragmatic relations between successive clauses, 'discourse connectives' or 'discourse markers' may integrate or separate more global units of discourse) the significance of discourse markers may obtain also at a level below that of the sentence, e.g. at the word-group level the significance of discourse markers may also be extratextual, i.e. pertaining to an aspect of the communicative situation rather than to a unit of text.

The last observation implies that discourse and conversation approaches to particles make a less strict distinction between connective and non-connective (e.g. modal) particles. I will come back to this point in chapter 4. The use of 19. Levinson (1983) can be mentioned as the first study in which discourse markers are put together in a separate functional group (cf. Levinson, p. 87-88).

46

§3.4

the word 'significance' in the above list of observations is intentionally vague: studies on discourse and conversation functions of particles are usually rather cautious in the description of the precise 'semiotic' status of particles. Instead of assuming that particles express meaning or meaning relations (the common viewpoint in formal and semantic studies on particles), discourse studies regard particles for instance as instructions for the hearer how and where to integrate an upcoming unit of speech within the current discourse. If particles express anything at all, it must be a ptocedural meaning, rather than a lexical or referential meaning. On account of considerations like these discourse and conversation studies on particles prefer to use verbs like signal or indicate (instead of express) when describing the semiotic status of particles. From the host of recent discourse and conversation studies on particles I will discuss only two in more detail here: Schiffrin (1987) on discourse markers in English, and Roulet et al. (1985) on discourse markers in French20 • Both present their empirical analyses of discourse markers in the context of a more general theory of discourse coherence, assuming that the words in question somehow play a role in marking coherence in a discourse or conversation. In chapter 2 (section 2.6) I argued that this type of study is most likely to offer valuable theoretical starting points for the analysis of the group of words under discussion in the present study. Schiffrin (1987) contains a data-based analysis of eleven English discourse markers: oh, well, and, but, or, so, because, now, then, y'know and I mean. The data are taken from sociolinguistic interviews of Jewish Americans in an urban Philadelphia neighbourhood. On the basis of her empirical results Schiffrin inductively comes to a theoretical model of discourse coherence, in terms of which the distribution of the items can be described and explained. The essence of Schiffrin's model is that coherence obtains on five separate 'discourse planes', and that the function of the discourse markers involved is described in terms of these planes21 • The planes involved (which Schiffrin sees as some kind of 'structures') are called ideational structure, action'

20. Special mention is deserved also by Franck (1980), whose research on modal particles in German can be regarded as a pioneer study in the field of conversation-based particle research. Other works, which deal more specifically with 'discourse markers' are e.g. Schourup (1985) (on the English discourse markers like, well and y'know), and Blakemore (1987) (working from the Relevance-framework proposed by Sperber and Wilson 1986). Fraser (1990) contains an overview of the state of the art in research on discourse markers, as well as a discussion on the delimitation of the category of words involved. 21. More extensive treatments of Schiffrin's model are given in Schourup (1988), Owen (1989), Fraser (1990), and especially Redeker (1991).

47

discourse and conversation approaches to particles

structure, exchange structure, participation framework and information state. Ideational structure pertains to relationships between the ideas (propositions) found within the discourse; action structure is involved with relationships between communicative acts; exchange structure has to do with the system of conversational turn taking; participation framework involves the attitudes of the speech participants with regard to the utterances, as well as with regard to one another; information state pertains to the organization and management of speaker/hearer knowledge and metaknowledge. Now the discourse markers are regarded as contextual coordinates for utterances in that they (i) locate utterances on particular planes of talk, and (ii) 'index' utterances to proximal and distal participant coordinates (speaker/hearer) and proximal and distal textual coordinates (prior text/upcoming text). In combination with the linguistic (i.e. lexical and grammatical) properties of a particular marker this suffices to define its entire communicative value. In other words, Schiffrin describes a particle's function(s) in terms of the interplay of an invariant basic meaning (dependent on lexical and, in the case of and, but and or, grammatical features) with fluctuating characteristics of the context (viz. the 'discourse slot' in which it is used). Her approach can hence be characterized as a rather strict minimalist one. Characteristic of the markers furthermore is that they may relate to more than one plane, but that for each marker a 'primary plane of use' can be established22 • On account of the great number of 'discourse functions' that are generated by (or rather incorporated in) this model, functional equivalents among the markers are expected to be rare. By way of illustration I present here an overview of Schiffrin's contrastive analysis of the English markers so and because (ibidem, p. 202): discourse plane

because

so

ideational structure fact based causality

information state knowledge-based causality

action structure action-based causality

'cause' 'result'

'warrant' 'inference'

'motive' 'action'

22. Discourse markers can also, according to Schiffrin, relate simultaneously to more than one level at once.

§3.4

48

Because and so both have an invariant semantic meaning: because conveys a meaning of 'cause' and so a meaning of 'result'. These meanings are realized not only at a sentence level, but also at the level of the discourse. The discourse planes involved are ideational structure (fact-based causality), information state (knowledge-based causality), and action structure (action-based causality). This is illustrated with the following well-known constructed examples: (3)

a John is home because he is sick b John is home because the lights are burning c Is John home? Because the lights are burning

So far Schiffrin's account resembles other pragmatically oriented accounts of English because or so (cf. also chapter 2, section 2.4.1, figure 4). However, in addition to their semantic meaning also the 'grammatical features' of because and so (viz. their grammatical capacity as markers of subordinate and main clauses) playa role, according to Schiffrin, in their discourse use. For because this means that in discourse it can be used as a marker of subordinate discourse units (e.g. descriptive background material within a narrative), whereas so can be considered a complementary marker of main units. Both can work at both local and global levels of talk. The latter is illustrated in (4), in which a speaker tells a story about a friend who received medical advice from relatives and neighbours which was superior to that received from a series of medical specialists (ibidem, p. 195): (4)

Sometimes it works. Because there's this guy Louie Gelman. He went to a big specialist. [story] So doctors are- well they're not God either!

In (4) because marks a subordinate part of a larger discourse unit (viz. a story that is told to support an opinion), while with so the speaker returns to the main line of discourse. Because and so can thus be said to have each a (different) structural role in this discourse fragment. Furthermore Schiffrin claims a separate function for so on the discourse plane that she calls participation framework: on account of examples such as (5) below she argues that so may mark turn transitions, for instance at the completion of adjacency pairs. Such transitions are potential shifts in the participation framework. According to Schiffrin the 'result' meaning is a basis for the function of so in participation transitions, considering that possible participation transition often coincides with explicit or implicit 'result' on an

discourse and conversation approaches to particles

49

ideational level. Because does not seem to have an analogous function on the plane of the participation framework. (5)

(The speaker has been asked whether her primary school teachers ever hit the students (Schiffrin, p. 220» a Yeh. I had one teacher, her name was Frank, b We used t'call her Frankenstein c So, yeh, she would hit kids with a ruler

According to Schiffrin (but see below) the speaker in (c) returns to answer the question, marking with so her completion of the adjacency pair, and thus, a turn-transition point. Schiffrin's study has had a great impact on research on discourse markers and on particles in general23 • It contains many interesting observations on the use of particles in conversation texts, and the general idea behind the theoretical framework (coherence in discourse is established at multiple planes; the function of discourse markers should be described in terms of these planes) is appealing. Its major flaw, as has been observed also by Fraser (1990) and Redeker (1991), is that the empirical and theoretical parts do not match consistently, and that the theoretical notions used are not sufficiently explained and defined. Especially the theoretical status of the five planes of discourse (why these and no others?) and their interrelatedness remain vague24 • Furthermore there are only few, if any, objective criteria mentioned for deciding the actual discourse plane on which a particle works. This can be illustrated for instance in Schiffrin's treatment of so in examples such as (5) above: Schiffrin's decision to describe the function of so in this example as obtaining on the discourse plane called 'participation framework' appears to be quite arbitrary. In my opinion the function of so in (5) does not essentially differ from so in (4), which Schiffrin appears to view as pertaining to the ideational 23. Articles on discourse markers which have been influenced to a greater or lesser extent by Schiffrin are e.g. Luke (1989), Maynard (1989; 1992), Schleppegrell (1991) and for Latin Hilton (1989). 24. Redeker (1991) rightly draws attention to the fact that the components information state and participation framework are not on a par with the other three planes, and that it is not at all clear what the difference is between Schiffrin's ideational structure and action structure. Redeker proposes instead a model of discourse coherence which distinguishes only three components of coherence, which grosso modo resemble Schiffrin's ideational, action and exchange structures. The speaker's cognition and attitudes (which are captured by Schiffrin's information state and participation framework) should be accounted for in another way, which Redeker leaves as yet more or less open. In the analytical framework that I propose in chapter 4 they are regarded as belonging mainly to what I call the 'interactional level' of discourse.

50

§3.4

structure: both instances indicate returns to the main discourse line after a 'digression'. I would argue rather that so is highly compatible with turn transition points, rather than (as Schiffrin does) that so can explicitly signal these points as one of its functions. Assigning some of the discourse functions of so to the participation framework obscures, in my opinion, the similarity between the various uses of so. It is my belief that Schiffrin's study would have gained considerably from the incorporation of some of the better worked out concepts of the discourse model set out in Roulet et al. (1985), also known as the 'Geneva model of discourse'25. This model has in common with Schiffrin's model that it is to a large extent eclectic, making use of insights from various fields of pragmatically oriented linguistic research. The Geneva model avows its indebtedness to inter alia speech act theory, conversation analysis, Goffrnan's 'interactionism' (cf. e.g. Goffrnan 1981), tagmemics (cf. Pike 1967), argumentation theory as developed by Anscombre and Ducrot, and the Sinclair and Coulthard model (1975) for classroom interactions. The Geneva model started as an attempt to describe the speech acts which constitute authentic (French) conversations and texts, and the relations between these speech acts. As such it can be said to belong to the pragmatic studies on text relations I discussed in section 2.4.2. The basic assumption or hypothesis of the model however is that discourse is more than a series of linearly ordered speech acts: in describing the relations underlying a coherent discourse structure one should have recourse to other discourse units as well, which are connected to one another in a hierarchical structure. This general hypothesis is 'translated' into two specific research goals. One of these goals concerns the study of the words (the 'markers') that play an important role in the linking of discourse units. The other goal pertains to the study of the discourse units that are 'articulated' by these markers, and of the relationships between them. To begin with the latter goal: following Sinclair and Coulthard (1975), the Geneva model postulates a hierarchical discourse structure composed of various 'levels' of units, the most important of which are called, in increasing order of complexity, act ('acte de langage'), move ('intervention') and exchange ('echange'). Each conversation can be analysed into one or more exchanges. Each exchange can in turn be analysed into a number of (initiating

25. Although both models were developed from the late seventies onwards, and provisional versions appeared in print before the central 1985 (Roulet) and 1987 (Schiffrin) publications, there seems to have been no mutual influence.

discourse and conversation approaches to particles

51

and reactive) moves, linked by illocutionary junctions which are in other theoretical approaches generally attributed to speech acts (e.g. question and response?6. Moves can usually be analysed into a main act ('acte directeur') and one or more units that are subordinated to it. The main act and the subordinate element(s) which constitute the move are linked by what is called, confusingly, interactive relations: justification, argumentation, preparation and the like. It is to this type of relationships (and their matching markers) that Roulet et al. (1985) give most attention, on the basis of the consideration that the vast majority of speech act studies have been concerned with the 'illocutionary' rather than with the 'interactive' aspects of speech act connection. These 'interactive' relationships are defined in argumentative terms: they may involve argumentation (e.g. causality, result, justification) or counter-argumentation (e.g. concession or contrast)27. This leads to the following basic scheme of a hierarchical discourse structure:

Every A (act) can be supported by an argument, which is represented by [+ X], or by the rejection of a counter argument, represented by [- Xl The main and subordinated elements together constitute a move (M). The order and number of the arguments are free. This basic scheme allows for numerous variations: the constituent units X can be acts, but also embedded moves or exchanges, which brings in an element of recursion in the model. The constituent units can moreover also remain implicit. The following constructed examples serve as an illustration of the way in which the structures of stretches of authentic discourse are analysed in the Geneva model (Roulet et aI., p. 38 and p. 40). Note that the bien que- and car-clauses in (6) and (7) count as independent acts (A), although they do not constitute independent clauses. I will come back to the concept of 'act' in chapters 4 and 5. The small letter m in the structure underlying (7) refers to an embedded move: 26. In the Geneva model relations between discourse constituents are specified by the functions that the constituent units have in the discourse. 27. Roulet et al. admit the existence of other (i.e. non-argumentative) types of interactive relations (such as reformulation, recapitulation, correction, preliminary), but in the original 1985 version of the model they are not very specific about them. Some of them were initially viewed as belonging to the 'argumentative' relations.

52 (6)

§ 3.4 Bien que Ie temps soit maussade, je sortirai, car il ne fait froid

A A A (7)

Bien que la meteo ait annonce une amelioration, Ie temps est maussade, car la depression a atteint la Suisse, mais je sortirai quand meme

Comparable hierarchical-functional analyses are possible for exchanges:

EC: From the above illustrations it emerges how the Geneva model of discourse (which characterizes itself as hierarchical-functional) combines insights from the work of Sinclair and Coulthard (1975) on units of discourse, with the argumention theory of Anscombre and Ducrot. Its most innovative characteristic, however, is that it is meant not only to apply to conversation, but also to monoiogai, written texts: in the Geneva model monoiogal text is viewed as a move which is in fact a constituent of an (implicit or fictitious) exchange28 • Accordingly Roulet et al. propose also that the traditional distinction between monologal discourse, produced by a single speaker, and diaiogal discourse, produced by two or more speakers, be complemented by a distinction between monological discourse (displaying the hierarchical-functional structure of a move) and dialogical discourse (displaying the structure of an exchange). Moreover they integrate the notion diaphony (or polyphony) in their model, which is meant to capture the fact that in the essentially mono logical contribution of a speaker, 'voices' may be integrated of real or fictitious conversation partners. In chapters 5 and 12 I will come back to these useful provisions in the Geneva discourse model for the analysis of monologal, written texts.

28 Roulet et al. (1985) owe this insight to the Soviet linguist Bakhtin, according to whom monologal discourse is always to be considered within the framework of a dialogal structure.

discourse and conversation approaches to particles

53

The second research goal formulated in Roulet et al. (1985) is the study of the different ways in which the discourse constituents are 'articulated' at different levels of the hierarchical structure and of the linguistic markers of these 'articulations'. It is in this context that in the second part of the book a number of 'pragmatic connectives' are described and subclassified. Although Roulet et al. distinguish three main types of discourse markers (markers of the illocutionary function, markers of the interactive function and markers of the structure of the conversation), most attention is given to the latter two types, which together are referred to as 'connecteurs pragmatiques'. As markers of the structure of the conversation alors, ben, pis, quoi, bon, oui mais and non mais are discussed. They have in common that they are all involved in marking off separate discourse constituents, and in giving information as to the position of a particular constituent within the hierarchical discourse structure. The interactive markers that are treated (car, parce que, puisque, comme, en effet, d'ailleurs, me,ne and au mains) are characterized by the fact that they signal interactive relations between a main act and a subordinate unit. Although the theoretical concepts of the hierarchical-functional discourse model serve as a starting point for the subclassification and description of the 'markers' in part 2 of Roulet et al. (1985) (questions are raised such as 'what is the type of relationship marked by the connective?', 'what is the status of the constituents between which it creates a bond?'), these concepts do not suffice for the complete description of a particular marker: for each subgroup of markers a grammatical description of the distribution of the individual members, as well as several other (semantic, pragmatic and discourse) features are given in addition. These additional distributional properties are not systematized and appear to be rather ad hoc. This is probably due to the fact that the descriptive parts concerned are based largely on a number of different, earlier studies on pragmatic connectives in the French pragmatic tradition29 . The status of the study of discourse markers within the Geneva model is, however, not unequivocal. On the one hand it seems that the study of particles is used as a heuristic means for analysing the complex structure of authentic discourse: discourse markers may give an indication of the hierarchical status of a discourse constituent, or of the specific relationship between two constituents. As such the study of discourse markers is viewed as a contribution to the testing and gradual development of the general discourse model proposed. On

29. Cf. e.g. Ducrot et at. (1975), Ducrot (1983), Ducrot et at. (1980), and various articles in the Cahiers de Linguistique Franryaise.

54

§3.4

the other hand one gets the impression that this discourse model is meant especially as a theoretical background for the analysis and description of discourse markers, which has always been (and still is) an important, independent research aim in modern French linguistics. Accordingly it is difficult to characterize the work of Roulet et al. in terms of 'top-down' or 'bottom-up' approaches to the issue of text relations and their linguistic marking (see ch. 2). What is, however, clear is that Roulet et al. (1985) does not provide a ready-made analytical model for the description of discourse connectives, but that the work (as well as other studies of the Geneva School of linguistics) does contain a number of important foundations for such a model. For more extensive overviews of the Geneva model of discourse I refer to Roulet (1984) and Drescher & Kotschi (1988). Since the introduction of the model in Roulet et al. (1985) a number of revisions and extensions have been proposed, which are summarized in Roulet (1991).

3.5 Evaluation In this chapter I have discussed the ways in which particles in various languages are approached in the literature. For the description of individual particles I argued that strictly syntactic accounts are insufficient and need to be complemented by a study of the semantic and pragmatic features of the particle concerned. Semantic accounts of particles are the most common, and in fact the ones we are most familiar with from dictionaries and reference books. This holds good also for Latin: for a long period accounts of Latin particles were predominantly semantic, as appears from the treatment of particles in the authoritative grammars by Kuhner-Stegmann and Szantyr, in the Thesaurus Linguae Latinae, and in a host of articles on Latin particles from the late 19th and early 20th centuries. These accounts have all been influenced to a greater or lesser extent by the still very useful handbook on Latin particles by Hand (1829-1845), which unfortunately never reached completion30 •

30. Hand's alphabetically ordered handbook consists of four volumes and ends with the description of the interjection-like pula ('imagine'). The work contains descriptions not only of 'modal', 'focus' and 'connective' particles, but also of other types of invariables, such as prepositions and interjections.

evaluation

55

One of the flaws of many of the older semantic enquiries into Latin particles is, as is also pointed out in Rosen (1989), their impressionistic way of approach. Rosen herself suggests two heuristic means which may help to determine the semantic value of Latin particles in a more systematic way3l: the study of particles in Latin translations of Greek texts, and the distribution of particles over a number of different sentence 'patterns' or 'models'. Rosen mentions (without much discussion) twenty-three possible 'patterns', but gives no actual descriptions of particles. Another drawback of the conventional semantic approaches is that one is often forced to distinguish a fairly large number of different (and apparently unrelated) meanings of a particle, which in turn obscures the differences between individual particles. This is most apparent in a study by De la Harpe (1923), in which she discusses, from a diachronic perspective, the 'causal' functions of the primarily adversative connectives tamen, sed and autem, and the 'adversative' functions of the primarily causal connectives nam and enim. At several points in the second part of the present study I will come back to the weaknesses of De la Harpe's account. Over the last two decades semantic descriptions of particles in Latin and other languages have become more pragmatically oriented32 . As to the connective particles this led to the useful distinction between paratactic markers of semantic relations between denoted facts ('semantic connectives'), and paratactic markers of pragmatic (rhetorical, functional, argumentative) relations between speech acts ('pragmatic connectives'). This distinction can be regarded as a first step towards more appropriate accounts of the particles concerned. However, Latin connectives can be only partly accounted for in terms of (semantic or pragmatic) relations between successive clauses (or, for that matter, between successive speech acts). This is due on the one hand to the fact that most of them have functions also in other (non-connective) distribution classes: e.g. et has a connective use ('and') as well as a use as a focus particle ('also'); enim and vero appear to have 'connective' as well as 'modal' features; and so on. On the other hand it appears that the 'connective' uses of the particles concerned are not confmed to the marking of (semantic or pragmatic) relations between successive clauses. The units of discourse that 31. Already DeWitt (1937) advocates a more systematic way of studying the semantics of Latin particles, without however being very specific about what such a more systematic approach might look like. 32. It is worth mentioning that Hand's handbook on Latin particles, composed in the first part of the 19th century, already contains a number of observations that could be called 'pragmatic'.

56

§ 3.5

are connected by the particles may, for one thing, vary from single constituents to rather extensive stretches of text. Moreover, in the latter case the types of relationship concerned are moreover hard to describe in terms of the conventional semantic-pragmatic relation taxonomies (see chapter 2). Latin sed ('but'), for instance, can be said roughly to have at least three uses or functions: (i) marker of coordination; (ii) marker of 'interclausal' relationships, indicating that the preceding clause or speech act contrasts with the sed-clause; (iii) marker of unexpected turns of events in the story line, or unexpected turns in the argumentation or conversation. In order to account, in a description of the entire functional potential of sed, for the last of these three functions, one has to have recourse to descriptive frameworks that explicitly take discourse into account. The same holds for instance for the description of the Latin connectives nam, autem and igitur, which also, as we will see later on, appear to have 'discourse' functions. An approach in which one assigns 'discourse' functions to the particles concerned is more elegant, and has in my opinion greater explanatory power, than the conventional semantic (and also semantic-pragmatic) approaches, in which one tries to explain all occurrences of a particular connective particle in terms of interclausal relationships. In the latter approach one is usually forced to assume (sometimes highly unlikely) ellipses (see also ch. 6). In order to emphasize that the group of particles investigated can be interpreted only in relation to the discourse context in which they occur I prefer to use the term discourse connectives (or discourse particles) instead of pragmatic connectives. As to the types of discourse functions that might possibly be assigned to particles, studies conducted in the field of Discourse Analysis and Conversation Analysis provide useful information. I discussed in some detail two such studies (Schiffrin 1987 and Roulet et al. 1985), which have in common that they embed their observations on what they call 'discourse markers' in a larger theoretical framework. Advantages of these studies are that they are better equipped for revealing the full force of individual particles, and hence for distinguishing more clearly between near-synonyms. The strongest point of Schiffrin's 1987 study is without doubt her multidimensional approach to particles, which offers a nice solution of the polyfunctionality problem. Roulet et al. (1985) proves to be useful especially for its insightful inventory of discourse units and their combinations, and has the advantage over Schiffrin's model that it is applicable not only to conversational, but also to monologal, written texts.

evaluation

57

Schiffrin's mainly inductive study falls short however of offering pointers to the linguistic implementation of the theoretical concepts discussed, while the Geneva model appears to be too general to function as a ready-made descriptive framework for the Latin connective particles investigated in the present study. One specific problem associated with the Geneva model appears to be that it focuses almost exclusively on the hierarchical-functional organization of discourse, with a near-total neglect of the thematic organization of discourse. In consequence it is difficult to account for the Latin connective autem in terms of this model (see chapter 10). In the next chapter I will propose an alternative framework for the description of connective particles. The actual descriptions of the Latin connectives on which this framework has been based can be found in the second part of this study (chapters 6-12).

4

A FRAMEWORK FOR THE DESCRIPTION OF CONNECTIVE PARTICLES

Chapters 4 and 5 present an analytical framework for the description of Latin discourse connectives to be applied in the remainder of this study. This framework, which should be regarded as a rough working model, is based on the view that coherence is a relevant concept on at least three separate levels of discourse, which I call the representational, the presentational and the interactional level of discourse. Coherence-indicating particles (,relators ') may accordingly be described and classified in terms of the level of discourse at which they have their primary coherence-signalling function. A more subtle subclassification of the particles involved is achieved by an intersecting parameter which serves to differentiate between semantic-functional and structural aspects of coherence.

4.1 The descriptive model: coherence, connectives and levels of discourse Connective particles may in the broadest sense be viewed as one of the linguistic means involved in signalling or emphasizing the coherence of a stretch of discourse. A systematic account of the concept of discourse coherence would therefore seem to be a prerequisite for making more specific statements about the various functions of individual connectives in Latin. Such a unified account is not readily available, as we have seen in chapters 2 and 3. However, while no individual taxonomy or model has so far been proposed' that is wholly adequate for the task of describing and subclassifying (Latin) discourse connectives, there are a number of studies that offer valuable points of departure for such a descriptive framework. The descriptive model that I will put forward in chapters 4 and 5 is the result of an examination of a number of Latin discourse connectives combined with the evaluation of the theoretical and heuristic instruments offered by literature on similar or related problems. The theoretical assumptions underlying this model are in keeping with the relatively recent insight that coherence exists at several levels of

59

descriptive model

discourse, and that the coherence-marking devices of a language such as connective particles (but also other types of particles) may accordingly be described and subclassified in terms of these levels. This view is advocated notably by Schiffrin (1987) in her study on English discourse markers, which I discussed in the preceding chapter. For the actual conception of the framework proposed here, and for the linguistic implementation of the concepts distinguished, the hierarchical-functional model of discourse structure known as the 'Geneva model' (Roulet et al. 1985, see also ch. 3) has been an important source of inspiration. 4.1.1 the first parameter: level of discourse

Although there is a growing agreement among linguists that coherence exists at several levels of discourse, there is no consensus as to what those levels are. One of the most influential ideas in this respect dates back to 1934, when Buhler contended that every utterance contains three components: a representation of ideas, an expression of these ideas by a language user, and an appeal to a language receiver. Buhler's ideas can be found in various forms in much of today's linguistic research. One of the best-known adaptations, and the most relevant one for the present study, is provided by the work of Halliday. To him we owe the view that clauses (and language in general) have three 'metafunctions', which correspond to three components or levels in a grammatical system!. These functions are called ideational, interpersonal and textual. The ideational function of language is concerned with the representation of content, with lithe function that language has of being about something (Halliday and Hasan 1976: 26). The interpersonal function of language comprises the social, expressive and conative functions of language, that is, it pertains to language in its function of establishing and maintaining interactional relationships between interlocutors. The textual function is concerned with the linguistic resources a speaker has available for creating a coherent discourse. These include the lexico-syntactic devices of reference, substitution, ellipsis, lexical cohesion and conjunction, for which Halliday uses the collective term cohesion2 • While Halliday's metafunction theory was applied originally to structures on the level of the clause or clause complex, in later studies by Burton II

I. Cf. e.g. Halliday (1973; 1985). 2. Cf. Halliday and Hasan (1976).

60

§ 4.1.1

(1980), Berry (1981) and Blum-Kulka (1983) it figures, although in a rather unsystematic way, in discussions on discourse coherence. The application of Halliday's metafunctions to the levels of discourse at which coherence may obtain involves, however, a number of problems. Not only is there a lack of adequate definitions of the individual functions, but there is also the feeling that the metafunctions are not all of the same order3 . In Halliday's system we have, on the one hand, an ideational and an interpersonal component of the language, both of which are mind-based, extra textual systems, each employing separate types of coherence relations. One may, accordingly, speak of coherence of content and coherence of the interactional situation. On the other hand, however, there is a language-based, intratextual system of the language (the textual component), which too, by definition, is involved in establishing coherence relationships. Now the inconsistency appears to lie in the fact that, from a Hallidayan point of view, cohesion (as part of the textual component of the language) is to be considered one of the three essential devices that bring about coherence, whereas in my opinion cohesion rather 'reflects' (i.e. expresses) or 'underlines' coherence than causes it4. It may therefore be useful, at least for our present purposes, to distinguish between an extralinguistic system of coherence and a linguistic system of cohesion (see also ch. 2, p. 12). These systems are interrelated in that for every level of coherence (to be specified below) a specific set of conventional linguistic devices may be identified, including the group of connective particles. This revised view of Halliday's metafunction model has two advantages. In the first place it avoids the paradox of a metafunction of language with an essentially intralinguistic character - as appears to be the case with the textual metafunction. Furthermore it may provide the basis for a model of discourse coherence in which connective particles do not all end up in one and the same categoryS, but in which they can be distributed over various 'cohesion categories' which, in turn, are all associated with a particular (non-linguistic) dis-

3. Halliday seems himself to be aware of this fact: at several places in his work he describes the textual function as an enabling, instrumental function in relation to the other functions. 4. In other words it appears to be more adequate to bring the language-based system of cohesion into relation with the mind-based system of coherence (cohesion = the linguistic reflection of coherence), than to make it an equal and integral part of it. 5. Note that this is the case in Halliday's system, in which conjunctions are all part of the textual component.

level of discourse

61

course level. The need for such distinctions is suggested by the Latin material, as I will demonstrate in the descriptive part of this study. The descriptive model I propose here is based on three levels at which coherence relations may obtain. I call them the representational, the presentational and the interactional levels of discourse. The representational level corresponds more or less to Halliday's ideational metafunction and the interactional level can be compared partly with the interpersonal metafunction (but see my remarks below). The presentational level, however, is essentially different from Halliday's textual metafunction: it is concerned not with the expression (i.e. the linguistic form) of the information, but rather with the presentation and organization of the information. The presentational level captures the fact that a language user imposes an organizing and rhetorical perspective on the ideas conveyed. He chooses, for instance, the order in which to present the content (the depicted world); he may indicate which parts of the information are central, and which are less central; he may choose to elaborate or comment on a certain information unit. He may freely broach new discourse topics, or continue old ones; he may indicate how the information of one stretch of discourse is functionally related to another stretch of discourse; and so on. In short, the presentational level of discourse is concerned with the speaker's 'staging' of information units with respect to one another, hierarchically as well as functionally. It should be noted here that Halliday did not simply overlook coherence relations of the presentational type, but rather that he did not see them as belonging to a separate component of the language system. Some of them, presumably, he would presumably consider to be textual in nature, in accordance with the fact that his textual component comprises not only cohesion devices but also two organizational aspects: thematic structure, which is concerned with the theme-rheme organization of the clause, and information structure, which involves the distribution of old and new information over the clause. There are, however, indications also that Halliday would assign another part of what I call 'presentational' relations to his interpersonal component. This can be deduced from Halliday (1985: 317), where with regard to conjunction he speaks of "external (ideational) relationships between phenomena of experience", and of "internal (interpersonal) relationships between steps in an argument" (emphasis mine, CK)6. The latter part of this statement can be taken to imply that Halliday's interpersonal component is not 6. For the notions 'internal' and 'external', see chapter 2.

62

§ 4.1.1

fully equivalent with my interactional level, just as his textual component does not correspond to my presentational level. It is for this reason that I will use a different terminology, which, besides, seems to be more in keeping with the essentially dynamic nature of discourse. The following figure shows the outlines of the proposed descriptive framework for the subcategorization of particles in general and of Latin connectives in particular; the details of this framework will be supplied in the subsequent sections of this chapter. The order in which the discourse levels are presented in the lefthand column is arbitrary and is not meant to reflect the actual cognitive process taking place in the mind of a language user or receiver. The righthand column is reserved for the specific linguistic devices involved in reflecting coherence relations on the various discourse levels 7 , and will be gradually filled in in the course of the remainder of this study. figure 1: outline of the theoretical framework levels of discourse

expression of coherence relations (cohesion devices)

representational presentational interactional

4.1.2

the second parameter: type of relationship

In the preceding paragraph I proposed that the discourse function of coherence-marking devices like particles should be stated in the first place in terms of the discourse level with which they are primarily concerned. In order to arrive at an analytical framework that is geared more specifically to the description and subclassification of the individual Latin connective particles, I propose now to introduce a second parameter, which can be said in a sense to intersect the first one. This parameter captures the fact that the function of particles may be narrowed down further according to the type of relationship involved. As to the type of relationship two observations are of interest.

7. Linguistic devices are not, of course, the only instruments that reflect discourse coherence: in spoken discourse they are usually accompanied by paralinguistic means like tone of voice, and by extra-linguistic means such as gestures. In written discourse the same effects may be achieved by typographical means (initial line indentation and the like). None of these nonlinguistic devices can be taken into account in the study of Latin.

type of relationship

63

4.1.2.1 intratextual and extratextual relationships First it should be noted that relationships may obtain not only within the text between verbally expressed units of discourse, but also between a unit of the text and an extratextual concept, idea or situation. On this point I refer to my broad conception of particles as given in chapter 3, p. 35: the common characteristic of particles is that they fit their host unit into a wider perspective. This wider perspective may be provided by the surrounding verbal context, in which case the particle may usually be labelled 'connective'. It may however concern also an implicit idea or event evoked by the verbal (con)text, or some aspect of the communicative situation. I use the term 'communicative situation' to refer to the social relationships between the interlocutors, their interactional goals and intentions, their general and situation-bound knowledge, the views of the interlocutors about the knowledge, attitudes and communicative intentions of their discourse partners, and so on. Particles that somehow fit their host unit into the extratextual reality (such as modal and focus particles, see ch. 3) I call, for lack of a better term, situating particles: they 'situate' (or 'evaluate'), so to speak, their host unit against the background of (some element of) the extratextual reality. Although they are not connective in a strict sense, these particles can still be seen as relational devices, since they function as a trait d'union between the textual and the extratextual/situational. A common treatment of connective and situating particles as members of one and the same functional category is methodologically attractive in view of the fact that in various languages the categories of situating particles and connective particles cannot, synchronically, be clearly distinguished. There are, moreover, also statements in the literature about diachronic developments in which situating particles gradually adopt a connective role, or the other way rounds. Indeed, one of my conclusions about the Latin discourse connectives investigated will be that synchronically only some of them are connective in a strict sense (e.g. nam, igitur) , whereas others reveal predominantly (enim, vera) or partly (at) properties of the group of situating particles. Note, moreover, that the Latin particle et has a situating use as an additive focus particle (comparable with etiam 'also'), and connective uses as a coordinator and a discourse connective ('and'). Observations like these argue in favour of

8. cr. e.g. Van der Auwera & Vandeweghe (1984); Franck (1980: 65); K1inig (l991a: 16-17; 65; ch.8). 9. For igitur cr. Kroon (1989) and my remarks on p. 86 and n. 42 below.

64

§ 4.1.2

a more general framework that can account also for particles which are less prototypically connective. 4.1.2.2 structural and semantic-functional relationships

A second remark about types of relationships concerns the observation that relations between verbally expressed units of text may be viewed from a structural as well as from a semantic-functional perspective, and that particles may be narrowed down further in accordance with this distinction.

structural relationships As I stated in section 2.5, texts may from a structural point of view be described in terms of sequences and embeddings of linguistic constituents of variable size, ranging from single words up to whole texts. On the level of the (complex) clause the constituents involved are defined in formal terms; they are grammatical units, such as main clause, subordinate clause, participle. The matching structures may be called coordination and subordination (see ch. 2, section 2.5.1). I will not elaborate on this formal type of structural relationships. On the level of the discourse the constituents involved or discourse units, as I propose to call them, are defined in thematic and/or communicative terms, rather than in formal terms 10. Thematic units are composed of one or more thematic chains which are based on the recurrent mention of certain information units (referent continuity, event continuity, and the like). In my discussion of autem in chapter 10 (see esp. 10.4) I will come back to the thematic organization of discourses. Communicative units, by contrast, are defined in terms of the linguistic action structure of the discourse. The communicative units that make up the discourse do not follow one another in an uninterrupted monolithic sequence, but are usually ordered in a hierarchical manner. Following Sinclair and Coulthard (1975), who were among the first to recognize hierarchical structures of units of discourse larger than the clause, we could call these communicative units act, move, exchange, transaction and interaction (see also my discussion in ch. 3 of the Geneva discourse model, in

10. Cf. the discussion in section 2.5.2, where it is argued that it is hard to define discourse units in formal terms.

type of relationship

65

which these units play an important role)l1. Every transaction between interlocutors may be said to consist of one or more exchanges. Every exchange can be analysed into a restricted number of (initiating and reactive) moves. Every move is composed of one or more acts, which may be central or subsidiary with regard to the communicative intentions of the speaker. The following serves as an illustration of the hierarchical-structural analysis of a rather straightforward stretch of discourse in its component communicative units. I use an adapted form of the schematic representation of Roulet et al. (1985)12: (1)

A:

B:

I've got an extra ticket for the Santa Fe Chamber Orchestra tonight. Are you interested? Yes, wonderful

figure 2: discourse structure and discourse units I've got an extra ticket for the Santa Fe Chamber Orchestra tonight - - - - act - - -.....~ move Are you interested? - - - - - - - - - - act _ _ _...J

J exchange

Yes, wonderful - - - - - - - - - - - - act - - - - - - move

Acts can be defined as the smallest identifiable units of communicative behaviour. In contrast to the higher order units called moves they do not necessarily further the communication in terms of approaching a conversational goal13. This appears for instance from the fact that acts often need other, II. In Sinclair and Coulthard's original model, which is exclusively based on classroom interaction, the largest unit is called 'Lesson'. The distinctions have been taken over in e.g. Edmondson (1981) and in the Geneva discourse model put forward in Roulet et al. (1985), for which see ch. 3 and below. 12. For the present study the discourse units act, move and exchange are of especial interest. For reasons that will become clear later on (4.2.2.1) it is important to distinguish between discourse act and illocutionary act. For the moment it suffices to say that an ilIocutionary act is defined by the total of the lexical and grammatical properties of the utterance itself, whereas a discourse act is defined by its position within a series of other discourse units, for instance by its combination with another (subsidiary or central) discourse act. \3. The definition is adapted from Edmondson (1981: 6). As far as their internal syntax is concerned, acts may be internally complex themselves. This is e.g. the case in a sentence like He did it because he loved her, which is a combination of two clauses. In the terminology introduced above the relation between these two clauses (which is explicitly signalled by because) can be described as obtaining at the representational level of discourse, since it involves a relation between two states of affairs in the represented world. In chapter 5 I will adduce a number of linguistic arguments for not considering such because-clauses independent discourse acts.

66

§ 4.1.2

subsidiary acts to call forth the intended reaction from the addressee. The higher-rank unit called nwve is defined as the minimal free unit of discourse that is able to enter into an exchange structure. Unlike the act, a move is defined not only in terms of communicative unity, but also in terms of thematic unity. A move usually consists of a central act (which is the most important act in view of the speaker's intentions and goals) and one or more subsidiary acts, which also cohere thematically with the central act. In the above example (1) the move by speaker A consists of a central act (are you interested?), and a preceding subsidiary act (I've got an extra ticket for the Santa Fe Chamber Orchestra tonight). The reactive move by speaker B (yes, wonderful) might perhaps be regarded as formed by a central act only, in which case move and act coincide. It could, however, also be defended that that the element yes is a separate act, in which case wonderful is to be regarded as an expressive, subsidiary comment. Relations between acts within a move may be described as relations on the presentational level of discourse. An exchange, lastly, consists ideally (but not necessarily) of an initiating move by speaker A and a reactive move by speaker B, to which in Conversation Analysis the term adjacency pair is applied. The relation between the constituent moves of an exchange may be described in terms of a relation on the interactional level of discourse. Very few discourses, however, have a communicative structure that is as transparent as in example (1). A language user may for instance choose also to link up a series of coordinated moves (with varying degrees of thematic relatedness) before yielding the floor for a reaction of the addressee 14 • He may also interrupt an ongoing move to start a new intervening move, before having completed the prior one to which he mayor may not return later on. Likewise, a move may be interrupted by an embedded exchange; and so on. I will come back to more complex structures of this sort in 4.2.2.3 below. Many particles (both in Latin and in other languages) appear to be involved predominantly or partly in marking the linear or hierarchical structure of a discourse as outlined above. This means that they mark out the separate units of discourse by indicating how these are structurally tied up with other units of the same discourse, both linearly (i.e. involving relations between units of

14. The relationships between consecutive moves within one and the same tum of speaking (which is the typical characteristic of monologal texts), as opposed to moves which make up an exchange, can be described as obtaining (like relationships between acts within amove) on the presentational level of discourse.

type of relationship

67

equal rank) and hierarchically (i.e. involving the relationships that units maintain with 'higher' or 'lower' units). As such these particles can be said to have a mainly organizational function.

semantic-functional relationships Other particles, however, have a less neutral 'load', in that they do not merely signal the linear or hierarchical position of units of text within a particular structural framework, but also (or predominantly) indicate the more specific semantic-functional relationships holding between the constituent units. As such, they add, in a sense, to the meaning conveyed by the text. Like the organizational particles they may be at work at all three levels of the discourse: on the representational level they mark 'semantic relations' between actions or events in the represented world (cf. because in the example cited in note 13, and the Latin equivalents quia and quod); on the presentational level they mark 'rhetorical relations' between acts or clusters of acts (e.g. nam, see ch. 7)15; and on the interactional level they mark 'interactional relations' between the constituent moves of an exchange (e.g. at, see ch. 12). The rough outlines of the descriptive framework given at the end of 4.1.1 can now be replaced by the following more specific figure: figure 3:

analytical framework for the description and classification of 'connective' particles TYPE OF RELATIONSHIP

LEVEL OF DISCOURSE

STRUCTURAL

SEMANTIC-FUNCTIONAL

representational presentational interactional

clause combining organization of the text conversational structure

semantic relations rhetorical relations interactional relations

The figure should be read as follows: Connective particles may be described, first of all, in terms of the level of discourse at which they primarily display

15. Note that I use the tenn 'rhetorical relation' in a quite specific way, viz. to indicate functional relations (preparation, justif!cation, conclusion, background, etc.) between acts or clusters of acts within a mono logical stretch of text.

68

§ 4.1.2

their 'relational' function, thus yielding an initial subdivision of the category into representational, presentational and interactional particles. A further differentiation of each of these groups is then arrived at by determining the more specific type of relationship involved. On this point we can distinguish roughly between particles that signal predominantly the structural relationship involved (linear or hierarchical), and particles that signal mainly a semantic-functional type of relationship. 'Clause combining', 'organization of the text' and 'conversational structure' refer to structural relationships on the representational, presentational and interactional levels of discourse, respectively. Likewise the terms 'semantic -', 'rhetorical -' and 'interactional relation' capture the various semantic-functional relationships at each of the levels of discourse. These six types of relationships correspond to six groups of particles, the individual members of which are to be distinguished ultimately in virtue of a different semantic primitive or constant (see ch. 5, section 5.1). For the sake of clarity figure 3 incorporates only the intratextual type of relationships (see section 4.1.2.1 above). When we want to take also extratextual relationships into account, we must add to the rubrics 'structural' and 'semantic-functional' a third rubric, 'situating'. It is to be noted that the 'situating' type of relationship pertains predominantly (although perhaps not exclusively) to the representational and interactional levels of discourse. As to the units of the text that are typically associated with each level of discourse we might say that relationships on the representational level of discourse are typically (though not exclusively) act-internal, relationships on the presentational level move-internal, and relationships on the interactional level typically pertain to the unit of discourse called exchange. Recalling our discussion in chapter 2 we might say that the essential additional insight of the present chapter is that the question of the properties of relation, relator and conjunct needs to be raised separately on each of three levels of discourse. In the next section I will take a closer look at these levels and at the other notions distinguished in figure 3. In view of the type of particles to be discussed in the second part of this study most attention will be devoted to the presentational level of discourse.

69

representational level of discourse

4.2. Comments on the analytical framework 4.2.1

the representational level of discourse

figure 4: relations on the representational level of discourse TYPE OF RELATIONSHIP LEVEL OF DISCOURSE

STRUCTURAL

SEMANTIC-FUNCTIONAL

representational

clause combining

semantic relations

As stated above, a first function of coherent language is to portray or represent some real or imaginary world outside the language itself, and often, but not necessarily, also outside the immediate discourse situation. The specific relations between the states of affairs (actions, events or facts) that make up the represented world (which may be temporal, causal-conditional, contrastive, etc.) mayor may not receive overt expression 16 • In many languages the job of expressing such relationships is fulfilled by for instance subordinating conjunctions. Compare (2) and (3): (2)

The cook plucked the chicken. Feathers were flying about everywhere.

(3)

When the cook plucked the chicken, feathers were flying about everywhere.

Latin examples of subordinating conjunctions with a function on the representational level are, among others, si ('if), ubi ('when'), postquam (,after') and quia ('because'). There are also a few coordinating conjunctions which may function on the representational level, including et ('and') and sed ('but'). Examples of representational et and sed are (4) and (5) (cf. also section 5.2.1 below): (4)

ego hanc amo et haec med amat ('I love her and she loves me', PI. As. 631)

16. The representational level of discourse is concerned not only with relations between states of affairs but, of course, also with relations between concepts or entities within a semantic frame: from a clause-internal viewpoint one could speak e.g. of the relations of the concepts with regard to a specific predicate, and of the linguistic devices (the 'relators') that are associated with these relations, such as case marking (cf. Pinkster 1990: ch. 5). In Functional Grammar (cf. Dik 1989) such relationships are captured by semantic functions (in a narrow, clause-internal sense). I do not take them into account in my discussion.

70

§ 4.2.1

(5)

Nihil Sequani respondere, sed in eadem tristitia permanere ('The Sequani made no reply, but continued in the same sullen silence', Caes. Gal. 1.32)

In addition to their 'representational use', however, et and sed may also have a 'presentational' use. Making use of the newly introduced concept of (discourse) act we may formulate the difference between the representational and the presentational use of the items as follows: Et and sed may have a connective function on the representational as well as on the presentational level of discourse. In the former case they usually link information units (often in the form of clauses) within one and the same act, whereas in the latter case they relate two acts within a move, or two subsequent independent moves. Et in (4) thus serves to coordinate two clauses which together form a close semantic and syntactic unity, but are part of only one single (discourse) act (viz. one in which the author wishes to inform us of a mutual love). Likewise in (5) sed seems to indicate a contrast between two states of affairs in a certain represented world (expected reply versus continued silence), rather than, for example, an opposition of two independent discourse acts. Representational et and sed share with the subordinating conjunctions discussed above the capacity to connect clauses within a single discourse act. In (6) and (7) below, on the other hand, it is more likely that et and sed signal a presentational relation between two individual communicative units 17. In (6), for instance, we are dealing with two acts, the first of which has the rhetorical function of a preliminary with regard to the second, which is introduced by et. Example (7) contains an instance of presentational sed: the two acts 'linked' by sed belong to different moves, sed indicating the transition to a new move which breaks off a still ongoing one: (6)

Versae inde ad Tiberium preces. Et ille varie disserebat de magnitudine imperii, sua modestia (,Then all prayers were directed towards Tiberius. And he delivered a variety f reflections on the greatness of the empire and his own diffidence', Tac. Ann. 1.11)

(7)

(Palaestrio to his master Pyrgopolinices) moderare animo, ne sis cupidus. Sed eccam ipsam, egreditur foras (' get yourself under control, and don't be too eager. But there she is herself, coming out of the house', PI. Mil. 1215)

17. I sometimes use the term 'presentational (representational, interactional) relation' as a convenient shorthand for 'relation on the presentational etc. level of discourse'. Likewise, I will sometimes speak of 'presentational (representational, interactional) particles'.

representational level of discourse

71

It should be noted here that it is not always easy to determine whether et and sed are used representationally or presentationallyl8. In section 5.2.1. I will

come back to this problem, and to the related issue of the linguistic criteria for assigning to a unit of text its status as discourse act. When we now extend our scope beyond connective items proper, and take also situating particles (see p. 63) into account, there appears to be yet another group of particles with a function on the representational level of the discourse. These particles are not so much involved in marking relations between two verbally expressed states of affairs in the represented world; rather they evaluate (or 'situate') an event or concept in terms of the (implicit) norms and expectations that hold in the depicted world; or they relate concepts and events of the represented world to an (implicit) possible alternative world. Examples of such situating particles in English are already, still, even, only and really. Latin examples are etiam ('also'), vero ('really', 'in truth'), modo ('just', 'only'), utique ('certainly, particularly') and demum ('at last', 'only')19. Situating particles with a function on the representational level of discourse (some of which are known under the name of 'focus particles', 'logical particles' or, in German, 'Gradpartikeln') may be seen as counterparts to those with an interactional function (to be discussed in 4.2.3.3 below). The latter relate the content of the host unit to some aspect of the communicative situation, thus being involved in the interaction management. It is notable that etiam, vero and modo all have a representational as well as an interactional use. In my discussion of vero in chapter 11 I will come back to situating particles in more detail (section 11.1). The following figure shows the distribution of a number of Latin representational particles over the various categories distinguished in the analytical framework set out in this chapter:

18. Cf. Pinkster (1990: 257-258) for the analogous problem of distinguishing between constituent connection and sentence connection. 19. Etiam, vero and modo are representational in only part of their uses. For vero see chapter 1I; for modo cf. Risselada (1994).

72

§ 4.2.1

figure 5:

Latin particles with a function on the representational level of discourse

TYPE OF RELATIONSHIP LEVEL OF DISCOURSE

STRUCTURAL

SEMANTIC-FUNCTIONAL

representational

clause combining

semantic relations

parataxis: et ('and')

adversative relations:

hypotaxis: cum (,whereas')

causal relations: quia ('because')

sed ('but') ..................................................................................................................................................

temporal relations:

ubi ('when') SITUATING

modo ('only') vero ('really') etiam ('also')

The ranking of sed, quia and ubi under the rubric 'semantic types of relationship' does not mean that these particles do not signal structural relationships, but only that their 'semantic load' is more prominent than is the case with e.g. et or cum. The provisional classification of the subordinator cum ('whereas' or 'while') under the rubric of the structural types of relationship is meant to capture the fact that cum often does not indicate a specific semantic relationship between two states of affairs in the represented world, but seems to signal primarily the interdependency (viz. hypotactic ordering) of two coextensive states of affairs; the more specific type of relationship involved (causal, concessive, temporal etc.) is then to be inferred from the surrounding context and the grammatical properties of the host unifo. Compare (8) - (10): (8)

dolo erat pugnandum, cum par non esset armis (cum causale) ('It was necessary to resort to a ruse, since he was unequal to his opponent in arms', Nep. 23.10.4)

(9)

Socrates ... cum facile posset educi e custodia, noluit (cum concessivum)

20. For Latin cum cf. Hale (1887) and Lavency (1975-1976; 1985; 1989).

73

representational level of discourse

('Socrates ... , though he could easily have been removed from prison, refused', Cic. Tusc. 1.71) (10) dies triginta ... in navi fui, cum interea semper mortem expectabam miser (neutral

cum) (' A whole month or more I was aboard ship and all the time I was looking forward to death, poor devil', Ter. Hec. 421-422)

Likewise the classification of et ('and') among the primarily structure- indicating devices (as a parataxis marker) is meant to capture the fact that et occurs with a variety of semantic relations, the specific interpretation of which appears to rely on the meaning of the content of the related units rather than on the semantic load of the particle21 • 4.2.2

the presentational level of discourse

figure 6: relations on the presentational level of discourse TYPE OF RELATIONSHIP LEVEL OF DISCOURSE

STRUCTURAL

SEMANTIC-FUNCTIONAL

presentational

organization of the text

rhetorical relations

The presentational level of discourse has been described in 4.1.1 above as being concerned with the manner in which the ideas conveyed are 'staged' by the language user. It accounts for the fact that the ideas represented do not follow one another in an arbitrary, linear order, but are arranged and put into a functional perspective vis-a-vis one another in accordance with a particular plan-structure of the speaker. The presentational level of discourse portrays the speaker, so to speak, in the role of a stage director whose main task is the coherent and communicatively effective presentation of the ideas that make up

21. Cf. KUhner-Stegmann (p. 3-4): "Das Bindewort et ... ist das einfachste und allgemeinste kopulative Bindewort, indem es weiter nichts ausdriickt als die Anreihung eines Satzes oder eines Wortes, wahrend que und atque (ac) ausser dieser einfachen Bedeutung auch noch eine Nebenbedeutung in sich schliessen". Compare Schiffrin (1987: 189) with regard to English and: " ... the content of ideas and actions are not themselves part of the meaning of and; rather they are contextual parameters which work with and to organize ideas and perform actions." Cf. also Sweetser (\ 990: 86-93).

74

§ 4.2.2

the thematic backbone of the play22. More technically speaking we can say that the presentational level of discourse concerns the internal (thematic and communicative) structure of a complex move. The linguistic relevance of distinguishing relations on a presentational level of discourse is suggested by the existence of connective particles which typically point to the way in which an individual discourse unit is connected from a hierarchical-structural or a semantic-functional viewpoint - to other discourse units in the surrounding verbal context. Other linguistic phenomena that might be accounted for on the presentational level of discourse include, for instance, certain word-order phenomena, and the choice between syntactically different but semantically equivalent construction types23 . The cognitive relevance of a presentational level of discourse is suggested by the possibility of making the presentational relations explicit by means of metadiscursive expressions like This argument leads to the following conclusion: ... ; Let us now return to our main point; What is more, ... , etc. The difference between relations on the representational and the presentational level of discourse may be formulated in terms of inter alia a difference between the conjuncts involved. In the preceding section representational relations were defined as objective semantic relationships between states of affairs in the depicted world, which are typically associated with the clause or clause complex as the basic formal vehicle for their expression. Presentational relations, on the other hand, do not hold between states of affairs, and usually obtain independently of any formal structure. The units involved in the relationship are rather discourse-pragmatic in nature: they are thematic and/or communicative, rather than grammatical units. Such discourse units, as I call them, often do not coincide with the grammatical clause24 .

22. It should be noted that communicative plan-structure differs from the type of structure that the mind more or less unconsciously imposes on a series of ideas or concepts. The latter type of structure relies largely on knowledge of the world, e.g. such knowledge as that certain effects are preceded by certain causes. The former is rather dependent on the specific communicative goals of the language user. 23. For an overview of a number of text-structuring devices in Latin cf. Pinkster (1990: 254-7), and the literature referred to there; the examples of syntactically different but semantically equivalent variants include the nominativus cum infinitivolaccusativus cum infinitivo-constructions (Bolkestein 1983), and the different case patterns of certain three-place verbs (Bolkestein 1985; Bolkestein & Risselada 1985; 1987). The discourse-pragmatic explanation that is called for with respect to these text-structuring phenomena is formulated in terms of a continuity of perspective. Cf. also the 'packaging phenomena' discussed by Chafe (1976): givenness, contrastiveness, definiteness, subjects, topics and point of view. 24. See also ch. 2, section 2.5.

presentational level of discourse

75

As we see from figure 6, issues of coherence on the presentational level of discourse may be subsumed under at least two headings, in keeping with the distinction between semantic-functional relationships and structural relationships. First we may distinguish rhetorical discourse relations between the individual discourse units, which involve the way in which a discourse unit (an act or a cluster of acts) is related functionally (e.g. as a comment, preparation, justification, conclusion, or background) to another discourse unit of equal rank in the hierarchy. Structural relations on the presentational level, on the other hand, involve the position of the individual discourse units within the overall hierarchical structure of a stretch of text. In figure 6 I use the label organization of the text to refer to this structural type of presentational relationship. Both types of relationship may receive explicit expression by means of a connective particle. In 4.2.2.1 and 4.2.2.2 I will discuss rhetorical discourse relations and their expression, while 4.2.2.3 deals with organization of the text and markers of the organization of the text2 s. Among these markers I would contend that also the so-called topic/focus particles might have to be taken into account. Although these particles are not connective in the conventional sense of the word, they can be said to have certain text-organizational features in that they are indicative of the thematic organization of a discourse. I will demonstrate this point in chapter 10 when discussing the particle autem. Another instance of a Latin topic/focus particle with text-organizational features is quidem, which has been treated at length by Solodow (1978). Compare also the much-discussed Japanese particle wa26 •

4.2.2.1. rhetorical discourse relations Rhetorical discourse relations obtain usually between the constituent acts of a move. The relations are 'rhetorical' in that they involve the function of an act vis-a-vis another act, viewed from the perspective of the communicative goals of the language user2 7 • It is to be noted that the propositional contents of the 25. A comparable distinction is made by Redeker (1990; 1991), who speaks of rhetorical versus sequential relations. 26. Cf. Hinds et al. (1987). 27. lowe the term 'rhetorical relation' to Mann and Thompson (see my discussion of Rhetorical Structure Theory in ch. 2.). Other terms found in the literature for the same phenomenon are 'functional relations', 'discourse relations', 'pragmatic relations', 'coherence relations', or 'argumentative relations'. Roulet et al. (1985) use the confusing term 'interactive relations'. Because of the wide applications and connotations of these terms I avoid using them for the rather

§ 4.2.2

76

connected acts involved may simultaneously maintain a relation on the representational level of discourse. That is, in a sequence of clauses such as I was late for the meeting. On the highway my car broke down we may discern a representational relation of 'cause' between the propositional contents of both clauses, as well as a presentational relation of 'justification' between the subsequent communicative acts. As a rule, the connected acts within a move differ as to how central they are to the communicative goals of the language user. Accordingly we can speak of central and subsidiary acts, or, in the terminology of Rhetorical Structure Theory, of 'Nuclei' and 'Satellites'. This can be illustrated with the same example that was used in 4.1.2.2 (example 1):

figure 2: discourse structure and discourse units I've got an extra ticket for the Santa Fe Chamber Orchestra tonight - - - - act,

---""'r-.-

move

Are you interested? - - - - - - - - - act2 ---....I

J exchange

Yes, w o n d e r f u l - - - - - - - - - - - act - - - - - move

The move of speaker A consists of two acts which are connected by a rhetorical discourse relation which could be called 'preparation' or 'orientation'. Another way of saying this is that the subsidiary act, has the rhetorical function of preparation or orientation with regard to the central aC1i. The combination of both acts results in the higher-rank unit called move, to which we may, in an analogous way, assign the function of invitation. In order to distinguish between the function of an act and the function of the higher-order unit move, I will use the term interactional function in connection with the latter concept: an interactional function is assigned to a move as a whole; whereas the term rhetorical function applies to its constituent acts28 • The initiating move in example (1) may now be visualized as follows:

specific phenomenon under discussion in the present section. 28. For the moment I will leave open the question whether a rhetorical function should be assigned to both central and subsidiary act, or to only one of them (notably the subsidiary act).

presentational level of discourse

77

figure 7: interactional functions and rhetorical functions Speaker A: moveInvitation ( (act,)pceparation (aclz) )

!

interactional function

!

rhetorical function

It should be emphasized that the linguistic concept known since Austin (1962) and Searle (1969) as illocutionary force differs from both the interactional function and the rhetorical function. Whereas an illocutionary force is determined by the lexical and grammatical properties of the isolated, individual act, the interactional and rhetorical functions are defined in terms of the relation between units of equal rank in the hierarchical discourse organization. Thus, in (1), the subsidiary act! and the central ac1:z each has its own illocutionary force (to be called e.g. assertion and question, respectively)29. But the more specific communicative intention of the speaker, which is taken up on the level of the exchange by the corresponding move of the interlocutor, is captured by the interactional function of the entire move. It is true that for the final interpretation of the interactional function of the move within the exchange the illocutionary force of the central act is commonly a more important factor than the illocutionary force of the subsidiary act( s). However, the interactional function and the illocutionary force of the central act are not necessarily identical, as is proved by example (1), in which the specific interpretation of the interactional function of the move is co-determined by, and dependent on, the presence of the subsidiary act. Other support for the usefulness of distinguishing between illocutionary force and interactional function is provided by examples such as the constructed example (11): (11)

A: B: A:

The roof of my house needs repairing, but I have no time to do it myself My neighbour is an excellent roofer. Besides he's out of work at the moment Okay, I'll give him a call

B's move consists of two subsidiary acts, both of which have the illocutionary force of an assertion. The function of the entire move (the interactional function), however, is suggestion or advice, as is proved by the occurrence of the consentient discourse marker okay in the reactive move by A.

29. An important - but not the only - indication for the iIlocutionary force of an act is sentence type. Cf. Dik (1989) and, for Latin, Pinkster (1990) and especially Risselada (1993).

78

§ 4.2.2

It is on account of considerations like these that I differentiate between illocutionary force, interactional function and rhetorical function, and that I use the term discourse act or communicative act instead of illocutionary ace o.

4.2.2.2 types of rhetorical relationship

The theoretical excursus in 4.2.2.1 above has been intended to give the concept of rhetorical (functional, discourse, pragmatic, etc.) relation a betterdefined place in a discourse system than is usually the case3 !. In view of my doubts as to the usefulness of semantic-pragmatic and discourse-pragmatic taxonomies of text relations for the study of discourse connectives (see the discussion in chapter 2), I will not give here a fully worked-out list of possible rhetorical relationships. Instead I will confine myself to some generalizing remarks about the major types of relationship. Rhetorical discourse relations are often discussed in the context of argumentative discourse, and sometimes put on a par with argumentative relations32 • Relations may be called argumentative in a strict sense if they are based on inferential calculations33 • Typical examples of such argumentative strings are 'argument-conclusion' and 'position-support', which may both be described as 'causal' relations on the presentational level of discourse. Similarly the argumentative string 'position-counterargument' might be taken as an 'adversative' relation on the presentational level of discourse. Argumentative relations hold usually between acts in the form of independent clauses. An exception to this tendency is illustrated in (12)34: (12)

Since you are wearing your new tennis shorts, you aren't going to the library, I

conclude/guess/suppose

30. More details can be found in Kroon (1994b). 31. Favourable exceptions are Butler (1982) and Roulet et al. (1985). 32. This is especially the case in studies in the tradition of Anscombre and Ducrot, who consider argumentation to be the fundamental act of communication. Cf. Ducrot et al. (1980); Anscombre and Ducrot (1983); Ducrot (1982). In Moeschler (1985) the argumentative theory as developed by Anscombre and Ducrot is applied to conversational discourse. This theory has had great influence also on other studies of the Geneva school around Roulet. 33. Argumentative relations differ from logical relations in that they are based on inferential calculation rather than on deduction or proof. Argumentative relations are not subject to truth values: whereas logical relations may be evaluated according to the binary distinction between true and false, argumentative relations are to be judged in terms of the scalar notions weak and strong. Cf. Moeschler (1985). 34. The example is taken from Sweetser (1990: 80).

presentational level of discourse

79

Despite its subordinate nature the since-clause in (12) could be considered a full blown discourse act which provides argumentative support for the central act conveyed by the following main clause. The same appears to hold for Latin quoniam-clauses: compared with most other types of subordinate clauses they display a high degree of 'sententiality' (i.e. they display a number of mainclause features), which may provide support for assigning them the status of individual discourse acts. Subordinate clauses introduced by quia or quod ('because'), on the other hand, often do not count as separate discourse acts, but function as causal adjuncts within the main clause. This means that quia and quod usually signal semantic relations between states of affairs and are hence to be accounted for on the representational level of discourse. In chapter 5 (section 5.2) I will come back to the sometimes tricky distinction between argumentative relations (presentational level of discourse) and semantic relations (representational level of discourse). Rhetorical discourse relations obtain also outside the context of argumentation proper. An example to illustrate this point is (13)35: (13) When I saw her in the river I was frightened. Cos at that point the currents were dangerous.

The discourse act introduced in (13) by cos functions not so much as an argument that supports a position within an inferential calculation process, but rather supplies a piece of additional background information within a narrative. The same holds for sequences such as the one cited in (1), where I've got an extra ticket ... etc. (above, p. 65) functions as a preparation or orientation with regard to the following central act Are you interested? A subsidiary discourse act may also be outside the thematic development constituted by the central acts. This is the case in the constructed example (14), where the subsidiary act introduced by cos has a metatextual character in that it functions as a justification or motivation for uttering the preceding central act: (14) When I saw her in the river I was frightened. Cos I promised to provide you with all the embarrassing details!

The cos-clause in (14) can be said to 'regulate' the discourse process rather than to contribute to the informational or thematic development of the dis-

35. The example is adapted from Thomson, A.J. and A.V. Martinet, A Practical English Grammar, Oxford University Press, 1986 (4th ed.), p. 291.

80

§ 4.2.2

course. In other words, the central and the subsidiary act belong to different layers of the text: the first sentence is obviously part of a narrative layer, whereas the second pertains to the reality of the discourse situation in which the narrative is being told36 • 'Regulatory' discourse units often have the form of a parenthesis37 • Note, incidentally, that in (14) for is unequivocally a signal of a coherence relation on the presentational level of discourse, as there is obviously no semantic relation of cause involved. 4.2.2.3 organization of the text

The term organization of the text refers to the segmentation of a text on the internal level of the (complex) move. A move (i.e. the autonomous unit of discourse that is able to enter into an exchange structure) may have various degrees of complexity. In some cases one single act can have the status of a full-fledged interactional move, that is, the single act is deemed to require no further support or expansion in order to call forth the intended corresponding reactive move. In most cases of longer stretches of monologue, however, a move consists of a configuration of discourse units of varying sizes and hierarchical orders (e.g. acts, but also lower order, embedded moves) of varying size and hierarchical order, which together constitute an extended monological stretch of text. Complex move structures (or, in more conventional terminology, extended monologues) depend on the employment of three structural principles, which I call Dependence, Listing and Recursion. There is a partial analogy between these discourse concepts and the syntactic concepts of subordination, coordination and embedding. A simple instance of a Dependence structure on the level of the move is the combination of a central and a subsidiary act, which often involves also a strong semantic-functional relationship (see above, 4.2.2.1). A typical marker of Dependence structures in Latin is the particle nam. In chapter 7 I will argue that nam is a connective particle with a function on the presentational level of 36. The layers of a text should not be confused with the levels of discourse. For the former I refer to chapter 5, for the latter to section 4.1.1 of the present chapter. 37. Other examples of discourse units that have little informative value, and function mainly to facilitate the transmission of the message conveyed by the central discourse units (rather than being a new component in the message proper) are discourse units that reformulate, recapitulate or summarize information supplied earlier in the discourse. The connective particles that may occur in such units (like Latin denique, 'in short', or ergo: see 12.5.2) cannot always be distinguished from the textual organizers to be discussed in the next section.

presentational level of discourse

81

discourse, which does not so much signal a specific type of rhetorical relationship, but rather indicates that its host unit has a subsidiary status vis-a-vis a more central discourse unit38 • Because of this capacity nam can function, in certain contexts, as a marker of the organization of the text. More complex moves may be attained by combining a Dependence structure with one or more List structures. By 'List structure' I mean the coordination of two or more discourse units of equal rank and function. A central act can be supported by a list of subsidiary acts instead of by one single subsidiary act. This is illustrated in figure 8: figure 8: organization of the text on a local text level central act - - - - - - - - - - - ,

_ +-_____

_ subsidiary act l ~ subsidiary act2 subsidiary act3

.J~ I complex move

The issue of organization of the text, and its linguistic reflections, is relevant especially for extended monological written texts, which can be regarded as series of (complex) moves, usually performed by one single speaker/writer. The recurrent extension of a monological stretch of text may come about in various ways. These are again based on the principles of Dependence and Listing, but now applied in a more global way (i.e. between more complex units: see the definitions of local and global scope in section 2.5) than is illustrated in figure 8 above. Figure 9 visualizes an extended monologue which is based on a local as well as on a global Dependence structure: figure 9:

organization of the text: Dependence relations on a more global text level

central act - - - - - - - - - ,

I~

subsidiary act l - - - - , subsidiary act2 _ _ _+--__...J subsidiary act3 - - - - '

central m o v e ] extended monologue

subsidiary move

38. The rhetorical relationship involved may be 'evidence', 'justification', 'explanation', 'elaboration', 'background', 'exemplification', 'orientation', etc.

82

§ 4.2.2

In the hypothetical discourse structure represented in figure 9 a virtually complete move (of the type illustrated in fig. 8) is expanded by the addition of an extra unit of information. This additional move (which, in turn, may be internally complex itself) has a subsidiary status with regard to the central move which captures the central discourse theme. From a semantic-functional viewpoint the subsidiary move is related to this central discourse unit in a rather vague way (e.g as an afterthought or supplement). The effect of adding such a loosely connected subsidiary discourse unit is that the closing of the current move (and the beginning of a new independent move) is temporarily postponed, eventually yielding the impression of an uninterrupted coherent stretch of text even when the text is in fact thematically discontinuous. English by the way and French it propos might perhaps be adduced as markers of the particular discourse structure (i.e. Dependence on a global text level) involved. The relevance of identifying such a type of discourse structure will be demonstrated in chapter 7, where it is argued that some of the ostensibly deviant instances of Latin nam can be explained in this way. A second type of discourse structure we may come across in extended monologues is based on the principle of global Listing. figure 10: organization of the text: List relations on a more global text level central act subsidiary act

~

;J

move!

central act - - - - - movez

central act

extended monologue

move3

subsidiary act

subsidiary act etc.---'

The structure involves a list of functionally equal and (communicatively as well as thematically) complete moves: each completed move (which may also itself be internally complex) is followed, not by a corresponding counter-move of the addressee, but by a new, independent, 'coordinated' move by the same

presentational level of discourse

83

speaker, thus yielding an extensive monologue. Figure 10 above represents the structure of an extended monologue in which the individual moves maintain a List relation. Note that the more local variant of this structure is used in the internal organization of move3 , which comprises two listed subsidiary acts. In anticipation of chapter 10 I mention here that the particle autem is one of the linguistic means in Latin of signalling explicitly the introduction of a new 'coordinated' move in an extended monologue 39 • This implies that nam and autem are more or less complementary devices: both particles can have an organizational function on the presentational level of discourse, the former being a typical Dependence marker, the latter a typical List marker. When used as markers of the global organization of the text, however, they may both have the same effect, to the extent that they both may advance the monological discourse in a forward direction40 • The third concept involved in the organization of a complex move is the principle of Recursion. This principle accounts for the fact that complex moves do not only evolve in a linear way. As a matter of fact, a move is often to be described as a hierarchical network of a number of moves which are all embedded each into another. This principle of embedding brings in an element of recursivity41. The structural principles hitherto discussed (Dependence, Listing and Recursion) can be illustrated in the following passage from Sallust's monograph on the Jugurthine war. With the arrival on the scene of the new discourse topic Sulla, a new episode in the narrative starts. This important break in the thematic development of the story is explicitly signalled by ceterum. The constituent discourse units of the structurally complex fragment are numbered from 1 to 7. For ease of exposition units [4] and [5], each of which consists of a list of coordinated acts, have not been broken down further:

39. A typical marker of Listing structures in Ancient and New Testament Greek is the particle Of. As to the role of Of in the segmentation of a text into macro-units which each represent a new stage in the development of the story, I refer to the discussion in chapter 2, p. 27. 40. Hence in conversational texts nam and autem can function more or less as floor-holding devices. 41. The concept of recursion is most successfully integrated and implemented in the Geneva discourse model put forward in Roulet et al. (1985). Special attention is drawn by Roulet to the type of recursion that consists in a retroactive reevaluation of a central unit as being subsidiary within a new move with a higher position in the hierarchy. One of the examples adduced to illustrate this phenomenon is: On a dit beaucoup de bien de ce film a sa sortie. En jait, c 'est un navet (1985: 41). In English the particle anyhow might perhaps be described as a marker of retroactive Dependence structures. Note that the concept of retroactive reevaluation is of interest particularly in that it captures the essentially dynamic nature of discourse.

84

§ 4.2.2

(15) Sallust, Bellum Jugurthinum, c. 95-96

1.

Ceterum dum ea res geritur, L. Sulla quaestor cum magno equitatu in castra venit, quos uti ex Latio et a sociis cogeret, Rornae relictus erat.

2.

Sed quoniam nos tanti viri res admonuit,

3.

idoneum visum est de natura cultuque eius paucis dicere;

4.

neque enim alia loco de Sullae rebus dicturi sumus et L. Sisenna, optume et diligentissume omnium qui eas res dixere persecutus, parum mihi libero ore locutus videtur.

5.

Igitur Sulla gentis patriciae nobilis fuit, familia prope iam extincta maiorum ignavia, litteris Graecis atque Latinis ... eruditus, animo ingenti, cupidus voluptatum sed gloriae cupidior; otio luxorioso esse, tamen ab negotiis numquam voluptas remorata; nisi quod de uxore potuit honestius consuli. facundus, callidus, et amicitia facilis, ad simulanda ... negotia altitudo ingeni incredibilis, multarum rerum ac maxume pecuniae largitor. Atque illi felicissumo omnium ante civilem victoriam numquam super industriam fortuna fuit, multique dubitavere fortior an felicior esset.

6.

Nam postea quae fecerit, incertum habeo pudeat an pigeat magis disserere.

7.

Igitur Sulla, uti supra dictum est, postquam in Africam atque in castra Mari cum equitatu venit, ...

presentational level of discourse

1.

During the attack on the fortress the quaestor L. Sulla arrived in camp with a large force of cavalry, which he had mustered from Latium and the allies, having been left in Rome for that purpose.

2.

But since the event has brought that great man to our attention,

3.

it seems fitting to say a few words about his life and character;

4.

for we shall not speak elsewhere of Sulla's affairs, and Lucius Sisenna, whose account of him is altogether the best and most careful, has not, in my opinion spoken with sufficient frankness.

5.

Sulla, then, was a noble of patrician descent, of a family almost reduced to obscurity through the degeneracy of his ancestors. He was well versed ... in Greek and Roman letters, of remarkable mental power, devoted to pleasure but more devoted to glory. In his leisure hours he lived extravagantly, yet pleasure never interfered with his duties; except that his conduct as a husband might have been more honourable. He was eloquent, clever, and quick to make friends; he had a mind deep beyond belief in its power of disguising ... his purposes, and was generous with many things, especially with money. And, before his victory in the civil war the most fortunate of all men, his fortune was never greater than his deserts, and many have hesitated to say whether his bravery or his good luck was greater.

6.

As to what he did later, I know not if one should speak of it rather with shame or with sorrow.

7.

Now Sulla, as I have already said, after he came with his cavalry to Africa and the camp of Marius, soon became ...

85

86

§ 4.2.2

Unit [1] in the passage in (15) has the status of a virtually complete move, after which in principle a new move could be started, connected with (by means of a List relation) and thematically continuing the previous one. The author deems it necessary, however, to insert first a subsidiary text part in which he elaborates on the person and the qualities of the main character of the story Sulla. This complex subsidiary section, comprising units [2] - [6], is introduced by the connective particle sed ('but'), which has the function of breaking off the current line of the narration and of drawing attention to some new point. This subsidiary segment clears the way for unit [7], in which the author finally returns to the main story line, which is signalled by the connective particle igitur ('so'; here better translated with 'now') and the recapitulative subordinate clauses 42 • In discourse unit [7] a new independent move is started, which, again, may turn out to be internally complex. The subsidiary move [2-6] can in turn be analyzed into a central move [5-6] and a subsidiary move [2-4]: [2-4] functions as a preparation and justification for the catalogue of Sulla's qualities in [5-6]. On the deepest level of the hierarchical structure, finally, units [5-6] and [2-4] are to be described as configurations of central and subsidiary acts (i.e. the smallest units of discourse). Note that the digression in [2-6] also involves a switch in 'layer' or 'perspective', viz. from the narrated world to the 'here and now' of the conception of the text. In chapter 8 I will argue that this is the preferred environment for enim to occur. For the phenomenon of 'layering' in narrative texts see chapter 5. The hierarchical structure of the text fragment in (15) may be schematized as given in figure 11 below. Figure 11 shows nicely that the monological text fragment in (15) is to be regarded as a collection of (initiating) moves, which are connected to one another in various ways by means of the text-organizational principles of Listing, Recursion and Dependence.

42. Igitur ('then') usually signals the transition to a new central step in the argument or narrative, following on preparatory, procedural, digressive or otherwise non-central material. It can be compared with French alors. Cf. Kroon (1989).

87

presentational level of discourse figure 11: organization of the text in Sallust, Jug. 95-96

move l ------------------------~)

~[1-6]~ central move (= act) [I]

central acts [5]

[7]

subsidiary move

~[2"1

central move

/

mov~

~

subsidiary move

/[24

1 ""

[5+6]"" subsidiary act [6]

central move

/[2+31~ central act [3]

subsidiary acts [4]

subsidiary act [2]

Listing: The move that is constituted by unit [1-6] is continued by a move starting with unit [7]: they are to be regarded as subsequent moves of the same order on the main story line, which is indicated in figure 11 by the horizontal arrow. Recursion: Moves may contain lower-order moves which are 'embedded' into the higherorder move. Thus, in the fragment in (15), at least three (and perhaps even four) hierarchical layers of moves can be discerned: e.g. move [1-6] contains the lower-order move [2-6], which in turn contains the lower-order move [5 +6]. The layering of moves is in essence indefinite, but at any moment the language user may choose to return to a higher (or the highest) level of the structure. Dependence: Moves may (like acts) have a subsidiary function with regard to another, more central move of the same rank order. This is for instance the case with unit [24] which as a whole functions as a preliminary to the following, more central unit [5 + 6]. Decisions as to the subsidiarity or centrality of a discourse unit

88

§ 4.2.2

can be made only in retrospect, when the overall structure of an episode has become sufficiently clear. Finally, from the passage in (15) it emerges that the 'joints' of a complex discourse structure tend to get highlighted by means of a connective particle. The particles ceterum (unit 1), sed (unit 2), igitur (unit 5 and unit 7) and nam (unit 6) can be said to have a text-organizing function in the passage concerned, rather than to signal, for iilstance, semantic relationships between subsequent clauses. To conclude this section on organization of the text I wish to emphasize that it has in no way been my aim in thi~ section to give an exhaustive overview of all structural possibilities on the level of the move. The above discussion and the ensuing analysis of the Sallust fragment served merely to illustrate how the combined principles of Dependence, Listing and Recursion may yield an almost infinite number of complex move structures. As a matter of fact, the situation may be even more complex, because the embedded units may take the form not only of lower-order moves but also of inserted lower-order exchanges 43 • In general, languages appear to have a number of devices at their disposal to highlight the organization of a stretch of text and mark off its constituent units. As Schiffrin (1987: 36-7) points out, the boundaries of units of discourse are often marked in some way. The specific markers involved might be thought of as signposts which help to keep track of where we are in the overall structure of a complex monological unit. They are especially prone to occur at those points of a discourse where there is a shift or reorientation of the thematic development, for instance when a particular topic of discourse is discharged and another one introduced, or when a side path is taken or a main

43. An example of an embedded exchange can be found in PI. Trin. 778-782. The sequence tenes iam? # propemodo atque ausculto perlubens counts as an exchange which is inserted in Megaronides' extended move, which continues after the embedded exchange: Me: Seque aurum ferre virgini dotem a patre dicat patremque id iussisse aurum tibi dare Tenes iam? Ca: Propemodo atque ausculto perlubens Me: Tum tu igitur demum adulescenti aurum dabis, ubi erit locata virgo in matrimonium (' He's to say he's bringing a sum of gold from the father for the girl's dowry and that the father ordered him to give this gold to you. Do you get the point now? # Pretty much, and I'm mightily interested # And then you'll finally give the gold to the young man after the girl is actually married')

89

presentational level of discourse

line resumed44 • Among the markers of the organization of the text we may distinguish certain paralinguistic devices (e.g. tone of voice, intonation, pauses and typographical devices in written discourse), all kinds of metadiscursive expressions, a number of clause internal devices (e.g. tense and modality marking, and the use of resumptive participle constructions), and, notably, the use of connective particles. In the present section I have mentioned the Latin particles nam, autem, sed and igitur as potential markers of the organization of the text, as they appear to be capable of creating links across broad spans of discourse. Later on we will see for nam (ch. 7) and autem (ch. 10) how this specific use is connected with other (more local and semantic-functional) uses of the same particle.

4.2.3 the interactional level of discourse figure 12: relations on the interactional level of discourse TYPE OF RELATIONSHIP LEVEL OF DISCOURSE

STRUCTURAL

SEMANTIC-FUNCTIONAL

interactional

conversational structure

interactional relations

SITUATING evaluation in tenns of the communicative situation

The interactional level of discourse, finally, is concerned with the coherence or incoherence of a unit of discourse in the light of the conversational exchange taking place between two or more interlocutors within a particular communicative situation. Whereas particles of the presentational type indicate how a discourse unit fits into the (complex) move, interactional particles signal how an utterance fits into an interactional exchange, or is evaluated in terms of the communicative situation. Another way of putting this is to say that through - among other things - interactional particles speaker and addressee, and their mutual relationships, become 'visible' in a text.

44. Cf. e.g. Polanyi and Scha (1983). For the linguistic devices involved in signalling the transition into or out of an embedded unit they use the evocative tenn 'push and pop markers'.

90

§ 4.2.3

In accordance with the classificatory parameters introduced in section 4.1.2 of this chapter, coherence relations on the interactional level may be viewed from a semantic-functional perspective (interactional relations, see 4.2.3.1 below), as well as from a more structural perspective (conversational structure, see 4.2.3.2). A separate group of 'relationships' is characterized by the fact that they do not obtain between two verbally expressed discourse units, but rather involve the way in which a verbally expressed unit fits in, or is evaluated in terms of, (some aspect ot) the extratextual, communicative situation (evaluation in terms of the communicative situation, see 4.2.3.3). For the concept 'communicative situation' I refer to p. 63 above. 4.2.3.1 interactional relations

Interactional relations involve the functional relationships between the constituent moves of an exchange, that is, the relationships between the discrete contributions of different speakers to the construction of an interactional exchange. The simplest form of an exchange is the adjacency pair, which consists of an initiating move by speaker 1, and a satisfactory reactive move by speaker 2. Consider example (16), in which an initiating move with the interactional function of a request is taken up by a reactive move which counts as a compliance45 : (16) Br: Sed tu ... orationes nobis veteres explicabis? Ci: Vero, inquam, Brute (' "Won't you go over with us those early speeches?" # "Certainly Brutus", I said', Cic. Brut. 300)

The construct of a move and a matching counter-move may involve various pairs of interactional functions, the three main categories of which are listed in figure 13. The specific interactional function of a move is determined by the interplay of a number of linguistic (i.e. grammatical and lexical) and nonlinguistic (contextual and situational) factors.

45. The simple pattern of move and counter-move may be optionally extended by an evaluating third move of the first speaker. Cf. Edmondson (1981). It should be noted, moreover, that the individual moves may be much more complex (i.e. consisting of several lower-rank units) than is the case in (16). For a general introduction to the field of Conversation Analysis (with special attention to the tum-taking system and adjacency pairs) I refer to Levinson (1983: 284-370).

interactional level of discourse

91

figure 13: types of move initiating

reactive

assertion directive elicitation

acknowledgment compliance reply

The pair request - compliance illustrated in (16) is to be seen as a subtype of the category directive - compliance. Assertive moves have the function of providing information; directive moves propose courses of non-verbal action to be carried out by the addressee; eliciting moves request some verbal action of the addressee. After the completion of a satisfactory reactive move, the interlocutors may start a new exchange, which may again consist of one initiating and one reactive move46 • In actual discourse, however, exchanges seldom show the straightforward binary construction of move and counter-move exemplified in (16). The reason for this is that reactive moves are not always fully satisfactory or adequate: often they do not fulfil, or only partially, the expectations set up by the initiating move. An assertive move may for instance be criticized, challenged or refuted instead of being straightaway acknowledged, while a directive move may be followed by a refusal instead of by a compliance. More generally, expected or 'preferred' counter-moves may be temporarily delayed and supplanted by corrections, requests for explanation or confirmation, and the like. The issuing of such a 'non-preferred' reactive move has the effect of expanding the initiated central exchange by an embedded subsidiary exchange, which 'holds up' the discourse rather than carrying it forward47 : prior to the potential broaching of a new or shifted topic of discourse in a new exchange, the first speaker is compelled to adjust the current exchange. Non-preferred counter-moves therefore count as both reactive and initiating at the same time. Consider the first counter-move of Argyrippus in example (17): (17) Li: Ar:

vehes pol hodie me, si quidem hoc argentum ferre speres

Ten ego veham?

46. The terms applied here to the individual types of move remind one inevitably remind of the labels used in classical speech-act theory. As stated earlier (4.2.2.1), however, the illocutionary force of a speech act does not necessarily coincide with the interactional function of a move. 47. Note that the distinction between central and subsidiary exchanges is analogous to the distinction between central and subsidiary acts on the level of the move.

92

§ 4.2.3 Li: Tun hoc feras argentum aliter a me? Ar: Perii herc1e ... inscende ('You'll carry me on your back today, if you count on getting this cash # I carry you on my back - I?? # See any other way of getting this cash, do you? # 0 damnation! Well, get up', PI. Asin. 699-702)

The above observation that exchanges do not always consist simply of initiating moves and satisfactory reactive moves thus gives rise to a multiplication of the list of interactional functions of moves in figure 13. One might consider adding to the list such (reactive) interactional functions as challenge, objection, refutation, correction, request for confirmation, request for subsidiary information, repetition, and the like. If one wished to draw up an exhaustive list of types of moves, one should moreover take into account that there are also exchanges that pertain to the opening or closing of an interaction. Such 'peripheral' exchanges may have a strongly ritualized or metadiscursive character. Examples are the summons answer pair, or the exchange of greetings. An elaborate overview of possible (sub)types of moves would however go beyond the scope of the present discussion. For an attempt at such an overview I refer to the classification of Butler (1982), which is based to a large extent on Burton (1980). As to the role of (connective) particles in signalling interactional relations between moves, attention can be drawn especially to those particles which tend to occur in reactive moves. In Latin, for instance, the particles at ('but') and immo (,rather') are used typically at the introduction of a 'non-preferred' reactive move, signalling objections and corrections, respectively. The particle sane, in contrast, expresses the speaker's positive reaction to the preceding move of a different speaker, and hence may be regarded as a consenting reactive particle48 • For at I refer to chapter 12, for sane to Risselada (1994). An example of immo is: (18) Me: Quid apud hasce aedis negoti est tibi? So: Immo quid tibi est? (,What business is there for you at this house? # Well rather (immo), what business is there for you?', PI. Am. 350)

Another, quite intriguing Latin particle which may' serve to indicate interactional relations between subsequent moves is ergo. In dialogue, ergo is found to signal reactive moves which provide or require an adjustment or confirma48. Depending on the nature of the preceding initiating move, sane may introduce a consent, compliance, confirmation, acceptance, etc.

interactional level of discourse

93

tion of the (implications) of a preceding move49 • Such moves (see also ten ego veham in example (17)) play a role in the management of the conversation, and expand the discourse as it is rather than carrying it forward. Consider (19), in which ergo introduces a request for confirmation: (19) Tr: Th: Tr:

Ait venisse ilium in somnis ad se mortuom Nempe ergo in somnis? Ita. Sed ausculta modo Ait ilium hoc pacto sibi dixisse mortuom (,He said that in his sleep that dead man came to him! # Oh, so it was in his sleep, then? # Yes. But just you listen! He said that dead man spoke these words to him', PI. Mos. 490-492)

It should be noted in this context that not every (connective) particle occurring after a change of speaker should be assigned a function in the marking of interactional relations. Consider the following example in which two interlocu-

tors are responsible for the construction of one single move, and in which the connective particle namque (= nam) signals a relation on the presentational rather than on the interactional level of discourse. The example illustrates the necessity of differentiating between moves and turns: although there are two turns of speaking, we are dealing here with only one single move50 • (20) Ac: Per epistulam aut per nuntium, quasi regem, adiri eum aiunt Mi: Namque edepol vix fuit copia adeundi atque impetrandi ('They say that he is addressed only by dispatch or envoy, just like a king' # Yes, cos, by Pollux, there was hardly an opportunity to approach him, and to win his consent!', PI. Mil. 1225-1226)

With regard to the overt signalling of the interactional function of initiating moves we might perhaps adduce particle-like expressions like quaeso (lit. 'I ask') and obsecro (lit. 'I beg'), both of which may indicate the interactional function of a request or plea. Although expressions like these are less clearly involved in marking discourse-sequential relationships than the reactive particles mentioned above, they might be taken as forward-linking to the extent that they set up expectations of a preferred counter-move. The same holds for the Latin interrogative particles num and nonne, each of which sets up the expectation of a specific answer (viz. a negative and a positive answer, respectively) .

49. Cf. Kroon (1989). 50. I will come back to this example in chapter 7.

94

§ 4.2.3

4.2.3.2 conversational structure

For the sake of completeness I draw attention here to markers of 'structural' relationships on the level of the interactional exchange, although this may turn out to be a less relevant category as far as the Latin data are concerned. In principle, languages may have particles at their disposal whose main function is to demarcate the constituent units of an exchange, and to regulate the alternation of the individual speech contributions, rather than to indicate the more specific interactional (i.e. semantic-functional) relations involved. In this respect one could think particularly of markers that play a role in the turntaking system, such as turn-initiators (cf. e.g. Italian aUora and the English response marker well), or turn-yielders (cf. the English tag right, or the. French bon)51. Markers of conversational structure cannot always be kept strictly apart from markers of interactional relations, since both functions are often reconciled in one and the same item. This appears to hold for the particle ergo for instance: on account of its capacity to signal counter-moves with a 'check'function (e.g. requests for confirmation: see ex. 18 above) it may at the same time signal the transition from a central exchange to a subsidiary exchange, hence being indicative of the structural organization of the discourse.

4.2.3.3 'evaluation' in terms of the communicative situation A separate type of relationship on the interactional level of discourse is concerned with the way in which the transmitted content is related to, or evaluated in terms of, the non-verbal interactional context in which the text is integrated. Particles associated with this type of relationship are referred to in the literature as 'attitudinal particles', 'modal particles', 'modal adverbs', 'illocutionary adverbs', 'pragmatic particles', or 'phatic particles '52. They have in common that they anchor their host unit to a certain aspect of the communicative situation, and are not connective in a strict, intratextual sense. This interactional or communicative situation includes the intentions, beliefs,

51. For Italian allora as a tum-initiating device (as opposed to the English near-equivalent then) cf. Bazzanella (1990). The English response marker well is discussed in Schiffrin (1987), while Traugott (1982) draws attention to the tum-yielding right. For bon cf. Roulet et al. (1985: Ill). 52. The type of particle considered here has been studied especially for German (,Modalpartikeln') and Russian. Cf. e.g. Weydt (1977; 1979; 1983; 1989), and Franck (1980) for German, Rathmayr (1985) for Russian. See also ch. 3.

interactional level of discourse

95

attitudes, emotions and knowledge (both general and situation-bound) of the discourse participants, their opinions of one another, and the social relationship between them, in short, the interpersonal and situational context within which the message conveyed is integrated, evaluated and interpreted. The particles concerned (called 'situating particles' in this study) may be speaker-oriented or addressee-oriented. The former type pertains to the commitment of the speaker to the transmitted content or to the communicative act conveyed, revealing the speaker's personal attitude and/or emotions (attitudinal and expressive particles). Latin examples are tandem and etiam (in some of their uses), both of which may express the indignation, resentment or impatience of the speaker. Another example is vero ('really', 'for sure'), one of whose functions will be described in chapter 11 as that of highlighting the personal commitment of the speaker and the sincerity of his interactional intentions. Addressee-oriented particles, on the other hand, point to the involvement of the addressee in the discourse. They may, for instance, appeal to the attention or cooperation of the addressee, or indicate the addressee's expected or desired behaviour (linguistic or non-linguistic). Latin enim is a typical instance of an addressee-oriented 'situating' particle, as I will show in chapter 8. Still other particles are neutral as to speaker/addressee orientation, and seem merely to modify (intensify, mitigate, etc.) the illocutionary force of the utterance in which they are used. A Latin example is modo ('juSt')53. The above observations are not more than a rough indication of how we might account for the group of situating interactional particles within the theoretical framework proposed for the description of discourse connectives. The integration of this category appears to be necessary on account of the observation that at least two of the investigated Latin discourse connectives (viz. enim and vero) pertain primarily to the evaluation of a discourse in terms of the communicative situation, rather than to the linking of subsequent verbal discourse units. It is obvious, however, that for an adequate description of the entire group of 'modal' particles many more subdistinctions of much greater refinement would be needed than I have attempted to give here.

53. Cf. Risselada (1994).

96

§ 4.3

4.3 Concluding remarks

In this chapter I have sketched the outlines of a framework for the description and classification of Latin particles in general, and of connective particles in particular, taking as a starting-point the idea that connective particles are in essence coherence-signalling devices. A number of general classificatory criteria have been discussed, the most important of which is the level of discourse at which a coherence relation obtains. A second parameter is concerned with the fact that coherence relations may be viewed from a structural as well as from a semantic-functional perspective. The application of both parameters yields six categories of coherence relations (see figure 3 on p. 67) and six categories of matching particles, which may be multiplied if we take into account that text units may be 'related' also to non-verbal, extratextual concepts, ideas and situations. The resulting classification of coherence relations differs considerably from the typologies of text relations discussed in chapter 2. The proposed classification of coherence relations may admittedly appear to be artificial, given that in actual discourse no clearcut boundaries can be observed between the various categories, and that there is no simple match between the distinctions drawn and the particles to be described. As a matter of fact, particles are notoriously polyvalent: some of the particles under discussion in this study will turn out to have functions on more than one level of discourse (see also my remarks on sed and et in 4.2.1 above); furthermore, a particular function on one discourse level may take different forms according to the specific linguistic and situational context in which the particle is used. It should be noted, however, that the model serves a rather specific analytical and descriptive goal, and in no way purports to be a full-fledged, general model for the production and interpretation of coherent discourse: the analytical framework is intended above all to 'map' as systematically as possible the quite complex functional potential of a given particle in contrast with other (sometimes very similar) particles. In the next chapter I will be concerned mainly with the question of how to implement the proposed theoretical framework in the study of Latin discourse connectives, that is, how in practice a Latin particle is to be assigned its primary discourse function(s). With this end in view I will discuss a number of heuristic 'tools', and address the issue of the relevance of the interactional level of discourse to non-conversational texts.

5

HEURISTICS

Chapter 5 is concerned with bringing the distinctions of the theoretical framework discussed in the preceding chapter into operation in the actual description of Latin 'connective' particles. As such it prepares the ground for the detailed analyses of individual particles in the remainder of this study. I sketch first (5.1) the 'descriptive format' to be used in the individual descriptions, that is, the various components that should be present in the description of a particle. In 5.2 I address the issue of the linguistic relevance (and the related issue of the linguistic manifestation) of the levels of discourse distinguished in chapter 4. The chapter ends (5.3) with an overview of various types of linguistic clues that may be helpful in determining the specific functions of individual discourse connectives.

5.1 Components in a description of particles: basic meaning, discourse function, actual use and side-effects

As was pointed out in chapter 3, particles commonly resist a conventional syntactic and semantic analysis, that is, they cannot be defmed simply in terms of their syntactic function, or of their lexical (referential) meaning. A related problem in the description of particles is their extreme versatility: particles generally appear to have a host of different functions or meanings, the relations between which are not always immediately clear. Generations of scholars have consequently treated particles as highly polysemous items, restricting their discussions to mere listings of seemingly unrelated meanings or uses (or, in the case of Latin, to etymological explanations). This way of proceeding referred to in chapter 3 as the maximalist approach to particle research. More recently, under the influence of pragmatically oriented currents in linguistic research, particle studies have tended to proceed from a minimalist position, which is characterized by a minimizing of semantics and a maximizing of pragmatics. In consequence of its severe reduction of a particle's proper

98

§ 5.1

meaning-potential, a minimalist approach is more likely than a maximalist approach to reveal the connection between the various uses of a particle . In chapter 3 I argued that a moderate variant of the minimalist position appears to be the most viable approach to particle research. Characteristic of such a moderate approach is that one aims at maintaining a minimalist position as long as possible, without however excluding the possibility of a restricted number of distinct (though related) meanings or functions operating under the same form. Starting from a moderate minimalist postition, and making use of the analytical framework developed in chapter 4, I now propose that the various uses of a particular (connective) particle be mapped along the lines indicated in the following flow chart: figure 1: 1. basic meaning

< --- > levels of discourse ...

2. discourse function < --- > properties of context

J. 3. actual use .j.

[4. side-effects]

Figure 1 shows the various components involved in the description of a particle, ranging from very general (stage 1) to rather specific (stage 3)1. An adequate description of the use of a (connective) particle has to incorporate in the first place an indication of its basic meaning (stage 1). By this I mean the semantic invariant or constant which all different uses or functions of the particle have in common, the 'common denominator' so to speak2. Except perhaps for some of the focus and modal particles, the basic meaning usually cannot be synchronically associated with (or derived from) a concrete lexical root. Basic meanings rather are abstract and general features which capture a set of universal (semantic and pragmatic) concepts. Basic meanings are often hard to determine: usually they can be deduced Orily from the inventory of all

1. Cf. Foolen (1989: 309) for a broadly comparable proposal for a multi-stage description of modal particles in German, and Risselada's adaptation of this proposal for the description of the Latin modal particles modo and sane (Risselada 1994). 2. If no such 'common denominator' can be isolated for the different manifestations of the same form we are dealing with a case of homonymy.

components in a description of particles

99

possible uses (and matching distributional properties) of the particle concerned. As far as connective particles in a narrow sense are concerned, the set of basic meanings might include for instance 'semantic primitives' such as connexity, dependency, implication, equivalence, distinctiveness and polarity, whereas in the case of situating (e.g. focus and modal) particles the basic meanings may involve concepts such as scalarity, restrictivity and exclusiveness. In addition to these more 'semantic' basic meanings we might perhaps discern also a group of pragmatic or conversational primitives, such as centrality, cooperation, frustration, and the like. Most basic meanings are sufficiently general to be applicable on more than one of the levels of discourse defined in chapter 4, or, in other words, to obtain in more than one 'reality ,3 . The general basic meaning of stage 1 is sensitive to, and interacts with, a specific level of discourse, giving rise to a more specific basic discourse function (stage 2). Discourse function is the pivotal element in the characterization of a particle, and hence qualifies for inclusion in a lexicon or reference book. The discourse function of a particle might be thought of as a pragmatic instruction from the speaker to the addressee as to how the host unit relates to, or is evaluated in terms of, the verbal or non-verbal context. As such it is concerned with discourse-procedural meaning rather than with referential meaning. On account of its discourse function a particle is assigned to one of the global categories distinguished in the analytical framework of chapter 4 (representational, presentational, or interactional particles; strictly connective or situating particles). Besides, the discourse function determines the more distinctive features of a particle in contradistinction to other particles of the same category: discourse functions should be defined in such a way that it is clear for instance that at is not equivalent to ergo, although both are interactional particles of a connective type (see section 4.2.3.1 and ch. 12). If a particle synchronically has functions on more than one level of discourse (and this is quite often the case), we should accordingly assign two or three different discourse functions to this particle, each with a separate functional definition. In chapter 11 for instance we will see that Latin vero ('in truth', 'really') may function synchronically on both the representational and

3. From a strictly minimalist viewpoint it might perhaps be possible to describe pragmatic and conversational properties like centrality, cooperation and frustration ultimately in terms of semantic or logical concepts such as dependency, connexity and polarity. It remains to be seen, however, whether it is of any practical use to reduce the set of basic meanings to a very limited number of maximally abstract and general concepts.

100

§ 5.1

the interactional level of discourse4 • Likewise we can say that at, in addition to its primary discourse function on the interactional level of discourse (viz. indicator of challenging reactions), also has a secondary discourse function on the presentational level of discourse (viz. indicator of marked contrasts). If a particle has more than one discourse function, the different functions are not necessarily equally characteristic of the particle involved. Accordingly we may speak of primary and secondary discourse functions of a particle. The qualification 'primary' is assigned to the discourse function that is synchronically predominant or most frequent. In the case of at for instance (see chapter 12) the interactional function of marking a challenging reactive move appears to be much more pervasive than its presentational function of marking contrasts, the latter being part of the idiolect of only a few individual authors 5 • Note that in a strictly minimalist approach there is no room for the co-existence of two or more discourse functions of a particle, because on that view the differences concerned directly follow from the application of the basic meaning (stage 1 in diagram 1) to a specific context (stage 3). The interplay of the basic discourse function(s) of a particle with properties of an actual discourse context in which it is used (especially of its host unit) leads, in stage 3 of the analysis, to the assessment of the particle's more specific actual use: in stage 3 a context-free discourse function takes the form of a context-bound actual use. Whereas stage 2 accounts for the variation of a particle's uses across the levels of discourse, stage 3 accounts especially for the variation within one and the same discourse level. Taking a different perspective we might also say that stage 3 of the analysis is concerned with the pragmatic motivation for using a particular particle in a specific discourse context. We may illustrate the impact of stage 3 of the analysis with the presentational particle igitur ('so'). As pointed out in Kroon (1989), the primary discourse function of igitur is presentational, more specifically the signalling of the introduction of a main unit of discourse after a subsidiary (viz. preparatory) unit. Depending on its specific context, however, igitur may have a local, rhetorical use (marking the conclusion in an argumentative

4. The same holds for the Latin particle modo ('just'), as is pointed out by Risselada (1994). 5. Note that secondary functions may develop into primary functions, the original primary function being potentially downgraded to a secondary function and eventually disappearing altogether. In this study I confine myself to synchronic observations.

components in a description of particles

101

sequence), as well as a global, text-organizational use (marking the transition to the main discourse line after a digression, cf. the use of igitur in the Sallust fragment quoted in section 4.2.2.3)6. The actual use of a particle in a specific context may, finally, entail certain side-effects (Le. contextual exploitations of the actual use), which are incidental and should hence not be integrated into the functional analysis of the particle concerned. It is for this reason that I bracketed stage 4 in figure 1. The effect brought about by the actual use of a particle (stage 3) differs from a side-effect in that the former is essentially (co)dependent on the presence of the particle, whereas a side-effect is to a large extent (or exclusively) to be accounted for in terms of properties of the context, and may remain operative even when the particle is omitted. In the more traditional Latin grammars and reference books there is a tendency to present properties of the context as if they were discourse functions or uses of a particle. In their treatment of the primarily interactional particle vero, for instance, reference books sometimes point to a presentational function of marking transitions in the discourse structure (see section 4.2.2.2: organization of the text, and ch. 11). In chapter 11 I will argue that this alleged presentational function is in fact no more than an accidental side-effect, which depends crucially on features of the context and is not inherent to vero itself. Likewise we can say that some of the particles under discussion in the present study are admittedly highly compatible with sequences of causally or contrastively related clauses, but that they do not themselves express or indicate such semantic relations of cause or contrase. Their causal or adversative force is, in other words, a side-effect. By way of illustration of the stepwise analysis of a particle consider the following preview of the analysis of enim, to be discussed in more detail in chapter 8:

6. Cf. also Risselada (1994), who argues that in the case of modo ('just') in directive utterances the interplay between the interactional discourse function of modo and certain contextually determined properties of the host utterance leads to such different uses as reinforcement of the illocution as well as mitigation of the illocution. 7. I do not exclude the possibility that a side-effect that consistently co-occurs with a particular particle may in the end become part of the functional 'load' of that particle.

102

§ 5.1

stepwise analysis of the particle enim 1. basic meaning: consensus 2. discourse function: appeal to interpersonal consensus 3. actual use: (i) toning down challengeability (ii) irony (iii) soliciting empathy 4. side-effects: (i) causality (ii) reinforcement Enim is a consensus particle which operates exclusively on the interactional level of discourse. On account of its straightforwardly interactional character the basic meaning (stage 1) and the discourse junction (stage 2) more or less coincide. When applied to a particular context at least three interrelated types of actual uses (stage 3) may be discerned: (i) toning down the challengeability of a polemic or controversial utterance; (ii) displaying irony in the case of a patently false statement; (iii) soliciting empathy, e.g. in lively reports or in eyewitness accounts. Actual use (i) may eventually give rise to a side-effect (stage 4) of causality, actual use (iii) to a side-effect of reinforcement. As to the analysis of particles the question remains, of course, how in practice the components of the functional definitions are to be determined. This question is all the more relevant for a language like Latin, for which we cannot rely on the intuitions of native speakers. The obvious method is to start the analysis of a particle by surveying its recurrent, different uses in a sufficiently large and heterogeneous corpus, on the evidence of which the basic discourse function(s) can be deduced and, in the last resort, the basic meaning. This inventory of uses is not a random affair, and should make appeal to observable criteria. Hence, in addition to the interpretation of the content of the host unit and the immediate context (which is to a certain extent subjective and ad hoc), the inventory of uses of a particle should rely particularly on objective linguistic clues. In the last section of this chapter (5.3) I will give an overview of possible linguistic clues, based on the observed distributional properties of the investigated Latin particles. Before doing so, however, I will first address the related heuristic issue of the linguistic manifestations of the levels of discourse in a text. In section 5.2 I will be concerned with the problematic distinction between (markers of) representational and presentational relations (5.2.1), and especially with the relevance of the interactional level of discourse in non-conversational, written texts (5.2.2).

distinction between representational and presentational particles

103

5.2 Levels of discourse and their linguistic manifestations

5.2.1

distinguishing particles

'representational' from

'presentational' connective

When investigating the role of connective particles (subordinators, coordinators and connectives, see ch. 3) in actual texts it is not always easy to decide whether we are dealing with a 'representational' or a 'presentational' connective particle, especially not in the case of interclausal connection: subsequent (main or subordinate) clauses that are connected to each other in the presentational domain cohere most often in the representational domain as well (see also my remark in 4.2.2, pp. 75-76), and both types of coherence relations may in principle be signalled explicitly by a connective particle. One thing that is clear, however, is that, as far as interclausal connection is concerned, presentational relations obtain as a rule between independent acts (or clusters of acts), while within-act relations are always representational (the reverse is not true, i.e. it is not the case that representational relationships obtain exclusively within acts; see below). One relevant question to be asked in potentially ambiguous cases might therefore be whether we are dealing, in the case of a combination of two clauses, with a connection of two independent discourse acts (in which case we might expect the presence of a presentational particle), or of two grammatical clauses which together constitute only one single discourse act (in which case only representational particles may occur). This, in turn, raises the question of the formal characteristics of a discourse act. As to this point consider (1) and (2): (1)

He did it because he loved her

(2)

It's been raining, because there are puddles on the pavement8

Example (1) may be regarded as an instance of a within-act (i.e. representational) relationship signalled by a connective particle (viz. the subordinator because). Example (2), in contrast, is to be described as an instance in which because signals a presentational relation between two independent discourse acts. Although at first sight there are no apparent formal differences between the complex clause in (1) and the complex clause in (2)9, such differences emerge when we check each because-clause for a number of distributional 8. The example is from Davies (1979). 9. Perhaps except for prosodic differences, which I leave out of account here.

104

§ 5.2.1

properties. The following tests (see also section 3.2) yield well-formed results for the representational because-clause in (1), but not for the presentational because-clause in (2)10:

* *

* *

*

a presentational because-clause cannot be the focus of a cleft-construction: ? It is because there are puddles on the pavement that it has been raining a presentational because-clause cannot be the focus of a question: ? Is it because there are puddles on the pavement that it has been raining? a presentational because-clause cannot be anaphorically or cataphorically referred to: ? Because there are puddles on the pavement, therefore it has been raining a presentational because-clause cannot be the answer to a why?-question: ? Why has it been raining? # Because there are puddles on the pavement a presentational because-clause cannot be in the scope of a modal adverb or some other focalizing adverb/particle: ? It has been raining, perhaps (of course, without doubt, etc.) because there are puddles on the pavement

What the tests have in common is that they all point to the relatively independent status of a presentational because-clause (example 2) as compared to a representational because-clausell. From the tests it appears that a representational because-clause behaves as a constituent of the main clause (with the function of 'adjunct', cf. Quirk et al. 1985 and Pinkster 1972; 1990), to which we can assign the semantic function of cause, and which as a whole can have a focus function within the entire clause. The 'disjunctive', presentational because-clause, on the other hand, is not an integrative part of the main clause, but may (despite its subordinate form) be regarded as a message on its own account, or, within the terminology of this study, as an individual 10. The tests derive from a number of relevant studies; they are found, in one form or another, in Ducrot et al. (1975), Anscombre (1984) and Moeschler (1987) for the different types of 'causal' subordinate clauses in French; idem for German in Dunbar (1985) and Rosengren (1986); for Ancient Greek in Rijksbaron (1976); for Latin in Fugier (1989), Bolkestein (1991) and Mellet (1995); for English in Davies (1979). The tests hold also for the distinction between representational and interactional particles. II. The list is not exhaustive. Another type of test may e.g. involve intonation (the representational because-clause is intonationally integrated with the preceding or following main clause, while the presentational because-clause is preceded by a pause and new intonational contour). In English and Latin there appear to be moreover systematic variations in word order: representational cause clauses tend to follow the main clause, whereas presentational cause clauses usually preceed the main clause. Cf. e.g. Fugier (1989) for Latin quoniam-clauses.

distinction between representational and presentational particles

105

discourse act. Thus, while the complex clause in (1) counts as only one discourse act, the complex clause in (2) consists of a central and a subsidiary discourse act, the latter of which can be assigned the rhetorical function of, for instance, 'evidence'. The discourse act status of a presentational subordinate clause is furthermore clear from the fact that it has its own illocutionary force i2 • This is not immediately clear in (2), where main and subordinate clause both have an assertive illocutionary force. Consider however the following example, in which the because-clause is clearly outside the scope of the illocutionary force of the main clause 13 : (3)

What are you doing tonight, because the Concertgebouw Orkest is playing Brahms' First Symphony

All in all, as far as causal subordinate clauses are concerned, there appears to be ample linguistic support for distinguishing between 'representational' and 'presentational' clauses i4 . In English, both types of clauses may be marked explicitly by the connective particle because, although for the presentational variant alternative means of expression are available in the form of the subordinator since or the connective loriS. In other languages, among them Latin, the difference between both types of causal relationships appears to correspond more systematically with separate causal particles (cf. section 2.4.1 and note 10 of the present section). From articles by Fugier (1989), Bolkestein (1991) and Mellet (1994; 1995) it appears for instance that in Latin the subordinator quia is typically used as a marker of representational cause relations, whereas the

12. 'Presentational subordinate clause' is short for 'subordinate clause which is explicitly marked for its connection with a main clause within the presentational domain'. 13. In (3) the illocutionary force of the main and subordinate clause are question and assertion; the overall interactional function of the entire move is however suggestion, or maybe invitation. For the difference between illocutionary force and interactional function, see chapter 4. 14. Other examples of subordinate clauses behaving as discourse acts (and being potentially sensitive to a comparable set of linguistic tests) are conditional and final subclauses of the type if you are hungry, there are hamburgers in the fridge, or to tell you the truth, I'm not rich at all. For Latin examples of this type of 'disjuncts' cf. Pinkster (1990: 34-6), who speaks of pseudo-causal, pseudo-conditional and pseudo-purpose clauses. Consider also concessive clauses like: tomorrow he will be forty, although he doesn't show it. Examples of pseudo-conditional clauses (viz. the socalled quotative si-clauses) can be found in ch. 12 and in 5.2.2.1 below, example (6). For pseudoconditionals in Latin cf. also Van de Griend (\989). For the same phenomenon in Ancient Greek, cf. Wakker (\994). 15. For the various functions of English because cf. e.g. Sweetser (1990) and most recently Schleppegrel (\ 991).

§ 5.2.1

106

subordinator quoniam is reserved mainly for the marking of presentational cause relations 16 • The subordinator quoniam shares this feature with the discourse connective nam (see ch. 7)17. Interclausal connection may involve not only 'subordinating conjunctions', but also 'coordinating conjunctions'. In section 4.2.1 I argued that the 'coordinating conjunctions' et ('and') and sed ('but') appear to have representational as well as presentational uses, but that in actual instances it is not always easy to determine which of the two uses is involved. To illustrate this point I repeat the examples of et cited in 4.2.1, but this time without punctuation marks. On mainly intuitive grounds I consider et in (4) as a coordinator which marks an act-internal relationship on the representational level of discourse. It may be significant, in this respect, that the relationship involved is symmetrical, that is, the segments connected by et can be reversed or rearranged without a change in meaning. In (5) we are, in my opinion, dealing with a discourse connective which marks a presentational relation between two independent discourse acts. The first clause of (5) functions as a preliminary with regard to the second, which by means of et is explicitly introduced as a new, important stage in the development of the story. The segments connected by et are obviously not symmetrical. Note, furthermore, the difference in tense in both clauses (perfect in the first clause: versae; imperfect in the second clause: disserebat). (4)

ego hanc amo et haec med amat (' I love her and she loves me', PI. As. 631)

16. In Bolkestein (1991: 435) it is observed that in colloquial and postclassical Latin the difference between the two types of causal subordinators has disappeared. This phenomenon (which can also be observed in French and, as we have seen, especially in English) could perhaps be explained with reference to the fact that in actual discourse representational and presentational cause relations often' coincide. Mellet (1994; 1995) argues that there are cases in which quia seems interchangeable with quoniam and that quia should therefore be considered the unmarked element of the Latin system of causal connectives. According to her, only quod can be considered an unambiguous marker of representational cause relations. 17. Although quoniam and nam can both be regarded as markers of 'presentational' cause relations, they are by no means identical. Quoniam-clauses usually have the informational status of shared information, which implies that the clauses do not function primarily as new, independent steps in the informational development of the discourse. In the case of nam-clauses, in contrast, both connected clauses contribute to the informational or thematic development of the discourse (i.e. contain a piece of new information), although they differ as to how central they are in view of the ultimate communicative goals of the speaker. A comparable difference might perhaps also be assumed for French puisque and car, German da and denn, English since and for, and Ancient Greek i7rd and "yap.

distinction between representational and presentational particles (5)

107

Versae inde ad Tiberium preces et ille varie disserebat de magnitudine imperii, sua modestia ('Then all prayers were directed towards Tiberius. And he delivered a variety of reflections on the greatness of the empire and his own diffidence', Tac. Ann. 1.11)

The discriminating linguistic tests adduced above with regard to the subordinating conjunctions are however not, or less well, applicable to the coordinating conjunctions. I mention here two other possible indications that might be useful for differentiating between the representational and the presentational functions of et and sed. A first indication may be a change in sentential mood: a shift from for instance an indicative to a subjunctive mood would point to a lack of semantic and syntactic parallelism between the two connected segments, and hence to a succession of two independent discourse acts. Consider the constructed Latin example (6): (6)

ego hanc amo; et tu mihi minime invideas ('I love her; and he is not to begrudge me this')

A second indication could be provided by the presence of modal adverbs and comparable devices in the first segment: in the case of representational et (i.e. the coordinator) the entire complex would be within the scope of such an expression, whereas in the case of presentational et the connected clauses would both employ their own modal modifications. Thus far I have mentioned two types of representational relationships that may be explicitly marked by connective particles: main-subordinate relationships, and coordination. These types have in common that they can obtain within a single act. This is not to say, however, that semantic relations on the representational level of discourse are excluded between clauses with the status of independent acts. In practice in Latin, however, such semantic relationships turn out to be signalled not by connective particles, but - given that they receive explicit marking at all - by for instance anaphoric adverbs such as ideo (,therefore'). For the different grammatical behaviour of anaphoric adverbs and presentational particles I refer to section 3.2. The considerations put forward in this section are not more than a rough indication of the type of linguistic arguments that might be adduced in favour of distinguishing between 'representational' and 'presentational' connectives. Although not every single instance of a Latin connective particle in context is provided with specific linguistic clues for its particular connective function (representational or presentational), it appears to be possible, by extrapolation,

108

§ 5.2.2

to distinguish quite precisely between representationally used and presentationally used Latin connective particles.

5.2.2

distinguishing interactional from other particles: the importance of the concept of discourse type

For the distinction between interactional and other (especially presentational) particles we cannot make use of the same - mainly syntactic - type of criteria and tests discussed above, but have to rely mainly on the discourse distributional properties of the particles involved. An important criterion appears to be the discourse type in which a particle occurs. As to this point one might simply assume that interactional particles are to be found especially in dialogal, oral discourse, whereas presentational particles will occur normally in monologal, written texts. In a language like Latin, for which we have to rely on written texts of a predominantly monologal nature, this simplified view of discourse type might easily lead to the misconception that interactional particles are confined largely to the few purely conversational texts that have been handed down to us. Such a view seems to be responsible for the fact that the interactional characteristics of some of the Latin particles investigated (esp. enim, vero and at) have not been sufficiently recognized or appreciated. In this section I will give a more sophisticated description of the concept of discourse type, leading to the conclusion that the interactional level of discourse is relevant not only for strictly conversational texts. This view is inspired by studies of the Geneva school of linguistics, in which monologal written texts are commonly viewed from a dialogal and conversational perspective (see section 3.4). In more theoretical terms this means that a monologue is considered to be a move which is in fact a constituent of an exchange, even though this exchange may be implicit or fictitious 18.

18. Cf. e.g. Roulet (1984: 35).

discourse type

109

5.2.2.1 discourse type: the oppositions monologal - dialogal and monological dialogical

I reserve the term discourse type for the communicative mode associated with a stretch of text, for instance monologue, dialogue and polylogue. Discourse type should be distinguished from text type. Whereas text type (e.g. narrative, procedural, argumentative, expository) is a global characterization which depends on factors such as subject matter and speaker/author's aims and goals, discourse type is a more local, dynamic phenomenon, which may change several times within the same text1 9 • Discourse type is determined on the basis of two interrelated parameters 20 • The first parameter is concerned with the opposition monologal/dialogal, which captures the number of speech participants actually involved in the phrasing of a stretch of text. A monologal text is a text which is phrased and produced by one single speaker or writer who has full structural and topical control; a dialogal text is phrased by at least two speakers who share the structural and topical control. Most Latin texts have of course a predominantly monologal form, that is, they have one central speaker who phrases the text, usually coinciding with the author. Exceptions include Latin comedy and Ciceronian dialogue, which have an inherently dialogal form. The second parameter has to do with the status of a text segment in the wider discourse structure, and is stated in terms of the opposition mono logical/dialogical (not to be confused with the opposition monologal/dialogal introduced above). A dialogical discourse segment consists of alternating moves of distinct discourse partners, which are related by their corresponding interactional functions and together constitute an interactional exchange. A monological discourse segment, on the other hand, is not composed of independent initiating moves and corresponding reactive moves, but consists of a single move, whose constituent discourse units (if any) are tied together by

19. In theory any discourse type may be used for a specific text type, but in practice every text type has one or more preferred discourse types. 20. lowe the distinction between monologalldialogal on the one hand, and monological/dialogical on the other, as well as the important concept of diaphony to be discussed in the next section, to Roulet et al. (1985). The ideas formulated there are influenced by Ducrot (passim), and originally go back to the Soviet theoretician Bakhtin. The treatment of discourse type that I give in this section can be considered an elaboration of the rough outlines sketched by Roulet et al. (1985).

110

§ 5.2.2

rhetorical functions rather than by interactional functions 21 • The interplay of the two parameters yields four different discourse types, as we can see in figure 2: figure 2: discourse types

1. 2.

3. 4.

PARAMETER 2

PARAMETER 1

dialogical monological dialogical monological

dialogal dialogal monologal monologal

The discourse types dialogical dialogal (1) and monological monologal (4) are more or less self-explanatory. They represent, so to speak, the 'default' situation, referring as they do to dialogues and monologues in the more conventional sense of the word. However, when a text is phrased by a central reporter (in written discourse: the writer), but simulates or reports a conversational exchange, we may speak of a dialogical monologal discourse (3). This discourse type is for instance regularly found in Cicero's letters and orations, when Cicero simulates a conversation between himself and a fictitious or real discussion partner. Another example of discourse type 3 can be found in reported direct speech, which typically occurs in Latin historical prose 22 . Discourse type 2, monological dialogal discourse, is relatively rare. It is sometimes found in Plautus when two conversation partners supplement each other's utterances (hence together constituting one single move, cf. ex. 20 in ch. 4), or when a conversation participant restricts his contributions to minimal responses, so that there is no real exchange taking place. We will come across instances of this discourse type in the discussion of nam in chapter 7.

21. For the concepts 'exchange', 'move', 'interactional function' and 'rhetorical function', see chapter 4, section 4.2.2. In ch. 4 it was also argued that moves (on account of the principles of Listing and Recursion) may have any degree of complexity (see section 4.2.2.3). 22. In publications of the French pragmatic school of linguistics the distinction between 'reporter' and 'speaking character' is conveniently captured by the opposition locuteur vs. enonciateur. The locuteur is the person who physically produces the utterance, while the enonciateur is the actual author of the utterance.

111

discourse type 5.2.2.2 diaphony

The two pairs of oppositions monologal-dialogal and monological-dialogical do not suffice to capture all possible discourse types. An important additional insight (which, as we will see later on, has considerable relevance to the study of Latin particles) concerns the observation that an essentially monological discourse (i.e. a stretch of discourse with the form of a complex move) may contain 'embedded voices'. This is for instance the case when a central reporter ('1ocuteur') phrases two distinct voices or opinions, not formally set apart as in a strictly dialogical discourse type, but inhabiting the same utterance or move. An example is (7), in which the conditional si-clause counts as a quotation of the words or thoughts of someone other than the 'locuteur' ('reporter') of the main clause: the si-clause contains an 'embedded voice' to which the reporter formulates a reaction in the main clause. That the si-clause should be regarded as containing an 'embedded voice' is proved by the presence of at in the main clause, a particle that is typical of reactive moves. I will come back to this example in the discussion of the particle at in chapter 12. (7)

Si vos urbis, Quirites, si vestri nulla cura tangit, at vos veremini deos vestros ab hostibus captos ('If you feel no concern, Quirites, for your city, or for yourselves, yet fear your gods, whom the enemy hold captive', Liv. 3.17.3)

In figure 3, which is a schematic reproduction of the information contained in figure 2 above, I refer to this additional, hybrid type of discourse as diaphonic discourse. figure 3: discourse type and diapbony

dialogal

dialogical / (1)

/""

"""

DISCOURSE TYPE

""

monologal

/

monological

dialogical

(2)

(3)

"""

monological

/

diaphonic (5)

(4) "-..

monophonic

112

§ 5.2.2

The label 'diaphonic' (discourse type 5) can be attached to any monological stretch of text that somehow displays the features of a communicative interaction, without having all formal characteristics of a dialogical discourse type (i.e. without having an actual exchange structure). Such 'dialogical traits' in a monological environment may be explained in the first place as a reflection of the fact that monological discourses consist in essence of two separate 'worlds', or 'layers': (i) the layer of the transmitted information pertaining to the topic discussed (that is, the actual narration, argumentation, exposition, etc.), which could be called the 'reportive' (,ideational', 'demonstrative') layer; and (ii) the communicative (or rather: metacommunicative) layer of the speaker and the addressee (or writer and reader) who from the outside may comment on (or react to) the transmitted information as well as on the communicative process taking place (the hie and nunc of the speech or writing situation)23. Whereas the reportive layer is always present24 , texts may differ as to how sensitive they are to the communicative aspect of discourse. The communicative frame in which a text is integrated may shine through repeatedly in a text, as is the case in the letters and speeches of Cicero for instance, which are directed to (and often incited by) a specific addressee or audience25 . In other text types, however, the communicative frame of writer and reader (or speaker and addressee) may come to the surface only sporadically, or not at all. A common phenomenon in narrative and expository prose, for instance, is the temporary interruption of the ongoing reportive discourse for a metacommunicative comment. This means that the prevalent reportive (ideational, demonstrative) layer of a text is temporarily deserted by the speaker/author for a remark that belongs to the expressive and/or communicative aspects of the discourse. Such interruptions may contain the author's personal reflections and comments on the related facts, his justification of the words chosen or procedure followed, all kinds of programmatic statements, appeals to the attention and involvement of the addressee, the monitoring of the addressee's under-' standing (are the shared understandings that are being assumed actually

23, The 'layers' of a text (reportive and (meta)communicative, or narratio and expositio) should not be confused with the levels of discourse, nor with the rank order of discourse units within the discourse structure. For the two latter concepts see chapter 4. 24. Only highly ritualized expressions such as greetings could perhaps be regarded as devoid of any ideational content, 25. Addressee-incited letters and speeches may, in a sense, be regarded as reactive moves (or as series of coordinated reactive moves) in reply to unexpressed initiating moves.

discourse type

113

available?), and so on. Not surprisingly such interruptions often take the form of a parenthesis, although they may also be more extensive and have a more individual status. An illustrative example of the alternation of the reportive and the communicative layer of a text is (8): (8)

Nam tum cum ex urbe Catilinam eiciebam - non enim iam vereor huius verbi invidiam ... ('Inasmuch as at the very time I was trying to drive Catilina out of Rome - for, mark you, I am not afraid now of the odium attaching to this expression .. .', Cic. Cat. 3.2.3)

Summarizing we can say that when a mono logical stretch of text contains explicit references to the communicative frame in which it is integrated we are dealing with a diaphonic discourse type. Of a slightly different order is the kind of diaphonic discourse that has the form of an indirectly rendered conversation. This type of discourse, in which the speech of internal characters is mediated by a central reporter, may show, to varying degrees, traces of a conversational mode, without having all the formal characteristics of a strictly dialogical discourse type. Indirectly rendered conversation differs from direct speech in that in the latter case the words or thoughts of the conversation partners are quoted verbatim (dialogical discourse), whereas in the former the deictic orientation of the reporter is (partially or entirely) retained (monological discourse with a diaphonic character). Thus, in indirect speech the conversation participants may continue to be designated by third person pronouns instead of by first or second person pronouns. Likewise, verbal tense (past vs. present) and time/place adverbs (then vs now; there vs. here) may conform to the deictic orientation of the reporter instead of being adapted to the point of view of the internal characters. The most indirect form of a reported conversation is that in which a reporter merely supplies the gist of a discussion between the internal characters, instead of a more or less accurate account of what was actually said26 . This appears to be the case in example (7) above. The above discussion on discourse type and diaphony was shown to be necessary by the practical experience of studying the behaviour of Latin particles. It turned out that, in order to arrive at a unified description of the discourse function of some of the Latin particles investigated, it is essential to

26. Cf. e.g. Toolan (1988) for comparable observations about the linguistic reflections of the various types of indirect speech, as well as for a discussion of the formal differences between direct and indirect speech.

114

§ 5.2.2

realize that there are several discourse types that could be called basically 'conversational', and as such are in essence capable of hosting interactional particles. These 'conversational' discourse types (italicized in figure 3) could be taken together under the name of dia-discourse, and differ only in the extent to which they show the formal properties of a conversational exchange. In the second part of this study it will be shown that the Latin particles enim, vera and at -in addition to their use in a straightforwardly conversational, dialogical context - are to be associated with a discourse type that is not simply monological, but monological with a diaphonic character. This observation provides strong evidence for a discourse function of these particles on the interactional level of discourse. In monophonic discourse, on the other hand, the central reporter merely phrases and organizes the information concerned with the current topic of discourse, with the minimum of subjective involvement, and without overt traces of his being involved in a communicative exchange with an audience. It is in this type of discourse that presentational particles such as autem (see ch. 10), nam (see ch. 7) and igitur (cf. Kroon 1989) are likely to occur. The following list contains a number of 'dialogical traits' on the basis of which a diaphonic discourse type can be diagnosed. Usually these traits will occur in clusters. They all count, in one way or another, as explicit references to the interactional frame of writer and reader in which a (monoiogicai) text is integrated. Note that the other type of diaphonic discourse discussed in the present section (viz. indirectly reported conversation between internal characters) can usually be recognized quite easily, for instance on account of the presence of a 'communication' verb.

features typical of diapbonic discourse:

* The

presence (in the host-unit or immediate context) of first- and secondperson pronouns and verb forms. * The use of present tense verb forms in a passage that is otherwise presented in the past. * The presence of metadiscursive expressions; such expressions somehow direct the communicative process and may therefore be regarded as reflections or manifestations of the interactional rather than of the ideational function associated with the discourse (they refer to the flow of discourse itself, and not to the topic discussed). Examples of metacommunicative expressions are performatives (such as I tell you that ... ; I ask you whether ... ), metadirectives

discourse type

115

(tell me; be sure that; remember that ... , believe me), and evaluative or procedural expressions of the type I have to admit, it should be stressed, now that we have come to this point in our discussion, and the like. Compare also Latin expressions such as mihi crede ('believe me'), dicendum est saepius ('it can't be stated often enough'), dicam ('I will tell you'), sicut nosti ('as you know'), ut supra demonstratum est ('as has been demonstrated above'), nolite putare ('do not think that .. .'). * The immediate presence of subjective evaluation verbs like arbitror, opinor, credo, puto, mihi videor ('I mean', 'I think', 'I believe', and the like), and of other expressions of subjective mood. * The use in the immediate context of questions (both rhetorical and nonrhetorical) or of directives, either of which presupposes the involvement of an addressee in the speech event (more than e.g. assertions do). * The presence of extraclausal interactional elements such as interjections, swear words and vocatives.

5.3 Overview of linguistic clues

In this section I give an overview of possible linguistic clues for determining the discourse function of a particular discourse connective. One of these clues (discourse type) has already been dealt with extensively in the preceding section. The heterogeneous (though not necessarily complete27 ) list of clues presented here is based on the observed distributional properties of the Latin discourse connectives to be discussed in the remainder of this study (nam, enim, autem, vero and at) and of a number of comparable items such as ergo, igitur and sed. These clues involve a set of possible syntactic, discoursepragmatic and lexical features of the immediate context in which a particle occurs, rather than features of the particles themselves: since discourse connectives (as well as particles in general) usually cannot be defined on the basis of their own syntactic features or of some transparent lexical meaning (cf. ch. 3), we are to rely on the systematic co-occurrences (and constraints on

27. I left out of account e.g. the factor of word order, at least as far as the position of the discourse connective in the clause is concerned (the particles involved are more or less automatically put in a certain (usually first or second) position. For an explanation of this phenomenon, cf. Pinkster (1990: 169-70).

§ 5.3

116

co-occurrences) of the particles with certain features of the context. Part of these features of the context are assumed somehow to reflect discourse functions shared with the particle in question (given the assumption that (dis)coherence devices tend to cluster), and hence to form an indirect indication of the particle's proper discourse function. Usually the discourse function of an individual particle is revealed not only by one single clue from the list, but by the interplay of a number of clues, in contrast with other particles which involve different sets of distributional properties. It is to be noted moreover that not all of the clues are equally relevant for each particle. Although some of the clues are clearly interrelated they have been divided, for the purpose of this exposition, into nine categories: (i)

(ii) (iii) (iv) (v) (vi) (vii) (viii) (ix)

discourse type collocations with other particles illocutionary force of the clause in which the particle occurs communicative structure information structure syntactic status of the clause in which the particle occurs tense in narrative discourse genre and text-type content

Under each heading I will offer some illustration and some explanatory remarks. For further details I refer to the descriptive chapters in the second part of this book. (i) discourse type

See the inventory of clues given at the end of section 5.2.2.2.

linguistic clues

117

(ii) collocations with other particles28 Examples: * Frequent collocation with particles that point to a high degree of challengeability of the host utterance. Examples of such particles are at and immo (which introduce objections and corrections, respectively), and the particles of negation non, nec and neque, which on account of the fact that they often involve the rejection of an implicit opinion or idea can be regarded as challengeability markers. This feature (i.e. frequent collocation with particles that point to a relatively high degree of challengeability) is displayed especially by enim (cf. ch. 8) and vera (cf. ch. 11), both of which will be described as capable of toning down a challengeable utterance. * Frequent collocation with the particle quasi ('as if), which is an important distributional property of the particle vera (cf. ch. 11): the frequent collocation of vera with quasi does not only form an indication of vera's proper discourse function ('actuality marker'), but also distinguishes vera from the particle enim, which in other respects behaves identically (cf. 11.5.2), but lacks this specific distributional peculiarity. * Frequent co-occurrence with enumerative adverbs and particles such as deinde ('next') and parra, post, praeterea, tum (all meaning 'then'), which is a typical feature of the thematic discontinuity marker autem (cf. ch. 10, esp. 10.3.3.1 and 10.4.2). * Restrictions on co-occurrences, which may serve as a confirmation of discourse functions that have been determined from other clues. The fact that nam cannot be combined with igitur is e.g. explicable in terms of nam's assumed discourse function of signalling subsidiary discourse units (cf. ch. 7), which is incompatible with the discourse function of signalling central discourse units, which is the typical function of igitur (cf. Kroon 1989). Likewise, igitur cannot be combined with autem (cf. ch. 10). The explanation lies again in the discourse functions of each connective: whereas autem is a typical discontinuity marker, igitur commonly involves thematic continuity, in virtue of its function as an introducer of central discourse units after a preparatory unit.

28. Systematically collocated particles (such as enim vero, et enim, at enim, iam vero and ergo etiam) may sometimes start to function as a complex particle with a specific idiomatic meaning. This holds especially for enim vero and et enim. In my discussion of enim in chapter 8 I will not take the complex particle etenim into account. I will regard the combination enimvero as consisting of two separate (though highly compatible) items. For the combination of ergo with etiam, cf. Kroon (1989).

118

§ 5.3

(iii) illocutionary force of the clause in which the particle occurs Given the observation that the distribution of the particles over the various types of illocution is quite uneven, we may also take illocutionary force into account as a linguistic clue for the function of a particle. I take the notion 'illocution' or 'illocutionary force' to refer to pragmatic concepts such as assertion, question and directive (and their various subtypes). Although assertions are prototypically expressed by means of declarative clauses, questions by means of interrogative clauses, and directives by means of imperative clauses, there is no strict one-to-one relation between the essentially formal notion of sentence type and the more pragmatic notion of illocutionary force. For instance, directives may be expressed not only by means of the imperative sentence type, but also by means of interrogative and declarative clauses (cf. Risselada 1993). Illocutionary force should be distinguished also from the notions 'interactional function' and 'rhetorical function', as I argued in section 4.2.2.1: whereas an illocutionary force is determined by the lexical, semantic and grammatical functions of the individual act, the interactional and rhetorical functions are defined in terms of the relation between discourse units of equal rank in the hierarchical discourse organization. Interactional function and rhetorical function also may serve as linguistic clues for determining the function of a discourse connective (cf. category (iv) below). * Illocutionary force appears to be a useful criterion especially in determining the discourse function of enim (cf. ch. 8, section 8.3.4). It turns out that the use of enim is virtually confined to assertions, regardless of whether these assertions have the form of a declarative or of an interrogative (viz. in the case of rhetorical and ironic questions). The only other type of illocutionary force that is found with enim-units is a quite specific subcategory of the directive speech acts, viz. metadirectives and directives that are to the benefit of the addressee and are 'binding' to a relatively low degree (notably proposals and. advices). This distributional peculiarity of enim forms one of the indications of the particle's discourse-function of stressing consensus. * In the same way we can take the uneven distribution of nam over the various types of illocution as indicative of its proper discourse function: the observation that nam is virtually restricted to assertions29 is fully in line with the description of nam as a marker of subsidiary discourse units, given that 29. For the marginal use of nam in metadirectives cf. 8.3.4 on p. 194. For the small group of cases of nam in a highly specific type of question, cf. 7.4.

linguistic clues

119

questions and directives are prone to occur in central discourse units, whereas assertions are much more neutral in this respect. * Not all particles exhibit a bias in favour of one specific group of illocution types. The particle autem (cf. ch. 10), for instance, appears to have no restrictions at all as far as the factor illocutionary force is concerned. This can be explained by the fact that the force of autem pertains to thematic rather than communicative units, which implies that autem is not sensitive to illocutionary force (illocutionary force being a typical feature of communicative units (viz. acts) and not of thematic units). Even lack of sensitivity to the factor illocutionary force may then count as a linguistic clue. * Indirect clue: the factor illocutionary force is an important clue also for determining the discourse type of a particular stretch of text (cf. 5.2.2.2 and category (i) above): questions and directives commonly propose a course of action to be carried out by the addressee, whose (real or implied) presence is an essential feature of so-called dia-discourse. Assertions on the other hand, as they do not require a specific type of reaction, are much more neutral with regard to the factor discourse type: they are equally frequent in mono- and diadiscourse. (iv) communicative structure An important clue for determining the function of a discourse connective is offered by the position of its host unit in the hierarchically-organized communicative structure (cf. ch. 4). Issues relevant to this point are: * The type of communicative unit to which the force of the particle pertains (e.g. act, move or exchange). * The more specific interactional or rhetorical function of the communicative unit functioning as the particle's host unit, or of the unit to which the host unit is connected. E.g.: the particle nam frequently occurs in discourse units with the rhetorical function of 'justification' or 'explanation'; the particle igitur is prone to occur after a discourse unit with the rhetorical function of 'preparation'; the particles vero and ergo are typically found in reactive moves with the interactional function of 'request for confirmation' or 'verifying question'. * The position of the host unit in an argumentative pattern (e.g. the major or minor premiss of a syllogism). E.g.: Latin igitur is often used in the conclusion of a syllogism, whereas autem and (to a lesser degree) at typically occur in the minor premiss. It is also significant that autem's alleged synonym vero does not display this distributional peculiarity.

120

§ 5.3

* (Interrelated with category (v) information structure, below) Type of connection of the host unit to the surrounding context: backward connection, forward connection (e.g. units that form the opening clause of a text, or convey a sudden change in the thematic development), or isolation (e.g. asides in comedy). A Latin instance of a discourse connective that typically occurs in forward-oriented discourse units is quidem (cf. Solodow 1978). Typical backward-oriented particles are nam and igitur. Autem displays forward- and backward-linking capacities at the same time. Vero and enim can appear in all three types of discourse units (with forward connection, backward connection, or in isolation). This apparent insensitivity to the factor 'connection of the host unit to the surrounding context' may be taken to imply that vero and enim are not markers of sequential discourse relationships (i.e. connective markers in a strict sense, cf. ch. 4), but should rather be regarded as 'situating particles'. * The above discourse clues are revealed, in turn, by a number of linguistic properties, such as illocutionary force (see above), tense and combinability with other particles, and, of course, by the content conveyed by the discourse units involved. (v) information structure a) global: thematic structure

Especially in written discourse thematic boundaries are often marked by the presence of a connective particle that is somehow indicative of the thematic break concerned. Usually such a particle co-occurs with one or more of the following other devices for indicating thematic breaks: * Theme constructions30 * Fronted NPs * Presentative constructions * Resumptive participle constructions * Particular anaphoric/demonstrative pronouns * Metadiscursive expressions which overtly indicate the thematic structure of a discourse (functioning as an announcement, recapitulation, transition formula, etc., see also above on pp. 114-115). 30. Theme constituents are noun phrases or adverbial phrases which are, so to speak, 'taken out' of the clause they semantically belong to, and placed in front of this clause. For Theme constructions in Latin, cf. Hoffmann (1989) and Somers (1994); for the notion Theme in general, cf. Dik (1989). In chapters 7 and 10 I will come back to Theme constructions. It is to be noted that not all fronted NPs are Theme constituents.

linguistic clues

121

The above list is based on an investigation of the distributional properties of the Latin particle autem (cf. ch. 10), which appeared to be a marker of thematic discontinuity. As to the co-occurrence with particular anaphoric pronouns it is significant that autem often co-occurs with the so-called demonstrative pronoun ille ('he'), but only rarely with anaphoric is ('he'). The situation of the discourse connective igitur is the reversed: is igitur is a quite common combination, whereas ille igitur is rare. These observations can be explained on account of the fact that ille and autem are both typically used in cases of thematic discontinuity, whereas is and igitur are usually associated with thematic continuitY! . On the roles of anaphoric and demonstrative pronouns in Latin discourse I refer to Pinkster (1987), Bolkestein (1992) and Bolkestein and van de Grift (1994). b) local: focus marking

The local information structure of a discourse is dependent largely on the alternation of focal (or salient) and non-focal information. Most languages have particles at their disposal that may playa role in marking out (or highlighting) the focal parts of a discourse. Focus-marking particles in Latin may co-occur with one of the following other devices for indicating focus: * Strong deixis * Focus constructions, such as the correlative patterns cum ... tum (,when ... then') and et ... et ('both ... and') * First and second person personal pronouns (esp. in the nominative and accusative cases), which on account of their being optional are inherently focal elements in Latin. Of the discourse connectives discussed in the present study only autem (and perhaps also vero, cf. 11.4.3) can be said to have to a certain extent focus marking capacities. (vi) syntactic status of the clause in which the particle occurs The occurrence of a discourse connective in a relative or subordinate clause (instead of in a main clause), or in parentheses (inc!. asides), may sometimes

31. Examples of the combination ille autem can be found in ch. 10. For is igitur cf. Cic. Rep. 2.4: Qui (i.e. Romulus) patre Marte natus - concedamus enim famae hominum ... - is igitur ut natus sit, cum Remo fratre dicitur ab Amulio ... ad Tiberim exponi iussus esse ('He was the son of Mars - for we may grant that much to the popular tradition ( ... elaboration) - so after his birth they say that Amulius ... ordered him, with his brother Remus, to be exposed on the banks of the Tiber')

122

§ 5.3

be indicative of its status as a 'situating' particle instead of a connective particle in a narrow sense. For instance, the connectives (in a narrow sense) igitur, ergo and nam do not occur in relative clauses that follow the main clause, whereas the 'situating' particles enim and vero do occur in such constructions (cf. Pinkster 1990, p. 253). (vii) tense in narrative discourse The grammatical category tense may be considered (at least as far as narrative discourse is concerned) to have coherence-establishing functions on the same three levels of discourse on which particles are assumed to have their coherence-indicating functions (cf. ch.4)32: * The role of tense on the representational level of discourse can be regarded as purely deictic, that is, as being concerned with the ordering of events on a time axis. * The role of tense on the presentational level of discourse is concerned with the hierarchical and functional organization of the discourse units. In Latin, for instance, the imperfect tense appears to be used predominantly for events that play a subsidiary role in terms of the communicative goals of the speaker/writer ('background information'), whereas the perfect tense characteristically occurs in more central segments which convey the successive events on the main story line ('foreground information'). Cf. Pinkster (1990: 254, and the references given there). * Tense may playa role also on the interactional level of discourse, inasmuch as the use in a narrative text of a present tense instead of a preterite may be indicative of a dialogical (or diaphonic) discourse type. In narrative texts there are at least three types of environments in which a present tense may occur: speaker comment or evaluation, embedded (direct or indirect) speech and vivid eyewitness accounts. The use of a present tense in speaker comment and embedded speech is quite explainable: it falls outside the narrative proper and refers to the hic and nunc of the communicative situation in which author and audience (or the reported characters) are having a 'conversation'. The use of the present tense in vivid eyewitness accounts can be explained as an attempt

32. Cf. also Fleischmann (1991), whose proposed framework for the description of tense and aspect shows some interesting resemblances to the framework for the description of connective particles expounded in the present study. Whether aspect is a relevant category for Latin is a question much discussed in Latin linguistics (cf. Pinkster 1990, ch. 11).

linguistic clues

123

of the narrator to get the audience involved in the recounted events. Such passages usually contain also other signals of a diaphonic discourse type (such as interactional particles). For a more extensive discussion of the use of the present tense (as well as of the historic infinitive) in narrative discourse I refer to chapter 11 on vero. In my discussion of the Latin discourse connectives later on in this book, I will only occasionally make use, in my argumentation, of the phenomenon of co-occurrence with particular tenses; this is due to the fact that only a small part of my sample was taken from narrative texts. Moreover, the discourse function of tense in Latin texts has as yet not been studied in all its facets 33 • A more systematic study of the factor tense, in correlation with the use of particular particles, seems likely, however, to be rewarding34 • (viii) genre and text-type Stylistic factors such as genre and text type may play an important role with regard to the selection or avoidance of a particular discourse connective in a particular text. Statistical research on the distribution of a certain connective over the various types of text might thus provide useful information about the particle involved. For instance, when a certain discourse connective proves to be more frequent in conversational texts than in non-conversational texts (especially in comparison with other discourse connectives), this may be taken as an indication that we are dealing with a particle with a function on the interactional level of discourse. The factor genre/text-type is however seldom indicative of the more specific discourse function of a particle. In my descriptions of the individual particles I will only occasionally refer to the factor genre/text-type, mainly as a corroboration of analyses based on other clues. By way of illustration, however, I mention here a few significant observations with regard to the distribution of the particle enim over various genres and text-types: * Enim appears to be relatively rare in poetry (cf. Axelson 1945), with the exception of didactic poetry (Lucretius). This can be explained by the fact that poetry is usually strongly speaker-oriented (i.e. subjective and expressive) and therefore not very likely to host addressee-oriented particles such as enim. An exception to this rule is didactic poetry, which is essentially addressee-oriented

33. For an extensive study of the Latin imperfect cf. however Mellet (1988). 34. For a first attempt, cf. Kroon (J994a).

124

§ 5.3

and therefore preeminently apt to host a particle such as enim, which is characterized by the fact that it appeals to consensus and common ground35 • * In Cicero's letters to his intimate friend Atticus the frequency of occurrence of enim and its alleged synonym nam is in the proportion of about 5: 1. In the correspondence as a whole however (i.e. including the letters to his family, colleagues and acquaintances) the ratio is only 2.5:1. This higher incidence of the use of enim in the letters to Atticus might be explained by a difference in the 'text-type' concerned: personal letters to intimate friends are more likely to contain appeals to shared knowledge and addressee involvement than e.g. business letters and other types of more formal correspondence. Cf. also 8.4.3. * Enim is relatively rare in the works of the Latin historiographers. When it occurs, it is restricted largely to direct or indirect speech. In Sallust, for instance, there are only four instances of enim, three of which occur in embedded speech. Of the 26 examples of enim in the historical works of Tacitus, 20 are found in speeches. Enim's alleged synonym nam is however used quite often in these works, also outside the speeches. The figures for Curtius Rufus' History of Alexander the Great are also of interest. In this work 76% of the instances of enim, but only 42% of the instances of nam, occur in (in)direct speech. These observations are all in line with the proposed discourse functions of enim and nam: enim is primarily an interactional particle and hence prone to occur in conversational texts; nam, on the other hand, is a presentational particle which is mainly involved in the rhetorical organization of a discourse and therefore more compatible with non-conversational (e.g. narrative) texts than enim. (ix) content

As to the factor of content no generalizations are possible of course. A few observations that appear to be of use for determining the discourse functions of individual particles are: * Enim is prone to occur with contents that count as shared or known information: enim-units often appear to convey e.g. information that has been supplied earlier in the discourse or that counts as a generic truth. * Autem occurs very often in the environment of lexical opposites and other semantic pairs (e.g. studium 'activity' vs. otium 'leisure'; ego T vs. tu 'you'). 35. Axelson, however, seems to ascribe the high number of enim-instances (and of other connective particles) in Lucretius to the relatively early date of the work.

linguistic clues

*

125

Vero has a preference for occurring in climactic contexts, e.g. in the last member of a cumulative string (cf. 11.4.1). * Furthermore, we can consider the group of metadiscursive expressions as content-based clues for determining the functions of the particles involved, especially of particles with a text organizational function.

PART II: DESCRIPTION

6

CAUSAL RELATIONS AND CAUSAL CONNECTIVES: CONVENTIONAL ACCOUNTS OF NAM AND ENIM IN THE LITERATURE

Chapter 6 functions as a preliminary to the discussion of nam and enim in chapters 7 and 8. It gives an overview of the way nam and enim have conventionally been described, and discusses in this context the contentious notion of causality.

6.1 Causal relations

Ever since the ancient Latin grammarians, nam and enim have been described as markers of causal relationshipsl. The adequacy of this designation (which will be under discussion in the next two chapters) depends for a large part on how one defines the notion of causality. Unfortunately there have been very few (if any) attempts among Latin grammarians to explain in their sections on causal conjunction what they mean by 'cause', 'causality', 'causal relation' and the like. This is not surprising, considering that causality is a notoriously difficult concept, not only in language, but also in philosophy2. Nonetheless, despite the lack of a generally accepted definition of causality in language, the concept is regularly used and understood in a range of highly differing contexts: apparently we rely, in recognizing causal relations, on very strong, common intuitions. The following is a concise overview of the most pervasive of the linguistic intuitions about causality, which is intended to give

1. The categories of conjunctions distinguished by the ancient Latin grammarians are variable. Nonetheless, five categories can be mentioned which remain more or less stable in the ancient grammatical tradition: the copulativae, disiunctivae, expletivae, causales and rationales. It is noticeable that the early grammarian Palaemon, who influenced many later grammarians, includes both nam and enim in his list of causales ('causal conjunctions') (cf. Keil 1961, Grammatici Latini I, 225. 17-8). For an overview of the treatment of conjunctions by the ancient Latin grammarians see especially Baratin (1989). 2. For attempts to connect the linguistic notion cause to the logical/philosophical concept cf. Rudolph (1981) and Anscombre (1984).

130

§ 6.1

an impression of what is generally understood by the designation 'causal conjunction' or 'causal connective'. It is only then that we can start a fruitful discussion on the aspect of causality that is allegedly part of the meaning definitions of nam and enim. In general terms causality in language can be taken in a broad and in a narrow sense. In a broad sense 'cause' or 'causality' applies to any dependency relation with the minimal requirement of asymmetry (i.e. irreversibility or unidirectionality): the second phenomenon involved in the relation (the consequent or effect) is derived from the first (the antecedent or cause) and not vice versa 3 • In logical-semantic approaches to text relations (see sections 2.1. and 2.2.1) this quite general type of causal relationship is represented by the symbol '>'4. It applies to such varying notions as effective cause, result, condition and purpose. In the narrower sense the terms cause and causality are reserved for the type of relationship in which a cause is linked to a particular effect. This narrow type of causal relationship differs from the conditional relation on account of the factivity of the cause conjunct. Further it differs from the result relation in that the cause is marked as such instead of the effect. In the same respect cause in the narrow sense differs from the purpose relation (which, in turn, differs from the result relation in that in the latter there is no intentionality involved). In everyday language intuitively-defined notions such as motivation, explication, reason and justification are all associated with cause in the narrow sense. These are in fact also the terms one comes across in the accounts of causal clause linkage in conventional grammars. In chapter 2, however, we saw that in recent decades proposals have been made for a more empirically based sub categorization of causality in the strict sense, which is in accordance with the relatively recent insight that causal relations (as all kinds of basic relation types) may obtain in various 'realities'. This insight provoked the provisional distinction between 'objective cause' or 'external cause' on the one hand, and 'subjective argumentation' or 'internal cause' on the other (or, in

3. The common feature of such relationships is referred to in the literature by terms such as 'kausale Konstante' (Rudolph 1981) or 'regelhafte Beziehung' (Klein 1987). 4. In logical-semantic approaches the category involved is often referred to by the term causeconditional. Nowadays it is more or less agreed that it is not very useful to define the linguistic cause relation in purely logical terms. A more realistic view would be to think of cause-conditional as being only parallel to the logical relation, given that in language the implication relation symbolized by > is based on culturally-based plausibility rather than on law-like necessity.

causal relations

131

the terminology introduced in chapter 4, between representational and presentational cause relations)5. More sophisticated accounts in this line are for instance those of Davies (1979), Anscombre (1984), Sweetser (1990) and Vandepitte (1993). Sweetser (1990) distinguishes, for instance, causality in three realities or dimensions: (i) (ii) (iii)

sociophysical causality epistemic causality conversational causality

Sociophysical causality refers to the causation of one represented event by another. A characteristic of clauses that maintain a sociophysical cause relation with other clauses is that they can provide the answer to a why?-question. Epistemic causality is used to refer to the causation of a conclusion by a premise. Epistemic causality relations may be verbalized in an overt way by adding phrases such as 'I know this because ... '. The specific causal relationship that is based on epistemic causality is often designated 'evidence'. Conversational causality, lastly, is the type of causation that holds between the content of the second clause and the performance of the speech act expressed by the first clause. This type of causality can be rendered explicitly by insertion of the phrase 'I say/ask/etc. this because ... '. Causal relations based on conversational causality often receive the informal designation 'justification' . Examples of the three causality types discerned by Sweetser are quoted under (1). See also the examples (7) - (9) in 2.4.1. (1)

a He loves me because I remind him of his first love b He loves me, because he wouldn't have proofread my whole thesis if he didn't c What are you doing tonight? Because there's a good movie on

The linguistic relevance of such subdistinctions of causality appears from the fact that they are apt to capture the differences between various causal connectives in a number of languages (see the discussion in chapter 2, section 2.4.1). In Latin the subdistinctions correspond roughly to the differences between quia/quod (sociophysical causality), quoniam (epistemic causality) and nam/enim (epistemic + conversational causality)6. 5. The ancient Greek grammarians seem to have been aware already of the difference, witness the fact that they differentiate between O/L'7tOAO"(LKO[ (' causal conjunctions ') and (JUAAO,,(LUTLKOL' (' inferentials '). 6. For a discussion, along these lines, on the interchangeability of the various causal connectives in Latin cf. Bolkestein (1991).

§ 6.1

132 figure: a tripartite subdivision of Latin causal expressions level of reality

causality type

relator

external

sociophysical

quia; quod

internal

epistemic

quoniam

conversational

nam; enim

In anticipation of the discussion of nam and enim in chapters 7 and 8 it should be remarked that for our present purposes the above picture is not fully satisfactory. Although the linguistic interpretations of the notion 'cause' may be useful for marking off nam and enim from other connectives with a more manifest 'causal load' (esp. subordinating conjunctions), they do not say much about the more particular functions of nam and enim. One problem that will arise for instance is the fact that only in some of their occurrences are nain and enim associated with contexts that are similar to (la-c) above. If one nonetheless wants to describe the full range of uses of nam and enim in terms of the notion cause, this would require a widening of this notion to such an extent that it would become so abstract and unspecific as to be devoid of any practical value. In the next section I will give a survey of existing accounts of nam and enim and discuss how Latinists have tried to cope with these problems.

6.2 Previous accounts of nam On account of instances like (2) and (3), nam is commonly ranked among the causal coordinating conjunctions. In (2) the clause introduced by nam contains the reason for the action described in the preceding clause. In (3) the nam-' clause supplies a motivation for the preceding directive clause. In both cases there are two subsequent clauses which cQntain two causally related states of affairs. (2)

Adibo ad hominem, nam turbare gestio ('I'll approach the man, for I'm aching for a row', PI. Men. 486)

(3)

Date operam, nam nunc argumentum exordiar ('Pay attention, for now I begin with the plot', PI. Mil. 98)

previous accounts of nam

133

However, as I mentioned above, in both preclassical and classical Latin there are also numerous instances of nam that resist a treatment in terms of the establishment of a causal relation between subsequent clauses. In (4) - (7), for instance, a description of the function of nam as a marker of a causal relation seems to be rather far-fetched or even nonsensical, however loosely this relationship be defined: (4)

Aristoteles putat causam tribus modis dici: "prima" inquit "causa est ipsa materia, sine qua nihil potest effici; secunda opifex; tertia est forma, quae unicuique operi inponitur tamquam statuae". Nam hanc Aristoteles 'idos' vocat. "Quarta quoque" inquit ... (,But Aristotle thinks that the word "cause" can be used in three ways: "the first cause" he says, "is the actual matter, without which nothing can be created; the second is the workman; the third is the form, which is impressed upon every work, - a statue for example". This last is what Aristotle calls the 'idos'. "There is, too, " says he "a fourth, .. .', Sen. Ep. 65.4)

(5)

Sed quaeso, ut scribas quam saepissime, non modo si quid scies aut audieris, sed etiam si quid suspicabere, maximeque quid nobis faciendum aut non faciendum putes. Nam, quod rogas, curem, ut scias, quid Pompeius agat, ne ipsum quidem scire puto; nostrum quidem nemo ('I entreat you, write to me as often as possible, not only what you shall know or hear, but even anything you may suspect; and especially give me your opinion as to what 1 ought or ought not to do. As to your request (nam, quod ragas) for information on Pompey's policy, 1 don't think he knows himself; certainly none of us knows', Cic. Au. 7.12.1)

(6)

Sa:

Leno sum, fateor, pernicies communis adulescentium, periuriu', pesti'; tamen tibi a me nulla est orta iniuria Ae: Nam herc1e etiam hoc restat (,Slave-dealer 1 am, the common bane, 1 admit it, of youth, liar, and nuisance; still 1 didn't start outraging you # (sarcastically) That's to come, is it?', Ter. Ad. 188190)

(7)

Allobroges diu in incerto habuere quidnam cons iii caperent ('The Allobroges for a long time were in doubt what course to pursue', Sal. Cat. 41.1)

In (4) the clause introduced by nam contains a comment of Seneca in which he adds the actual Greek word Aristotle used for what Seneca himself has rendered by means of statua in the preceding clause. It is obvious that there is no representational relationship of cause involved between the nam-unit and the clause preceding it. At most the clauses can be said to be joined by some

134

§ 6.2

vague rhetorical (i.e. presentational) relationship7. A causal relationship is still harder to trace in cases like (5), where nam appears to be used at the beginning of a new paragraph, without there being a link with the content of the preceding context. In the dialogical stretch of text in (6), where nam obtains between clauses that are uttered by different speakers, an interpretation of the nam-clause as being causally linked to the preceding clause would be equally forced. As we will see later on, instances like (6) have led many grammarians to distinguish a 'modal' particle nam with an affirmative or asseverative force, alongside the more common causal connective. In (7) we see an example of the enclitic use of nam with question words, which is also obviously of a different order from the causal connective use. Studies of nam have not failed to notice this range of different uses of the particle, and they usually devote considerable space to the question whether and how the uses are related. For quite some time the prevalent view has been that nam should be described as an explicative or causal conjunction in all its occurrences, including cases like (3) - (6) above, which are to be explained as involving some sort of ellipsis. This means that the explanandum for which the nam-clause provides the explanation has to be supplied from the context: '[verum est], nam ... (this is true, for ... ); [id dico], nam ... ('I say this because'); and the like8 . The approach can thus be characterized as an attempt to describe nam as a marker of the representational relation of cause in all its occurrences. The ellipsis-solution has been firmly rejected by Hand (1829-1845, vol. IV: 12): "scilicet erroris causa in eo sita est, quod nam reddunt per denn, nec subtilius quaerant, qualis sententia legitime possit omitti, nec respiciunt naturam particulae, quae etiam affirmandi vim in se habet." ('the mistake is obviously caused by their rendering denn for nam, and they fail to make sufficiently clear what kind of sentence it is that might be left out. Moreover they overlook the true nature of the particle which also comprises an affirmative aspect. '). Instead Hand assigns to nam a 'designating force' (vis designandi) which may lead to an explicative value of the host unit (expositio; 7. Sweetser (1990) would speak of 'conversational causality'in cases like these, see ex. (I) c above. 8. Cf. Schiwy (1932: 12; 25-34), who refers e.g. to Draeger (1878, II, p. 154 ft). In Lodge's Lexicon Plautinum (1904-1933) about 80% (!) of the instances are described as 'cum quadam ellipsi'. Conjectures and the assumption of lacunas in the text have also been popular solutions. Note also the common use of commentaries to use paraphrases in the description of the function of nam, e.g. Pease ad Cic. ND 1.27; Friedlander ad Petr. 38.4; Petersmann (1977) ad Petro 44.3; Langen (1880: 262) ad PI. St. 572-3.

previous accounts of nam

135

German 'namlich'), a causal value (comprobatio; German 'denn'), or an affirmative value (asseveratio or affirmatio; German 'ja'). The explicative value, however, is still seen as the prevalent or central one. What Hand means by a 'vis designandi' is not very clear. Hand's viewpoint has been adopted, with a few changes, by KuhnerStegmann, who introduce a diachronic aspect into the description of nam. Unlike Hand, Kuhner-Stegmann assign historical priority to the affirmative use of nam, which abounds in the preclassical period, and continues to be used in classical Latin - though much more sparingly and only in highly specific contexts. From this original affirmative use, according to Kuhner-Stegmann, an explicative ('erklarend') and a motivating ('begriindend') use develop, which, in turn, give rise to a host of other uses, including the use of nam in emphatic interrogative clauses (cf. ex. 7), and its use in the occupatio and praeteritio (see section 6.2.3 below, and 7.3.2). A more thoroughly diachronic treatment of the particle can be found in Schiwy (1932). In this monograph on nam Schiwy attempts to demonstrate that the causal use of nam is not primary, but should be regarded as a later development of an originally affirmative particle. This affirmative particle incorporates what Schiwy calls an 'anaphoric' and a 'deictic' aspect: it refers back to (an element of) the previous utterance, and at the same time confirms it. According to Schiwy the 'anaphoric' aspect gave rise to the strictly affirmative use ('indeed', 'really', etc.), the 'deictic' aspect to the explanatory-causal one ('for', 'because'? Both developments - which are considered to have taken place independently from each other - must have reached completion already in preliterary times. Schiwy differs from Kuhner-Stegmann in that he traces all other uses of nam (e.g. the neutral, 'copulative' use, see ex. 5; or the use of nam in emphatic interrogative clauses, see ex. 7) back to the affirmative use, and not to the causal one, thus avoiding the inelegant solution of the assumption of a syntactic ellipsislO. An important implication of Schiwy's view is, therefore, that the uses exemplified in (4) - (7) above are not necessarily to be assigned to a stage in the development of nam later than the rise of the 'causal' use.

9. Cf. also Poyser (1952): "for some considerable time the meanings 'indeed' and 'for' existed concurrently and occasionally almost indistinguishably". 10. Hand, KUhner-Stegmann and Szantyr succeed only partially in avoiding the employment of an ellipsis.

§ 6.2

136

In summary we can say that Latin reference books unanimously discuss nam in the sections on coordinating causal conjunctions, even while admitting that there is little support for the view that 'causality' is the central or most characteristic feature of the particle. This inconsistency should probably be considered in the light of the longstanding tradition in Latin grammars to assign a concrete semantic meaning to particles in general, and the connective ones in particular, the causal meaning being undoubtedly the one to be grasped most easily!!. Schiwy's monograph on nam, on the other hand, not only offers a useful collection of instances, but also provides us with a more plausible alternative to the traditional description, because he regards the causal use of nam as only one of many different uses that developed from its originally affirmative force. His approach too, however, has some apparent flaws. For one thing, he fails to give a good description of the specific type of affirmative particle we might be dealing with here 12 • The term 'affirmative particle' appears to be used by Schiwy as the common wastebasket term for any 'modal' particle with a highlighting or corroborating effect on the host unit, including Latin particles such as enim, vero or modo. Without a proper definition based on sound linguistic criteria the term 'affirmative particle' is as little revealing of the actual function of nam in a text as is the term 'causal conjunction'. With regard to (8), for instance, it is not very insightful to state that nam should be interpreted affrrmatively rather than causally!3: (8)

Quid negoti est? Nam occupatus sum ampliter ('What do you want? For I am an extremely busy man', PI. Cist. 598)

The shortcomings of Schiwy's approach are even clearer in his discussion of the category of 'affektschwaches affirmatives' nam!4, illustrated by example (9). Not much appears to be gained here by describing nam as having a

II. The treatment of conjunctions by the ancient Latin grammarians has in tum been influenced strongly by the Greek grammatical tradition. Cf. Baratin (1989) for a discussion. 12. Another problem with Schiwy's approach (as well as with Kilhner-Stegmann's and Szantyr's) is that the proposed diachronic developments are hard to prove, especially those which are assumed to have taken place in pre literary times. Observations such as that the strictly affirmative use of nam predominates in pre classical Latin, whereas the causal use is typical for the classical period, do not necessarily point, in my opinion, to an affirmative origin of nam: the uneven distribution of the various text types over the period of Latinity should also be taken into account as a potentially distorting factor for the diachronic picture. 13. Cf. Schiwy (1932: 45), category I, 'affirmatives nam, affektstark, versichemd, beteurend'. 14. Schiwy (1932: 51-61).

previous accounts of nam

137

slightly weakened affirmative value 15 : (9)

Amicum castigare ob meritam noxiam immoene est facinus, verum in aetate utile et conducibile. Nam ego amicum hodie meum concastigabo pro commerita noxia ('Castigating a friend even when his offence deserves it, is a thankless job, but at times it's useful and expedient. Now here am I (nam ego) - with a friend I mean to castigate thoroughly, as his offence thoroughly deserves', PI. Trin. 23-26)

The conclusion must be that in terms of minimalistic and maximalistic approaches (see chapters 3 and 5) not only Hand, Kuhner-Stegmann and Szantyr, but also Schiwy have a position at the strongly maximalistic end of the scale. They provide us with an extensive list of uses of nam, the interrelatedness of which remains - at least from a synchronic point of view rather vague 16 •

6.3 Previous accounts of enim The Latin particle enim is usually described in much the same way as the particle nam, as an originally affirmative particle which gradually develops into a causal connective. Compare Kuhner-Stegmann (p. 119): "Das Bindewort enim [... ] stimmt in Bedeutung und Gebrauch im allgemeinen mit nam uberein. Es hat ebenfalls von Haus aus nur versichernde Bedeutung, die sich bei ihm indes in ausgedehnterem Umfange erhalten hat als bei nam; im Altlatein wiegt sie noch durchaus vor. Daraus hat sich dann weiterhin sein Gebrauch in erkliirenden und begrundenden Satzen entwickelt". An example of what could be considered a 'prototypical', 'causal' instance of enim is (10): (10) lam eum, ut puto, videbo; misit enim puerum se ad me venire ('I think I shall see him soon, for he has sent a servant to announce his coming', Cic. Au. 10.16.5)

The problems one encounters in dealing with nam (see 6.2 above: how are we to account for the non-causal uses of the particle?) hold also - and a fortiori for the treatment of enim. Not only in early Latin, but in classical Latin as well, there are numerous instances of enim that defy a description as a causal

15. See p. 148 for an alternative description of this example. PI. Trin. 23-26 is a much-discussed example. In Cicero Inv. 1.95 and Rhet. Her. 2.35 this example is adduced to demonstrate an injirma ratio ('weak reason'). 16. Illustrative in this respect is the way in which Hand, KUhner-Stegmann and Szantyr treat the use of nam in the occupatio and praeteritio (see chapter 7, pp. 157-158).

§ 6.3

138

connective. This can be seen from the following examples. The tentative translation of enim in these examples will be commented on in chapter 8: (11) (Caesar has given orders to arrest Durnnorix) si vim faciat neque pareat, interfici iubet, nihil hunc se absente pro sano facturum arbitratus qui praesentis imperium neglexisset. Ille enim revocatus resistere ac se manu defendere suorurnque fidem implorare coepit, saepe cIamitans ... ('if he (sc. Durnnorix) offered force or refused to obey, he (Caesar) ordered to put him to death; for he supposed that a man who had disregarded his command before his face would do nothing rightminded behind his back. And indeed (enim) when Durnnorix was summoned to return he sought to resist and to defend himself by force, entreating the help of his followers and crying repeatedly .. .', Caes. Gal. 5.7)

(12) Artifices scaenici ... hoc indicio imitantur verecundiarn. Deiciunt enim vultum, verba summittunt, figunt in terram oculos et deprimunt (' Actors in the theatre ... imitate bashfulness by means of the following signs. They hang, as you know (enim), their heads, lower their voices, and keep their eyes fixed and rooted upon the ground', Sen. Ep. 1l.7)17 (13) verum est, quod apud Athenodorum inveni: "Tunc scito esse te omnibus cupiditatibus solutum, cum eo perveneris, ut nihil deum roges nisi quod rogare possis palam". Nunc enim quanta dementia est hominum! ('It is a true saying which I have found in Athenodorus: "Know that you are freed from all desires when you have reached such a point that you pray to God for nothing except what you can pray for openly". But, as you know (enim), how foolish men are now!', Sen. Ep. 10.5)

In (11) the events referred to in the enim-unit (resistere ... defendere ... impiorare coepit) take place at a later stage of the chronological development than the event described in the preceding clause (interjici iubet). Therefore, it is not easy to take the enim-unit as providing a natural or logical cause for the event referred to in the foregoing clause. In other words, in terms of the theoretical framework outlined in chapter 4, enim in (11) does not mark a representational relation of cause. Likewise in (12) the cataphoric hoc indicio ('by dint of the following indication') prevents us from interpreting the states of affairs referred to in the enim-unit as being causally linked to the state of affairs in the preceding clause 1B • Moreover, on much the same grounds (chronological posteriority of the event in the enim-unit in a string of subsequent events; cataphoric reference) it seems inappropriate to regard the enimunits in (10) and (11) as providing subsidiary information (explanation, 17. A comparable example is Cic. Sen. 18.65. 18. Comparable examples are Cic. Fin. 1.18-9; Sen. 65; 77.

previous accounts of enim

139

justification, background, and the like) with regard to the preceding text. In other words, (11) and (12) plead also against a description of enim as a marker of some more 'vague' type of causal relationship on the presentational level of discourse - a contention one might perhaps uphold with regard to nam, as we will see in chapter 7. In (13), lastly, the utterance containing enim seems to be connected contrastively to the preceding text, rather than causally. The conventional treatment of enim in the literature runs, as I stated above, largely parallel with the treatment of nam. The instances of enim are in general divided into three groups: (i) affirmative enim; (ii) causal and explicative enim; (iii) transitional (or copulative) and adversative enim 19 • These different uses of the particle are usually presented and explained in terms of a diachronic development2°: it is observed that in Plautus enim is used predominantly (or exclusively; cf. Langen 1880) as an affirmative sentence particle, whereas Terence already shows the beginnings of a more firmly established causal or explicative use. The latter is said to become the normal - or even the only - use of enim in classical Latin, while traces of the original affirmative force are thought to be present in a number of fixed combinations only, such as et enim, at enim and enim vero. Furthermore it is sometimes assumed that in later stages of the language, probably as early as Petronius, enim loses its specific 'causal' potential, and comes to be used as a merely transitional, or even adversative particle 21 • Since the non-causal, affirmative use of enim, at least in early Latin, appears to be more pervasive and less disputable than the affirmative use of nam22 , grammars seem to be less inclined to resort to an ellipsis solution (cf. p. 134 with regard to nam) which would force the 'deviant' uses of enim into the straitjacket of a causal connective23 • However, despite this awareness of a non-causal use of enim we invariably come across the particle in those parts of

19. See e.g. the exposition of enim in the TLL: (i) enim affirmativum; (ii) enim rationale; (iii) enim = autem. 20. Cf. e.g. Clement (1897), Langen (1880: 261-71), KUhner-Stegmann (p. 119-29) and Szantyr (p. 507-508), who speaks of a "nachplautinische Bedeutungsverschiebung". Cf. also Petersmann (1977: 254) in his study on the language of Petronius: "Zu Plautus' Zeit war enim noch fast ausschliel3lich Beteuerungspartikel im Sinne von 'ja, wahrhaftig' ( ... ). In der klassischen und spateren Latinitat ist dieser Gebrauch mit Ausnahme weniger Verbindungen wie at enim (... ) bereits altertUmlich (er fehlt daher bei Caesar und Cicero)". 21. Cf. Szantyr (p. 508-509), and especially De La Harpe (1923: 85 ff.). Cf. also the OLD sv enim, no. 6 and the TLL sv enim, p. 589, 68ff. 22. Cf. Priscian XVI.l5; for a discussion of this passage, cf. Fontenrose (1944: 193-4). 23. Exceptions are e.g. Petersmann (1977: 254-5, ad Petro 72.5) and Clement (1897: 406, ad Pl. Poen. 604). Hand (1829-45: II, 376) explicitly rejects the ellipsis solution.

140

§ 6.3

the grammars which deal with causal conjunctions, 'causality' being considered the most characteristic feature of enim in classical Latin, and classical Latin being the main concern of most grammars. The more general drawbacks of the traditional expositions of the 'causal coordinating conjunctions' have been touched upon already in the previous section on nam. For the most part these drawbacks are related to the rather strict maximalistic method of description employed24 • With regard to enim the shortcomings are very apparent in Hand's discussion of the particle (182945, vol. IV), and in the discussion by Kuhner-Stegmann (1912) that is based partly on Hand; both expositions provide a considerable list of different uses of the particle, the interrelatedness of which, whether synchronic or diachronic, is not very clear. In Kuhner-Stegmann moreover this impression of heterogeneity and lack of unity in the description is enhanced by the rather confusing terminology employed: in the discussion of the various properties of enim Kuhner-Stegmann speak by turns of the meaning of the particle (p. 120: "die affirmative Bedeutung hat enim wohl noch durchweg bei Plautus ... "); its proper translation (p. 121: "Wenn der Satz eine Erklarung oder Erlauterung enthalt, so IaJ3t sich enim durch 'namlich' ubersetzen"); its junction (p. 122: "Ferner dient enim ... zur Einleitung erklarender Parenthesen"; p. 123: "ferner wird enim wie nam zur Angabe eines Grundes gebraucht); or its use (p. 125: "der rhetorische Gebrauch von enim findet sich ... "). It may be clear that the question of the interchangeability with nam cannot be solved satisfactorily within the traditional maximalistic approach: the various 'meanings' distinguished for enim are so numerous and varied that in general they seem to apply to nam as well. In the literature this gives rise to observations such as the one by Kuhner-Stegmann cited above on p. 137. Compare also Hand (1829-1845, II: 374-5) ad Cic. Off 2.12.42, and Szantyr (p. 508): "Seit klassischer Zeit ist enim in Bedeutung und Verwendung von nam nicht wesentlich verschieden; nur der rhetorische Gebrauch von enim ist seltener als der von nam". A second problem I want to draw attention to is that one gets the impression from the literature that the non-causal use of enim is virtually confined to early Latin, except for a few fixed combinations such as et enim or at enim. That

24. The most important flaws of strongly maximalistic approaches to particle research are: (i) the 'meaning', 'function' and 'use' of the particle are confused; (ii) the function of the particle is systematically confused with the function of its host unit; (iii) the characteristics of the context are mistaken for the proper function of the particle. See also ch. 3 and 5.

previous accounts of enim

141

this yields too simplistic a picture has been proved already by the examples (11) - (13) above. A diachronic change in the 'meaning' of enim cannot be considered to be a full explanation of classical Latin examples like these. As a final defect of most treatments of enim in the literature we may mention the gratuitous character of the label 'affirmative particle' that is often attached to the non-causal instances of enim - a point brought up also in the discussion on nam in 6.2 above25 • As long as no attempts are made to specify further, on the basis of sound linguistic reasoning, the type of affirmative particle involved, it seems to me to be simply an instance of taking the easy way out2 6 . There is one, more recent study that deserves special mention here. In Fugier (1987) the 'causal connectives' nam and enim are described from a more modern discourse-pragmatic viewpoint, comparable to the one adopted in the present study. The main tenets of this article are that the observed polyvalence of nam and enim should be explained, not by a diachronic development involving a shift in the meaning of the particles, but by their relatively low degree of semantic fixedness. In Fugier's view, the function of nam and enim is to be described according to their conditions of use, which involve: (i) the fact that they figure at the beginning of the clause, and (ii) that they cannot open the discourse, but follow necessarily a preceding clause. Hence they should be regarded as connective items which present their host units as a comment with regard to the preceding clause. Their use may involve a number of quite different semantic relationships, the most stable one of which appears to be 'causality'. The more specific interpretation of these relations depends, however, on the context, and not on some presumed lexical meaning of the particles themselves. On account of this semantic non-specificity nam and enim are preeminently apt to be used as organizers of the textual development. As far as nam is concerned Fugier's conclusions correspond roughly to my own views to be expounded in chapter 7, although the restricted corpus on which her study seems to be based, as well as the lack of a systematic investigation of the material, allow for very general statements only27. I do not

25. For the same view, cf. Fontenrose (1944), who rejects the "perennial favorite 'indeed'" as the most appropriate rendering of enim, but in my opinion fails to come up with a viable alternative description. 26. Even the most detailed statement on this subject does not go beyond the observation that enim has more or less the same meaning as hercle, pol, edepol and profecto (Langen, 1880: 262; 267). 27. To judge by the examples cited, Fugier's corpus must have been based mainly on Cicero, and appears to consist exclusively of relatively unproblematic instances of backward-linking nam.

142

§ 6.3

subscribe, however, to Fugier's treating enim in all respects on a par with nam, an approach which she justifies as follows: " . .. en verite je ne vois entre eux aucune difference de valeur. Seules les distinguent de minimes differences distributionelles, avec les consequences qu'elles entrainent ( ... ). Pour la valeur semantique, discursive, pragmatique, nous les trouvons litteralement equivalents" (Fugier, p. 16) As I will attempt to show later, in chapters 7 and 8, there are many more and more essential - distributional differences between nam and enim than the relatively moderate scope of Fugier's investigations allows to be detected28 •

6.4. Interchangeability of nam and enim Few studies have gone systematically into the question of the interchangeability of nam and enim. Usually one assumes a near-synonymy (see for instance the discussion of Fugier's article above), without giving much attention to the matter. There are, however, a few casual attempts to put into words an intuitively felt difference between the two particles. Fontenrose (1944: 168), for instance, notes his view that nam is heavier and more formal, while enim is lighter and more casual. Another footnote is spent on the issue by Pinkster (1972: 173): "In my opinion enim appeals to the knowledge the speaker/hearer has, or rather is supposed to have, either on the basis of the accompanying context and situation or on the basis of his general knowledge. It does not really establish a causal relation. nam, on the contrary, does." I know of only one article that is devoted entirely to the subject of the interchangeability of nam and enim, and which - as early as the beginning of the century - gives a surprisingly modern, functional presentation of the question: on the basis of the observation in a Ciceronian COrpUS 29 of a significant difference in frequency of enim and nam, Barendt (1902) presumes a 28. According to Fugier (p. 16) the only essential difference between nam and enim is enim's preference for the second position in the clause, which makes it a more versatile particle than the rather strictly clause-initial nam. All other potential differences emanate from these placement features. E.g., enim's capacity to occur in protests depends on its combinability with clause-initial at. In my view the distributional differences between nam and enim are highly revealing for the specific discourse functions of both particles, and as such deserve closer attention. 29. Barendt's observations are based on the following works: de Orat.; Off.; Cati!'; Amic.; Sen.; Sui.; Div.; Ver.

interchangeability of nam and enim

143

contrasted force of meaning. A number of passages in which nam and enimsentences follow each other serve to illustrate that "Cicero uses enim as a corroborative and asseverative particle, only indirectly explanatory, and always with a more or less defined intimate appeal on the part of the speaker" (Barendt, p. 208). Nam, on the other hand, is employed by Cicero as an "explanatory and restrictive corroborative particle, which, while never introducing a full, direct reason (as quod does) for a preceding statement, yet qualifies its application" (Barendt, ibid.). Although Barendt does not provide proper (linguistic) argumentation for his conclusions (he considers the examples to be more or less self-explanatory), and although he makes use of a limited corpus, it will turn out that his intuitions are essentially valid and his conclusions more widely applicable. Lastly, in an article by Bolkestein (1991: 444-446) on the interchangeability of causal connectives in general (i.e. of both subordinating and coordinating ones), it is suggested that nam and enim may differ as to the type of focus employed in the preceding clause. On account of the instances of nam and enim in the third book of Seneca's Naturales Quaestiones Bolkestein argues that enim-clauses appear to be used to justify a preceding utterance containing a highly surprising or controversial focal part, whereas nam-clauses seem to follow on utterances containing a more neutral type of focus. Although this observation should be checked (as Bolkestein herself freely acknowledges) for a larger corpus, we can at least say that it is compatible with the discourse functions of nam and enim as they are to be formulated in chapters 7 and 8. We may conclude that the literature on nam and enim provides us with a number of quite varied observations about potential differences between the two particles. Although in general these observations turn out to be essentially valid, none of them is able to give more than a partial account of the matter. In the next two chapters I will attempt to show how the various statements on the interchangeability question are in fact all closely connected, and how the rather intuitive and subjective reasoning underlying them may be replaced by an argumentation that is more systematically based on linguistic criteria, which are applied to a larger and more heterogeneous corpus.

7

NAM

In the previous chapter I discussed, among other things, the unsuccessful attempts in the literature to describe nam, in all its occurrences, as a marker of causal relationships (i.e. on the representational level) between subsequent clauses. In the present chapter I will argue instead that the concept of subsidiarity, rather than that of causality, should be the central element in a description of nam. This concept (which typically belongs to the presentational level of discourse) is capable of uniting most of the seemingly unrelated usages of nam adduced in the literature. Furthermore the concept will prove to be useful in distinguishing nam from its alleged synonym enim, as will be demonstrated in chapter 8.

7.1

A discourse-pragmatic description of nam

From a discourse-pragmatic viewpoint, and in accordance with the theoretical outlines given in chapter 4, the Latin particle nam may be described as primarily a connective particle which functions on the presentational level of discourse l . More specifically, nam marks discourse units that provide subsidiary information with regard to another, more central discourse unit. In the following I will elaborate on this preliminary definition of the discourse function of nam by discussing systematically the various uses that are mentioned in the literature (see 6.2). For the sake of this presentation tQ.e material has been divided into three groups: 'backward-linking' nam 'forward-linking' or 'copulative' nam (iii) the so-called 'affirmative particle' nam

(i)

(ii)

I. For the distinction between primary and secondary functions and uses, see 5.1. For the concept 'presentational level of discourse' see ch. 4.

backward-linking nam

145

7.2 'Backward-linking' nam In chapter 4 I introduced the view that discourses may be conceived of as recursive systems of hierarchically ordered discourse units, which may be defined in communicative terms (communicative units) or in thematic terms (thematic units). One of the most essential 'building blocks' in such a system is the move. Moves typically consist of a central act, and one or more subsidiary acts 2 • The central acts might be thought of as forming the communicative backbone of the discourse: they are the units most relevant for the development of the discourse, and are usually highly formative of the interactional function of the move. The subsidiary acts accompanying the central act contain supporting material, and usually have a lower communicative impact and a more restricted scope than the central act they are attached to. A subsidiary act may, for instance, supply the evidence for the truth or validity of a central statement; it may justify the specific wording chosen in the central act, or give the motivation why the central act has been uttered at all (e.g. instead of some other, more expected utterance); it may contain the author's personal comment on the facts or events presented on the central discourse line; it may provide illuminating details or explanatory remarks, and so on. The above description of the nature of a subsidiary act perfectly matches the conditions of use of the Latin particle nam. In the majority of cases nam occurs in a monological 3 environment, introducing a unit of text that serves as evidence, justification, background, explication, comment, etc., with regard to a central text part. Most often the connection indicated by nam is backwardoriented, that is, the nam-unit provides subsidiary information with regard to a preceding discourse unit. Examples (1) - (19) in 7.2.1 - 7.2.5 below serve as illustration. For the sake of presentation the examples are arranged according to the common labels used in most taxonomies of text relations (see ch. 2). It should be noted, however, that these labels apply to possible functions of the host unit as a whole, and not to the discourse function of nam (discourse function in the sense defined in 5.1, p. 99).

2. Moves may also have a more complex nature, due to the application of the principles of Listing and Recursion. See 4.2.2.2. 3. For the term monological discourse see 5.2.2.

146

7.2.1

§ 7.2

the rhetorical relation 'evidence'

In (1) - (3) nam is used in discourse units that maintain a rhetorical relation of 'evidence' with respect to the preceding discourse unit4 : (1)

Sed mater rure rediit, nam video Syram astare ante aedis ('but my mother's back from the country, for I see Syra (her servant: CK) standing in front of the house', PI. Mer. 807-808)

(2)

De Hispaniis novi nihil. Magnum tamen exercitum Pompeium habere constat; nam Caesar ipse ad nos misit exemplum Paciaeci litterarum, in quo erat illas XI esse legiones (,There is no news of either of the Spains. It is certain, however, that Pompey has a large army; for Caesar himself sent me a copy of Paciaecus's dispatch, in which the number of legions there is stated to be eleven', Cic. Fam. 6.18.2)

(3)

Cl: Times ecastor Ly: Egone? Mentire hercle Cl: Nam palles male ('Good heavens, you're frightened # I? You're lying # For you're terribly pale', PI. Cas. 982)

7.2.2

the rhetorical relation 'justification'

In (4) - (6) the nam-unit does not so much provide evidence for the truth or validity of the content of the central act, but somehow justifies the fact of its having been uttered at all. In other words, the nam-unit does not pertain to the content of the preceding central act, but rather to its status as a communicative act (justification of an assertion, command and question, respectively): (4)

Nos autem (nam id te scire cupere certo scio) pubJicis consiliis nullis intersumus (,As for me (jor I am sure you want to hear about it) I take no part in public deliberations', Cic. Au. 2.23.3)

(5)

Eho, istum, puere, circumduce hasce aedis et conclavia. Nam egomet ductarem, nisi mi esset apud forum negotium (,Hey, boy! Take this gentleman over the house, all the rooms. I'd take you in myself, you see, except for having business at the forum', PI. Mos. 843-844)

(6)

Coges me? # Minume # Namque id metui ('Will you force me? # Not at all # For I was afraid of that', Ter. Ad. 193)

More or less comparable is example (7), in which Tacitus appears to add nam scriptores in eo dissentiunt in order to justify his indeterminacy in the pres en4. Some other examples chosen at random are: PI. Cisl. 659; Curc. 394; Truc. 657; Ter. Ad. 293.

backward-linking nam

147

tation of the facts (seu quis alius regum dedit): (7)

sedem earn acceperat a Tarquinio Prisco, seu quis alius regum dedit; nam scriptores in eo dissentiunt ('he had received the district as a settlement, either from Tarquinius Priscus or by the gift of another of our kings; for on that point the authors disagree', Tac. Ann. 4.65)

In (8), which resembles example (8) in chapter 6 (p. 136), the nam-unit justifies the use of a conversation-closing formula: sed numquid vis? (' anything else?'). (8)

7.2.3

sed numquid vis? Nam est, quod me transire ad forum iam oportet ('is there anything more I can do for you? (nam) Some business calls me down to the forum', TeL Ree. 272-273)

the rhetorical relations 'explanation', 'elaboration' and 'background'

In the above examples (1) - (8) the subsidiary nam-units have the function of enhancing the reliability of the content of the language user's central act (evidence), or underlining the appropriateness of (the wording of) this act (justification). The following examples, which are typical of the use of nam in narrative texts, are not essentially different: in (9) - (12) the nam-unit fills a potential information gap in the knowledge of the reader, that is it provides additional information which is considered necessary for the addressee to appreciate and understand fully the central information of the main story-line. The rhetorical relation obtaining between the nam-unit and the preceding central text-part could be labelled 'explanation', 'elaboration', 'background', and the like5 • (9)

is nunc Amphitruo praefectust legionibus,

nam cum Telebois bellum est Thebano poplo ('at present this Amphitryon is at the head of the Theban army, the Thebans being at war with the Teleboians', PI. Am. 100-101) (10) Is pagus appellabatur Tigurinus: nam omnis civitas Helvetia in quattuor pagos divisa est ('The name of the canton was the Tigurine: for the whole state of Helvetia is divided into four cantons', Caes. Gal. 1.12.4) (11) Mensa largiter instructa denique, ut dei cuiusdam adventus, sic expectatur adulteri. Nam et opportune maritus foris apud naccam proximum cenitabat 5. Some other examples are: PI. Rud. 70; Cic. Ver. 2.71; Liv. 3.44.6.

148

§ 7.2 ('Then with the table generously set she awaited the advent of her lover as if he were a god. For her husband conveniently happened to be dining out at the fuller's next door', Apul. Met. 9.22)

(12) (In the battle against Veji and Fidenae Tullus has concealed from the Romans the retreat of the Alban army) Dimicatum est enim non magis cum hostibus quam ... cum proditione ac perfidia sociorum. Nam, ne vos falsa opinio teneat, iniussu meo Albani subiere ad montes ('For you fought not only against your enemies, but ... against the treachery and the perfidy of your allies. For, to undeceive you, I gave no orders that the Albans should draw off towards the mountains', Liv. 1. 28.5)

Note that in the last example the function of the nam-unit (iniussu meo Albani subiere ad montes) with regard to the preceding text is nicely spelled out by the inserted clause ne vos falsa opinio teneat ('so that your opinion might not be wrong on this point').

7.2.4

the rhetorical relation 'exemplification'

A special group of backward-linking cases of nam is formed by instances such as (13) and (14) below (see also ch. 6, ex. 9, p. 137), where a general or summarizing statement, functioning as a central act, is followed by a particular instance (not seldom in the form of an extensive narratio) which clarifies or fills in the details of the general statement. This use of nam is highly characteristic of the narrative technique of Livy6. In the cases in question the host unit of nam appears to be larger than the clause it is part of. Since the sequence of a general statement plus a nam-clarification also involves the transition from preparatory remarks to the actual exposition, this particular use of nam appears to have certain text-organizational features as well. (13)

Ea desperatio Tuscis rabiem magis quam audaciam accendit. Nam cum ... (start of an extensive narratio) ('In desperation at this turn the Etruscans had been inflamed to the point rather of madness than of recklessness. For when .. .', Liv. 2.47.6)

(14)

Amicum castigare ob meritam noxiam immoene est facinus, verum in aetate utile et conducibile. Nam ego amicum hodie meum conc~stigabo pro commerita noxia, invitus (,Castigating a friend even when his offence deserves it, is a thankless job, but at times it's useful and expedient. Now here am I (nam ego) - with a friend I mean to castigate thoroughly, as his offence thoroughly deserves', PI. Trin. 23-26)

6. See e.g. Liv. 2.22.1; 2.43.7; 2.46.2; 2.47.2; 2.47.6; 3.26.3; 3.36.3; 3.43.4; 3.50.3; 5.10.2; 5.26.4.

backward-linking nam

149

More or less comparable is the use of nam after an introductory announcement of the theme of the ensuing discourse 7 : (15) Nunc hanc laetitiam accipe a me, quam fero. Nam filium tuom modo in portu Philopolemum vivom, salvom et sospitem vidi in publica celoce ('Now, sir, prepare for the ecstacy of which I am the vehicle. (Nam) A few minutes ago at the harbour I saw your son Philopolemus, alive, safe and sound, in a despatch boat', PI. Capt. 872-874)

7.2.5

additional remarks

Summarizing thus far we can say that the examples (1) - (15), which are representative of most of the instances of nam I investigated, have in common that nam introduces a discourse unit which provides subsidiary information with regard to a preceding, central discourse unit. Depending on the content of the connected units, and on the specific communicative context they are part of, the function of the nam-unit (not to be confused with the function of the particle itself, see p. 145) may be described as providing evidence, justification, explanation, elaboration, background, exemplification, and the like. The constellation of a central and a subsidiary act is typical of a mono logical stretch of text. Nam thus turns out to be an important linguistic means in Latin to signal rhetorical relations on the internal level of the move. A few additional points should be discussed here. A first remark concerns the fact that there is a relatively small group of instances for which the term 'backward-linking' is not fully appropriate, as the nam-unit is inserted parenthetically within the central text part (cf. example 4), or even precedes it, as seems to be the case in (16). Note, incidentally, that in cases like (16), in which nam is forward- rather than backward-oriented, it is hard to speak of a causal coordinating conjunctions: (16) Negotium magnum est navigare atque id mense Quinctili. Sexto die Delum Athenis venimus ... ; hinc (sc. Gyaro) Syrum, inde Delum, utroque citius, quam vellemus,

7. Schiwy (1932) speaks of 'nam expositionis'; other examples are Lucr. 2.67; PI. Poen. 1265; Capt. 872; Am. 120; Ter. An. 52; Cic. Fam. 4.13.6. 8. See also e.g. Liv. 4.43.4; Ter. Ad. 680. The forward-linking interpretation of nam in ex. (16) depends especially on the presence of itaque in the following context, which retroactively signals that the preceding context has the status of a preparation with respect to the unit introduced by itaque. Perhaps we have to assign also a backward-linking force to nam in this example, given that initially nam seems to look back to citius quam vel/emus in the preceding clause.

150

§ 7.2 cursum confecimus. Nam nosti aphracta Rhodiorum; nihil, quod minus fluctum ferre possit. Itaque erat in animo nihil festinare [Delo] nee me movere, nisi omnia Ct.KPCX rVpEWV vidissem (,Travelling by sea is no light matter, even in July. We have taken six days to get from Athens to Delos ... ; from there (i.e. Gyaros) we made our way to Syros, thence to Delos, and to both faster than we could have wished. You know these Rhodian open ships (nam nosti aphracta Rhodiorum); nothing makes heavier water. So I don't intend to hurry or to leave Delos until I have seen all the peaks of Gyrae plain', Cic. Aft. 5.12.1, translation Shackleton Bailey)

My second remark concerns the issue of the discourse type with which nam is typically associated 9 • The above description of nam as a marker of the internal structure of the move implies that the particle will have a strong preference for a monological environment. The instances cited earlier in (3) and (6) do not form real counterexamples to this expected tendency: the nam-units involved may be regarded as postponed subsidiary acts added to a prior central act by the same speaker, and urged by the challenging nature of the intervening reaction of the addressee. Hence there is no reason to speak in such cases of an 'affirmative' use of nam, as Schiwy (1932) proposes (see 6.2)10: (3)

Cl: Times ecastor Ly: Egone? Mentire hercIe Cl: Nam palles male (PI. Cas. 982)

(6)

Sa: Coges me? Ae: Minume Sa: Namque id metui (Ter. Ad. 193)

A more interesting group of seeming counterexamples to the preferred monological discourse type is illustrated by (17) and (18). Here nam occurs after a change of speaker, in a clause that confirms the preceding statement of the first speaker!!: (17) Ac: per epistulam aut per nuntiurn, quasi regem, adiri eum aiunt Mi: Namque edepol vix fuit copia adeundi atque impetrandi

9. For the concept of discourse type see 5.2.2. 10. See also PI. Cas. 197; Poen. 328; Ps. 970. In Tac. Ann. 14.44 the challenging intervening move by the addressee is introduced by at. For at see ch. 12. I!. Comparable with (17) are PI. Capt. 896; Cist. 69; Men. 1137; Mer. 508; Trin. 731; 747; 353; Ter. Eun. 487. Also example (6) in chapter 6 (Ter. Ad. 190) can be explained in this way. Parallels for (18) are PI. Men. 1137 and Cic. de Orat. 2.144. It is remarkable that in all these cases nam is combined with the interjection here/e, (ede)pol or ecastor. Namque may be considered a combinatory variant of nam which may be used instead of nam when preceding a word with an initial vowel (cf. KUhner-Stegmann, p. 213).

backward-linking nam

151

('They say that he is addressed only by dispatch or envoy, just like a king # Yes, for, by Pollux, there hardly was an opportunity to approach him, and to win his consent!', PI. Mil. 1225-1226) (18)

Ar: Solus te solum volo, Hegio He: Istinc loquere, si quid vis, procul. Tamen audiam Ty: Namque edepol si adbites propius, os denasabit tibi mordicus Ar: ... volo (,Hegio, I want a word with you # Say it from there, if there's anything you want from away off there. I shall hear it all the same # That's right, by Jove! For if you go any nearer, he'll bite your nose off # ... okay', PI. Capt. 602-607)

According to Schiwy we are dealing here again with an affirmative sentence particle nam. Such a description conceals, however, the similarities that are intuitively felt to exist between (17) and (18) on the one hand, and (1) - (16) on the other. One may account for these intuitions by assuming that the two groups of instances do not differ in the basic discourse function of nam (viz. the signalling of a subsidiary discourse unit), but only in the discourse type they are used in. Whereas the discourse type of the majority of the instances of nam may be said to be monological rrwnologal, we can speak of a monological dialogal discourse type in the case of (17) and (18). This means that one single move is constituted by two different speakers: speaker 2 adds supporting material (= subsidiary act) to the utterance of speaker 1 (= central act), which thus creates the impression of a strong affirmative reaction by speaker 212. That we are dealing here with the connection of a central and a subsidiary act within one and the same move is especially clear from example (18), since speaker 2 is not the addressee of the utterance of speaker 1, but merely enhances speaker l's words. The hierarchical organization of the various discourse units in example (18) may be schematized as in figure 1. In the final overview of the discourse function of nam in section 7.5 the alleged affirmative force of nam will accordingly be described as a mere side-effect of the particle's use in a dialogal context. 12. Another revealing example in this respect is PI. Am. 605-606, where on account of fateor ('you're right; lit. 'I admit') it is more obvious than in (17) and (18) that nam is used in its usual function as a marker of subsidiary information: Huic homini nescioquid est mali mala obiectum manu, postquam a me abiit # Fateor, nam sum obtusus pugnis pessume (,The fellow is bewitched somehow: the evil hand has been laid on him since he left me # you're right, for the way 1 got beaten to jelly was damned evil'). There is no essential difference between this example and (17) - (18): speaker 2 adduces material to support the utterance of speaker 1, but this time the side-effect of affirmation is made explicit by fateor. Cf. Schiwy (1932: 23), who regards nam here as an affirmative particle to be rendered 'in der Tat'.

152

§ 7.3

figure 1: discourse structure of PI. Capt. 602-607 1.

Hegio: istinc loquere procul; - - - - central a c t } tamen audiam

2.

Tyndarus:

initiating move

namque si adbites propius, os denasabit tibi - - - - - subsid. act 3.

Aristophontes: ... volo ...

exchange

reactive move

7.3 So-called forward-linking nam 7.3.1

possible instances of 'forward-linking' nam

Beside what I call the backward-linking function of nam - which is by far the most common - reference books and other studies tend to distinguish a 'forward-linking', or 'copulative' (see 6.2) function as well. In the cases involved, nam appears at first sight to signal a transition to a new theme (and hence to a new move) in the discourse, rather than a backward connection between a central and a subsidiary act within one and the same move. The frequent reference in the literature to such a 'copulative' function of nam seems to be based on instances like example (5) in chapter 6, and (19) - (26) to be discussed below. In the passage cited in (19) the slave Palaestrio is setting out the plot of the Miles Gloriosus, in which he himself is one of the main characters. He first mentions the name of the play and the location, and then introduces the first protagonist, his current master Pyrgopolynices, whose character he dwells on for some time. Then the slave comes to his own role in the play, and this section is introduced by nam l3 :

13. A comparable example is Mil. 145; cf. the comment of Donatus ad loc.:'''nam' inceptiva particula est". See also Brix (1866-1901) ad loc.: ''Nam; denn was mich betrifft, so geh!lrt meine Treue einem ganz anderen Herrn (vgl. 114 ff.). Damit erschuldigt er das wenig schmeichelhafte Bild, das er von seinem jetzigen Herrn entworfen hat, und leitet ungezwungen zur Erziihlung der Vorgeschichte ilber" (italics mine, CK).

so-called forward-linking nam

153

(19) (New section) Nam ego hau diu apud hunc servitutem servio; id volo vos scire quo modo ad hunc devenerim in servitutem ab eo cui servivi prius ('Now I (nam ego) have not been serving long in this service myself; and I want you to know how I came to be his servant and left my former service', PI. Mil. 95-97)

Other examples that might be adduced to illustrate a forward-linking use of nam are (20) and (21)14: (20) (Book 4 of the Verrines consists of a series of stories about Verres' habit of stealing works of art. In the preceding paragraph Cicero has dealt with Verres' theft of the ivory scenes on the door of the temple of Minerva in Syracuse. In 126 he starts a new chapter in the story of Verres' plundering of Syracuse) Nam Sappho quae sub lata de prytanio est dat tibi iustam excusationem, prope ut concedendum atque ignoscendum esse videatur ('Your theft of the Sappho from the town hall, of course, may very reasonably be defended: it almost seems an action that deserves to be allowed and overlooked', Cic. VeT. 4.126) (21) (Concerning people with magic powers) Quorundam hominum tota corpora prosunt, ut ex iis familiis quae sunt terrori serpentibus ... , quorum e genere sunt Psylli Marsique et qui Ophiogenes vocantur in insula Cypro ... ; signum eius familiae est ... vernis temporibus odoris virus. Atque eorum sudor quoque medebatur, non modo saliva. Nam in insula Nili Tentyri nascentes tanto sunt crocodilis terrori, ut vocem quoque eorum fugiant ('Of certain men the whole bodies are beneficial, for example the members of those families that frighten serpents ... ; in this class are the Psylli, the Marsi, and the Ophiogenes, as they are called, in the island of Cyprus ... ; a feature of this family ... is the foul smell of its members in spring. Their sweat also, not only their saliva, has curative powers. And (nam) the natives of Tentyris, an island in the Nile, are such a terror to the crocodiles that these run away at the mere sound of their voices', Plin. Nat. 28.30-31)

In the literature - that is, in those studies which do not assume ingenious ellipses - it has often been stated that this 'forward-linking', 'copulative' use of nam is comparable to the transitional particle autem, and may serve as the equivalent of the Greek particle OilS. The view that nam may be employed to signal a transition to a new thematic unit could be supported by the fact that the particle frequently combines with a so-called Theme constituent, giving rise to constructions such as nam quod

14. See also e.g. Plin. Nat 28.73; 28.24; 13.46. 15. Cf. e.g. Schiwy (1932: 66-88). De la Harpe (1923: 73-84) and Szantyr (p. 505-6) even speak of an 'adversative' use of nam in late Latin. For the transitional use of Oi in Ancient Greek, cf. Denniston (1954), Ruijgh (1971: ch. 5) and Bakker (1993); for New Testament Greek, cf. Levinsohn (1987; see also ch. 2). For autem, see chapter 10.

154

x ...

§ 7.3

('now concerning the fact that ... ') or nam X quidem ... ('now as regards

X')16. An illustrative example is (22). The passage cited is part of a catalogue

of Greek philosophers, in which their theological ideas are discussed. After a discussion of Thales, Anaximander, Anaximenes, Anaxagoras, and A1cmaeon, the section on Pythagoras is introduced by nam, the section on Parmenides by nam ... quidem l7 : (22)

Nam Pythagoras ... Tum Xenophanes ... Nam Parmenides quidem .. . ('As for Pythagoras ... Next, Xenophanes ... As for Parmenides ... ', Cic. N.D. 1.27-28)

The combination of nam and a Theme constituent is a device used often in Cicero's letters to indicate new paragraphs. The construction nam quod X is illustrated by (23) - (24)18; the construction nam X quidem by (25) - (26)19. Note that in both (25) and (26) the Theme constituent falls outside the scope of the utinam-clause. (23) semper tu ad me cum argentata accedito querimonia; nam istuc quod nunc lamentare, non esse argentum tibi, apud novercam querere ('Always trot up to me with any complaint that's silver-plated; as to your present lament that you have no money (nam istuc quod nunc lamentare), you're complaining to a stepmother', PI. Ps. 312-314) (24) Nam quod ... reprehendisti ... utinam ... vellent me labore hoc levare! ('And as regards the fault you find with me (nam quod reprehendisti) ... I should be only too glad if they ... would relieve me of this arduous task!', Cic. Planc. 84) (25)

Nam Opimi quidem calamitas utinam ex hominum memoria posset evelli! ('As regards Opimius (nam Opimi quidem), would that his sad story could be erased from the memory of men!', Cic. Planc. 70)

(26) Part of a recital of Trimalchio's immeasurable riches) Mel Atticum ut domi nasceretur, apes ab Athenis iussit afferri ... Ecce intra hos dies scripsit, ut illi ex India semen boletorum mitteretur. Nam mulam quidem null am habet quae non ex onagro nata sit ('He had bees brought from Athens to give him Attic honey on the premises, ...

16. For the notion Theme, cf. section 5.2.3, note 30. 17. The example is discussed by Poyser (1952). Poyser explains the use of nam here as half-way between the purely asseverative usage ('indeed') and the causal one ('for'), paraphrasing 'here is a further instance'. Comparable examples are Brut. 81; Div. 2.66-69; Tusc. 4.52; 71; N.D. 1.93. 18. Other examples of the construction nam quod X are: Cic. Cael.17; Pis. 66; Att. 3.10.2; 3.13.2; 3.15.2; 7.3.3; 7.12.1 (see ex. (4) in ch. 6); 8.9.1; Tac. Dial. 25; Ter. Hau. 16; 671; An. 395; Ad. 15. Cf. also the construction nam quod attinet ... in Livy, e.g. 5.30.2; 6.15.12. 19. For the particle quidem, see Solodow (1978); other examples of the construction nam ... quidem are: Cic. Clu. 187; Pis. 66.

so-called forward-linking nam

155

Within the last few days, I may say, he has written for a cargo of mushroom spawn from India. And (nam) he has not got a single mule which is not the child of a wild ass', Petr. 38.4)

In my opinion, however, the bipartition of the nam-material into nam with a forward-linking function and nam with a backward-linking function is not fully appropriate and gives too simplistic a representation of the facts. For one thing, it leaves unanswered the question whether both functions have enough in common to warrant the use of one and the same discourse connective. In this connection it is interesting to note that there are a number of remaining instances of nam in my material which strongly suggest that the forward-linking function of nam is more closely related to its backward-linking function than one would be inclined to assume on the basis of instances such as (19) - (26). The next section contains first a discussion of a number of instances of nam that appear to have forward- as well as backward-linking capacities, after which I return to the examples (19) -(26). This leads in section 7.3.3 to the conclusion that 'backward-linking' nam and 'forwardlinking' nam in fact involve different uses associated with one and the same underlying discourse function. 7.3.2

the relation between 'backward-linking' and 'forward-linking' nam

It turns out that there are no, or only very few, instances of nam in classical Latin that are strictly forward-linking in the sense that they merely indicate transitions to a new independent move, centered around a new topic of discourse (as autem does, see ch. 10). For the most part, the discourse unit introduced by 'forward-linking' nam also has the nature of an 'afterthought', containing information that is related to the preceding text in a rather associative way, as a kind of addendum. The occurrence of such a unit may be explained by the speaker's inclination to consider an issue in all its aspects and from all possible angles, in order to forestall reactions of the addressee such as 'shouldn't you say something about X at this stage of the discourse?'. On account of their relatively loose and associative connection to the preceding context, these nam-afterthoughts may well entail a minor shift in the topic of discourse, thus yielding the impression of a merely copulative, forward-linking use of nam. The corpus investigated contains several instances of backward-linking nam where the particle appears to be on the verge of a forward-linking use, that is,

156

§ 7.3

instances where the nam-unit not only provides subsidiary information with regard to a preceding discourse unit, but at the same time introduces a new thematic unit. One of them is cited in (27)20: (27) Haec eo facilius magnam partem aestatis faciebant (sc. Galli) quod nostrae naves tempestatibus detinebantur, summaque erat vasto atque aperto mari, magnis aestibus, raris ac prope nullis portibus, difficultas navigandi. Namque ipsorum naves ad hunc modum factae armataeque erant: ... ('They pursued these tactics for a great part of the summer the more easily because our own ships were detained by foul weather, and because the difficulty of navigation on a vast and open sea, with strong tides and few - or hardly any harbours, was extreme. Now the ships of the Gauls (namque ipsorum naves) were built and equipped in the following fashion: .. .', Caes. Gal. 3.12-13)

In the context preceding example (27) Caesar has recounted the military operations of the Gauls at sea, which are much more successful than those of the Romans. He continues with a description of the problems the Roman fleet encountered (quod nostrae naves ... ), and then shifts his attention to a description of the ships of the Gauls. The section introduced by namque ipsorum naves (which extends for several sentences) not only counts as a new start with a shifted topic of discourse (the fleet of the Gauls versus the fleet of the Romans), but at the same time functions as deferred background information (an 'afterthought') with regard to the preceding text. It forms an answer, as it were, to the question 'how is it that the Gauls were more successful at sea?', and hence seems to function as a subsidiary discourse unit with regard to the preceding text. More or less comparable is the use of nam in (28): (28) Cum maxime haec dicente Gaio puer [... ] Trima1chionis delapsus est. Conc1amavit familia, nec minus convivae, non propter hominem tam putidum, cuius et cervices fractas libenter vidissent, sed propter malum exitum cenae, ne necesse haberent alienum mortuum plorare. Ipse Trima1chio cum graviter ingemuisset superque bracchium tamquam laesum incubuisset, concurrere medici, et inter primos Fortunata crinibus passis cum scypho, miseramque se atque infelicem proc1amavit. Nam puer quidem qui ceciderat, circumibat iam dudum pedes nostros et missionem rogabat ('Just as Trima1chio was speaking the boy (puer) slipped and fell [ ... ]. The slaves (familia) raised a cry, and so did the guests (convivae), not over a disgusting creature whose neck they would have been glad to see broken, but because it would have been a gloomy finish to the dinner to have to shed tears over the death

20. Another example is PI. Poen. 849.

so-called forward-linking nam

157

of a perfect stranger. Trimalchio groaned aloud, and leaned over his arm as if it was hurt. Doctors (medici) rushed up, and among the first Fortunata, with her hair down, and a cup in her hand, calling out what a poor unhappy woman she was. As for the boy (nam puer) who had fallen down, he was crawling round at our feet by this time, and begging for mercy', Petr. 54.1-3)

In the scene cited here a boy inadvertently falls against Trimalchio's arm, causing a lot of consternation among the people present, whose reactions are subsequently described. With nam puer quidem Petronius finally focuses on the boy (whom we have lost sight of for a number of clauses), as if replying to the question 'but what happened to the boy?'. Thus the nam-unit not merely introduces a new element in the string of subsequent discourse topics (familia ipse Trimalchio - medici - Fortunata - puer), but also functions as an 'afterthought' (i.e. deferred subsidiary information) with regard to the preceding text. Note that the use of nam, as well as the use of the imperfect tenses circumibat and rogabafl, perfectly match the picture that has been built up thus far in the scene: the absolutely unimportant role of the boy is not only stated explicitly (non propter hominem tam putidum, cuius et cervices Jractas libenter vidissent, sed propter malum exitum cenae, ne necesse haberent alienum mortuum plorare), but also reflected in the chosen linguistic means of expression. Although the puer forms part of a string of subsequent discourse topics, it is made clear in more than one respect that the information concerning the boy does not belong to the central story line. The 'afterthought' status of the nam-unit is even clearer in (29), which Kiihner-Stegmann cite as an example of the rhetorical use of nam in what is called the occupatio22 : (29) Phoenices . . . Hipponem Hadrumetum Leptim alias que urbis in ora marutima condidere, eaeque brevi multum auctae, pars originibus suis praesidio, aliae decori fuere. Nam de Carthagine silere melius puto quam parum dicere ('The Phoenicians ... founded Hippo, Hadrumetum, Leptis, and other cities on the coast. These soon became very powerful and were in some cases a defence and in others a glory to the mother city. As to Carthage (nam de Carthagine), I think it better to be silent rather than say too little', Sal. Jug. 19.1-2)

21. For the relation between tense and the foreground/background distinction, see section 5.3. 22. The term occupatio is given to a part of text that serves as an anticipation of an expected objection by the addressee. For more examples of the use of nam in the occupatio and the related rhetorical device of praeteritio, see KUhner-Stegmann, p. 117-9. Comparable to example (29) are e.g. Cic. Flac. 35 and ad Brut. 1.3a.

158

§ 7.3

In an account of the cities founded by the Phoenicians Carthage is, of course, a highly expected subtopic. Sallust seems to feel obliged therefore to justify (in an afterthought-like manner) the omission of Carthage from this list. After their treatment of nam in the occupatio, Kuhner-Stegmann deal with the use of nam in the praeteritio, a device closely related to the occupatio23 • An example is (30)24: (30) (Concerning the demands of oratory) Accedat eodem oportet lepos quidam facetiaeque et eruditio ... Tenenda praeterea omnis antiquitas ... neque legum ac iuris civilis sci entia neglegenda est. Nam quid ego de actione ipsa plura dicam ... ? ('To this there should be added a certain humour, flashes of wit and culture .. , Further, the complete history of the past ... must be retained in the memory ... nor maya knowledge of statute law and our national law in general be omitted. (nam) And why should I go on to describe the speaker's delivery ... ? " Cic. de Orat. 1.17-18)

In my opinion25, there is no need to distinguish the use of nam in the occupatio and praeteritio as separate categories - as Kuhner-Stegmann do -, nor to qualify nam in these instances as having a mildly adversative force - as De la Harpe (1923) proposes. Just as in examples (27) and (28) nam can be described as introducing a new thematic unit (a new move) which at the same time has a subsidiary function with respect to the prior context: the addition, in an afterthought-like manner, of one more (though less important) subtopic in a string of SUbtopics which constitute a thematic whole. This amounts to saying that nam in the occupatio and praeteritio has both backward-linking and forward-linking capacities. At this stage of the argument we can now return to the apparently forwardlinking, 'Theme-indicating' instances of nam cited in the preceding section 7.3.1, to see whether here too nam can be shown to have a certain (if weaker) backward-linking force. My preliminary conclusion is that it does. On this point it is essential to note that in most of the examples discussed in 7.3.1 (which are repeated below for the sake of convenience) the new topic of discourse introduced in the nam-clause does not come out of the blue, but has

23. KOhner-Stegmann (II: 118-9) give the following description of the praeteritio: "der Redner stellt sich, als wolle er einen Punkt als nebensatzlich oder OberflOssig gar nicht berOcksichtigen, urn dann tatsachlich in dem anschlie13enden Satze darauf einzugehen. Manche Beispiele flir occupatio enthalten im Grunde schon eine praeteritio." 24. Other examples: PI. Am. 41; Capt. 906; Cic. Ver. 3.216; Sen. Ben. 6.10.1; Sal. Cat. 52.34. 25. For the same view, see Poyser (1952).

so-called forward-linking nam

159

been expected by the addressee/reader for quite some time as a necessary element in the course of the speaker's monologue: the nam-unit counts more or less as a constituent part of a series of subsidiary acts (a series of 'subtopics') which all support the central discourse theme26 • In (19) (see also p. 153), for instance, the audience more or less expects Palaestrio to come to speak at a certain moment about his own part in the play - if only to explain his very presence on the stage27 : (19)

(New section) Nam ego hau diu apud hunc servitutem servio; id volo vos scire, quo modo ad hunc devenerim in servitutem ab eo cui servivi prius (PI. Mil. 95-97)

Likewise, the topics of discourse introduced in the nam-clauses in (22) (Pythagoras and Parmenides, see p. 154 above) might be said to be to a certain extent 'expected' topics in a more extensive list of well-known Greek philosophers, Pythagoras being Alcmaeon's teacher, and Parmenides being Xenophanes' pupil: (23)

( ... ) A1cmaeon ... Nam Pythagoras ... Tum Xenophanes ... Nam Parmenides quidem ... (Cic. N.D. 1.27-28)

In the passage cited in (21) (p. 153) there is at first sight no indication that the people of the island Tentyris is an expected topic of discourse: (21)

(Concerning people with magic powers) Quorundam hominum tota corpora prosunt, ut ex iis familiis quae sunt terrori serpentibus ... , quorum e genere sunt Psylli Marsique et qui Ophiogenes vocantur in insula Cypro ... ; signum eius familiae est ... vernis temporibus odoris virus. Atque eo rum sud~r quoque medebatur, non modo saliva. Nam in insula Nili Tentyri nascentes tanto sunt crocodilis terrori, ut vocem quoque eorum fugiant (Plin. Nat. 28.30-31)

However, in book 8 of his Natural History (§ 92-93) Pliny has already discussed the curative powers of this tribe at some length, in a section on the natural enemies of the crocodile. In the same passage the Psylli also were mentioned. This earlier treatment of the people of Tentyris might be taken as support for the view that in (21) they are explicitly presented (by the use of nam) as a topic of discourse that is touched upon only in passing and added as a kind of afterthought.

26. The central discourse theme may be formulated in a central act, but may also be left implicit. 27. And his none-too-flattering description of his master shortly before. Cf. again (see also note 13) the comment of Brix (1866-1901) ad loc., which nicely spells out the dual nature (backwardlinking as well as forward-linking) of nam in this passage.

160

§ 7.3

The view that forward-linking nam commonly has some backward-linking features as well is further corroborated by the observation that most examples of ostensibly forward-linking nam are found in Cicero's correspondence. The nam-units involved have a highly stereotypical form, in that they contain a Theme constituent, and often also an explicit .indication that the topic of discourse broached has been raised by the addressee in an earlier letter: nam quod scribis ... ('now as to what you write about ... ') ; nam quod rogas ... ('now as to what you ask about ... '); and the like28 • Compare the earlier example (24), and the less explicit but comparable examples (25) and (26). Example (20) also may be described in this way: (20) (New section) Nam Sappho quae sublata de prytanio est dat tibi iustam excusationem, prope ut concedendum atque ignoscendum esse videatur (Cic. Ver. 4.126)

Although the constituent Sappho quae sublata de prytanio est ('the statue of Sappho that was stolen from the town hall') is not explicitly presented as an expected topic of discourse, it is likely that Verres' theft of the statue of Sappho is a fact well known to the audience (or at least is presented as such by Cicero). Its mention is exactly what the audience expects in the present corttext, which is completely devoted to Verres' various thefts in Syracuse. The same holds for (31), where Cicero with the following argument turns down Atticus' invitation to stay at the latter's residence in Epirus: (31) primum est devium, deinde ab Autronio et ceteris quadridui, deinde sine teo Nam castellum munitum habitanti mihi prodesset, transeunti non est necessarium ('First, it is too far out of the way; moreover, I should be only four days' march from Autronius and the rest; moreover you would not be there yourself. (Nam) yes, a fortified place would be useful to me if I were settling there, but it is unnecessary, when I am merely passing', Cic. Att. 3.7.1)

The sequence primum ('first') ... deinde ('then') ... deinde ('then') ... nam ('as to .. ') may be regarded as a series of coordinated subsidiary acts which support a prior central act of turning down the invitation. The conjoined subsidiary acts are, however, not of the same order. Cicero first raises three

28. For nam quod seribis ... see e.g. Aft. 3.13.2; 3.24.1; 7.7.5; 11.22.1; 11.24.3; 12.25.2; 12.40.4; 14.8.1; for nam quod ragas ... : Aft. 7.12.1; Fam. 2.16.5; other examples of nam quod in combination with a predicate in the 2nd person singular: Aft. 3.15.2; 7.3.3; 8.2.2; 9.2.A.2. It is noticeable that this use of nam in the correspondence is almost confined to the letters to Atticus. This might perhaps be explained by the fact that the letters to Atticus are more often a reply to, and pick up topics from, a letter by the addressee, than the letters to his family and colleagues. There are also some scattered instances in the orations, e.g. Vat. 6; Ver. 3.159; Plane. 84. Cf. also Schiwy (1932: 106-9).

so-called forward-linking nam

161

closely connected arguments against accepting the invitation (primum ... deinde ... deinde); then, as a kind of afterthought (nam), he takes the edge off an argument that Atticus himself appears to have used (presumably in a prior letter) in favour of accepting the invitation (nam castellum munitum?9. It is as if Cicero realises that in the present context Atticus still expects a reaction to the castellum munitum argument. Note, incidentally, that it is not possible to describe the nam-unit in (31) as being causally linked to the immediately preceding clause. 7.3.3

nam as a marker of the organization of the text

In the light of the examples discussed in sections 7.3.1 and 7.3.2 we may now venture that nam, on account of its common presentational discourse function of signalling subsidiary discourse units (usually establishing a backward connection with the immediately preceding sentence), may under certain conditions also help to move the discourse forwards. This happens when the communicative act hosting nam, though entailing a new discourse topic (and hence starting a new move), is also still attached (as a kind of addendum) to the previous move. Another way of saying this is: in actual discourse we may sometimes discern a text organizational use of nam on a global text level (i.e. between moves), alongside its more common rhetorical use on a local text level (i.e. between two subsequent acts within one single move). To distinguish the specific textual structure involved from, for example, the structure called Listing (see section 4.2.2.3), I will speak in the case of nam of the textual structure of Dependence. Consider figure 2 below, which is repeated from section 4.2.2.3. The figure represents schematically an extended monologue which consists of one 'central' move (built around a central act and one or more subsidiary acts), the closing of which is deferred by the adding of a 'subsidiary' move with a dependent status with regard to the 'central' move. It is the latter, subsidiary type of discourse unit in which we may come across text organizational nam30 • The 'central' move and the 'subsidiary' move

29. Cf. the comment of Shackleton Bailey ad loc.: "Atticus has no doubt pointed to the strength of his Buthrotian 'castle (castel/um munitum may be quoted from his letter) as one reason for Cicero's going there." 30. Following Roulet et at. (1985) the use of a nam-unit could be described as an attempt to produce a move with enough subsidiary material to achieve a 'compIetude interactive', that is, an attempt to establish a move with enough subsidiary material to warrant a successful counter-move

162

§ 7.3

together form a more complex move, which I characterized 'extended monologue'.

In

figure 2 as

figure 2: organization of the text: Dependence signalled by nam central act - - - - - - - - - ,

I~

subsidiary act, ---..... subsidiary act2 - - - - f - - - - - ' subsidiary act3 _ _ _...J

central m o v e ] extended monologue

subsidiary move (nam)

In the case of the textual structure called Listing, by contrast, an extended monologue is presented as a list of 'coordinated' and functionally equal moves. In chapter 4 I mentioned that the connective particle autem is one of the linguistic means used in Latin to signal the introduction of a new, 'coordinated' move in an extended monologue. For a graphic representation of this type of discourse structure I refer to chapter 4, p. 82. However, since the concepts of Listing and Dependence both involve the transition to a new thematic unit, the functions of autem and nam may seem to overlap - as has been noticed in the literature3!. It is important to note, though, that this shared use of autem and nam as markers of thematic transitions is associated with essentially different primary discourse functions. Nam pertains primarily to central - subsidiary relations between communicative units (i.e. between acts or between moves within a complex monological structure). Since moves (in contrast to acts) are defined in terms not only of their communicative unity but also of their thematic unity (communicative and thematic unity being interrelated notions), it is not surprising that nam, in some of its occurrences, appears to be used as a text-organizational marker of thematic breaks. This happens at the introduction of a new, subsidiary move. Autem, as we will see in chapter 10, is connected exclusively with thematic units, irrespective of the communicative organization of the discourse. The of the addressee. 31. See my remark on p. 153 above, and Schiwy (1932: 73). In his letters to Atticus Cicero seems to use nam and autem almost indistinguishably to signal thematic breaks; compare e.g. Aft. 11.16.2 quod autem scribis to the examples of nam quod scribis cited in note 28. See Schiwy (1932: 75) for more examples of the use of autem in Cicero's correspondence.

so-called forward-linking nam

163

primary discourse function of autem can hence be described as purely text organizational: it involves a neutral continuation of the discourse, without the connotation of subsidiarity (afterthought, supplementation) that is characteristic of nam. In this connection it should moreover be emphasized that in the material used for this study a strictly forward-linking use of nam (i.e. nam used as a mere marker of Theme constituents or thematic breaks in general) appears to be rare, if not absent. Of the examples cited in 7.3.2 probably only (26) may be regarded as such, given that the 'afterthought' status of the nam-unit is less clear here than in the other instances32 : (26) (Part of a recital of Trimaichio's immeasurable riches) Mel Atticum ut domi nasceretur, apes ab Athenis iussit afferri ... Ecce intra hos dies scrips it, ut illi ex India semen boletorum mitteretur. Nam mulam quidem nullam habet quae non ex onagro nata sit (Petr. 38.4)

Anyhow, it cannot be denied that in (27) too the nam-unit is backwardoriented, and that it supports a central text-unit in the prior contexe 3 . It is clear that this isolated instance cannot provide sufficient support for the assumption of two separate discourse functions of nam.

7.4 The so-called affIrmative particle nam I will discuss only briefly the question whether it is necessary to distinguish a separate function of nam as an (affirmative or asseverative) evaluation particle on the interactional level of discourse (see chapters 3 and 4). Although

32. Other examples of nam in Petronius that have been regarded in the literature as 'forwardlinking' are Petr. 48.8; 52.3; 56.4; 63.2. However, they all resemble (26) in that the nam-unit is part of a series of subsidiary acts ('subtopics'), and therefore also has 'backward-linking' capacities. Friedlander (1891) ad Petro 38.4 adds also 54.1 and 66f to this list. For 54.1 see pp. 156-157 (= example 28). In 66f a guest called Habinnas recounts the courses that were served at some funeral meal. As a kind of afterthought he adds that there was also ham, but that he skipped this course: NAM pernae missionem dedimus (,For we had let the ham go'). In my opinion the nam-unit has a clearly additional, subsidiary character here. Cf. for example (26) also Szantyr (p. 505), who rejects the possibility of a copulative use of nam, and appears to favour an ellipsis solution: "In Fallen wie Petr. 38.4 ... entspricht nam zwar einem farblos anknUpfenden 'was ... betrifft', ist aber in Wirklichkeit als Einleitung eines Begrundungssatzes zu einem unterdrUckten Zwischenglied gedacht ... ". 33. The central idea for which the nam-unit supplies support has been formulated explicitly in § 37: ipse nescit quid habeat, adeo saplutus est ('he is so enormously rich that he does not know himself what he has').

164

§7.4

especially Schiwy (1932) assigns a central role to such a function of nam, its actual occurrence in Latin after Plautus and Terence appears to be relatively marginal. Besides, a reduction of all uses of nam to an originally affirmative or asseverative particle has, from a synchronic viewpoint, a very limited explanatory value. There are two groups of instances of nam that might in essence qualify for the status of an affirmative or asseverative evaluation particle. First, Schiwy (1932: 40-51), Kiihner-Stegmann (p. 114, no. 3), and Szantyr (p. 505) point to an an affirmative use of nam in Latin comedy, with a few isolated instances in classical Latin. The instances concerned involve for the most part declarative clauses in which nam is accompanied by the interjections hercle ('by Hercules!'), (ede)pol ('by Pollux!'), or ecastor ('by Castor!'). They occur typically in a dialogal discourse type after a change of speaker34. In 7.2 above (pp. 150-152), however, I pointed out that these instances of nam can be better accounted for by a description in terms of a backwardlinking function of nam, that is, as introducing a subsidiary discourse unit35 . They involve cases in which the nam-unit has the function of a postponed subsidiary act, urged for instance by the challenging nature of an intervening reaction of the addressee (see the examples (3) and (6), and the instances cited in note 10); or they occur in a context in which the subsidiary and central unit involved are uttered by different speakers, but nonetheless belong to one and the same move. In chapter 5 I referred to this type of context as a monological dialogal discourse type. Examples are (17) and (18) in 7.2 above, and the instances cited in note 11. A second group of instances, which might provide a stronger case for the existence of an interactional particle nam, is formed by occurrences of the particle in questions. On this use of nam the handbooks are remarkably silent. Szantyr (p. 584, n.l) remarks: "die Verbindung von nam mit interrogativen Partikeln und Pron .... ist noch zu untersuchen". Kiihner-Stegmann's comment (p. 116) that interrogative nam-clauses are in essence cause-indicating clauses is interesting, but tantalizingly brief. Not very informative furthermore is the

34. It should be noted here that the addition of interjections like herc/e, edepol and ecastor is not to be taken as an indication of a merely affirmative or asseverative (i.e. interactional, evaluating) character of nam. This is especially clear from Cic. Fam. 5.10a.l, where the combination nam mehercule occurs in a text-unit that clearly supplies subsidiary information with regard to another text-unit. 35. I found only one example that seems to fit less well with this description: Cic. Ver. 1.133. In Cic. Ver. 3.196 the reading of nam is disputable.

so-called affirmative nam

165

common statement in other reference books that nam merely strengthens the interrogative clause, or that it lends the clause a more urgent characte2 6 • Most examples cited are from Plautus37 , but according to Szantyr (p. 584, n.l) combinations such as ubinam ... ?, quisnam ... ? and quidnam ... ? (where nam is added as an enclitic to an interrogative adverb or pronoun) occur also in Cicero, Nepos, Sallust, Livy, and - sporadically - in Augustan poetry. However, the examples adduced in the literature do not provide sufficient clues for a decision whether nam can indeed function as an unambiguous interactional particle with 'situating' capacities (a 'modal' particle, in a more conventional terminology), or is rather a variant of the connective particle discussed in the foregoing sections of this chapter. In any case, the common description of nam in questions as an affirmative or asseverative particle (a definition that seems to be applicable to almost all 'modal' particles) would be too vague to be fully satisfactory. For instance, it does not do justice to at least two remarkable distributional properties of interrogative nam-clauses: (i) the fact that nam is practically confined to interrogative clauses with the status of a reactive move; and (ii) the fact that nam combines only with num among the question-opening particles, and not with nonne or an. As to the latter of these properties, I cannot think of a reason why a merely affirmative or asseverative particle (as nam in some of its occurrences is held to be) would be compatible with the interrogation particle num, as in (32) below38 , but not with the interrogation particles an and nonne: (32) Hoc vide; dentibus frendit, icit femur! Num obsecro nam hariolust, qui ipsus se verberat? ('Look there; he's grinding his teeth and wacking his thigh! Good Lord! Is he some mad dervish, to beat himself that way?', PI. True. 601-602)

36. See e.g. Riemann (1942: 578); Hale and Buck (1903: 137); Emout-Thomas (1951: 379); cf. also Hand's remark (IV, p. 18: "interrogatio, quae ex animo magis commoto proficiscitur, significantior fit per particulam nam"), Janson (1979, p.105-6: "In questions the word is sometimes used with no other meaning than denoting the speaker's emotion (impatience, indignation, surprise etc."», and the Oxford Latin Dictionary, s. v. nam, no. 7: "in lively or impatient questions". In my opinion the OLD wrongly includes Sal. Cat. 52.33 under this heading: we are in fact dealing here with an instance of text organizational nam of the praeteritio type (see 7.3.2). 37. Gonzalez Lodge's Lexicon Plautinum yields 93 instances of nam in interrogative clauses, with the exclusion of rhetorical nam-questions (the latter are, in my opinion, assertions rather than questions, and hence of a different nature). 38. Other examples are: PI. Truc. 352; Cic. Part. 26 (num quid nam ... ?; some of the mss. read iam instead of nam). For the combination numnam, see PI. Am. 321; 1073; Aul. 389; Cist. 658; Poen.976.

166

§7.4

This distributional peculiarity clearly calls for a more specific description of the interactional function of nam than the one offered in the literature - a description yet to be determined. A more thorough examination of the use of num, nonne and an would perhaps shed more light on the issue. As to the first group of nam-instances referred to on p. 164, consider examples (33) and (34)39: (33) Tr: Bonum animum habete Pa: Nam, obsecro, unde iste animus mi invenitur? ('Keep up your courage, girls # But, tell me, where can I find any courage?',

PI.

Rud.686)

(34) Am: Scelestissimum te arbitror So: Nam quamobrem? Am: Quia id quod neque est neque fuit neque futurum est mihi praedicas ('It's my opinion that you are a damned rascal # Oh sir, why? # Because what you tell me is not so, never was so, never will be', PI. Am. 553-555)

Although admittedly the use of nam in the interrogative clauses in (33) and (34) differs from the commoner use of nam in declarative clauses, I would contend that there is at least one point of resemblance between them. In the case of a declarative clause the nam-unit supplies a piece of subsidiary information, whereas in the interrogative counterpart such subsidiary information is requested. This amounts to saying that both uses of nam at least share the same basic meaning of subsidiarity. Thus in (33), after Trachalio has given the girls the advice to 'keep up their courage', one of them reacts with a mock-serious request for subsidiary information, in order to be able to comply with Trachalio's directive: 'In order to be able to do so you should first tell me where I could find such courage'. In more technical terms we could say that the main sequence (an exchange consisting of a directive initiating move preparing for a complying or rejecting counter-move) is temporarily interrupted for a side sequence (an embedded exchange consisting of a request for subsidiary information and a matching reply), because the original initiating move has not been sufficiently explicit to call forth the intended (verbal or non-verbal) reaction on the part of the addressee. This description holds good also for (34). The above description may be applied regardless of whether nam is used as an enclitic or not, as appears from the following example:

39. See also PI. Bac. 1114; I found no examples of requests for background information introduced by nam in later Latin. Cf. Hand (1829-1845) s.v. nam, p. 9-20.

so-called affirmative nam (35)

167

Credo ego istoc exemplo tibi esse pereundum extra portam, dispessis manibus, patibulum quom habebis # Quamnam ob rem? ('You'll soon have to trudge out beyond the gate in that attitude, I take it - arms outspread, with your gibbet on your shoulders # Because of what?', PI. Mil. 359360)

So far I have raised a few arguments against the assumption of a separate, interactional function of nam alongside its common presentational function. One might, however, also come up with arguments in favour of such a function of nam in the context of interrogative clauses. I found two passages, for instance, in which the particles nam and igitur occur in the same clause, which would be a highly unexpected combination if both particles were used in their common presentational, connective function. For whereas nam typically introduces subsidiary text parts, igitur primarily signals the introduction of a new, central step in the unfolding discourse4o • This would imply that nam and igitur are mutually exclusive in one and the same clause, unless one of them is not used in its common connective function on the presentational level of discourse41 • Another indication for a separate discourse function of nam on the interactional level of discourse might be its occurrence in certain types of indirect questions, for which consider example (36)42: (36)

Allobroges diu in incerto habuere quidnam consili caperent ('The Allobroges were for a long time uncertain what course to pursue', Sal. Cat. 41.1)

It is clear that more research is needed on the subject of the ostensibly affirmative, interactional particle nam. For the moment I confine myself to the preliminary - probably not very satisfactory - conclusion that the use of nam in interrogative clauses may be associated partly with its common basic function of signalling an upcoming subsidiary text-part, and that they share at least the same basic meaning of subsidiarity. However, especially in those cases where it is added as an enclitic to an interrogative pronoun or adverb, nam appears to be used rather mechanically and without a clear connective force. In such

40. Cf. Kroon (1989). 41. The two examples are Cic. Ver. 4.36.80 (Quisnam igitur tuebitur P. Scipionis memoriam mortui), and PI. Epid. 116 (nam quid te igitur retulit beneficium esse oratione, si ad rem auxilium emortuum est). In the latter example the correlation of igitur and si may probably be taken as an indication that igitur is not used as a presentational particle here. 42. Cf. also Ter. An. 235; Cic. N.D. 1.10.24. According to Bodelot (1987: 65) nam is much more frequent in direct than in indirect questions.

§ 7.5

168

cases we should perhaps speak rather of a separate, interactional particle nam2> which is related only diachronically with the connective presentational particle nam/3 • The specific discourse function of such an interactional particle nam2 remains still to be investigated.

7.5 Conclusions A discourse-pragmatic description of nam may run as follows:

stepwise analysis of the particle nam 1. basic meaning: subsidiarity 2. discourse function: (i) presentational level of discourse, connective:

(ii)

3. actual use:

(i)

indicating the subsidiary status of a particular communicative unit with regard to some other, more central communicative unit [interactional level of discourse, situating: affirmation or reinforcement, see 7.4] local, rhetorical use (most frequent): the marking of rhetorical relations between the constituent acts of a non-complex move ('backward-linking nam')

(ii)

global, text-organizational use: the introduction of a new move with a subsidiary status with regard to the prior context (,forward-linking nam)

4. side-effects:

(i) (ii)

indication of the semantic relation of causality indication of the semantic relation of adversativity (see p. 153, note 15) (iii) affirmation (see pp. 150-152)

43. In accordance with Denniston's suggestion (1954: 86-7) for 'Yap, the Greek near-equivalent of Latin nam, examples (17) and (18) on pp. ISO-lSI might be regarded as transitional cases in the development of the connective particle nam into an affirmative or asseverative sentence particle: from nam in affirmative reactions it is but a short step to the use of nam as an assentient particle per se.

conclusions

169

From the examples discussed in this chapter we may conclude that in the majority of cases nam is not involved primarily in causal clause combining, in the sense of marking semantic (notably causal) relationships between consecutive clauses. It appears to be more adequate to say that nam signals the occurrence of a discourse unit which has a subsidiary role with regard to another, more central discourse unit. On account of this basic discourse function, nam can be employed for two different connective jobs on the presentational level of discourse. In other words, to use the terminology proposed in chapter 5, there are at least two actual uses of nam that may be connected to the same basic discourse junction of the particle: a local, rhetorical use (which is by far the most common), and a global, text-organizational use. In its local use nam marks relations between the constituent acts of a noncomplex move. It signals that the communicative act it is part of enhances, supplements or prepares another, more central communicative act, thus constituting a fully developed, complete move. By a 'fully developed', or 'complete' move I mean a move with enough subsidiary material to warrant a successful counter-move on the part of the addressee. Depending on the position of the subsidiary act with regard to the supported act (following or, sporadically, preceding), the nam-unit has the discourse function of an elaboration (= evidence, justification, explanation, background, and the like) or orientation, respectively. Since the supporting nam-unit usually follows the supported unit, I have referred to this use of nam as 'backward-linking'. Examples are to be found mainly in section 7.2. When used globally, on the other hand, nam signals that its host unit functions as a supplement to a prior move, thus being involved in one of the strategies for creating an extended monologue (viz. the strategy of Dependence). As such nam may be said to playa role in the global organization of the text. Since the relation with the preceding discourse is often rather vague, and since nam in a few cases appears to signal a mere change of the topic of discourse (more or less comparable to autem), I have referred to this use of nam as 'forward-linking'. Examples of the forward-linking use of nam have been discussed in 7.3. In addition nam should perhaps be assigned also a secondary discourse function on the interactional level of discourse, in order to account for the use of nam in reactive interrogative clauses.

170

§ 7.5

The causal flavour that may be perceived in the majority of the instances of nam should be explained in terms of a high degree of compatibility of nam with sequences of causally-related clauses (a so-called side-effect), and should not be part of the functional definition of the particle. Likewise the alleged affirmative force of nam (see pp. 150-152 above) is to be seen as a side-effect of the particle used in a dialogal context.

8

ENIM

This chapter contains a discussion of the junction of the Latin particle enim, with special attention to the following two interrelated issues: (i) its status as a (causal) connective particle; (ii) its alleged interchangeability with nam. On these two points it will be argued that enim is primarily a situating interactional particle, whose basic junction differs considerably from the discourse junction of nam. This conclusion will be arrived at by a systematic comparison of the distributional properties and conditions of use of both particles.

8.1 A discourse-pragmatic description of enim

I propose the following definition of the discourse function of enim:

Enim is a situating consensus particle which indicates an appeal to the involvement and cooperation of the addressee in the speech-event, and can thus be said to playa role on the interactional level of discourse The next two sections will be concerned mainly with providing the linguistic evidence in support of the description proposed. In 8.2 I will go into the first part of the definition, that is, the 'non-connective' (viz. situating) and noncausal character of enim. The conclusion that enim is not connective in a narrow sense will be reached by an investigation of its syntactic features on the local level of the clause, and of its behaviour and position within the wider context of the discourse structure. In 8.3 and 8.4 the negative approach ("what enim is not") will make way for a positive one, as I will attempt to establish, in a stepwise manner, the specific type of situating particle we are dealing with here. Section 8.5 contains a number of evaluating remarks and a few conclusions on the issue of the interchangeability of enim with nam.

§ 8.2

172 8.2 The 'non-connective' (situating) nature of enim

8.2.1 syntactic distributional properties As to the syntactic distributional properties of enim at least two observations are worth mentioning. First, the fact that enim easily combines with various types of unequivocal connective particles should be regarded as an obstacle to taking enim itself as a connective in a narrow sense. The following combinations are attested:

enim + 'coordinating' connective particles (i)

(ii)

(iii)

'adversative' connective particles (,but', 'rather') at enim immo enim sed enim verum enim 'additive' connective particles ('and'; 'and not') et enim nec/neque enim 'causal' connective particles nam( que) enim

enim + 'subordinating' connective particlesl (i)

(ii)

'causal' connective particles ('because', 'since') quia enim quod enim quoniam enim quando enim 'final' connective particles ('in order that (not»' ut enim ne enim

I. It should be noted that in preposed subordinate clauses (especially concessive and conditional clauses) it usually cannot be decided whether enim has scope over the subordinate clause only, or operates on a more global level and rather belongs to the subsequent main clause or to the whole complex. As such, the combination of enim and a subordinator cannot always be adduced immediately as support for the non-connective character of enim. For quia enim see below on p. 176. The combinations quod enim, quoniam enim and quando enim are rare. For quod enim cf. e.g. Var. R. 1.13.4; 1.55.1; Apu1. Met. 9.11; 9.25.

'non-connective' nature of enim (iii)

173

other (e.g. 'concessive', 'conditional' conjunctions: 'although'; 'if) etsi enim quamvis enim quamquam enim si enim

Examples (1) - (3) serve as an illustration of the combinatorial possibilities of

enim: (1)

Matronae magi' conducibilest istuc, mea Selenium, unum amare et cum eo aetatem exigere cui nuptast semel. verum enim meretrix fortunati est oppidi simillima: non potest suam rem obtinere sola sine multis viris ('That's more profitable for a fine dame, Selenium dear, to love just one man and pass her days with him, once she's married him. But (verum) a courtesan, you know (enim), is much like a prosperous city: she can't get along by herself, without plenty of men', PI. Cist. 76-80)

(2)

(Charinus is thinking about going into exile) Eu: Cur istuc coeptas cons ilium? Ch: Quia enim me adflictat amor ('What makes you think of taking such a step as that? # Because (quia), of course (enim), I suffer so from love', PI. Mer. 648)

(3)

Quando enim me in hunc locum deduxit oratio, docebo ... (' As (quando) my discourse has led me to this topic, I will show that .. .', (enim: 'as you see') Cic. N.D. 3.43)

As to the combinatorial possibilities of enim it should be noted that not all connective particles combine with enim. There are, for instance, no attestations of the combinations igitur enim or ergo enim in the investigated part of the Ibycus corpus (see chapter 1). Furthermore, the combinations with enim that do occur in Latin literature are not all equally common. For instance, at enim is very frequent, whereas sed enim and verum enim have a much more limited use2 • Besides, the distribution among the various authors is not always even-

2. The selected part of the Jbycus corpus (see chapter I) contains 190 instances of at enim. The combination is especially common in Plautus (22 occurrences) and in the speeches of Cicero (46 occurrences). There are, however, only 9 occurrences of the combination verum enim, 6 of which are found in comedy; the remaining instances (Sal. Cat. 2.9; 20.10; Cic. Ver. 3.194) are all of the type verum enim vera. The combination sed enim (64 occurrences), although more frequent than verum enim, has a rather uneven distribution in Latin literature, as it is used mainly by the Augustan poets and their imitators, and by Apuleius. The only instance in Cicero is de Drat. 1.16, the only instance in Plautus Bac. 1081. The rarity of the combination sed enim (64 occurrences) as compared to at enim (190 instances) is even more striking if one takes into account that sed is about five times more frequent than at in the investigated corpus (5536 instances of at, 28861

174

§ 8.2

balanced, as is indicated for instance by the fact that quia enim is almost confined to the comedies of Plautus and Terence, while, conversely, the typically Ciceronian combination et enim is rare in Plautus and Terence. I take most of these combinatorial preferences and incompatibilities of enim as indicative of the discourse function of the particle, and will return to them as such on occasion in the following sections. For the moment suffice it to say that on account of the combinability of enim with unadulterated connective particles of various types (adversative, additive, causal; both coordinating and subordinating) enim is not a very likely candidate for the status of a connective particle which indicates the linking of two causally related states of affairs. A second 'syntactic' indication of the essentially non-connective nature of enim is provided by instances such as (4) - (5), in which enim is part of what might be called an apodotic main clause, following a temporal subordinate clause. Note that in both cases enim is accompanied by vero (see also below, p. 186 and n. 16): (4)

cum gladii abditi ex omnibus locis deverticuli protraherentur, enimvero manifesta res visa ('as they pulled out the hidden swords from every corner of the inn, well really (enimvero), then the crime seemed to be evident', Liv. 1.51.8)

(5)

Quae ubi obiecta spes est, enimvero indignum facinus videri ('When this hope had been held out to them, they felt that they were indeed (enimvero) ill-used', Liv. 6.14.12)

Another example illustrative of this point is (6), taken from a passage in Plautus' Miles Gloriosus in which the soldier Pyrgopolynices is recounting his virtues. The last all-encompassing virtue is introduced by post ('further'; 'and finally'), a connective item which establishes the more global link between the passage quoted and the preceding context. A role of enim as a marker of a (causal) relation between two states of affairs is not very likely here (nor, for that matter, is a role as a marker of constituent connection3):

instances of sed). For a discussion of the combination sed enim see Fontenrose (1944). Immo (,rather') enim is found 13 times in the investigated corpus. From the group of additive connective particles both et ('and') and neclneque ('and not') frequently combine with enim. The only instance of namque enim is PI. Trin. 61; cf. also Quint. Inst. 1.5.38, where the combination is given as an example of a 'solecism by addition'. In section 8.5.2 I will come back to this example. 3. In Latin the coordination of opposed constituents is commonly not marked explicitly by a connective particle. Textbooks refer to this phenomenon as asyndeton adversativum. Cf. for instance Kiihner-Stegmann (p. 156-7).

'non-connective' nature of enim (6)

175

post Ephesi sum natus, non enim in Apulis ('and finally, I was born in Ephesus, not, y'know (enim) in Apulia', PI. Mil. 648)

8.2.2 discourse type With respect to the behaviour of enim within the more global structure of the discourse, and to its preferred discourse type, several observations appear to be of interest. When we first take a look at enim in the comedies of Plautus and Terence we may conclude that the position of the enim-unit within the wider structure of the discourse does not point to a discourse connective with backwardlinking capacities (like nam), nor to a causal conjunction which establishes a link between two causally or logically related states of affairs (like quia). In other words, enim does not appear to be a marker of sequential discourse relationships. Connective particles of both the presentational and the representational type, and especially the 'causal' ones, typically occur in a monoiogicai stretch of text in which a central act and one or more subsidiary acts together constitute the monological discourse unit called move (see ch. 4 and 5). The discourse environment with which enim appears typically to be linked, however, is hardly compatible with move-internal linking-devices 4 • Some of the data consist of enim-units that are not connected with the preceding context within the same move: they function as asides, opening clauses of a new scene, sudden changes in the ongoing discourse, changes of speaker, and the like. Compare examples (7) - (8)5: (7)

(Mercury in an aside, interrupting Sosia's soliloquy:) Certe enim hic nescioquis loquitur ('Surely (certe) someone is speaking here' (enim not translated here), PI. Am. 331)

(8)

(Stratippoc!es is purchasing the girl he is in love with from a money-lender) St: Accipe argentum hoc, danista. Hic sunt quadraginta minae ... Da: Bene fecisti, bene vale St: Nunc enim tu mea es

4. In Plautus, for instance, only 15% of all instances of enim occur in a context that is compatible in essence with the occurrence of causal or backward-linking connective particles. For Terence the figure is about 30%. 5. Other examples of enim in an aside are PI. Capt. 534; St. 88; Trin. 1134; for enim at the beginning of a new scene or at turning-points in the development of the discourse, cf. also Ter. An. 206; Hau. 1045; Ph. 465. Note that in these examples enim is accompanied by the particle vera (see chapter II, and p. 186 below).

176

§ 8.2 (,Take the money, usurer. Here are forty minae ... # Thanks. Good-bye and good luck to you, sir [exit usurer] # Now you are mine, aren't you?', Pl. Epid. 646-648)

Another indication that enim is not a clear move-internal linking-device is furnished by the fact that the majority of the enim-units in Plautus (about 60%), and 50% of the instances in Terence, count as reactive moves and occur after a change of speaker. In these cases the discourse type in which enim occurs is dialogical rather than monological. Most often the reactive enim-unit forms a reply to a preceding question6 , as is illustrated by (9) and (10): (9)

Th: Quid tute tecum? Tr: Nihil enim (,What are you saying to yourself? # Nothing, of course', Pl. Mos. 551)

(10)

AI: Tun te abiisse hodie hinc negas? Am: Nego enim vero ('Do you deny that you left today? # Of course (enim), for sure (vera), I deny it', Pl. Am. 758-759)

An interesting observation in this respect is that all 27 instances of the combination quia enim ('because of course') in Plautus and Terence (see above, pp. 172-174) occur after a change of speaker as an answer to a why-question, and also that the only two instances of ne enim ('lest, of course') in Latin literature occur at the beginning of a reactive move 7 : (11)

He: Quid tu per barbaricas urbes iuras? Er: Quia enim item asperae sunt ut tuom victum autumabas esse (,What are you swearing by foreign cities for? # Why (enim) , because (quia) they're the same as you said your meals were - perfect terrors', Pl. Capt. 884-885)

(12)

Tr: Ego interim hanc aram occupabo Th: Quid ita? Tr: Nullam rem sapis. Ne enim illi huc confugere possint (,Meanwhile, I'll just occupy this altar # Why so? # You have no sense, sir. Why (enim), lest (ne) the others take refuge at it', Pl. Mos. 1094-1095)

6. 54 of the 160 instances of enim in Plautus occur in answers to a question (wh-questions as well as yes/no-questions). Cf. e.g. Paen. 387; Men. 162; Per. 185; Ps. 979; Trin. 987. In Terence 8 out of the 45 instances of enim occur in an answer to a prior question. 7. A comparable example with ut ('in order to') is PI. Epid. 277: Quam ad rem istuc refert? # Rogas? ut enim praestines argento priu' quam veniat filius (' What is the good of that? # You ask (ragas)? Why (enim), so that (ut) you may have her bought and paid for before our son comes'). Cf. also Cas. 268 and Paen. 855.

'non-connective' nature of enim

177

(13) Pa: Quid metuis? Sc: Enim ne < nos> nosmet perdiderimus uspiam (,What are you afraid of? # Why (enim) , that (ne) we've lost our identity somewhere', PI. Mil. 429)

Although enim occurs most frequently in replies to preceding questions it is in no way confined to this type of reactive move. In (14), for instance, the enimunit forms a reaction to a preceding statement/assertion, in (15) to a preceding

directive: (14) La: Gaudeo tibi mea opera liberorum esse amplius De: Enim non placet. Nil moror aliena mi opera fieri pluris liberos ('I'm glad to report, sir, that your family has been increased by my efforts # WelI (enim) , I don't like that! I don't care to have other people's efforts increase my family', PI. Cist. 777-778) (15) Ch: Adsequere, retine, dum ego huc servos evoco De: Enim nequeo solus: accurre (,After him, hold him, while I calI out the servants # I clearly (enim) can't by myself: come and help', Ter. Ph. 983)

Finally, enim's predilection for co-occurring with at ('but', see note 2 above) perfectly matches the picture of the particle built up thus far, at being a typical introducer of challenging reactive moves. The same holds for the combination of the corrective particle immo ('no rather') with enim, which is, however, much less commons. For an extensive discussion of the particle at I refer to chapter 12, and to Kroon (1994c). From the discussion so far, I draw the conclusion that in Latin comedy enim in contrast with nam, plays no part in marking sequential discourse relationships, and cannot be considered a connective particle in the narrow sense of the term. This conclusion is in accordance with the view of most earlier studies on the subject (see 6.3), but is based on more systematic argumentation, which will, moreover, turn out to be a good starting point for adjusting the very general, if not vague, description of early Latin enim as an affIrmative or corroborative sentence particle: in view of the fact that enim occurs typically in a dialogical discourse type (notably in reactive moves) we can now hypothesize that enim, at least in Latin comedy, somehow plays a role in the

8. There are altogether 6 instances of immo enim in Latin comedy: PI. Capt. 608; Ps. 31; St. 704; Ter. An. 823; Eu. 355; Ph. 338. Outside comedy it is found only in Apu1eius (5 occurrences) and in Col. 12.25.1, which is an example of a self-correction: quidam, immo enim fere omnes, Graeci ... (,certain, no rather all, Greeks .. .').

178

§ 8.2

management of the communicative interaction between the discourse participants. When we now turn our attention from Latin comedy to later, less 'interactional' texts, it can be argued that there are still plenty of indications that the discourse type in which enim occurs, although perhaps not dialogical in a strict sense, can still be said to have some dialogical 'traits'. This means that a number of conversational features can be found within the essentially monological environment of a move, without all the formal characteristics of a dialogical discourse being present. For this hybrid type of discourse I introduced in chapter 5 the term diaphonic discourse. Examples of diaphonic discourse are reported conversations between the internal characters of a narrative, or stretches of discourse that somehow reflect the communicative interaction taking place between the speaker/author and his implied audience. The classical Latin corpus investigated contains a handful of instances of enim within an indisputably dialogical environment, and thus reminiscent of the majority of instances of the particle in Plautus and Terence. Few though these instances are they are significant in that they prove that we should be careful in assuming a diachronical development in the function of enim: from instances such as (16) - (18) below it may be inferred that at least in Ciceronian times enim still has a non-connective and non-causal use. In (16) the enim-unit forms a response to a preceding question: (16) Quae cum Scipio dixisset, "Verene" inquit Manilius "hoc memoriae proditum est, Africane, regem istum Numam Pythagorae ipsius discipulum ... ? ... neque vero Tum Scipio: satis id annalium publicorum auctoritate declaratum videmus" "Falsum est enim Manili" inquit "id totum ... " ('At this point in Scipio's discourse Manilius said: "Is there really a tradition, Africanus, that this king Numa was a pupil of Pythagoras ... ? ... yet we are quite certain that it cannot be definitely proved by reference to our official recoreds" And Scipio replied: "This story is indeed (enim) entirely false, Manilius, ... " " Cic. Rep. 2.28)

In (17) as well the enim-unit is not connected to a preceding clause which belongs to the same move, but counts as a new (reactive?) move after a change of speaker. Note that the words of the previous speaker (i.e. the host) are not cited verbatim: (17) (At a dinner party one of the guests urges the host to send in his young daughter to 'entertain' the revellers. When the father refuses, another guest repeats the request)

'non-connective' nature of enim

179

Hic tum alius ex alia parte, "Enim vero ferendum hoc quidem non est; vocetur mulier!" ('At this someone in another part of the room called out: "But really (enim vero) , this is intolerable: let the woman be sent for!" " Cic. Ver. 1.66)

In the passage cited in (18) enim is combined with the reactive particle at, a combination that is frequent also in Latin comedy, as we have seen above on pp. 173 and 177: (18) Bene Pericles, cum ... casu formosus puer praeteriret dixissetque Sophocles: "0 puerum pulchrum, Pericle!" "At enim praetorem, Sophocle, decet non solum manus sed etiam oculos abstinentes habere" ('How pertinent was Pericles's reply when ... a handsome boy chanced to pass and Sophocles said: "Look Pericles; what a pretty boy!" # "But (at), you will agree (enim) , Sophocles, a general should keep not only his hands, but also his eyes under control" " Cic. Off. 1.144)

As stated above, there are only a few examples in classical Latin of enim in a clearly dialogical context after a change of speaker9. The more important observation, however, is that in the majority of the other, more regular instances of enim the discourse type associated with it is not just simply monological (as appeared to be the case with nam), but monological with a diaphonic character (see ch. 5, section 5.3). That is, in the particular monological stretch of text in which enim is used it is suggested (rather than formally expressed) that an actual conversation is taking place, either between the internal characters of a narrative, or between the speaker/author and his audience. Indications of such a diaphonic discourse type have been listed in chapter 5, section 5.2.2.2. They include the presence (in the enim-unit or immediate context) of first and second person pronouns and verb forms; the use of present-tense verb-forms in a passage that is otherwise presented in the past; the presence of metacommunicative or performative expressions; the immediate presence of subjective expressions; the use in the immediate context of questions and directives; the presence of interactional elements like interjections, swear words and vocatives. Enim-units (or their immediate context) contain significantly more often than nam-units one or more of these elements which pertain to the interactional frame in which the text is embedded. The 9. One obvious explanation of this is that the classical Latin texts that have been handed down to us are in general predominantly mono logical and non-conversational. Examples other than those cited in the text are Cic. Caec. 3.8; Div. 1.123; de Orat. 2.40; Sest. 110; Ver. 1.25; Plin. Nat. 36.5; Quint. Inst. 6.3.63.

180

§ 8.2

interactional frame may surface repeatedly in a text, or only sporadically. A common phenomenon in narrative and expository prose, for instance, is the temporary interruption of the ongoing discourse for a metacommunicative comment. This means that the prevalent narrative or demonstrative 'layer' of a text is temporarily deserted by the speaker/author for a remark that pertains to the expressive and/or communicative aspects of the discourse, rather than to the reality of the represented world. An illustrative example is (19) (the translation is from Barendt 1902): (19) Nam tum cum ex urbe Catilinam eiciebam, - non enim iam vereor huius verbi invidiam - ... ('Inasmuch as at the very time I was trying to drive Catilina out of Rome - for, mark you (enim), I am not afraid now of the odium attaching to this expression .. .', Cic. Cati!. 3.2.3)

(20) is an example of enim in indirect discourse, where enim appears to belon~ somehow to the communicative interaction between the 'reported' characters. Compare its direct 'counterpart' in (17) above: (20) Cum ... dictator unus nihil nee famae nee litteris crederet, ut vera omnia essent, secunda se magis quam adversa timere diceret, tum M. Metilius tribunus plebis id unum enim vero ferendum esse negat ('When ... only the dictator refused to credit either rumour or dispatch, and declared that even though the story were all true, he feared success more than adversity, then Marcus Metilius, tribune of the plebs, cried out that this (id unum enim vero) was past all bearing', Liv. 22.25.2-4)

Finally I want to draw attention to the use of enim in examples like (21). See also (2) above: (21) Ecce tanto periculo malum maius insequitur. De summis enim tectis ... rusticani illi saxa super nos raptim devolvunt (,And look, this great danger is followed by even worse troubl6. From the rooftops, mark you (enim), ... the farmers furiously begin to hurl rocks down on' us .. .', Apul. Met. 8.17)

Example (21) is illustrative of the frequent use of enim in narratives that somehow have the appearance of a vivid eyewitness account. In addition to enim (which, as I will argue in the next section, marks an appeal to the empathy and involvement of the reader, who is e.g. invited by the speaker to imagine himself as being present at the scene of the action) usually the presence of an historic infinitive or an historic present tense also contributes to the impression that the narrated events are taking place before the eyes of the

'non-connective' nature of enim

181

author and his audience. In (21) the use of ecce ('look!', 'mark this!') provides another clue. All in all it appears to be justified to assume that enim in examples like (21) provides the text with an added value of 'interactionality' which would have been lacking if nam had been used instead of enim in the same context. In 8.4.3 below I will return to the use of enim in eyewitness accounts 10 • Significant additional support for the view that enim is associated with a dialogical or diaphonic discourse type (and thus presumably in some way or other plays a role on the interactional level of discourse) comes from the distribution of the particle among the various authors and text-types, and in contrast with nam. Revealing in this respect is for instance the distribution of enim in Livy as compared with that of nam: whereas about 30 % of the instances of enim in Livy occur in a direct or indirect speech, this is the case for only 12 % of the instances of nam. Much the same picture is provided by the comparable - as far as text-type is concerned - historiographical work of Curtius Rufus, where 76 % of the instances of enim, but only 42 % of the instances of nam occur in (in)direct speech. Another remarkable observation is that enim occurs only four times in the historical works Bellum Jugurthinum and Catilinae Coniuratio of Sallustll . Three of these instances occur in an embedded speech (Cat. 51,17; 51,25; Jug. 14,23). The only other instance (Jug. 95,2) belongs to a metacommunicative, 'programmatic' passage. This example has been discussed already in chapter 4, section 4.2.2.3: (22) Ceterum, dum ea res geritur, L. Sulla quaestor cum magno equitatu in castra venit, quos uti ex Latio et a sociis cogeret Romae relictus erat. Sed quoniam nos tanti viri res admonuit, idoneum visum est de natura cultuque eius paucis dicere; neque enim alio loco de Sullae rebus dicturi sumus et L. Sisenna ... parum mihi libero ore locutus videtur. Igitur Sulla gentis patriciae nobilis fuit, ... (Sal. Jug. 95.2)

Note that after the 'programmatic' intermezzo Sallust returns by means of igitur to the narrative layer of his exposition. Another distributional property of enim suggesting a function of the particle on the interactional level of discourse is its frequent co-occurrence with first person forms of what could be called 'subjective evaluation verbs' such as 10. In my opinion the interpretation given here is to be preferred over the ad hoc explanation of the passage in the OLD (s. v. enim, no. 7), where this instance of enim is characterized as "expository, following the announcement of what will be said" (a description which, as a matter of fact, is more appropriate for a specific use of nam, as we have seen in chapter 7, p. 149). II. There are 104 (!) instances of nam in these texts.

182

§ 8.2

credo, opinor, puto ('I think', 'I believe'). In the complete corpus of Cicero's Letters I found 28 instances of enim-clauses containing credo, opinor, or puto, while in the same corpus only 4 nam-clauses turned out to have such a predicate 12 • As a corollary to this section on discourse type we may say that in classical Latin the specific discourse type associated with enim (i.e. predominantly monological with a diaphonic character) highly favours the occurrence of interactional particles. This observation helps to explain the seemingly deviant cases of the particle, that is, those cases in which a strictly connective, presentational interpretation appears to be excluded or forced, and a situating, interactional interpretation seems more reasonable. Compare for instance example (23), where enim is combined with the causal conjunction quia ('because'). The diaphonic character of the surrounding context is suggested by, among other things, the directive expression vide modo, which explicitly refers to the interactional level of the speaker/author (Trimalchio) and his audience: (23)

(Trimalchio relates the story of a workman who has invented unbreakable glass and shows this invention to the emperor; in response the emperor asks the man whether there are also other persons who know how to blow glass like this) Vide modo. Postquam negavit, iussit illum Caesar decollari: quia enim, si scitum esset, aurum pro luto haberemus ('Just see what happened. He said not, and then Caesar had him beheaded. Why? Because (quia), of course (enim), if his invention were generally known we should treat gold like dirt', Petr. 51. 6)

It is interesting to note, in this respect, that the only two instances of quia enim in Latin literature that do not occur in a dialogical context after a change of speaker, obtain in a context with clear diaphonic elements 13. The same could be maintained for the only instance in Latin literature of quando enim, cited above under (7), where the enim-unit is part of a metacommunicative statement.

12. On this point it is to be noted that in Cicero's correspondence enim is only 2.5 times as frequent as nam. The four nam-clauses concerned are in Aft. 8.11.1; 11.9.2 (+ opinor); Fam. 11.9.1; 10.33.5 (+ pula). 13. The other example is Cic. Div. 1.93, where the expression ut tu sales dicere ('as you are wont to say') counts as an explicit indication of the diaphonic discourse type.

'non-connective' nature of enim

183

Another example of enim that has been said to show a deviant (i.e. nonconnective, non-causal) behaviour, but can in fact be perfectly well described in the way suggested in this chapter, is (24)14: (24) Ecce autem subitum atque oculis mirabile monstrum ... / procubuit viridique in litore conspicitur sus / quam pius Aeneas tibi enim, tibi, maxima Iuno,/ mactat sacra ferens et cum grege sis tit ad aram (,But look, a portent, sudden and wondrous to see, ... lay outstretched on the green bank before their eyes - a sow: which the good Aeneas offers in sacrifice to thee (fibi enim: to you of course), yes to thee, most mighty Iuno', Verg. A. 8.81-85)

The apostrophe in which enim occurs clearly refers to the interactional level of the speaker (Vergil) and an addressee (luno), a passage which temporarily interrupts the narrative layer of the text. I will come back to this example on p.204. All in all, the conditions of use of enim discussed in this section (notably the discourse type of the immediate context) suggest strongly that in early as well as in classical Latin the function of enim belongs with the expressive and communicative aspects of a text rather than with the representation of its content, or with the presentation/organization of its content. In other words, enim appears to conform to the definition of an interactional particle as given in chapter 4. In the next section we will take the argument one step further and use these findings in determining the more specific type of interactional particle we are dealing with here.

8.3 The interactional nature of enim

The discourse type with which enim is typically associated suggests strongly that the particle functions on the interactional level of discourse. However, this observation is still rather vague, as interactional particles form a motley collection of words that have somehow to do with the management of the interaction between a speaker/author and his addressee, or in other words, with the interactional frame in which the text is integrated. In 4.2.3 (figure 12) I proposed a subclassification of interactional particles into three main categories, according to whether the particles concerned are

14. Cf. the comment of Servius ad loc.: vacat enim et tantum ad omatum pertinet.

184

§ 8.3

involved in (i) the marking of structural relationships (markers of the conversational structure, e.g. turn-taking devices), (ii) the marking of semanticfunctional relationships (markers of interactional relations between the constituent moves of an exchange, e.g. objection markers), or (iii) evaluation of (part of) the extratextual, communicative situation (e.g. 'modal' or 'phatic' particles). The last category is the most heterogeneous one. I mentioned one parameter for further differentiation within this category, viz. speaker/addressee orientation: situating interactional particles may pertain to the speaker's attitude and emotions, or to the involvement of the addressee in the discourse, or they may be indifferent as to speaker/addressee orientation. In my opinion, enim belongs to the group of situating interactional particles that are addressee-oriented. More specifically, the particle signals an appeal to the involvement and cooperation of the addressee in the speech-event. By using enim the speaker affirms or suggests that there is a certain consensus between himself and the addressee; he pictures the addressee, so to speak, as being on the same 'wavelength'. This consensus aspect has been captured in various ways in the translations and paraphrases of enim used hitherto in this chapter: 'of course', 'as you know', 'y'know', 'indeed', 'clearly', 'you will agree', 'mark you', and the like. In some cases a tag-question appears to be particularly appropriate for rendering the function of enim in English. For want of a good English equivalent I sometimes have recourse to the translation 'for', which, however, lacks the interactional consensus-aspect and is too much associated with sequential discourse relationships to count as a full equivalent of Latin enim15 • Several considerations have led to the proposed description of enim as a consensus particle, some of which have already been touched upon in the preceding paragraphs. The discussion in the next few paragraphs will fall under the headings shared information (8.3.1), co-occurrence with other particles (8.3.2), evidence from question-answer pairs (8.3.3), illocutionary force (8.3.4) and meta-communicative comment (8.3.5).

15. A good Dutch equivalent of Latin enim is immers.

interactional nature of enim

185

8.3.1 shared information

Firstly, the view presented above is in line with the observation that an enimunit often contains information that counts as already-shared knowledge of the speaker and addressee, whether because the host unit contains a general truth or a generic statement (as is the case in ex. 25 below, for which see also 6.3), because it refers to a situation in the past at which speaker and addressee were both present (example 26), or because it refers to something that has already been dealt with or experienced earlier in the current discourse situation or on a previous occasion (examples 27 and 28): supplying shared information entails more or less automatically the consensus of the addressee, which may well be expressed explicitly in the form of a particle. To put this somewhat differently, consensus-particles are highly compatible with the presentation of common knowledge. (25) Artifices scaenici ... hoc indicio imitantur verecundiam: deiciunt enim vultum, verba submittunt, figunt in terram oculos et deprimunt (,Actors in the theatre ... imitate bashfulness by means of the following signs: they hang, as you know (enim) , their heads, lower their voices, and keep their eyes fixed and rooted upon the ground', Sen. Ep. 11.7) (26) Maiores nos res scribere ingressos ... e cursu ipso revocavit voluntas tua. Cum enim mecum in Tusculano esses ... incidisti in Aristotelis Topica quaedam, quae sunt ab illo pluribus libris explicata ('I had set out to write on a larger subject ... when 1 was recalled from my course by your request. You will remember that (enim) when we were together in my Tusculan villa ... you hit upon certain Topics of Aristotle which were expounded by him in several books', Cic. Top. 1) (27) Quae res (sc. tempestas) magnas difficultates exercitui Caesaris attulit. Castra enim, ut supra demonstratum est, cum essent inter flumina duo ... neutrum horum transiri poterat ... ('This (Le. the storm and the consequent breaking down of the bridges) caused serious difficulties to Caesar's army. For (enim) the camp being situated, as has been explained above (ut supra demonstratum est), between two rivers ... neither of these could be crossed .. .', Caes. Civ. 1.48) (28) mirabiliter enim moratus est, sicut nosti ('for (enim) he's a strange person, as you know (sicut nosti)" Cic. Aft. 2.25.1)

186

§ 8.3

8.3.2 co-occurrence with other particles Other support for the view that the content of an enim-unit is often presented as obvious and indisputable comes from the observation that there is one specific, well-defined group of particles and adverbs that regularly co-occur with enim, namely epistemic particles or adverbs that emphasize the truth or actuality of the transmitted content of the utterance involved. This group includes items like certe ('without any doubt', 'certainly'), prafecto (,undoubtedly', 'assuredly'), videlicet ('evidently', 'of course', 'no doubt'), nimirum ('without doubt', 'evidently', 'of course') and 'vera (certainly', 'without doubt', 'really', 'indeed')!6. Especially the combination enim vera is extremely common. For examples of this combination see (4), (5), (10) and (17) in this chapter. Examples of some of the other combinations are!7: (29) Ar: Oh melle dulci dulcior tu es Ph: Certe enim tu vita es mi. Complectere (,Oh, you're sweeter than sweet honey! # And you, without doubt (certe) y'know (enim), are my very life. Embrace me!', PI. As. 614) (30) Sed animadvertendum est diligentius, quae natura rerum sit. Nimirum enim inops ilIe, si bonus est vir, etiamsi referre gratiam non potest, habere certe potest (,But we should observe more carefully how the matter really stands: of course (nimirum), you will agree (enim), the poor man of whom we spoke cannot return a favour in kind, but if he is a good man he can do it at least in thankfulness of heart', Cic. Off. 2.69)

16. Videlicet originated from videre licet, lit. 'one can see'. Nimirum literally means 'unless a miracle has occurred' (ni(si) mirum est). The group contains particles that indicate the degree of certainty of occurrence of the state of affairs referred to in the clause (,objective modality markers'), as well as particles that express a high degree of commitment of the speaker to the truth of the propositional content ('subjective modality markers', or, in the terminology of Greenbaum 1969, 'attitudinal disjuncts'). It is not always clear whether an epistemic particle/adverb is used. objectively or subjectively. The particle vero is even more complex as it may also emphasize the sincerity of the illocutionary intention of the speaker. For a discussion of the subject and for a description of the various uses of the particle vero I refer to chapter II, where I will argue, among other things, that in the fixed collocation enim vero the element vero pertains to the involvement of the speaker in the speech event, whereas the element enim concerns the involvement of the addressee, thus giving rise to a strong assertion. Note that enim and the epistemic particle in question are usually adjacent, but that this is not a strict rule. 17. Other examples are: + certe: PI. Am. 331; 658; Aul. 811; Cic. Alt. 16.3.1; Ac. 1.13.8 + profecto: Cic. Alt. 4.16.9; 4.17.4; Cati!. 3.19.1; Agr. 2.3.8; Liv. 1.15.7; 3.21.5; Sen. Ep. 89.1; Plin. Nat. praef. 16; 18 + videlicet: PI. As. 598; Cic. Alt. 13.31.3; Brut. 289; Cati!. 2.12; Font. 29; Ver. 2.137 + nimirum: Cic. Arnie. 52

interactional nature of enim

187

With regard to the above examples it should be noted that enim is not itself a truth-emphasizing particle, merely highly compatible with such a particle. This is proved for instance by the fact that there are several instances in Latin literature of the combination of enim with the epistemic adverb fortasse ('perhaps', 'possibly'), which qualifies the propositional content as having only a moderate degree of certainty. Compare example (31)18: (31) Raras tuas quidem (fortasse enim non perferuntur), sed suavis accipio litteras (,The letters I get from you are few and far between (perhaps (fortasse enim) they are not reaching me), but they are charming', Cic. Fam. 2.13.1)

Given instances such as (31) it is not very likely that enim is a modality marker of certainty, pertaining to the representational discourse level. An interpretation of enim as a consensus marker, pertaining to the interactional framework in which a clause is embedded is, however, very plausible. Now if we compare enim with nam with regard to their co-occurrence with other particles or particle-like words, we find a remarkable difference between the two: whereas enim regularly co-occurs with various kinds of epistemic, truth-emphasizing particles or adverbs, the combinatory possibilities of nam are almost confined to combinations with imprecations like edepol ('By Pollux!'), hercle (,By Hercules!') and ecastor (,By Castor!'), a combination which, in its turn, is not found with enim l9 • One possible explanation for the latter point might run like this: the imprecations edepol, hercle and ecastor are in origin appeals to the authority of a deity, employed to enforce one's words, for instance in the case of strong oaths or utterances with a highly surprising and unexpected content. Such a context might not be very compatible with a consensus-particle like enim, which presents the information contained in the utterance as obvious and indisputable.

18. Other examples of enim andfortasse occurring in the same clause: Cic. Tul. 55; Fam. 14.20.1; Sen. Nat. 2.53.2; Cels. 3.9.1. 19. This is not to say that the combination of nam with eerte, profeeto, and the like is excluded; cf. e.g. Cic. Fam. 3.9.3 and Aft. 4.7.4 (both examples of nam profeeto), and by Cic. Inv. 1.32 (nam ... nimirum). The important observation, however, is that in the immediate environment of nam one comes across these particles only sporadically and quite incidentally, while their combination with enim appears to be a much more systematic (and sometimes almost mechanical, cf. enim vero) phenomenon. The combinations nam edepol, nam herde and nam eeastor are quite frequent in Latin comedy. Examples are PI. Capt. 464; 604; 896; Cist. 69; 662; Men. 1137; Mil. 1228; Poen. 328; Ps. 970; 1248; Trin. 731; 747; Ter. Ad. 190; Eu. 487. Outside comedy these combinations occur only in a few dialogical passages in Cicero: de Orat. 2.144; Ver. 1.133; 2.72; Fam. 5.10a.1.

188

§ 8.3

8.3.3 evidence from question-answer pairs A third distributional property of enim that may provide an indication of its specific interactional consensus function comes, again, from its use in Latin comedy. It has been observed already in 8.2 above that in Plautus and Terence enim-units are often answers to preceding questions. A closer inspection of the material now reveals that we are dealing here not simply with neutral questionanswer pairs. Rather, it is regularly indicated that the relevance of the question preceding the enim-unit, or the sincerity of the questioner in putting this question, is being queried, the answer (containing enim) being regarded as obvious. Consider the following examples: (32)

Ep: Quam ad rem istuc refert? Pe: Ragas? Ut enim praestines argento, priu'quam veniat filius (,What is the good of that? # You ask (ragas)? Why (enim) , so that before your son comes you may have her bought and paid for', PI. Epid. 276-277)

(33)

Ag: Sicine ego te orares iussi? Mi: Quo modo ergo orem? Ag: Ragas? Sic enim diceres, sceleste: ... ('Was that the way I told you to entreat her? # How should I, then? # You ask? (ragas ?). This was, of course (enim) , the way to say it, villain', PI. Paen. 386387)

(12)

Tr: Ego interim hanc aram occupabo Th: Quid ita? Tr: Nullam rem sapis. Ne enim illi huc confugere possint ('Meanwhile, I'll just occupy this altar # Why so? # You have no sense, sir. Why (enim), lest the others take refuge at it', PI. Mos. 1094-1095)

(34) Me: Sed quid ais? Pe: Egone? Id enim quod tu vis, id aio atque id nego (,But what do you say? # I? Why (enim), whatever you want, that's what I say and unsay', PI. Men. 162) (35)

Ae: Quid ipsae? Quid aiunt? Mi: Quid Was censes? Nil enim (,And the ladies? What do they say? # What do you expect them to say? Some nonsense, of course (enim)', Ter. Ad. 656)

interactional nature of enim

189

8.3.4 illocutionary force A fourth clue to the proposed discourse function of enim comes from the distribution of enim-units among the various illocution types (see section 5.3). The important observation in this respect is that enim-units, although they may occur in all sentence-types, are confined largely to utterances with an assertive illocutionary force, which provide an environment most congenial to a consensus-stressing particle. It is not surprising that in most cases the sentence type of a clause containing enim is declarative. A more interesting observation is that enim is frequent also in interrogative clauses that count not as actual questions but as rhetorical or ironic ones. Since such interrogative clauses are effectively assertions rather than questions, the instances of enim concerned are not inconsistent with (and rather corroborate) the description of enim as a consensus particle which emphasizes the common knowledge or view of speaker and addressee2o • What is more, the use of irony and rhetorical questions is a typical trait of a dialogical or diaphonic discourse type, which, as we have seen in 8.3.1, appears to enhance the use of enim. Examples (36) and (37) illustrate the use of enim in rhetorical and ironic questions: (36) (Alexander to his soldiers) Subest nimirum altius malum quod omnes avertit a me. Quando enim regem universus deseruit exercitus? Ne servi quidem uno grege profugiunt dominos ('Undoubtedly there is some deeper evil which turns you all from me. For (enim) when has a whole army abandoned its king? Not even slaves run away from their masters in a single body', Curt. 10.2.20) (37)

"Tu me" inquis "mones? lam enim te ipse monuisti, iam correxisti? Ideo aliorum emendationi vacas?" Non sum tam improbus ut curationes aeger obeam, sed tamquam in eodem valitudinario iaceam ('''What?'' say you "Are you giving me advice? You have indeed (enim) advised yourself already, already corrected your own faults? Is this the reason why you have leisure to reform other men?" No, I am not so shameless as to undertake to cure my fellow men when I am ill myself', Sen. Ep. 27.1)21

20. For the difference between sentence type and illocution in the context of Latin cf. e.g. Pinkster (1990) and Risselada (1993). See also my remarks in 5.3. 21. Other instances in Seneca of enim in ironic questions can be found in Ep. 77.17-18, which contains four instances of this use of enim, and in de Ira 3.27.3. Cf. Marouzeau (1948: 249): "On trouve encore chez Seneque enim employe pour souligner la valeur ironique d'une interrogation". A possible example of an ironic enim-question outside Seneca is Cic. Phil. 10.15. Cf. also Liv. 3.39.9. It is noticeable that ironic questions lack the formal properties that are typical of the Latin interrogative sentence type, i.e. a wh-word or interrogative particle: the clauses are ambiguous in

190

§ 8.3

In view of my definition of enim as a consensus particle one might well expect that enim will not be very compatible with a directive illocutionary force, that is, with utterances by means of which the speaker attempts to get the addressee to do something. Interestingly enough, such a hypothesis turns out to be untenable. Although relatively rare, compared with (declarative and interrogative) assertions, the use of enim in directives is not totally ruled out: in fact I gathered 31 instances. However, I will argue here that the specific type of directives with which enim occurs corroborates the view of enim presented in this chapter, rather than conflicting with it. On this point it is crucial, first, to realize that there are various types of directives, such as orders, requests, supplications, prayers, advice, suggestions and proposals, which differ each from the others in a number of respects. Following Risselada (1993) I assume that these subdistinctions correspond to a number of classificatory criteria, the most important of which appear to be the 'benefit' criterion (in whose interest is the content of the directive?: in that of the speaker, of the addressee, or of both?), and the 'bindingness/optionality' criterion (to what extent is the addressee bound to performing the action referred to in the directive utterance?). Categorical orders (Give me the money.'), for instance, are strictly to the benefit of the speaker, and count as maximally binding for the addressee. As such they are diametrically opposed to, for example, suggestions (wouldn't it be a good idea to deposit your money with me ?), which primarily benefit the addressee and leave the addressee the option of not complying. Now it may be predicted that consensus-particles like enim, although incompatible with e.g. orders and requests, might not be totally excluded with directives that are to the benefit of the addressee and have a relatively low degree of bindingness. Hence suggestions and advice (both to the benefit of the addressee), as well as proposals (to the benefit of both speaker and addressee) might in theory be possible candidates to host a consensus-particle: in such cases the speaker might appeal, by adding enim, to the common sense of the addressee, thus creating an atmosphere of general understanding of the usefulness and appropriateness of the action to be performed. Such an atmosphere of cooperation called forth by enim would obviously lead to a paradox in the case of strictly-binding and speaker-benefitting directives, which are essentially one-way communications.

form (interrogative/decJarative)as well as in illocutionary effect (question/assertion).

interactional nature of enim

191

The prediction formulated above is indeed in accordance with what we actually find. There are no examples of categorical 'orders' or 'requests' with enim in my material, whereas I found five 'proposals' and five 'pieces of advice'. A much discussed example of enim in a piece of advice comes from Vergil's Georgica, a didactic poem on agriculture22 : (38) (concerning cattle-breeding) Semper erunt quarum mutari corpora malis:/ semper enim refice ac, ne post amissa requiras,/ ante veni ... ('Ever will there be some kine whose mould you would wish to change; ever, indeed (enim), renew them and, lest too late you regret your losses, keep in advance .. .', Verg. G. 3.69-71)

I will come back to this example in 8.4.1 below. My contention that enim in its function as a consensus-particle is compatible with the illocutionary force of a piece of advice (or its negative counterpart 'warning') is sustained by the existence of at least four parallel cases, all from Latin comedy23. An example of enim in a proposal is (39?4. The specific function of enim here (stressing the cooperativeness of speaker and addressee in a joined action) is hard to render in the English translation: (39) (Cicero is discussing the Peripatetic notion of 'limit') Etenim quis erit tandem modus iste? Quaeramus enim modum aegritudinis, in qua operae plurimum ponitur (,For what, I ask, will the suggested 'limit' be? Let us inquire (quaeramus; enim hard to translate) into the limit of distress to which they devote most attention', Cic. Tusc. 4.40)

A second interesting observation with regard to the use of enim in directive utterances is that almost all of the other instances in my material (18 out of the

22. Ever since Servius (5th century) commentators on Vergil have racked their brains over this obviously non-causal instance of enim. Whereas Servius remarks ad lac. that enim has no value at all, Page paraphrases with the deontic assertion semper enim refieienda [corpora matrumJ, explicitly rejecting Pierius' comment that enim in this passage equals the consecutive connective itaque ('and so'). A consecutive value is assumed also by Hand (1829-1845, s.v. enim, p. 381; enim = igitur); cf. also Norden ad A. 6.28 ('und so .. .'). Szantyr (p. 508) speaks, with respect to this example, of a "zwischen Versicherung und Begrilndung schillemde Bedeutung", and qualifies the use of enim here as an archaism. 23. Pl. Ep. 94; Mil. 293; St. 616; Ter. Eu. 751. 24. Other examples of enim in a proposal are: Cic. Tuse. 5.68 (Sumatur enim nobis quidam praestans vir); Div. 2.76 (externa enim auguria ... videamus); Rep. 2.4 (eoneedamus enim famae hominum); N.D. 2.32 (audiamus enim Platonem). I do not agree with the comment of DouganHenry ad Cic. Tuse. 4.40 (cited as ex. 41), which ascribes to enim the function of introducing an example.

§ 8.3

192

total of 31) can be characterized as metadirectives25 • Metadirectives are not just another subtype of directives, but are introduced by Risselada (1993) as a separate category in order to capture the fact that directives (more or less independently of their specific subtype) are often used to express the immediate communicative actions required of the addressee26 • A few explanatory remarks are in order here. By performing a speech act the speaker inherently puts a claim on the addressee. For instance, in the case of a question the inherent interactional claim on the addressee is to provide an answer, while in the case of an assertion the addressee is required to believe (or at least to take into consideration) the transmitted content of the speech act. These inherent interactional claims may remain implicit, but can also be expressed explicitly in the form of a directive, as in the English expressions tell me, believe me, and the like. Similarly, the speaker may explicitly direct the addressee to retrieve knowledge that is already stored in his knowledge base (remember that ... ), or to perform the 'instrumental' communicative actions of listening and paying attention (listen!; now pay attention to this point:). Note that metadirectives are, in a sense, the addressee-oriented (or 'perlocutionary') counterparts of the speaker-oriented (,illocutionary') performative expressions, such as I tell you that ... ; I promise you that ... ; I ask you to .... My material contains the following examples of enim in metadirectives: sic enim tibi ... persuade 'be convinced that' enim, mihi crede 'believe me' puta enim 'suppose that' cave enim putes 'beware of thinking that'

Cic. Fam. 3.1.1 Cic. Fam. 4.10.1 Sen. Ep. 94.23; 113.19 Cic. Leg. 2.7; Tusc. 5.19

25. The three remaining 'directive' instances of enim in my material are all on the border between directives and assertions: enim non ibis nunc (PI. Per. 236: 'you shan't go now, you can be sure of that (enim)); non enim ibis (PI. Mos. 1133); non enim tu hie quidem occupabis omnis quaestus quos voles (PI. Rud. 989: 'you shan't, you can be sure (enim), monopolize all the trades you want in these parts'). The examples are all strictly speaking assertions in which the speaker refers to a future event, the irrevocable occurrence of which is presented as obvious for both speaker and addressee ('you won't go, and you know it as well as I do'). As it involves a future behaviour of the addressee, however, it may be interpreted ultimately as a directive. 26. Risselada (1993: 258): "Metadirectives are expressions by means of which the speaker explicitly mentions the perlocutionary effects that are systematically connected with the speech act that is being performed and 'directs' the addressee to realize these effects."

interactional nature of enim nolite enim idl ita putare 'do not think. that' scito enim 'know that' fac enim 'imagine that', 'suppose that' da [enim] 'imagine that', 'suppose that' attendel attendite enim 'pay attention to the fact that ... ' accipite enim 'now hear about the fact that .. .' id enim tenetote 'you must keep in mind, that ... facite enim ut ... recordemini 'do bring back to memory that .. .' repete enim exemplum quod paulo ante .. .' 'repeat the earlier mentioned example that ... '

193 Cic. Har. 62; Pis. 46 Plin. Nat. praef. 22 Cic. Ver. 3.160 Cic. Sen. 69 Cic. Phil. 2.31; Ver. 2.82; 3.37 Cic. Sen. 39 Cic. Rep. 2.57 Cic. Clu. 111 Sen. Nat. 1.2.7

A few comments may be made with regard to the expressions listed above. First, it seems rather that enim affects the embedded assertion than that it pertains to the metadirective main predicate. This would imply that there is little point here in raising the 'compatibility' issue of enim and directive host unit, since the actual host unit might in effect be assertive rather than directive. Thus for instance in (40) I would argue that enim somehow has a bearing on the (content of the) embedded clause carissimum te mihi esse (,that you are most dear to me') rather than that it affects tibi persuade ('be convinced that'): (40) sic enim tibi persuade, carissimum te mihi esse cum propter ... tum ... ('you must convince yourself that you are very dear to me, both on account of ... , and also because .. .', Cic. Fam. 3.1.1)

In the above example enim and tibi persuade may even be said to be functionally comparable expressions, to the extent that they both relate the actual content that is transmitted (carissimum te mihi esse ... ) to the interactional framework in which it is integrated (i.e. they both playa role in the management of the interaction). More specifically, enim and tibi persuade both have the function of soliciting the addressee's acknowledgement of the communicated message. On this point, it may be noted that the use of enim with metadirectives is another clear indication of enim's preference for a dialogical or diaphonic discourse type, that is, its preference for a context in which the

194

§ 8.3

speaker/writer's awareness of his audience is made relatively explicit27 • Compare in this light the difference in communicative effect between the - in point of content material - equivalent messages 'you are most dear to me' and 'you are, y'know (enim), most dear to me, be sure of that (tibi persuade)'. A second remark involves the fact that the discussion about enim in combination with metadirectives applies also mutatis mutandis to nam. Compare the following example28 : (41) (At the introduction of a new topic in the discussion) Nam, quaeso, redite in memoriam, iudices, quae libido istius in iure dicendo fuerit ('Now (nam) recall, I pray (quaeso) , to your memories, gentlemen, the wanton character of Verres' administration of the law', Cic. Ver. 1.120)

On account of its function of signalling subsidiary (often explanatory) discourse units it may be predicted that nam is incompatible with a directive host utterance. This indeed turns out to be the case, the only counterexamples involving metadirectives. Note, incidentally, that nam in the passage quoted above occurs in a context with a high degree of 'interactiveness', in which, moreover, an explicit appeal is made to shared knowledge of speaker and audience ('bring to your memories'). Although such an environment is much more typical of enim than of nam there are good, independent grounds for the use of nam in this example29 • On this point I refer to chapter 7, section 7.3 ('forward-linking' use of nam). In conclusion to this section we may say that the distribution of enim among the various illocution types is quite uneven. Whereas the particle is most common in declarative sentences with the illocutionary function of an assertion, it does not occur in questions (except in rhetorical and ironic questions, which are inherently assertive), and only in a highly specific group of directives (viz. in directives that are in the interest of the addressee, and in metadirectives). On the one hand this particular distribution supplies additional support for the more specific discourse function of enim proposed in this chapter. On the other it shows that the description of enim as a mere reinforc-

27. See the overview of dialogical and diaphonic elements in 5.2.2.2. 28. Comparable instances of nam in metadirectives are: PI. St. 519 (nam facito ut cogites); Mi!. 915 (nam hoc cogitato); Capt. 711 (nam cogitato); Cic. At!. 13.14.1 (nam scito); Agr. 3.11 (nam attendite); Ver. 3.218 (nam videte et prospicite); Leg. 1.16 (nam sic habetote); Leg. 3.12 (idem); Fam. 4.4.3 (nam sic fac existimes). 29. A comparable example of nam in an environment that is at first sight more typical for the use of enim is Cic. Cati!. 3.19 (Nam profecto memoria tenetis ..,). The use of nam here can again be explained by the fact that the sentence functions as a transition to a new narrative episode.

interactional nature of enim

195

ing (illocution-modifying?) particle (see section 6.3) is inadequate. Such a description would for instance wrongly predict the occurrence of enim in urgent orders and requests 30 • The 'reinforcing' value of enim is better taken as a side-effect (see section 5.1 and below, p. 202). 8.3.5 meta-communicative comment

In 8.2.2 above it has already been pointed out that enim is prone to occur in passages that have the function of a metacommunicative comment. Such passages are characterized by the occurrence of metadirectives (see above), performatives and other expressions pertaining to the interactional layer of the discourse rather than to the layer of the transmitted content. First and second person pronouns and verb forms 3!, as well as subjective expressions like arbitror, opinor, credo, puto, mihi videor (,I mean', 'I think', 'I believe', and the like) are not uncommon in the passages concerned; in addition, presenttense verb forms outnumber the historic tenses. In predominantly narrative or expository texts metacommunicative passages often have the status of a parenthesis, which can be regarded as a formal reflection of their belonging to a different (i.e. interactional rather than ideational) layer of the text. The following examples illustrate the widespread use of enim in metacommunicative passages32 • For lack of adequate English equivalents I leave enim untranslated or use a paraphrase. The function of enim in these examples can in general be characterized as 'agreement soliciting': (42) Dicam enim vere ... quod sentio: ... ('I will tell you truthfully... what 1 think: .. .', Cic. de Drat. 2.365) (43) - puto enim etiam atque etiam mihi dicendum esse ('- it has to be stated, 1 think (puto) , you will agree (enim) , again and again -', Cic. Fam. 13.28.1) (3)

Quando enim me in hunc locum deduxit oratio, docebo ... ('As (quando) my discourse has led me to this topic, 1 will show that .. .', Cic. N.D. 3.43)

30. Cf. e.g. Hand (s. v. enim, p. 392), whose description is too general: "asseverationem particula addit imperativo et futuro, qui pro imperativo ponitur" ('the particle adds a reinforcement to an imperative or a future that is used as an imperative'). 31. In Cicero there are e.g. 16 instances of the combination tu/ego enim, whereas there are only 3 attestations of nam in combination with tu/ego. Note that in general enim is only 2.5 times more frequent than nam in Cicero. 32. A few other examples are PI. Cas. 372 (dicam enim); Mil. 810 (ego enim dicam); Mos. 888 (ego enim dicam); Cic. Aft. 1.18.6; 3.9.1; 3.15.4; Amic. 77; 85; Sen. Ep. 30.8.

196

§ 8.3

8.4 Actual use: pragmatic motivations for the use of enim At this point in the discussion the description of the particle enim may be extended further by raising the issue of the specific pragmatic motivations underlying its use. There are at least three pragmatic factors that may trigger the interactional strategy of ascertaining the consensus and commitment of the addressee. In other words, the discourse function of interpersonal consensus may give rise to at least three different actual uses (see ch. 5, section 5.1): (i) toning down a challengeable utterance; (ii) irony; and (iii) empathy.

8.4.1 actual use 1: toning down a challengeable utterance A first motivation for the use of a consensus-particle appears to be the presence of an utterance with a high degree of challengeability, that is, an utterance that might potentially fail to be accepted or understood. Typical examples of challengeable utterances are polemic or controversial statements, and utterances with a highly surprising content. The appeal to consensus or common ground in such utterances can be explained as an attempt by the speaker to enhance his credibility and forestall any negative reactions: the validity or truth of the utterance in question is not based entirely on the authority of the speaker/author, but is presented as more generally accepted. The challengeability factor provides an explanation for the frequent combination of the particle enim with at (see p. 173 and p. 177 above). In chapter 12 I will describe at as a marker of challenging reactive moves (such as protests and objections), which by definition have a high degree of challengeability33. An example of the combination is (44): (44) father:

Quotiens monstravi tibi, viro ut morem geras, quid ille faciat, ne id observes, quo eat, quid rerum gerat daughter: At enim ille hinc amat meretricem ex proxumo ('How many times have I explicitly told you to humour your husband and not keep watching what he does, where he goes, and what he is about? # But (at) he makes love to this strumpet next door, doesn't he? (enim)!', PI. Men. 787-790)

Moreover, the challengeability factor appears to account for one more remarkable distributional property of enim that I have left unmentioned so far, namely

33. The less frequent combination of enim with immo ('no rather') can be explained in the same way.

pragmatic motivations for the use of enim

197

the extremely high frequency of the combination of enim with the preposed negative particle non ('not'), or the negative connectives nec and neque ('and not')34. In fact such utterances are challengeable in the same way as protests and objections, since they commonly involve the rejection of a presupposition contained in the context or situation, or of a generally held opinion. They might thus be called 'reactive' in a diaphonic sense. An example of non enim is (45): (45) non enim pudendo, sed non faciendo id, quod non decet, impudentiae nomen effugere debemus (,not, of course (enim), by feeling shame at what is unbecoming, but in not doing it, we must escape the reproach of shamelessness', Cic. de Orat. 1.120)

In support of the proposed explanation for the use of enim in negative clauses we may compare the following two dialogical passages, in which non enim is used at the start of a protesting reactive move. The example cited in (46) is an undiluted dialogical passage, whereas (47) is on the borderline between a dialogical and a diaphonic discourse type: (46) An:

Cedo nunc porro: Phormio dotem si accipiet, uxor ducendast domurn: quid fret? Ge: Non enim ducet ('Tell me this now about the future: If Phormio gets the dowry, he must marry the wife: what then? # Why (enim), he won't marry her', Ter. Ph. 692-694)

(47) marmora invehi ... quae vetaret lex nulla lata est. Dicat fortassis aliquis: "non enim invehebantur " ('no law was ever passed forbidding us to import marble ... Perhaps someone might say: "Of course (enim) not, no marbles were being imported', Plin. Nat. 36.4.2)

As to the challengeability factor it may be noted fmally that consensus-particles may be used to 'counterbalance' the challengeability not only of the host unit itself, but also the challengeability of a preceding utterance: a potentially challengeable utterance will often be counterbalanced (or: supported) by an utterance with an obviously irrefutable character (for instance an utterance with a consensus particle). This is in accordance with Bolkestein's observation (1989: 444-6) that in Seneca's Naturales Quaestiones enim-clauses (in contrast 34. In the first ten books of Cicero's Letters to Atticus, for instance, about 10% of the sentences containing enim are introduced by the sentence negation particles non, nee or neque. Clauses with nam, in contrast, involve less often a sentence negation: in only 2% of cases does nam occur together with a sentence-negating non within the same clause (on account of its first position in the clause nam is excluded from combining with clause-connecting nee and neque).

198

§8.4

to nam-clauses) tend to follow utterances with a significantly strong (i.e. highly surprising or controversial) focus (see 6.4, p. 143). It is especially in such situations that enim seems to indicate a 'causal' relation of the type commonly signalled by nam. Compare (48), in which the potentially challengeable utterance quapropter ... explanabit is 'toned down' by a justificatory clause containing a common sense argument: (48) Quapropter ille tibi omnia explanabit; id enim mihi et ad brevitatem est aptius et ad reliquas res providentius ('That's why he [i.e. instead of me, what you might expect or hope for, CK] will make everything plain to you. For that, you will agree (enim), helps me to make my letter shorter, and is less risky in view of other matters ... ', Cic. Fam. 3.1)

With regard to instances such as (48) I would contend that, although the clause hosting enim has unarguably the rhetorical function of a justification, the function of enim itself is not (or, at any rate, not exclusively) to be described in terms of the explicit signalling of this relationship. Such a view would not do justice to the highly specific conditions for the use of enim recounted in the preceding sections, which appear to differ considerably from the conditions of use of nam. The marking of 'causal' relations (whether representational or presentational) is presumably only a very persistent side-effect (see section 5.1) of enim's proper discourse function, which I described as signalling interpersonal consensus: consensus particles (such as enim is argued to be) are highly compatible with subsidiary (i.e. justificatory, explanatory, etc.) text parts, but do not necessarily signal them. I will come back to this point in 8.5.1.

8.4.2 actual use 2: irony By using enim the speaker/author appeals to common knowledge or a shared view in order to get the addressee to consent to a certain utterance. This appeal to common ground may be unwarranted in the sense that the speaker may make it appear as if the view were already shared, when, in fact, it is not. This strategy is especially opportune in utterances with a highly challengeable content, as it may lend the utterance a higher degree of positiveness and hence, a greater acceptability. However, the appeal to common ground may be not only unwarranted, but even downright insincere. In such cases we can speak of an ironic use of enim. In an ironic enim-unit the speaker/author echoes a consensual truth which is patently false: it appears from the context or situation that the speaker actually

pragmatic rrwtivations for the use of enim

199

believes the opposite of the expressed view35 • In 8.3.4 I discussed the use of enim in ironic questions. Another example of an ironic enim-unit is (49)36. In the translation the irony of the statement is rendered by means of 'of course'. In the Latin original the ironic flavour is dependent mainly on the use of enim: (49) (One of Petronius' fellow-freedmen rebukes Ascyltos for laughing too loud) "Quid rides" inquit "vervex? An tibi non placent lautitiae domini mei? Tu enim beatior es et convivare melius soles ... " ('''What are you laughing at, mutton head?" he said, "Are your host's good things not good enough for you? Of course (enim) , you are richer and used to better living ... "', Petr. 57.2)

Note, incidentally, that in the above example enim cannot be replaced by its alleged synonym nam: the use of nam, which lacks the 'conversational' nuance typical of enim, would have lent the clause the function of genuine rather than ironic support37 • 8.4.3 actual use 3: empathy

A number of further pragmatic motivations for the use of enim can be grouped together under the rubric of empathy, by which I mean, in the context of this chapter, the addressee's identification with the speaker in a linguistic event, or his identification with a character in a narrated event38 • One typical instance of an appeal to the involvement of the addressee in the event is the dramatic technique of the aside in Latin comedy. Asides have the function of involving the audience directly in the plot, for instance by imparting to the audience something that the characters of the play do not know or are not allowed to know. The use of the consensus particle enim underscores the atmosphere of confidentiality. An example of enim in an aside is: (50) Nunc enim vero ego occidi ('Now it is all over with me', PI. Capt. 534)

35. For an account of the concept of irony in tenns of echoic utterances cf. Sperber and Wilson (1986: 237-43). 36. The ironic use of enim is explicitly mentioned by both Hand (1829-1845, s. v. enim, p. 391) and KUhner-Stegmann (p. 125), who give more examples of the phenomenon and also refer to the combination of enim and ironic videlicet in Cicero. For this combination see also note 17 above. 37. Cf. also Kroon (1992: 58-9) with reference to Liv. 31.7.2. 38. Note that this view is an extension of Kuno's concept of empathy, which he defines as "the speaker's identification, in varying degree, with a participant in an event" (1976: 431).

200

§8.4

More or less comparable is the appeal to the involvement of the addressee in exclamatory utterances, in which case the addressee is invited, by means of enim, to sympathize with the speaker's anger, surprise, despair, etcetera; an example is Sen. Ep. 10.5, cited in chapter 6 as example (12). Compare also: (51) Quae enim res umquam, pro sancte Juppiter! non modo in hac urbe, sed in omnibus terris est gesta maior (,For what thing, holy Jupiter! ever done, not in this city only but in all the world, was greater?', Cic. Phil. 2.32)

Appeals to empathy are especially in place in the atmosphere of intimacy that is characteristic of the personal letter. This explains the remarkably high frequency of enim in Seneca's letters to Lucilius and Cicero's letters to Atticus. In Cicero's letters to his intimate friend Atticus, for instance, the frequency of enim and nam is in the proportion of 5: 1, whereas in the Ciceronian correspondence as a whole (i.e. including the letters to family, colleagues and acquaintances) the ratio is only 2.5 : 1. In 8.2.2 above (with regard to example 21), I already hinted at yet another form of empathy, which is typically employed in narrative texts: the identification or sympathy of the speaker/author and the addressee with the characters of the narration (or at least the suggestion that speaker and addressee are present at the scene of the narrated action). Another instance of these 'vivid eyewitness accounts' in which enim is used is example (4) cited in 8.2.1. Compare also (52) and (53) below. Vivid eyewitness accounts are characterized by, among other things, the use of historic infinitives or historic presents (see also section 11.4.3 on the use of the equally interactional particle vera in narrative texts). (52) Accidit etiam repentinum incommodum biduo quo haec gesta sunt. Tanta enim tempestas cooritur ut numquam illis locis maiores aquas fuisse constaret (,There also happened an unforeseen disaster within two days of these occurrences. Imagine (enim), a storm of such intensity springs up that it was agreed that there had never been a greater rainfall in that district', Caes. Civ. 1.48) (53) Refertur eius sermo ad Apronium. Enim vero iste (sc. Apronius) ridere ac mirari Lollium nihil de Matrinio, nihil de ceteris rebus audisse (,His words were reported to Apronius. Apronius laughed, of course (enim), and wondered that Lollius had not heard about Matrinius and all the other affairs', Cic. VeT. 3.61)

The observation that the consensus particle enim is prone to occur in lively narrative passages may help to explain a few notoriously difficult (i.e. noncausal) instances of the particle, such as in (54) below. Enim in (54) has

pragmatic motivations for the use of enim

201

unjustly in my opinion been qualified in the Oxford Latin Dictionary and the Thesaurus Linguae Latinae as an instance of enim adversativum, marking a slight contrast and therefore being more or less comparable with autem. Compare also De la Harpe (1923: 85), who speaks with regard to this example of a mixture of affirmative and adversative features ("enim oppose ego a simia; ego autem serait parfaitement acceptable ")39 . I contend that there is no support for the assumption of an 'adversative' function of enim, at least not as far as early and classical Latin (until the second century) are concerned. In the passages concerned features of the context are obviously mistaken for the proper function of enim. (54) Interea ad me haedus visust adgredirier, infit mihi praedicare sese ab simia capram abduxisse et coepit inridere me; ego enim lugere atque abductam illam aegre pati (,Meanwhile a kid, so it seemed, came up to me with the announcement that he had taken that she-goat away from the monkey, and began to laugh at me; I - as you can imagine (enim) - was weeping and wailing at her abduction .. .', PI. Mer. 248-251)

The passage cited in (55) contains another instance of the use of enim in a vivid eyewitness account that has proved difficult to account for in terms of causal connection. Norden (1916: 129-30) and Austin (1977, ad loc) for instance speak of an 'asseverative' use of enim. 'Soliciting the empathy of the audience' is, in my opinion, a more adequate description of the function of enim in this rather emotional passage40 : (55)

Navita sed tristis nunc hos nunc accipit illos,! ast alios longe submotos arcet harena / Aeneas miratus enim motusque tumultu,l "dic" ait "0 virgo, quid vult concursus ad amnem?" ('But the surly boatman takes now these, now those, while others he thrusts apart, back from the brink. Then aroused and amazed by the disorder (enim: 'as you can imagine'), Aeneas cries: "Tell me, 0 maiden, what means the crowding to the river?" " Verg. A. 6.315-318)

39. The comment by Enk ad lac. offers a different solution: "enim hoc loco profecto, nempe significat". This view is apparently shared by Clement (1897), who ranks this example under the 'asseverative' use of enim. 40. A comparable example is A. 10.874.

202

§ 8.5

8.5 Conclusions In conclusion to this chapter we can say that there is ample linguistic support for the view that in early as well as in classical Latin enim is not a connective in a narrow sense, but should rather be described as a consensus particle which operates primarily on the interactional level of discourse. In this function it can be considered a near-equivalent of for instance German ja, English y'know, Italian sai, Japanese ne and Cantonese La in some of their uses 41 • I repeat from section 5.1 (p. 102), with a few additions, the full discourse-pragmatic description of enim: stepwise analysis of the particle enim

1. basic meaning: consensus 2. discourse function: appeal to interpersonal consensus (i) toning down challengeability 3. actual use: (ii) irony (iii) soliciting empathy (i) causality 4. side-effects: (ii) reinforcement (iii) focalization (iv) weak adversativity (v) result The account of enim in the present study might be attractive for at least two reasons. First, it provides a unifying concept (,interactional consensus') for a motley collection of different and at first sight unconnected uses. And second, it enables us to distinguish more clearly between the alleged near-synonyms enim and nam. Both points will be worked out in more detail in the next two subsections.

41. For German ja cf. e.g. Franck (1980), Hentschel (1986), Doherty (1987), Heringer (1988), Foolen (1989), Lindner (1991). Extensive treatments of y'know are given by Ostmann (1981), Schourup (1985) and Schiffrin (1987). Bazzanella (1990) discusses Italian (come) sai. Japanese ne is described by Suzuki (1990: 317-8) as seeking the listener's agreement with what the speaker is saying, or the listener's agreement on the speaker's holding the floor. For Cantonese la cf. Luke (1989).

conclusions

203

8.5.1 interactional consensus as a unifying concept As has become clear in this chapter consensus-particles like enim may occur in a range of different contexts, thus giving rise to a variety of uses and interpretations. By far the most pervasive is the 'causal' interpretation, which in the literature is accordingly taken as the central characteristic of the particle (see ch. 6, section 6.3). There are clear advantages, however, to regarding the putative causal connective function of enim as a side-effect of the particle's proper discourse function: whereas all instances of enim in early as well as in classical Latin can be taken equally well (or even better) as consensus particles which encourage the interactional involvement of the addressee, a description in terms of a causal connective is less universally applicable. On the basis of a systematic investigation of the syntactic and discourse distributional properties of enim it seems reasonable to acknowledge an interactional 'added value' even in those cases where enim seems to behave as a marker of sequential (causal) discourse relationships (see e.g. ex. 50). An important observation in this respect is that essentially monological texts (which make up the bulk of classical Latin literature) may to varying degrees be diaphonic, i.e. containing the voices of a speaker and an addressee. The putative connective function of enim in classical Latin can be ascribed to the high degree of compatibility of consensus markers with subsidiary discourse units: enim-units often function as common-ground arguments that support a challengeable central unit"2. Likewise the reinforcement of the illocutionary force of an utterance, or the signalling of a significant focus element in the clause, should be regarded as contextual side-effects of the use of consensus-marking enim, rather than as independent (additional) functions of the particle43 • Illustrative examples are (56) and (57) below, which are commonly adduced as instances of the 'affirmative', 'corroborative' or 'intensifying' sentence particle enim. In (56) enim rather signals, in my opinion, an appeal to the addressee to put himself in the position of a character in the narrated story (empathy, see 8.3.3; note the use of the historic present adpropero), which may have the inherent effect of a reinforcement of the illocutionary force of the utterance:

42. Less disputed consensus markers such as English y'know and Germanja have also been said to occur typically in the supportive parts of arguments. Cf. e.g. Schiffrin (1987: 279-81) and Foolen (1989: 314). An illustrative example ofja in this respect is: Kein Wunder daft er Unsinn redet. Der hat ja schon die halbe Whisky-jlasche geleert (Franck 1980: 232). 43. Alleged examples of enim as a 'focus particle' can be found e.g. in Hand (1829-1845, s.v. enim, p. 392f), and the OLD (s. v. enim, 3: "(enim) app. used to emphasize the preceding word").

204

§ 8.5

(56) (The slave Olympio gives an account of how he tried in vain to make love to Casina) ... principio earn savium posco, reppulit mi manum, neque enim dare sibi savium me sivit. Enim iam magis adpropero, magi' iam libet in Casinam inruere ... (' ... I ask her for a nice long kiss. She pushed my arm away, not a bit of a nice long kiss would she let me give her. Now I get, you can imagine (enim) , more urgent; now I'm more eager to have my Cas ina .. .', PI. Cas. 887-889)

Likewise in (24) (see also p. 183) I would argue that the putative focalizing function of enim arises directly from the consensus-marking function of the particle, which is in place because of the unexpectedness of the content (challengeability factor)44. Besides, enim seems to signal also an appeal to the addressee luno to sympathize with the narrated character Aeneas (empathy factor). In my view, calling enim here a corroborative focus particle does not do full justice to the dramatic impact of the passage: (24) Ecce autem subitum atque oculis mirabile monstrum .. ./ procubuit viridique in litore conspicitur sus! quam pius Aeneas tibi enim, tibi, maxima luno,! mactat. sacra ferens et cum grege sistit ad aram (Verg. A. 8.81-85)

Furthermore it has been pointed out several times in the literature that enim, in addition to its causal and affirmative use, may occasionally be employed as a slightly adversative or transitional particle, comparable to Latin autem. The examples adduced, however, all turn out to be rather straightforward instances of the interactional consensus function of enim45 • Nor is there any need to assume (as do e.g. Hand, s. v. enim, p. 381) and the TLL (s. v. enim, p. 573, 18ff), that enim is sometimes equivalent to the consecutive connectives igitur and ergo ('so'). Compare the earlier cited example (39), repeated here for convenience 46 : (39) (Concerning cattle-breeding) Semper erunt quarum mutari corpora malis:! semper enim refice ac, ne post amissa requiras,! ante veni ...

44. The goddess luno has been Aeneas' most formidable adversary. Nonetheless, on the advice of the prophet Helenus (A. 3.435ft) Aeneas now makes a sacrifice to her first of all. Tibi ('to you') is admittedly the most surprising (focal) element in the clause. 45. See the overview of the literature in 6.3. Examples of the alleged adversative/transitional enim are PI. Mer. 251; Ter. Ph. 113; Cic. N.D. 3.9; Plin. Nat. 14.15; 23.18. Cf. also De la Harpe's remark with regard to Cic. Aft. 1.16.1: "enim sert simplement a faire transition et a enchainer Ie n!cit" (1923: 85). With regard to the example at Petr. 56.6 (Apes enim ego divinas bestias puto), Szantyr (p. 508-509) speaks of an "ankniipfende Bedeutung von enim", while the OLD remarks that enim here introduces a slight contrast. Note, however, that ego and puto are diaphonic elements which tend to cluster with the consensus-particle enim. 46. Other alleged examples of enim in the role of consecutive igitur or ergo are PI. Aul. 500; St. 616.

conclusions

205

(,Ever will there be some kine whose mould you would wish to change: ever, indeed (enim) , renew them and, lest too late you regret your losses, keep in advance .. .', Verg. G. 3.69-71)

Conceiving of enim in the way advocated in the present study provides, in my view, a feasible alternative to the rather far-fetched ad hoc explanation that enim might sometimes signal a consecutive discourse relationship. The clause semper ... malis (,there will ever be cows whose mould you would wish to change') appears to echo the voice of the reader, and could hence be regarded as diaphonic. The ensuing enim-clause, couched in the form of a piece of advice, functions more or less as a consentient reaction to this voice. Enim may in this context be explained as suggesting that the usefulness of the recommended procedure is beyond dispute and in fact already agreed. 8.5.2 limitations on the interchangeability ofnam and enim It is arguably true that nam and enim occur partly in identical contexts and

positions in the discourse structure. Yet, as has gradually become clear in the course of chapters 7 and 8, there are also a considerable number of distributional differences to be accounted for, which show that ultimately nam and enim are essentially different pragmatic clues and can only rarely replace each other. The syntactic and discourse distributional properties of nam point in the direction of a connective particle concerned primarily with the presentation and organization of the information conveyed in the discourse (see chapter 7). More specifically, in its role as an introducer of subsidiary discourse units (background, explanations and other additional information-units) nam counts as a move-internal 'Dependency' device. Accordingly nam is tied up with a monological discourse type. Although the interactional elements which form an indication of a dialogical or diaphonic discourse type (see chapter 5) are not completely excluded from the immediate environment of nam, their occurrence is highly restricted and by no means systematic. Since it is itself a connective device, nam cannot be combined with other connective particles. For the same reason nam-units do not occur in isolation, and when they turn up after a change of speaker they involve a continuation - by a different speaker - of an ongoing move rather than a new independent (initiating or reactive) move. Moreover, nam-units are never ironic and not particularly 'challengeable'. In keeping with the particle's function as an introducer of subsidiary information

206

§ 8.5

the use of nam is largely confined to assertions; in contrast to enim, nam does not occur in proposals or advice, nor in exclamatory utterances 47 • The syntactic and discourse distributional properties of enim, on the other hand, point in the direction of a 'non-connective' (situating) particle concerned primarily with the management of the interaction, more specifically, with the involvement of the interlocutor in the speech situation. Accordingly enim is tied up with a dialogical or diaphonic discourse type. As shown above, enim is compatible with a large variety of contexts, only one of which is the subsidiary (explanatory, justificatory, etc.) discourse unit in a monological stretch of text48 • In view of the fact that the subsidiary discourse unit is the characteristic (and virtually exclusive) environment of nam, nam is potentially replaceable by enim, that is, the substitution of nam by enim will usually not result in an ill-formed stretch of discourse. The reverse (replacement of enim by nam) , however, will yield an acceptable result in only a limited number of cases. One of the examples in which enim seems at first sight to be interchangeable with nam is (57): (57) Habes res Romanas; non enim te puto de lustra, quod iam desperatum est, aut de iudiciis quae lege Clodia fiunt quaerere ('Well, there you have the news of Rome - I don't suppose (non puto; enim not translated) you are interested in the census, which has now been given up as a bad job, or the trials under the lex Clodia', Cic. Aft. 4.16.8, trans!' by Shackleton Bailey)

The passage cited, part of a letter to Atticus, is an instance of occupatio, a context in which nam is notably prone to occur49 • However, the use of enim instead of nam is motivated in several ways. The text part concerned is a transitional passage with a metacommunicative function; the addressee's involvement is explicitly expressed in the form of a second person pronoun (te). The preposed negation particle non and the subjective evaluation verb puto complete the picture of a typical enim-context. The same line of argument may be applied to the following pair of examples, in which nam and enim seem at first sight to have a fully identical

47. For the occurrence of nam in metadirective speech-acts, see p. 194. For the marginal use of nam in a highly specific type of question I refer to chapter 7, pp. 164-167. 48. Since enim does not function on the presentational level of discourse, its occurrence is not sensitive to the distinction between central and subsidiary discourse unit. Clear instances of enim in central discourse units are e.g. Cic. Ae. 2.52; Amie. 9. 49. See chapter 7, p. 157; cf. also Kiihner-Stegmann (p. 125-6) with regard to enim: "ganz selten in der sog. occupatio statt des gewohnlichen nam".

conclusions

207

function, viz. the introduction of an explanatory or justificatory parenthesis: (58) Habes legis prooemium; sic enim haec appellat Plato (,There you have the proem to the law; for that is, as you know, the name given to it by Plato', Cic. Leg. 2.16) (59) Fuit in iIIis nauarchis HeracIiensis quidam Furius, - nam habent iIIi non nulla huiusce modi Latina nomina, (,One of these captains was a certain Furius of Heraclia, - for those people have such Latin names as this, - " Cic. Ver. 5.112)

In (58) the speaker, by adding an elaborative parenthesis, anticipates a possible protest of the addressee that he does not understand the use of the Greek word prooemium. The insertion of enim might be seen as a subtle hint that the addressee is supposed to know his Plato. Note that the use of an interactional particle has already been prepared for by the preceding metacommunicative clause habes legis prooemium ('here you have the proem of the law'). In (59) the speaker elaborates in a parenthesis on the Roman name Furius, which in a description of Greek naval commanders might potentially cause surprise on the part of the audience. This time, however, there are no dialogical or diaphonic elements on account of which we might expect the occurrence of an interactional particle: the nam-parenthesis is part of a narrative in which the interactional frame of speaker and addressee is, so to speak, out of sight. Cicero's Letters, being a mixture of interactional and narrative/expository passages, lend themselves perfectly to further inquiry into the difference between nam and enim signalled by (58) and (59). A first investigation of the preferred environments of the two particles in this corpus shows that enim, as expected, prefers passages with a strong interactional character, so for instance those containing a metacommunicative comment. Nam-units, in contrast, belong usually to the 'reportive' or 'ideational' passages of the letters, and accordingly have no significant preference for interactional elements: namunits less often contain metacommunicative expressions, subjective evaluation verbs, and first and second person pronouns; moreover, historic tenses turn out to be more frequent in nam-units than in enim-units. Since enim-units regularly interrupt the current (narrative, argumentative or expository) line of the discourse, they often have the form of a parenthesis. Nam, as a matter of fact, also (though less often, cf. Kiihner-Stegmann, p. 122) occurs in parentheses, an observation which is in accordance with the function of nam as a marker of subsidiary discourse units. However, in contrast to the parenthetical enim-units, which may involve a switch to a

208

§ 8.5

different (i.e. interactional) layer, the nam-units remain within the current reportive (= narrative, argumentative or expository) layer. This can be illustrated with the following pair of examples: (60) Senatus vocatus in curiam. Pompeius domum. Neque ego tamen in senatum, ne ... in Pompeio defendendo (nam is carpebatur a Bibulo, Curione, Favonio, Servilio filio) animos bonorum virorum offenderem (,The Senate was called to the curia; Pompey went home. I did not myself, however, attend the Senate ... so as not, by defending Pompey (for he was being attacked by Bibulus, Curio, Favio and Servilius junior), to hurt the feelings of the loyalists', Cic. Q.jr. 2.3.2) (61) Tu vero sapienter (nunc demum enim rescribo iis litteris quas mihi misisti convento Antonio Tiburi) sapienter igitur quod manus dedisti quodque etiam ultro gratias egisti ('You really were wise (at last, you see (enim), am I answering the letter you sent me after meeting Antony at Tibur), well then, you were wise in giving in and even going so far as to thank him', Cic. Att. 16.3.1)

Compare also the earlier example (19), in which nam marks the transition to a new stage in the narrative, and enim occurs after a switch to the interactional layer of the text: (\9)

Nam tum cum ex urbe Catilinam eiciebam - non enim iam vereor huius verbi

invidiam - ... sed tum cum ... ('In as much at the very time I was trying to drive Catilina out of Rome - for, mark you (enim), I am not afraid now of the odium attaching to this expression ... then, I say, .. .', Cic. Catil. 3.2.3)

The 'added value' of interactionality that is typical of enim might, by extrapolation, be assumed also for the relatively few instances in which enim is the only indication of a diaphonic discourse type 50 •

50. On the basis of the accounts of nam en enim in chapters 7 and 8, Mieke Koenen (p.c.) investigated the use of these particles in the fourth book of Lucretius' De rerum natura. She concludes that the distinctive distributional properties of nam and enim discussed above can be clearly observed also in Lucretius. An additional distributional peculiarity of nam observed by Koenen in Lucretius is that nam is regularly found in combination with the verbs videre and videri, which may indicate that the information of the nam-unit is empirically based (in contrast to enim, which assumes shared or common, rather than empirically based knowledge). Furthermore, enim (in contrast to nam) is remarkably often found in highly poetical passages, as can be illustrated with Lucr. 4.189-190: suppeditatur enim confestim lumine lumen / et quasi protelo stimulatur fulgere fulgur ('in hot haste, you know, the place of light is taken by light, and as though driven in a team, one flash is goaded by another flash'). For the use of nam and enim in Lucretius cf. also Schrijvers (1970).

conclusions

209

Having established the difference in discourse function between nam and enim we may finally return to the only passage in Latin literature where both particles occur in the same clause: (62) Me: Yin commutemus, tuam ego ducam et tu meam? faxo haud tantillum dederis verborum mihi Ca: Namque enim tu, credo, me inprudentem obrepseris ('Would you like to exchange, then? you take my wife and I take yours? I wouldn't complain of the bargain. # For (namque) I, no doubt (enim) am such a mug that you may catch me by surprise!', PI. Trin. 59-6\)

Although namque enim is the undisputable reading of the manuscripts the passage has been subject to various emendations51 • The co-occurrence of nam and enim is, however, fully explicable. Both particles turn out to display their regular, distinctive behaviour: namque signals the addition of supporting material to the previous utterance which, exceptionally, has been voiced by a different speaker, thus giving the impression of a strong affirmative reaction. For this use of nam see the discussion in the preceding chapter, p. 150-152 with regard to exx. (17) and (18). Enim, on the other hand, is indispensable for the ultimate ironic interpretation of the clause. Its appropriateness is enhanced moreover by the presence of the emphatic personal pronoun tu and the subjective evaluation verb credo.

5\. Compare the discussion in Langen (1880: 261-71), who defends the reading of the manuscripts. Cf. also Hand's comment on the passage (s.v. enim, p. 391): "alii aut mutarunt enim, aut eiecerunt. Tamen praevalet auctoritas Iibrorum". The proposals for alteration of the text have been prompted no doubt by the widespread misconception that nam and enim are equivalent connective means. Editors and commentators who retain the text of the manuscripts usually assume an ellipsis (see ch. 6); cf. e.g. Brix-Niemeyer ad loco

9

ADVERSATIVE RELATIONS AND ADVERSATIVE CONNECTIVES: CONVENTIONAL ACCOUNTS OF AUTEM, VERO AND AT

Chapter 9 forms the counterpart of chapter 6. It briefly discusses the linguistic notion of adversativity and gives an overview of the ways in which the 'adversative connectives' autem, vero and at have conventionally been treated in Latin gramnuzrs and reference books. As such it forms a preliminary to the descriptive chapters 10, 11 and 12.

9.1 Adversative relations A second group of Latin discourse connectives to be discussed in this study comprises items that in grammars of Latin are usually taken together, for the sake of convenience, under the heading of 'adversative conjunctions'. Consider for instance Roby's 'definition' of adversative conjunctions (1874, II, p. 461): "[adversative conjunctions] contrast the meaning, while they connect the sentences. Such are sed, verum, ceterum, autem, vera, ast, at, atqui, quod, and in some uses quamquam, tamen, etsi, tametsi". Useful as this quite general designation may be for the particular aims of grammars and handbooks, it is, because of the vagueness of the notion 'adversativity', not very revealing of the highly distinctive characteristics of the individual particles themselves. In this study therefore I make use of a more sophisticated notion of 'adversativity', which is in accordance with modern linguistic views on text relations and with the theoretical framework outlined in chapter 4. The survey in this section of the various possible subtypes of the adversative or contrastive relationship (which in view of the complexity of the material has to remain very sketchy) is based partly on a study by Spooren (1989), which incorporates and critically discusses earlier studies on adversative relations. Spooren's account of adversative relations differs from many other studies on the subject in that it is not stated exclusively in semantic

adversative relations

211

terms, but takes also pragmatic and discourse aspects into account!. The particular aims of the present study require, however, a somewhat different account of the matter. For an adequate description of the function of 'adversative' connective particles, I regard as necessary at least the following five notions. It is to be noted that this list is not meant as a typology or subclassification of adversative relations (see also below, p. 216). (i) (ii) (iii) (iv) (v)

substitution (direct or indirect) concession semantic opposition discourse contrast rebuttal

The term substitution can be used to refer to the much-discussed type of adversative relationship that is involved in sentences such as: (1)

this is not green, but yellow

(2)

he is not only a coward, but also a liar

(3)

he is not just a singer, but an all-round performer

In German this particular type of adversative relation is marked by sondern, as opposed to aber, which is reserved for all other types of adversative relations2 • As this type of relation obtains typically intrasententially (and the matching markers are of the 'structural' type, see ch. 3, section 3.2), it falls strictly speaking outside the scope of this study3. Examples of a direct concession relation can be seen in the sentences: (4)

he is sixty-five, but he still plays soccer every week

(5)

although he is sixty-five, he still plays soccer every week

Characteristic of this relation is that an implication of the first conjunct (in this case e.g. something like 'he is not a sportsman anymore') is, so to speak,

1. Spooren's reason for this 'mixed' account is that no adequate semantic system seems available to express the conditions of use of English but. 2. The situation in German is parallelled by e.g. Hebrew and Spanish, which use the pairs ava/ ela and pero - sino. 3. Cf. Spooren (1989: 9-11) for a discussion of the linguistic features of the substitution relation.

212

§ 9.1

'frustrated' (i.e. denied to obtain in this particular case), leading to the expression of the opposite of the implication. Longacre (1983) uses the informative term 'frustrated implication' instead of concession. Direct concession is to be distinguished from indirect concession4 • The typical characteristic of an indirect concession is that the first conjunct gives an argument in favour of a certain conclusion, whereas the second conjunct gives an argument against this conclusion. An example is (6)

Marco is fast, but he has some knee problems

uttered in the context of a coach of a soccer team who is pondering over the line-up of his team: whereas the argument 'Marco is fast' might give rise to the conclusion that this player will be part of the selected team, the argument 'he has some knee problems' would lead to the opposite conclusion. That the distinction between direct and indirect concession is linguistically relevant appears from the fact that only in the former case is there the alternative of using although instead of but (cf ex. 5)5. The relations differ also in the way in which the discourse can be continued: only an indirect concession can be followed by a conclusion. The term semantic opposition is usually reserved for the type of adversative relationship in which two propositions are opposed which do not maintain a logical implication relation, and which are characterized by the fact that the conjuncts are fully symmetric. Examples are: (7)

Mike comes from California, but Fred was born in Utah

(8)

honesty is a virtue, but dishonesty is a sin

Note that the semantic pairs may be lexical opposites, but that this is not a necessary condition. Grammars usually speak of 'weak contrasts'. The relationship involved may be expressed also by and, or it may get no marking at all (as seems to be the normal situation in Latin). In chapter 10 I will make use of the concept of 'contrastive focus' in dealing with semantic opposition relations. In chapter 10 I will also argue that the borderline between semantic

4. A good explanation of the tenns direct and indirect concession can be found in e.g. BrauBe (1983). Other tenns used in literature on the subject are 'denial of expectation' and 'concessive opposition', respectively. In Longacre (1983) the difference between both types of concessions is described in tenns of differences in the expressed parts of what he calls 'the anatomy of frustration'.

5. The same holds for e.g. French, Dutch and Latin. For French cf. Moeschler and De Spengler (1981; 1982).

adversative relations

213

opposition and discourse contrast (more specifically thematic discontinuity, see below), is a fuzzy one. I use the label discourse contrast to refer to the use of adversative relations on the presentational (rather than e.g. the representational) level of discourse. This means that the contrast relations involved do not belong primarily to the referential content of a discourse (as appears to be the case in most examples discussed above), but rather to (i) its rhetorical (,argumentative')6, or (ii) its thematic structure. Examples of the first category (i.e. 'rhetorical contrasts' or 'counterarguments') are (9) and (10), taken from Sweetser (1990: 101). (9)

(Please) look up that phone number - but don't bother if it will take you more than a few minutes

(10) King Tsin has great mu shu pork, but China First has excellent dim sum

In (9) there is no contrast between the conjuncts in the 'content domain' (to use Sweetser's terminology), but between the communicative acts that together constitute a move (see ch. 4). The use of but signals the speaker's consciousness of presenting two at least partially discordant communicative acts side-byside. We could speak for instance of a (conditional) cancellation, which to a certain extent might be considered the discourse pendant of the 'structural' substitution relation (see above? In (10), in the preceding context of which someone has been asked for suggestions of a good Chinese restaurant, the initial indirect suggestion of going to eat at King Tsin apparently contrasts with the subsequent indirect suggestion of going to eat at China First. Notice that in this example a rhetorical contrast relation coincides with a semantic opposition (note the resemblance with ex. 7)8. Another example of a rhetorical contrast relation is (11): (11) I therefore conclude that people should be more concerned for their environment. But the reason why we gathered here tonight is to discuss the problems of mass unemployment.

In this example but draws attention to the fact that a new important discourse unit is being brought up, which counts as more relevant than the preceding

6. Argumentative contrast relations have been studied best by Anscombre and Ducrot (1977) and Ducrot (1980), in the context of an analysis of French mais. 7. Another type of substitution on a discourse level can be seen in sequences such as 'go and get a doctor; but do it quickly', in which the second conjunct functions as a correction or elaboration of the first conjunct (cf. the examples 2 and 3). 8. It is to be remembered that successive clauses are usually connected in both the representational and the presentational domain (cf. ch. 4 and 5).

§ 9.1

214

discourse unit. Another way of saying this is that a particular (communicative) implication of the first discourse unit (viz. that its expression is relevant in the current communicative situation) is frustrated. This description bears a certain resemblance to the definition of the direct concession relation given above. Perhaps we could go as far as to say that we are dealing here with a discourse use of the concession relation. The same may hold, mutatis mutandis, for example (10). The second type of discourse contrast is illustrated in examples (12) and (13): (12) Such was the situation in Gaul. (New section) But in Rome/ In Rome meanwhile preparations were made for the annual elections. (13) I haven't yet told you that my daughter is ... (lnterrruption) But isn't that John coming out of the building?

The examples have in common that they both contain two discourse units, the latter of which is contrastive in the sense of being 'thematically disruptive' or 'thematically discontinuous'. In cases like these, then, we are dealing with a contrast on a quite abstract level. The difference between this type of discourse contrast and semantic opposition (see above) appears to be largely a matter of scope: semantic opposition obtains on a local text level, thematic discontinuity on a global level. The borderline between these two types of relation appears to be not very clear-cut (cf. ch. 10 on autem). The last of the 'adversative notions' listed above, labelled rebuttal, is characterized by a contrast on the interactional level of discourse. Interactional contrast relations obtain typically between the constituent moves of an interactional exchange. They may involve objections, protests, corrections, and the like, which count as challenging reactive moves following an initiating move of a different speaker. An example is (14). Notice that the interactional contrast involved coincides with (or rather, perhaps, is based on) a concession relation: (14) A: B:

I happened to be in that hamburger restaurant But you are on a diet!

As a matter of fact, most concession relations (esp. the indirect ones) could be considered as belonging to the interactional level of discourse, because the conceded part of the sequence involved usually contains the opinion of a different speaker, which is countered in the consequence. For instance, the examples (4), (5) and (6) might in certain contexts be described as rebuttals

adversative relations

215

(i.e. as contrast relations on the interactional level of discourse), rather than as logical concessions (representational level of discourse) or rhetorical concessions (presentational level of discourse). For (6) such a context might be a situation in which a journalist has forced the coach of a soccer team to explain why his fastest player will not be selected for the team. In such a context the clause 'Marco is fast' counts more or less as an initiating move to which the but-clause forms a reaction (diaphonic discourse type with an embedded voice, cf. 5.2.2.2). The correctness of this view is proved by the behaviour of the typical Latin 'interactional contrast marker' at, as I will argue in chapter 12. On this point it is especially significant that at can apparently occur in a correlative pair with the concessive subordinators etsi and quamquam (lit. 'although ... but'; cf. ch. 12, section 12.2.3.3). A few additional remarks may be made with regard to the above overview of the concepts related to the notion of adversativity. First it is to be noted that there has been some dispute in recent literature about the necessity of discerning an 'external' contrast relation in addition to the (not disputed) 'internal' contrast relation, analogous to the distinction between 'external' and 'internal' cause relations (cf. ch. 2 and section 6.1)9. While not wishing to go deeply into this discussion, I would contend that in any case we may discern contrast relations on all three levels of discourse (representational, presentational and interactional). On this point I would recall that the representational level of discourse, as defined in chapter 4, is not fully coextensive with what is usually understood by 'external reality' (as opposed to cognitive and communicative reality, which are both regarded as 'internal'). Although representational concession relations involve a cognitively based implication relation, this implication relation can be described as an objective relation of cause between two states of affairs in the represented world. That is, the conjuncts involved are primarily related by their contrastive content, rather than by for instance their opposed rhetorical (argumentative) value. An example is (15), in which the concession relation is clearly semantically based: 'to succeed' is the natural consequence in the represented world of a state of affairs like 'to try'. (15) Although they tried hard, they didn't succeed

9. Cf. e.g. Spooren (1989: 2-3), who concludes that there is no external contrastive relationship, and Sweetser (1990: 102-104), who leaves the question undecided.

216

§ 9.1

We might use the term 'logical concession' for this type of relationship, as opposed to 'rhetorical concession' (concession on the presentational level of discourse) and 'interactional concession' (concession on the interactional level of discourse). For examples of the latter two types, see for instance (11) and (14), respectively. As we have seen above, we can in the same way speak of opposition on the representational level of discourse (semantic opposition) versus opposition on the presentational level of discourse (discourse contrast: thematic discontinuity); or of substitution on the representational level of discourse ('structural' substitution) versus substitution on the presentational level of discourse (cancellation, cf. ex. 9). So, although it has not been my aim in the present section to give a full-blown subclassification, in terms of the theoretical framework presented in chapter 4, of the adversative relationship, it may gradually have become clear by now what such a classification might look like: substitution, concession and opposition might be regarded as basic concepts; these might apply to distinct levels of discourse, giving rise to more specific concepts such as semantic opposition, rhetorical concession, thematic discontinuity, interactional concession, and the like. A further remark concerns the question of the necessity of subclassifications of contrast relations such as those alluded to here. Although the above distinctions may seem to be linguistically less relevant for a language like English (in which the various kinds of contrast relations seem to be marked almost exclusively by but), they turn out to be helpful for drawing distinctions between a series of Latin but-equivalents that are commonly treated in Latin grammars as more or less synonymous 10 . In anticipation of chapters 10, 11 and 12 we can say for instance that autem, at and vera (in grammars of Latin all commonly described as 'adversative connectives', cf. the citation from Roby given above) cannot be used 'structurally' in substitution relations: substitution relations commonly involve the use of sed or verum, which can hence be considered Latin equivalents of German sandemll. Sed can be considered to be the nearest Latin equivalent

10. Although there are many studies of English but, I do not know of any literature on other adversative connectives, such as however, anyhow, or as a matter of fact. II. In Cicero's Verrines I found a subtle difference between the use of sed and verum in substitution relations: whereas verum is used almost exclusively in corrective substitutions (cf. ex. 3), sed appears to be frequent in eliminative substitutions as well (cf. ex. 1). Such a difference in the conditions of use of verum and sed is however not to be observed in other authors. Menge (1960: 7) mentions also ceterum as an equivalent of German sondern.

previous accounts of autem, vero and at

217

of English but inasmuch as it has the widest applicability: it seems to be possible in all of the contrast relations referred to above, with the exception of semantic oppositions, where the particle autem is prone to occur, which is used also to mark thematic discontinuity (i.e. a particular type of discourse contrast). The Latin but-equivalent at, on the other hand, functions primarily as a marker of concessions on the interactional level of discourse (rebuttals)12. Vero, finally, does not seem to belong to any particular type of adversative relationship. In chapter 11 we will see that vero is not involved primarily in marking sequential relationships (such as the adversative relation), but is merely highly compatible with such relationships, its actual function being 'situating' (rather than strictly connective) in nature.

9.2 Previous accounts of autem, vero and at in the literature In contrast to enim and nam the 'adversative connectives' autem, vero and at have had few studies devoted to them. The only monograph on adversative particles I know of is Saur (1913), which is hardly more than an ordered collection of instances of the various uses of the particles concerned (at, atqui, autem, contra, immo, sed, tamen, vero and verum). The most elaborate accounts of autem and at can be found in Hand (1829-1845), which study formed the starting-point for the treatment of these particles in KlihnerStegmann and many other Latin grammars. Apart from this, the TLL provides useful material. The particle vero appears to be the most challenging object of investigation, since it has not been discussed by Hand, and has yet to be published in the TLL. The accounts of vero in Latin grammars are, accordingly, very brief and noncommittal l3 • Although Latin grammars and reference books provide us with a number of adequate and useful observations, they do, of course, not contain elaborated,

12. The only other Latin but-equivalent that appears to be capable of marking rebuttals is sed. The (relatively rare) use of sed in rebuttals seems however to be confined to a very specific subtype only, viz. to corrections that are incited by a preceding challenging move (which may, in tum, be marked by at). This implies that there is no full equivalence between interactionally used sed and at. Examples are Cic. Orat. 31; Tusc. 2.44; Sen. 65. Cf. e.g. KUhner-Stegmann (p. 74-5) and Saur (1913: 45). 13. In addition to the manuals mentioned I consulted Roby (1882), Nligelsbach (1905), Menge (1960), Szantyr (1965), the OLD, and a number of lexicons on individual authors. Scattered remarks on the particles concerned can be found also in De la Harpe (1923), DeWitt (1937), Petersmann (1977) and Janson (1979).

218

§ 9.2

systematic accounts of the particles concerned. As a rule they employ a strictly semantic definition of the notion 'contrast' or 'adversativity' (cf. again Roby's definition of adversative conjunctions: "they contrast the meaning, while they connect the sentences"). In those cases in which it is hard to detect a semantic contrast relation, the grammars tend to speak of a 'weak contrast', or of an altogether deviant use of the particle concerned, due, for instance, to diachronic changes. There are no attempts (or only very rudimentary ones) to relate the various different uses of a single particle to one 'common denominator' . Some grammars (including Kuhner-Stegmann) give rather detailed descriptions of the conditions of use and the various contexts in which a particular particle may occur, but seem to mistake these contexts for the particle's proper function. Observed differences between the various adversative connectives are usually described in terms of 'weak' versus 'strong' contrast relations: whereas at and vero are held to indicate relatively strong contrasts, autem is characterized usually as a marker of weak contrasts, while sed is believed to be neutral in this respect. Nonetheless the various particles are often considered to be synonyms or near-synonyms. 9.2.1

Previous accounts of autem

In Latin grammars the particle autem is discussed always in the context of adversative clause combining, although it is generally agreed that autem is the weakest of the adversative conjunctions: its value is described as somewhere between 'contrastive' and 'copulative', and hence as comparable to Ancient Greek O€. The copulative aspect of autem is moreover assumed to be responsible for a 'disjunctive' or 'transitional' use of autem. The most characteristic feature of autem (as opposed to other adversative connectives) is that "it introduces a different statement in continuation of the preceding, without really altering or limiting it" (Roby, 1874)14. The most adequate description of autem comes, in my opinion, in the TLL (s. v. autem, p. 1578)15:

14. Cf. also Nagelsbach (1905: 779): "(autem bezeichnet) den indifferenten Gegensatz eines Neuen gegen das Friihere zur Fortflihrung der Rede", and Menge (1960: 348): "sie will nicht das Gesagte aufueben oder den Gegensatz scharf betonen, sondern nur (wie das griech. M) das Neue als etwas von Vorhergehenden Verschiedenes anftihren oder in der Erzahlung fortfahren". 15. The definition in the TLL is based to a large extent on the unsurpassed study of particles by Hand (1829-1845, vol. I, p. 559).

previous accounts of autem

219

" ... sententiam vel alterum enuntiati membrum praecedenti opponit, non quo id quod praecedit reiciatur aut restringatur, sed ut alia res ab alia distinguatur. Particula vis non est adversativa, sed oppositiva et discretiva" ('it opposes the sentence or some other part of an utterance to what precedes, not in the sense that it rejects or restricts the preceding part, but that it distinguishes both parts from one another. The particle's value is not so much adversative as oppositive and discretive') The above statements amount to saying, in terms of the classification of adversative relations proposed in 9.1 above, that autem is not a marker of substitution relations, and is not very likely to be a marker of concessions either. The following examples are cited in the literature among illustrations of the contrastive (,however', 'but'), copulative ('and', 'moreover'), and transitional ('now') uses of autem, respectively. (16)

nam iniusta ab iustis impetrari non decet, iusta autem ab iniustis petere insipientia est ('It would be unfitting, of course, for unjust favours to be obtained from the just, but looking for just treatment from the onjust is folly,' PI. Am. 35-36)

(17)

abite tu domum et tu autem domum ('Go on home, you, yes, and home with you, too!', PI. True. 838)

(18)

(The beginning of the chapter on the events of the year 405 BC) Fuere autem tribuni T. Quinctius Capitolinus Q. Quinctius Cincinnatus ... ('Now the tribunes were T. Quinctius Capitolinus Q. Quinctius Cincinnatus, ... ', Liv. 4.61.1)

In addition to the adversative, copulative and transitional uses of autem (which are generally assumed to be closely connected 16) one usually mentions a number of other autem-usages which seem to be as odds with this general picture. Thus, it is assumed for instance that autem may also be used to

16. Cf. DeWitt (1937: 453): "... autem is properly a continuative particle, advancing the narrative or argument to a new stage: if what follows conflicts with what precedes, it is mildly adversative, 'however'; if it confirms what precedes, the translation is 'moreover'. Once more, since the force of particles fades out at times, the transitional 'now' is especially an adequate rendering". In some accounts it is assumed that the variations in use of the particle are due to diachronic change. Kuhner-Stegmann and Szantyr e.g. assume that the adversative value of the particle was not fully developed until the classical period. Cf. also Petersmann (1977: 247), who assumes that the possibilities of use of autem were gradually enlarged. I do not agree with this view.

220

§ 9.2.1

express indignation or surprise III questions 17, or to underscore the exclamatory force of an utterance. This seems to point to a 'modal' use of the particle autem. An example is (19): (19) (Jupiter to Mercury) Pergin autem? nonne ego possum, furcifer, te perdere? ('You still go on? Is there no choking you off, you jailbird?', PI. Am. 539)

Moreover, autem is regularly held to establish causal connections (explications, justifications, and the like), in a way more or less comparable with nam or enim 18 • One of the examples adduced in the literature is: (20) Ipse cum equitatu antecedit ad castra exploranda Cornelia, quod is locus peridoneus castris habebatur. Id autem est iugum directum eminens in mare, utraque ex parte praeruptum atque asperum, sed tamen paulo leniore fastigio ab ea parte, quae ad Uticam vergit. (,Curio himself goes on in front with his cavalry to explore the Cornelian camp, because that spot seemed particularly suitable for a camp. For this (id autem) was a straight ridge, projecting into the sea, abrupt and rugged on either side, but with a somewhat gentler slope on the side facing Utica', Caes. Civ. 2.24.3)

It is clear that the 'modal' and 'causal' uses of autem form a major problem in the traditional accounts of the particle. Usually they are merely added, without explanation, to the list of various uses of autem. This is the case also in the TLL, which fails to make clear how the 'modal' and 'causal' uses of autem fit into the (very adequate) general definition of the particle with which the TLL lemma begins (quoted above, p. 219). I return to (20) in ch. 10, p. 267.

9.2.2 Previous accounts o/vero In most Latin grammars and reference books vero is pictured as a versatile particle with two quite distinct main 'functions' or 'meanings': roughly speaking one discerns an affirmative or reinforcing particle ('sentence adverb') vero ('really'), as well as an adversative conjunction vero ('but'). Although there is no argument about the common lexical origin of these two uses of vero, and some studies assume a diachronic development from the affirmative

17. Cf. Hand (1829-1845) s. v. autem, p. 575: "atque primum autem exprimit quaerentis impetum et vehementiam sive ab admiratione proficiscatur, sive ab indignatione et iracundia, maxime in reprehensione. Graeci dicunt ori". Sometimes also the use of autem in indignant repetitions of a word in a preceding clause is assigned to this category. 18. Cf. e.g. De la Harpe (1923: 67-69) and TLL s.v. autem, p. 1588. Although the 'causal' use of autem is assumed to be particularly characteristic of late Latin (cf. Szantyr, pp. 490-491), one . usually adduces instances from early and classical Latin as well.

previous accounts of vero

221

use into the connective, adversative use 19 , the two functions are from a synchronic viewpoint felt to be sufficiently different as to warrant separate treatment. Thus, in Kiihner-Stegmann, vero is discussed partly among the "Modalitatsadverbien" (vol. I, 795-799), and partly among the conjunctions denoting an adversative relationship ("adversative Beiordnung", vol. II, 7981). In this latter use vero is often equated with autem. The following examples are adduced in the literature among the prototypical instances of the 'modal adverb' vero and the 'adversative conjunction' vero, respectively: (21) Ni: cape hoc tibi aurum, Chrysale, i fer filio Ch: non equidem accipiam ... Ni: cape vero ('Take this money, Chrysalus: go, carry it to my son ... # No indeed, I won't take it ... # Come, come now, take it', PI. Bac. 1059-1062) (22) Sed sunt haec leviora; illa vero gravia et magna, quod ... (,But these are comparatively trivial points; of greater weight and moment is however the fact that ... ', Cic. Planc. 86)

In examples such as (21) vero is usually said to corroborate the imperative. Other, related modal uses of vero that are generally pointed out in the literature are the so-called 'affirmative' use in answers, and the 'intensifying' use in combination with pronouns, for which consider (23) and (24). Very few specific comments or explanatory remarks are added to the discussions. (23) Br: tu orationes nobis veteres expJicabis? Ci: vero, inquam, Brute (' "Won't you go over for us those early speakers?" "Certainly Brutus", I said .. .', Cic. Brut. 300) (24) (The start of a new section) Mihi vero ne haec quidem notiora et illustriora carere vi divina videntur ... ('To my mind even better known and more famous fields of labour do not seem removed from divine influence', Cic. Tusc. 1.64)

The chapters in reference books on the use of vero as an adversative conjunction are not very illuminating either, especially not with regard to the question what type of adversative relationship is involved. The only clues given on this point are that vero is excluded from the substitution relation, and that it introduces its 'contrasted thought' with a certain emphasis.

19. Cf. e.g. KUhner-Stegmann, p. 80.

222

§ 9.2.2

Other observations that recur in the sections on 'adversative' vera are that vera often expresses an intensification or climax (,Steigerung', as KiihnerStegmann and Saur call it), and that it may also be continuative, signalling a transition to a new thought (notably in combination with the particle iam). Some Latin linguists mention autem as a near-equivalent of adversative vera. There are no real attempts to relate the adversative use of vero to its modal use. 9.2.3 Previous accounts of at

Of the three 'adversative' particles under discussion, at appears to be the bestdocumented in Latin grammars and reference books. It has received thorough treatments in Hand (1829-1845) and the TLL, and consequently also in the Latin grammars. There is also a short article on at by Rebert (1929). It is generally agreed in the literature (with the exception of Rebert, see note 20 below) that at is an adversative particle marking a 'strong' contrast. As such it is held to differ from for instance autem and sed, which are considered markers of 'weak' and 'neutral' contrasts, respectively. There are, however, many puzzling exceptions (it is observed that at may indicate also a weak contrast, or no contrast at all), which have often led to the conclusion that at (like so many other particles) is a highly polysemous word. There are only few attempts to search for a common denominator which covers the various uses 20 • Although the various treatments differ considerably, most accounts mention at least the following three uses of at: (i) adversative conjunction in dialogues (indicating an objection), (ii) adversative conjunction in monologues (indicating a strong or weak contrast), and (iii) sentence particle with a strongly emotional force. The use of at to mark objections in dialogues (which Hand calls the usus rhetoricus) is usually regarded as the most distinctive feature of the particle. This use, which abounds in comedy, is not absent from essentially monologal

20. Cf. however Szantyr (p. 488-489), who, following Hand, explicitly mentions 'opposition' (,Gegentiberstellung') as the basic meaning underlying all the different uses of at, but who gives no further comment on the apparently deviant cases. Rebert's proposal (1929) to abandon the idea of at being an adversative conjunction in all its uses is attractive, but his solution (at is in essence a particle with an emotional force, strongly related to the interjection attat, and only secondarily an adversative conjunction), which is based on Cicero's Philippics only, is far from convincing and cannot be regarded as a viable alternative to the traditional view.

previous accounts of at

223

text types either: all accounts draw attention to the 'objection' use of at in contexts in which the author enters into a discussion with a fictitious (often unidentified) interlocutor (as is often the case in Cicero's works, e.g. in the form of argument called occupatio). Hand (1829-1845, vol. I, p. 431-439) gives the most elaborate account of the various types of objections that can be introduced by at. I give two examples. (25) eia vero, age dic # at deridebitis # non edepol faciemus ('Dh really now! Come out with it # But you two will make fun of me # No, no, we won't, upon my word' , PI. Epid. 262) (26) at occidit Saturninum Rabirius. utinam fecisset! ('But (you say) Rabirius killed Saturninus. Would that he had!', Cic. Rab. Perd. 31)

The use of at as an adversative conjunction in monologues is called by Hand the usus grammaticus, by which he means that at in these cases has a 'simple meaning of contrast' (simplicem oppositionis signijicationem, ibid., p. 420). Under this and comparable headings the manuals discuss, among other things, instances of at which serve strongly to oppose two persons, objects, facts, etc. Consider the following two examples: (27) tempore illi praecepto, at hi numero avium regnum trahebant (' ... the one party laid claim to the honour from priority, (but) the other from the number of the birds', Liv. 1. 7 .2) (28) Metellus ille honoratis quattuor filiis, at quinquaginta Priamus (,The famous Metellus had four sons who became dignitaries of State, but Priam had fifty', Cic. Tuse. 1.85)

Sometimes the contrast involved is less strong, in which cases at appears to signal merely the continuation of the discourse (example 29). This use of at is found especially in historiography, and appears to have been developed to the full by Tacitus (example 30). (29) Postquam Caesar dicendi finem fecit, ceteri verbo alius alii varie assentiebatur. At M.Porcius Cato rogatus sententiam huiuscemodi orationem habuit: " .... " 'After Caesar had finished speaking, the rest briefly expressed their adherence to one or another of the various proposals. But M. Porcius Cato ... spoke to the following purport: " ... " " Sal. Cat. 52.1) (30) (Having reported a rebellion among the Roman legions on the Rhine, Tacitus continues as follows) At in Chaucis coeptavere seditionem praesidium agitantes vexillarii ...

224

§ 9.2.3 ('Among the Chauci, however, a detachment ... which was serving on garrison duty, made a fresh attempt at mutiny ... ', Tac. Ann. 1.38)

Apart from this oppositive use (which at seems to share with autem and Ancient Greek M), the manuals signal also a concessive use of at in monological texts. Examples are (31) and (32). In (32) at apparently occurs in the apodosis of a concessive clause introduced by etsi ('although'). There are comparable examples in which at apparently occurs in the apodosis of a conditional clause introduced by si ('if). (31) parum succedit quod ago; at facio sedulo ('My attempt is not successful; still I work hard', TeL An. 679) (32) etsi scelestus est, at mihi infidelis non est (,Rascal though he is, he's a loyal one to me', PI. Trin. 528)

The third use of at that is commonly distinguished in the manuals (sentence particle with a strong emotional force, or in Hand's terminology, usus patheticus) is obviously the most problematic. It is found in exclamations, orders, supplications, wishes, and the like. It is regarded as expressing not some sort of contrast with the preceding text, but merely the astonishment, indignation, despair, contempt, etc. of the speaker. An example is: (33) (In an aside) at ut scelesta sola secum murmurat! (,Hear the old criminal mumbling away to herself, though', PI. Aul. 52)

It has already been observed that Latin grammars and reference books are not very specific about what they mean by 'adversative relation' and 'adversative conjunction'. When we try, however, to translate their observations about at into the various 'adversativity concepts' discussed in 9.1 above, we might come to the conclusion that at covers almost the entire spectrum of adversativity, and even more besides (cf. examples such as 33). Its use as a marker of rebuttals (my concept v, p. 211 above) is undeniable. It appears, however, to be possible to use the particle also as a marker of semantic oppositions (concept iii, cf. examples 27 and 28), and as a marker of a specific type of discourse contrast (concept iv, cf. 29 and esp. 30). The manuals point more-

previous accounts of at

225

over to a concessive use of at (concept ii, cf. examples 31 and 32). An apparently 'structural' use of at in a substitution relation (concept i; but see my remarks in ch. 12, p. 348) is illustrated furthermore in (34)21: (34) non honestum cons ilium, at utile, ut aJiquis fortasse dixerit (,The trick (viz. Ulysses' ruse) was not morally right, but, some one may perhaps say, it was expedient ... ', Cic. Off. 3.97)

All in all, the picture of at we may receive from the literature is very diffuse. In chapter 12 I will attempt to show how the various uses are interrelated.

21. Another instance is Caes. Civ. 2.34 (erat vallis ... non ita magna, at difficili et arduo ascensu). Despite the existence of instances such as (34) the manuals do not hint at the possibility of using at in substitution constructions. Riemann (1886: 570) explicitly states that at, in contrast to e.g. sed and verum, cannot be used as an equivalent of German sondern.

10 AUTEM

Latin reference grammars and studies of Latin particles distinguish a large variety of uses of autem (see section 9.2.1). Although these accounts appear for the most part to be quite accurate, they leave a number of questions unresolved, especially concerning the interrelatedness of the various uses. Major problems are raised for instance by the alleged 'explanatory' use of autem (socalled autem in the role of enim or nam), and by its use as a 'modal' particle with a certain emotional force. How do uses like these fit in with the prevailing view of autem as primarily a marker of weakly adversative relationships? In the present chapter a more unitary description of autem is proposed along the theoretical (discourse-pragmatic) lines sketched in chapter 4, and parallel to the accounts of nam and enim in chapters 7 and 8. It starts by formulating the general discourse junction of autem (lO.I) and subsequently deals with the various specific uses associated with this discourse junction (10.2 and 10.3). The concluding section 10.4 is concerned with the pragmatic motivations that might be taken to underly the occurrence of autem in a text; in this section autem will also be contrasted with a number of comparable discourse connectives.

10.1 The discourse function of autem

Autem is a presentational particle which marks the discrete status of a piece of information with regard to its verbal or non-verbal context. Depending on whether the particle is applied locally (on the level of the sentence) or more globally (on the level of the text) it can be characterized as a 'highlighting' or 'focusing' device (see 10.2 below), or as a marker of the organization of the text (viz. of thematic discontinuity).

discourse function of autem

227

explanatory remarks From the above definition it appears that autem is not a presentational particle that marks rhetorical relations between communicative units such as acts and moves; the function of autem rather pertains to information units or thematic units, and hence to the thematic organization of a discourse.

10.2 The local use of autem: autem as a focusing device When autem is used with local scope, the discriminating and demarcating force which is central to the discourse function of autem works as a way of highlighting or emphasizing a particular piece of information coded by a noun, verb, adverb, adjective or a combination of these elements. As such, autem could be called a focusing device: it singles out constituents which count as salient in the immediate context or situation l . Most commonly the distinctive status and consequent saliency of the information unit marked by local autem is due to some kind of parallelism or contrast (in the broadest sense of the word) with a preceding, corresponding piece of information. Linguists usually apply the label 'contrastive focus' to such constituents which are salient by dint of contrast. To illustrate the type of contrastive focus at issue here2 consider the following simple English example. The 'opposed' constituents are in upper case. (1)

I know your FATHER; your MOTHER I don't know

1. Cf. Dik's definition of focal information (1989: 277): "The focal information in a linguistic expression is that information which is relatively the most important or salient in the given communicative setting, and considered by S (the speaker, CK) to be most essential for A (the addressee, CK) to integrate into his pragmatic information." Various subcategorizations of focality have been proposed. Dik distinguishes e.g. between completive (= assertive) and contrastive focus. The latter can be narrowed down further into parallel focus and counter-presuppositional (e.g. replacing, expanding, restricting and selecting) focus. 2. Within the purview of the description of Latin autem I am interested only in the relatively weak type of contrastive focus that could also be called 'parallel focus'. Stronger forms of contrastive focus in English are typically coded by stress, cleft or pseudo-cleft (Givon 1990: 704). Compare: (a) Joe will milk THE GOAT (stress-focus) (b) It's THE GOAT that Joe will milk (cleft) (c) What Joe will milk is THE GOAT (pseudo-cleft) Latin autem is not used as a marker of this stronger type of focality. I will therefore use the term 'parallel focus' rather than 'contrastive focus' when dealing with autem.

228

§ 10.2

In contexts like (1) Latin has the option of inserting the particle autem in the second member of the parallel construction, as can been seen from the following example: (2)

Cum deinde Dareum, ubicumque esset, occupare statuisset (sc. Alexander), ut a tergo tuta relinqueret, Amphoterum CLASSI ad oram Hellesponti, COPIIS autem praefecit Hegelochum ('Then, since he (i.e. Alexander) had resolved to catch Darius, wherever he might be, in order to leave everything behind him safe, he gave Amphoterus command of THE FLEET at the shore of the Hellespont, and Hegelochus of THE LANDFORCES', Curt. 3.1.19)

In this example the constituent copUs ('the land-forces') bears 'contrastive' (or, rather, parallel, cf. note 2) focus, being opposed to classi ('the fleet') in the first part of the sentence. Most parallel focus constituents are contextually or situationally 'given' or accessible information, and function as part of the thematic strands along which the discourse evolves. As such they involve not only contrast or parallelism, but also topicality3. In (2), for instance, the constituents classi ('the fleet') and copiis ('the land-forces') convey contextually accessible information: they are not only already implied by the verb occupare ('to occupy', 'to overtake'), but can also be regarded as stock elements in a historical work about a military campaign. Although Alexander counts as the main discourse topic in this example (a status which he maintains throughout the entire work), classi and copUs can be considered discourse topics as well - albeit with narrower scope and less persistence. Less often parallel focus constituents are contextually 'new' and non-topical information, as is the case in the early Latin example (3), where the status of discourse topic should be assigned to cibo ('food') or the complex cibo opus esse ('to need food'), while the parallel constituents puero ('a child'), matri ('the mother'), quae puerum lavit ('she who bathes the child') and nutrici ('the nurse') are all new focal elements: (3)

PUERO opust cibo, opus est MATRI autem, QUAE PUERUM LAVIT, opu' NUTRICI

3. There is no unequivocal definition of 'topic' and 'topicality' in the linguistic literature. In this study I use the term 'topic' or 'discourse topic' to refer to those constituents in the discourse which are characterized by (i) referential accessibility or givenness and (ii) thematic importance. For a general discussion of topic and topic devices cf. Giv6n (1990); for a treatment of topic within the framework of Functional Grammar cf. Dik (1989) and the evaluative remarks by Mackenzie and Keizer (1991). In 10.3 I will give more attention to the issues of thematic development and topical constituents.

local use of autem

229

('THE CHILD needs food; THE MOTHER needs it too; THE WOMAN THAT BATHES HIM needs it; THE NURSE has her needs', PI. True. 902-903)

In a restricted number of cases the distinctive status and consequent saliency of the information unit marked off by local autem is absolute, i.e. not due to some explicit contrast with a preceding parallel information unit. An example of autem as a marker of absolute focus is (4): (4)

Th: Ius dicis Ep: Me decet Th: lam TU autem nobis praeturam geris? ('You speak like a judge # The proper thing for me! # What, do YOU (tu autem) already hold the praetorship?', PI. Epid. 25)

In the following I will deal first (10.2.1) with the use of autem in parallel focus constructions, and next (10.2.2) with its use as a marker of absolute focus. 10.2.1 autem as a marker of parallel focus It was observed in the case of the examples (2) - (3) above that the distinguish-

ing and demarcating force of autem is regularly employed for the signalling of a parallel focus element. In the context in question the focus of attention shifts from constituent A to constituent B, A and B being (semantically or pragmatically) related, though, at the same time, explicitly demarcated constituents. Clear instances of the focalizing use of autem in the second member of a parallel construction are cited in (5), (6) and (7), which are all taken from the comedies of Plautus: (5)

Fiam, ut ego opinor, Hercules, TU autem Linus ('I'll be a Hercules, YOU, on the other hand, a Linus', PI. Bae. 155)

(6)

Matrem HlC salutat suam, HAEC autem hunc filium ('HE is greeting his mother, SHE, on the other hand, her son', PI. Poen. 1144)

(7)

TU magnus amator mulierum es, Messenio, EGO autem homo iracundus ('YOU, Messenio, are a great lover of the ladies, while I, on the other hand, am a choleric man', PI. Men. 268-269)

That the use of focalizing autem in parallel constructions is not confined to Latin comedy is clear from (8) below. In the passage cited Cicero describes Verres' motivation for letting the crucifixion of an enemy take place at a point on Sicily that looks over the Straits of Messina:

230 (8)

§ 10.2.1 Italiae conspectus ad eam rem ab isto delectus est, ut ILLE in dolore et cruciatuque moriens perangusto fretu divisa servitutis ac libertatis iura cognosceret, IT ALIA autem alumnum suum servitutis extremo summoque supplicio affixum videret ('This place with its view of Italy was deliberately picked out by Verres, that HIS VICTIM (ille: 'he'), as he died in pain and agony, might feel how yonder narrow channel marked the frontier between the land of slavery and the land of freedom, and that IT ALY, on the other hand, might see her son, as he hung there, suffer the worst extreme of the tortures inflicted upon slaves', Cic. Ver. 5. 169)

That autem in the parallel constructions illustrated above counts as a focalizing particle and not, say, as a coordinator, is proved by the occurrence of the combination et ... et autem ('and ... and on the other hand'), which is especially common in early Latin comedy. The construction is to be considered the Latin equivalent of Ancient Greek KCXt ... KOit ... Of. Consider the following two examples4 : (9)

Facile istuc quidemst, si et ILLA volt et ILLE autem cupit ('This (Le. her leaving) is easy enough, if SHE wishes it, and HE, moreover, is crazy for it', PI. Mil. 1149)

(10) currenti properanti hau quisquam dignum habet decedere. Ita tres Slmltu res agendae sunt, quando unam occeperis: et CURREN DUM et PUGNANDUM et autem IURIGANDUM est in via (,When a man's running in a hurry, not a soul sees fit to make way for him! so when you've begun on one thing, you have to do three things all at the same time: RUN and FIGHT and QUARREL, too, all the way', PI. Mer. 117-119)

A comparable example of the construction et ... et autem from later Latin is (11)5: (11)

Itaque utriusque bonum est merito laudari, tam mehercules quam bene iudicasse iudicantis bonum est et eius secundum quem iudicatum est. Numquid dubitas quin iustitia et HABENTIS bonum sit et autem sit ElUS CUI DEBITUM SOLVIT? ('It is therefore a good belonging to both sides, - this being praised when one deserves it - just as truly as a good decision is the good of him who makes the

4. Some other examples are PI. Poen. 841 and 927. A comparable combination with -que is found in Men. 1089-1090: Neque aqua aquae nee lacte est lactis, crede mi, usquam similius, quam hie tui est tuque huiius autem ('No drop of water, no drop of milk, is more like another, believe me, than

he's like you, yes, and you like him'). 5. The construction et ... et autem is rare in classical Latin prose. For the comparable construction neque ... neque autem ('neither ... nor, on the other hand') see however Cic. Fam. 5.12.6: neque enim tu es, qui ... ; neque autem ego sum ita demens, ut ... ; the latter construction is found also several times in Lucretius: 1.857; 3.561; 4.152; 5.366; 6.103; 6.779. An explanation for the scarce use of the et ... et autem construction in classical Latin prose may be the lack of conversational texts from this period, which form the preferred environment for strongly emphatic focus.

autem as a marker of parallel focus

231

decision and also of him in whose favour the decision was given. Do you doubt that justice is a blessing TO ITS POSSESSOR, as well as (et autem) TO THE MAN TO WHOM THE JUST DUE WAS PAID?', Sen. Ep. 102.19)

The expression et ... et autem is used to emphasize on the one hand the coordination of corresponding units of information, and on the other the individuality and mutual discreteness of the coordinated elements. This remarkable mixture of (formal) conjunction and (semantic or pragmatic) disjunction is, as we will see in the course of this chapter, a recurrent feature of autem, which distinguishes it from other, comparable particles. The effect of using strongly focalizing expressions like et ... et autem might become clear when we compare (9) and (10) above with their unmarked alternatives (9)" (9)" and (10)': (9)'

si ilia volt et ille cupit ('if she wants it and he desires it')

(9)"

si ambo volunt ('if they both want it')

(10)'

currendum, pugnandum et iurigandum est ('one has to run, fight and wrangle')

10.2.1.1

parallel constructions with lexical opposites

It hardly needs comment that the label 'adversative connective' (as it is used

by Latin grammars, that is, in a strictly semantic sense) does not fit very well with the use of autem illustrated in (5) - (11) above, especially not with the examples (9) - (11). However, in cases where the components of the parallel construction happen also to be lexical (i.e. semantic) opposites, such a description (more or less interchangeable with sed, 'but') might at first sight seem to be quite appropriate. Compare (12): (12) neque ego inter me atque te quicquam interesse umquam duxi praeter voluntatem institutae vitae, quod ME ambitio quaedam ad honorum studium, TE autem alia minime reprehendenda ratio ad hones tum otium duxit ('and I have never thought that there was any difference between you and me, except our choice of a career. A touch of ambition led ME to seek political advancement, while another perfectly laudable motive led YOU (te autem) to honourable ease', Cic. Au. 1.17.5)

It cannot, indeed, be denied that the clauses me ambitio quaedam ad honorum studium (sc. duxit) ('a touch of ambition led me to seek political advance-

§ 10.2.1

232

ment') and te alia ratio ad honestum otium duxit ('another motive led you to honourable ease') maintain some sort of semantic contrast relation. It cannot be denied either, however, that we are dealing here also with a parallel construction not different from the ones cited above in (5) - (11):

duxit

/ ~

[;] I ambitio quaedam

I ad honorum studium

~ I alia ratio

I ad honestum otium

I would hence contend that the semantic contrast interpretation that is intuitively given to the sequence in (12) is not so much dependent on the presence of autem, as on the lexical opposites studium ('activity') and otium (,leisure'), as well as on the anticipatory interesse ('to be different'). Under appropriate contextual conditions (viz. the presence of lexical opposites) parallel focus particles however seem to approach more strictly adversative connective items (such as sed 'but') in value. Another illustration is (13), in which the contrary reactions of two different groups of people (the frivolous and the more serious) are being described: (13) 0 rem, quam homines SOLUTI ridere non desinant, TRISTIORES autem sine maximo dolore audire non possint! ('What a situation, about which THE FRIVOLOUS cannot stop laughing, but which THE SERIOUS (tristiares autem) cannot hear of without the deepest mortification!', Cic. Dam. 104)

In (14) and (15) below I have cited the two examples of autem in my sample that would qualify best for a description in terms of the expression of a semantically based adversative relationship between two adjacent clauses. In (14) the contrastive focus element marked by autem is a verb instead of a noun or noun phrase, which obviously enhances such an interpretation. In fact, however, there is no real difference from (12) and (13): scribo ('I write') and [ego ('I read') function as parallel focus elements which happen to be lexical opposites. (14) Ego hic cesso, quia ipse nihil SCRIBO, LEGO autem libentissime ('I am idling here, because 1 am not WRITING anything myself; READING, on the other hand, is a great pleasure to me', Cic. Fam. 16.22.1)

Example (15) contains what at first sight seems to be an isolated instance of autem in a concessive construction (see section 9.1 for the various types of

autem as a marker of parallel focus

233

adversative relations). On this example KUhner-Stegmann (p. 92) remark that autem is used instead of sed or at ('but'): (15) (Croesus) hostium vim sese perversurum putavit; pervertit autem suam ('Croesus thought that he would overthrow his enemy's power; in fact he overthrew his own', Cic. Div. 2.115)

Although the above example has indeed the appearance of a concessive construction (in which autem might be said to play the role of an adversative, specifically a concessive, conjunction), one could however equally well maintain that (15) is a marked instance of a parallel construction, in which the pair of parallel focus elements is formed by two different temporal distinctions applied to one and the same state of affairs. Instead of for instance two nouns or two different verbal predicates, two possible tenses of the action pervertere ('to overthrow') (future and past, respectively) are being opposed. The resemblance between (15) and the parallel focus constructions in (6) - (13) becomes clearer when we paraphrase (15) as 'WHAT HE INTENDED TO DO was to overthrow his enemies' power, WHAT HE ACTUALLY DID, on the other hand, was to overthrow his own'.

10.2.1.2 change of addressee A quite specific type of parallel focus construction in which autem occurs is that involving a change of addressee. Here the discriminating force of autem appears to be employed in order to avoid any misunderstanding about the identity of the person addressed, in case for instance of a possible referential ambiguity, or when the speaker expressly wants to appeal to the personal involvement of each individual of a number of addresses. In the passage quoted in (16) below the slave Pythodicus issues two different directives to two different flute-girls. The resemblance to the parallel constructions in (5) - (10) is obvious. A similar example from Cicero's orations is cited in (17), where the first directive (recita) is addressed to the clerk of the court, the second (exsurge) to the author of the testimonium, who is personally present at the meeting6 :

6. Another example from Plautus is Poen. 1414, from Cicero, Ver.4.70.

§ 10.2.1

234 (16) Co: St:

(17)

Herc1e iniuria dispertivisti: pinguiorem agnum isti habent At nunc tibi dabitur pinguior tibicina I sane cum illo, Phrugia. TU autem, Eleusium, huc intro abi ad nos ('That's not a fair way to divide: they've got the fatter lamb # Oh well, I'll give you the fatter music girl. (turning to the girls) You go with him, Phrygia. YOU, on the other hand, Eleusium, you step over to our place', PI. Aul. 330-334)

quod (sc. testimonium) recita. TU autem, nisi molestum est, paulisper exsurge; perfer hunc dolorem commemorationis necessariae ('Read it. (The clerk prepares to read out the testimonium) And do YOU (tu autem) , sir, if I am not asking too much, stand up for a few moments and nerve yourself for this painful but indispensable recital', Cic. Clu. 168)

Another instance of autem after a change of addressee is (18): (18)

Solvite istas. Agite, abite TU domum et TU autem domum ('(to slaves) Release them. (to girls, loosed) Get along, go home, (to his own maid)

YOU, (to the other) yes, and home with YOU (tu autem) too', PI. True. 838)

The last example is remarkable in that after the second person plural imperative (abite) the speaker still thinks it necessary to address both girls individually. The intended effect of this redundancy (in combination with the discriminating force of autem) is presumably that the directive will have a greater impact on the girls (especially on the girl that is not the speaker's own maid and therefore less likely to obey his orders). Note the difference in the alternative, 'collective' expression agite, abite vos domum ('Get along, you two go on home'). Note that (18) may be regarded as a very straightforward example of a parallel construction, and that no (semantic) adversative relationship whatsoever is involved between the two parts of the sentence. As to the use of autem after a change of addressee I want to draw attention finally to a passage in Plautus' Pseudo Ius in which the pimp Ballio addresses his girls one by one. In two of the three changes of addressee autem is added to the form of address: (19)

Nunc adeo hoc factust optumum ut nomine quemque appellem suo ... : advortite animum cunctae ... Principio, Hedylium, ... Aeschrodora, tu ... Tu autem, Xytilis, ... Tu autem ... Phoenicium, ... ('Now then, the best thing I can do is address you each by name ... : attention, all of you. To begin with, Hedylium, ... Aeschrodora, you ... Now for you (tu autem) , Xytilis ... And now for you (tu autem) ... Phoenicium .. .', PI. Ps. 185227)

autem as a marker of parallel focus

235

The above passage shows that the local use of autem as a parallel focus particle, and its global use as a marker of the organization of the text (i.e. of thematic discontinuity, see section 10.3 below), are entirely compatible and are not to be regarded as two essentially different discourse functions: in (19) both aspects are represented. 10.2.1.3

additional remarks

A few additional remarks conclude this section on the use of autem as a parallel focus marker. First, it is a well-known fact that from Cicero onwards both elements of a parallel construction may be marked by a parallel focus particle, the first element being accompanied by preparatory quidem7. The construction concerned is a rarely used equivalent of the highly frequent Greek /l.Ev ... DE construction. I give two examples: (20) Fieri tamen potest, ut MORBUS quidem id desideret, CORPUS autem vix pati posse videatur (,But it can happen that SOME DISEASE, on the one hand, demands it (Le. bloodletting), but that THE BODY, on the other hand, seems scarcely able to bear it', Cels. 2.10.7) (21)

"ILLOS quidem senatus, ME autem tuebitur Tiberiolus meus" (' "with THEM the senate will side, with ME, on the other hand, my pretty little Tiberius" " Tac. Ann. 6.5)

A second remark concerns the high frequency of autem in combination with a personal pronoun, especially in conversational texts (note e.g. the combination tu autem or ego autem in the examples 5, 7, and 16 - 198). This phenomenon may be explained in terms of the compatibility of such pronouns with the specific focalizing function of autem: autem, as we have seen above, is apt to cluster with referents whose identity is salient, surprising or ambiguous in the particular context or situation. Now it has been observed by Pinkster (1986; 1987) that in Latin conversational texts first and second person pronouns (especially those in subject function) are not used unless specific focal factors

7. For the particle quidem cf. Solodow (1978). The combination quidem ... autem is mentioned only in passing, in the context of what Solodow calls "contrasting quidem", which happens to be followed regularly by an 'adversative' particle (1978: 32). 8. Cf. e.g Hand (1829-1845, vol. I s.v. autem, p. 561-2) for the combination of autem with pronouns in general, and p. 574 for ego autem.

236

§ 10.2.1

are at stake9 • Among the pragmatic motivations that may prompt the use of such a subject pronoun he mentions (i) the identification of the referent (in case of uncertainty or ambiguity), (ii) contrast, and (iii) absolute prominence or saliency of the subject referent (see below, p. 240). Subject pronouns form, in other words, a specific group of inherently focal elements in whose immediate context we may well expect to find the equally focalizing particle autem: the combination of a subject pronoun and the particle autem can be regarded as a cumulation of focus indicating devices. It is quite feasible, furthermore, that when a speaker shifts from one referent or discourse partner to another, the combination of (subject) pronoun and autem is accompanied by certain non-verbal signals (gestures, eye-movements, etc.) which give physical support to the verbal identification. Thus for instance in the scene from Plautus' Trueulentus cited in (18) above, it is quite possible that when the speaker addresses the second girl (et tu autem: 'and you too'), he points the finger to her, or turns his head or entire body towards her. The same may hold for the combination of autem and a demonstrative pronoun or adverb. Consider (22), in which the character Menaechmus, standing in front of his mistress' door, is afraid that neither his wife nor his mistress will let him in. In all probability hue autem ('to this place, on the other hand') will be accompanied by a deictic gesture of the speaker: (22) Quid ego nunc faciam? Domum ire cupio: uxor non sinit; hue autem nemo intromittit ('What shall I do now? I long to go home, but my wife won't let me, and as for this place (hue autem), no one will let me in', PI. Men. 963-964)

Slightly different is example (23): (23)

Ego consul ... cum ... Catilina in castris, in his autem templis atque tectis dux Lentulus esset constitutus ... interitu rem publicam liberavi (,During my period as consul, ... when Catiline had been made commander in the camp, and Lentulus commander in these very temples and dwellings, ... I freed the state from ruin', Cic. Sui. 33)

The use of focus indicating autem in (23) is motivated in more than one way. The constituent in his templis atque teetis ('in these temples and dwellings ') is not only salient on account of its being part of a parallel construction with in eastris ('in the camp'), but is also focused on in an absolute sense: the

9. Subject pronouns in Latin (e.g. ego, nos, tu, vas) are optional elements, the person and number distinctions being coded in the finite verb form of the clause.

autem as a marker of parallel focus

237

outrageous fact that Lentulus had been put in charge of the places most sacred for a Roman citizen, that is, the temples of their gods and their own private homes, warrants an unusually strong emphasis. This emphasis is achieved by a combination of the strongly local deictic pronoun his ('these', i.e. 'our very own') and the highlighting particle autem, and is rendered very adequately in the Loeb translation ('in these very temples and dwellings'). It is moreover not inconceivable that in the particular context of (23) the strongly deictic pronoun his is reinforced, not only verbally by autem, but also by some physical signal, for instance a pathetic gesture. The compatibility of strong deixis with autem may also help to explain the notoriously difficult example Pi. Poen. 1056-1057: (24) qui potuit fieri uti Carthagini gnatus sis? Hie autem habuisti Aetolum patrem ('How could you have been born in Carthage? Here (hie autem) you have an Aetolian father', PI. Poen. 1056-1057)

The presence of the putatively adversative connective autem in a clearly causal or explanatory clause has not failed to perplex commentators and grammarians. The most extreme position on this problem is that taken by De la Harpe (1923: 70). With regard to autem she argues that the particle came in the course of time to adopt a causal meaning, a process which reached completion only in late Latin. Among the four earliest examples of autem with an undisputably causal meaning she also mentions (24): "La nuance causale de autem est evidente: tu ne peux etre ne a Carthage, puisque tu as ici un pere etolien. " In my view, however, there is no need and no evidence for the quite unlikely assumption that an 'adversative' particle like autem may so easily adopt the role of a causal particle, at least not as far as (24) is concerned. We are obviously dealing here with the same combination of parallel focus (Carthagini vs. hic) and absolute focus (strong deixis: hic) as in (23), an environment that I have diagnosed as highly characteristic of locally used autem. The proper force of autem may, however, be obscured by the circumstance that (24) does not contain a parallel construction in a strict (syntactic) sense. However, if (24) had had the following form (24)' qui potuit fieri uti Carthagini gnatus sis, hie autem habeas Aetolum patrem? ('How can it be that you have been born in Carthage and here, on the other hand, have an Aetolian father?')

(in which also the autem-clause is governed by qui potuit fieri ut), one would probably have been less inclined to assign a causal meaning to autem, despite

238

§ 10.2.1

the fact that the meaning of (24) and (24)' is almost the same. The conclusion should be therefore that the two clauses in (24) do indeed maintain a causal relationship, but that this relationship is not explicitly signalled, and certainly not by autem. In the next section we will see that also some of the 'deviant' instances of absolute focalizing autem can be explained in this way, that is, as autem in combination with strong (verbal or physical) deixis. 10.2.2 autem as a marker of absolute focus Most reference books and grammars distinguish a separate type of autem in questions and exclamations (see section 9.2.1). The use of autem in interrogative clauses is exemplified in (25) and (4). Exclamations containing autem are quite rare. (25)

Le:

Etiam tu, ere, istunc amoves abs te atque ipse me adgredere atque ilia sibi quae hie iusserat mihi statuis supplicasque? Ar: Quem te autem divom nominem? Le: Fortunam, atque Obsequentem (,Come sir, get rid of that chap, won't you, and apply to me in person, yes, and let me have those statues and supplications he ordered for himself # Ah, and by what divine name do I address you (te autem)? # Fortune, or more precisely, Indulgent Fortune', PI. As. 714-716)

(4)

Th: Ius dicis Ep: Me decet Th: lam tu autem nobis praeturam geris? ('You speak like a judge # The proper thing for me! # What, do you already hold the praetorship?', PI. Epid. 25)

In the cases concerned, which for obvious reasons are found for the most part in the comic poets, autem does not seem to have a clear connective force. The labels and descriptions used in the literature concerned seem to point rather to an entirely different, 'modal' (or 'situating', see ch. 4) particle autem, the specific function of which - if this is indicated at all - is commonly stated in an intuitive, non-linguistic waylO: besides its general function of lending more

10. The literature consulted consists of the following works: the OLD; the Latin Dictionary by Lewis and Short; the TLL, s.v. autem; Hand (1829-1845, vol. I, s.v. autem); Langen (1880); Saur (1913: 24-5); the grammars by Kuhner-Stegmann and Szantyr; the lexicon of Plautus by Lodge (1904-1933); the lexicon of Terence by McGlynn (1963-1967); the study by Petersmann (1977) on the language of Petronius; Rosen (1989: 399).

autem as a marker of absolute focus

239

emphasis to the interrogative or exclamatory character of the clause, autem is held also to be responsible for the more specific emotions conveyed by the utterance, such as indignation, surprise, incredulity, admiration, rebuke, reproach, fear, anger, etc. On this point both the grammarian Priscianus (6th century AD) and Hand (vol. I, s. v. autem, p. 575) mention the similarity between autem and the Greek particle 07/ 11 • If this view could be shown to be correct (which I doubt), one would have to speak, within the theoretical framework of this study (see ch. 4 and 5), of two different particles autem: one with a presentational discourse function ('connective' autem), the other with an interactional discourse function ('modal' or 'situating' autem, signalling the attitude or specific illocutionary intention of the speaker). The major drawback of these accounts (apart from their vagueness as to the actual contribution of autem to the clause) is that they all sidestep the issue of how this particular use of autem is related to the connective, 'adversative' use which they all regard as primary or prototypical. In the following I will attempt to demonstrate that the discourse-pragmatic approach advocated in this study may have greater explanatory power on this point. I will argue, among other things, that the use of autem in questions and exclamations fits quite well with the above description of autem as a focusing device. At the beginning of this chapter I defmed autem as a presentational particle which marks the discrete status of a piece of information with regard to its verbal or non-verbal context. This means that autem is involved in singling out pieces of information which for some reason temporarily receive a special focus of attention. One of the preferred environments of a focusing device, as we saw in the preceding section, is the parallel construction in which a piece of information is compared to, or put into contrast with, another (in the case of autem a preceding) piece of information. We may accordingly speak of parallel or contrastive focus. Slightly different is the situation in which one of the contrasted elements remains implicit, for instance because its reference is self-evident in the context or situation. From this type of implicit contrastive (or parallel) focus it is only a small step to what I call 'absolute focus'. Absolute focus involves the singling out of pieces of information, not on account of their relational saliency (i.e. not on account of their relation with a specific corresponding or II. Prisc. in G.L. 3. 286-287: " or/ coniunctio tam completiva quam confirmativa invenitur apud illos ... quomodo apud nos 'vero' et 'autem'"

240

§ 10.2.2

contrasting referent which is present in the context or situation), but on account of their absolute prominence. Absolute prominence is assigned for instance to pieces of information that are contrary to expectation12 • In a way we could still speak of parallel or contrastive focus, the piece of information concerned taking its prominence from the implicit comparison with a more likely or preferred alternative 13 • This implicit alternative is often of a very general nature. We could also speak of a thwarted presupposition. Depending on the context the qualification 'contrary to expectation' may be further specified, giving rise to interpretations such as surprise, indignation, admiration, astonishment, impatience or anger. A related form of absolute prominence could be labelled 'emphatic identification'. The term accounts for strongly deictic expressions such as hie autem ('here, on the very spot, and nowhere else'), hoe autem ('this, and only this'), or tu autem ('you, and you alone'). The assumption that autem is a focusing device, a use resulting from the particle's basic function of discriminating and distinguishing, thus explains how autem can have both a relational (i.e. connective) and a non-relational value. Hence there is no need for the distinction of two different particles autem, one a presentational particle with a connective force, the other an interactional particle with an emotional force. A comparison with the personal pronoun tu, which is an inherently focal element, may again be illustrative in this respect. Latin tu (as well as the other subject pronouns, see above on pp. 235-236) may involve, just like the particle autem, relational as well as absolute focus: (i) relational (i.e. parallel or contrastive) focus: e.g. in case of ambiguity, when a 'you' is contrasted e.g. with an '1', or when there are more candidates who may be addressed by tu. (ii) non-relational (i.e. absolute) focus: in case of an emphatic appeal to the personal involvement of the tu-referent, with the meaning 'you, and no one else', 'you more than any other', and the like; or when the identity of the referent is contrary to expectation, in a context of surprise, indignation, anger, impatience, etc.: 'you, of all people ... ?'. The latter use of tu is found especially in interrogative clauses, cf. example (27) above.

12. Cf. the category 'counter-presuppositional focus' in Dik (1989). 13. Cf. also Giv6n (1990: 700): '''Contrast' is founded on the more general cognitive dimension of informational predictability or its converse, counterexpectancy".

autem as a marker of absolute focus

241

When we reconsider, in the light of the above, the instances of autem in emphatic interrogative clauses 14, we find that there is nothing remarkable or deviating about most of them. The majority of the examples cited in the literature appear to be rather straightforward instances of autem as a marker of contrastive focus in parallel constructions. In a further group of instances autem does indeed appear to behave as a genuine marker of absolute fOCUS I5 • I will discuss examples from both groups in turn. 10.2.2.1

autem as a marker of parallel focus in questions

An obvious example of parallel focus marking autem in questions is (26). The interrogative form of the clause containing autem is, in my opinion, irrelevant for explaining the role of autem here. (26) Quid MIHI ad defendendum dedisti, bone accusator? Quid HISCE autem ad suspicandum? ('What then have you given ME to refute, my worthy accuser? What grounds for suspicion have you given THESE GENTLEMEN?', Cic. S. Rosc. 58)

A comparable example of autem as a marker of parallel focus is the earlier mentioned example (25), although the underlying parallelism may seem at first sight (i.e. when the preceding context is not sufficiently taken into account) to be less apparent here: (25) (The slaves Libanus and Leonida are teasing their master Argyrippus: they are willing to hand over the money he needs, but only under certain, for the master quite humiliating, conditions): Ar: Datisne argentum? Li: Si quidem mihi statuam et aram statuis atque ut deo mi hie immolas bovem: nam ego tibi Salus sum Le: Etiam tu, ere, istunc amoves abs te atque ipse me adgredere atque illa sibi quae hic iusserat mihi statuis supplicasque? Ar: Quem TE autem divom nominem? Le: Fortunam, atque Dbsequentem (,will you give me the money? # Dh, well, if you put me up an altar and statue, yes, and offer me up an ox here the same as a god: for I'm your goddess Salva-

14. In the relevant literature (see note 10) altogether 75 instances are cited, most of them gathered from the comedies. 15. A third group of the question clauses concerned consists of instances of autem with a clear text-organizing function, i.e. autem as a marker of thematic discontinuity. Examples are Cic. Tusc.1.65; Ver. 3.208; Petro 95.1; 101.4; PI. Mer. 718. In 10.3 I will discuss this use of autem in more detail.

§ 10.2.2

242

tion, I am # Come sir, get rid of that chap, won't you, and apply to me in person, yes, and let me have those statues and supplications he ordered for himself # Ah, and by what divine name do I address YOU (te autem)? # Fortune, or more precisely, Indulgent Fortune', PI. As. 712-716)

After Libanus' announcement that he wants to be treated like a god with the name of Salvation, his companion Leonida declares that he, too, is a god. Argyrippus' reaction to Leonida's request can be paraphrased as follows: 'LIBANUS wants to be hailed as Salvation, but by what divine name do I address YOU?'. The role of autem in this example is fully in line with the general picture of autem as evolved in the previous section, and hence more feasible than e.g. the view of Langen (1880: 317), who with regard to (25) comments: "Zuweilen steht auch in solchen Satzen, welche inhaltlich, nicht bloss formal Fragen sind, autem, um der Frage Nachdruck zu geben (emphasis mine, CK); es lasst sich in diesen Fallen im Deutschen mit 'denn' wiedergeben" . Parallel focus may serve also as an explanation of the use of autem in the following three examples, which all involve a change of addressee. Compare the discussion on pp. 233-235 above. (27) (Chaerea enters in eunuch's dress; the slave Parmeno lists the opinions of the bystanders) Quid TU ais, Gnatho? Numquid habes quod contemnas? Quid TU autem, Thraso? ('What do YOU say, Gnatho? Anything to disprize there? And what do YOU (tu autem) say, Thraso?', Ter. Eu. 474-475)16 (28) (the soldier Antamonides to a man and woman who are in close embrace) Heus TU, tibi dieo, mulier, ecquid te pudet? Quid TIBI negotist autem cum istac? Dic mihi ('Hi, YOU! I mean you, woman! Have you no shame? (to man) And YOU! What is your business with that wench? Answer me!', PI. Poen. 1305-1306) (29) (Agorastocles and Lycus are having a discussion; Hanno interferes) Ag: Leno, eamus in ius Ly: Obsecro te, Agorastocles, suspendere ut me liceat Ha: (joining them) In ius te voco Ly: Quid TIBI mecum autem? (,Pimp, let us be off to court! # Do be merciful, Agorastocles, and let me hang myself # (joining them) I summon you to court! # You? And what have YOU (tu autem) got against me? (or: and what's that got to do with YOU?)', PI. Poen. 1342-1344)

16. Note that the paraphrase of the OLD (s.v. autem, 3a: 'moreover', 'furthermore') is not very revealing for the function of autem in this example.

autem as a marker of absolute focus

243

The strongest case for the assumption that autem has two widely differing functions (viz. not only a 'connective', but also a 'situating' or 'modal' function) could be provided by instances of autem-clauses in which no parallel or contrastive element whatsoever can be detected. The following example, which is quoted in almost all discussions on autem, does indeed seem to contain an unambiguous instance of such a clause: (30) (excerpt from a conversation between Alcmena and the god Jupiter, who has taken the shape of Alcmena's husband Amphitryon. Mercury every now and then tries to interrupt, thereby rousing the anger of Jupiter. The fragment starts when Jupiter is handing over a present to Alcmena) lu: Alcumena, tibi condono AI: Facis ut alias res soles Ecastor condignum donum, qualest qui donum dedit Me: Immo sic: condignum donum, qualest cui dono datumst lu: Pergin autem? Nonne ego possum, furcifer, te perdere? (' Alcmena, I give this as a present to you # (taking it) That is so like you! Oh, your gift just matches the giver! # Oh no, not the giver _ that gift matches the getter # You still go on autem? Is there no choking you off, you jailbird?', PI. Am. 536-539)

It is undeniably true that the expression pergin autem? does not contain an

explicit parallel or contrastive element. However, the example is very much comparable to the relatively large group of instances that are characterized by a marked change of addressee (cf. the examples 16 - 19, 25 and 27 - 29): Jupiter obviously breaks off for a moment his conversation with Alcmena and turns to Mercury. The difference with the more regular instances is that in (30) the expected 'contrastive' tu (pergin tu autem?) remains implicit. The correctness of this analysis is supported by the following fully identical example (31), in which the element tu does indeed occur: (31) (Demipho and Charinus are in love with the same girl. Charinus' friend Eutychus tries to talk Demipho out of it. Every now and then Euthychus' father Lysimachus interrupts the conversation between Eutychus and Demipho with a tormenting remark or some pedantic advice. At a certain moment Demipho gets fed up with this and interrupts his conversation with Eutychus to reprimand Lysimachus) De: Eutyche, ted oro, sodalis eius es, serva et subveni: hunc senem para cIientem; memorem dices benefici Ly; Ora ut ignoscat delictis tuis atque adulescentiae De: Pergin TU autem? Heia, superbe invehere (,Eutychus, I beseech you - you're his chum - save me, stand by me! Do take an old fellow under your protection; you'll say I remember a kindness # Beg him to overlook the vagaries of your hot young blood # You (tu autem) still go on? Ugh! The superior way you drop on me!', PI. Mer. 995-998)

§ 10.2.2

244

Summarizing thus far we can say that some of the 'deviant' instances of autem in interrogative clauses appear to be quite regular manifestations of the use of the particle as a marker of parallel focus. However, since one or both of the 'contrasted' elements may remain implicit this function of autem is often obscured 17 • The emotional (impatient, indignant, etc.) added value of the interrogative clause that is often felt to be present is due to all kinds of contextual factors and does not depend directly on autem (but is a side-effect; see 5.1 and 10.4.1 below). 10.2.2.2

autem as a marker of absolute focus in questions

The remaining instances of autem in emphatic questions can in my view all be taken together under the heading of 'marker of absolute fOCUS'18. There are various types of absolute focus, depending on the specific pragmatic motivation for singling out and highlighting a piece of information. In (32), for instance, autem helps to single out a particular element stated earlier in the preceding context by speaker 1, which is repeated and highlighted by speaker 2 on account of its surprising nature ('you really mean YOURS, you rascal?'Y9. We could speak of an emphatic or surprising identification: (32)

Ch: Ah, quid agis? tace Thr: Quid tu tibi vis? Ego non tangam meam? Ch: TUAM autem, furcifer? ('Ah, what are you about? Hold your tongue # What do you mean? Am I not to lay a finger on mine own? # YOURS (tuam autem), you scoundrel?', TeL Eu. 797-

798)

The element tuam in (32) bears emphatic focus because its reference is contrary to expectation. In the ironical context of (32) the emphatic repetition of part of the last speech of the other speaker leads eventually to the interpre17. Another example is PI. Aul. 819. A silly remark of the slave Strobilus is answered by Lyconides as follows: iamne autem, ut soles? ('now you're back (autem) to your old habits?'). The implicit idea with which the clause might be felt to be in contrast could be something like 'a few moments ago you appeared to be a sensible guy'. The example is clearly on the border between 'relational', contrastive focus and absolute focus. An identical example is PI. Truc. 695. 18. The unambiguous instances of autem as a marker of absolute focus are relatively rare: I diagnosed as such only 27 examples in the material provided by the relevant literature. 19. There are six more examples of this use in Terence: Ad. 185; 934; Hau. 251; Hec. 72; 100; Ph. 775. Plautus has only two such instances: Am. 901 and Ps. 305 (for the latter cf. Langen 1880: 316). Perhaps also Bac. 91 could be explained in this way, nihili picking up nul/us from the utterance of the former speaker. Note that autem does not necessarily cluster with the repeated element.

autem as a marker of absolute focus

245

tation of an indignant correction. The indignation has per se, however, nothing to do with autem20 • The use of autem exemplified in (32) is found also in classical Latin prose, in the rhetorical figure called epanorthosis (,self-correction'). Consider example (33?1: (33) Modo vos idem in A. Gabinium iudices sedistis. Numquis testis Postumum appellavit? Testis autem? Num accusator? ('You yourselves were lately empanelled in the case of Aulus Gabinius. Did any witness then mention Postumus? Witness, do I say (testis autem)? Did the prosecutor himself?', Cic. Rab. Post. 10)

Next, in the earlier cited example (4), we are dealing with another case of a surprising identification, this time without an echo of the preceding context. The element tu is contrary to expectation and hence receives full emphasis ('you, of all people, ... '): (4)

Th: Ep: Th:

Ius dicis Me decet lam TU autem nobis praeturam geris? (PI. Epid. 25)

The following example differs from (4) only in that it involves a 'wh-question' instead of a 'yes/no-question'. Again the full focus of attention is directed to the specific identity of one particular referent. There is also a certain resemblance to the epanorthosis example cited above in (33): (34) Ly: Sed adde ad istam gratiam unum Ph: Quid id est autem unum? ('But just one more favour, sir # Ah, and what is that one?', PI. Trin. 385)

Note that in (34) it is difficult to pinpoint the specific element to which the focalizing force of autem applies. Both the inherently focal quid ('what?')22 and the rather emphatic id unum (,that very one (sc. favour),) allow for being singled out in a special way by autem. It is not unusual, however, that autem, instead of clustering with one single, explicit parallel or contrastive constituent, appears to be related to several contrastive elements in the clause, or to an idea of contrastivity or parallelism that is present in a more general way. The

20. 21. Alt. 22.

For the opposite view cf. e.g. De la Harpe (1923: 66). Other examples of epanorthosis containing autem that are mentioned in the literature are Cic. 5.13.3; 6.2.1; 6.2.8; 7.1.4; Pis. fragm. 2; 91; Liv. 21.44.7; Plin. Pan. 28.1. Cf. e.g. Pi. Mos. 1016 qUid autem?

246

§ 10.2.2

same holds for (35)23: (35)

(Beginning of a new scene) Sed quid haec hie autem tam diu ante aedis stetit? (,But I wonder why this lady has been standing out in front of this house for so long?', PI. True. 335)

In (35) even more factors may be held responsible for the occurrence of autem. In addition to the inherently focal quid? ('why') and tam diu (,that long'), and the deictic hie ante aedis (,here, in front of this very house') and haec ('this woman here'), also the presence of transitional sed ('but') may help to explain the occurrence of autem24 • Latin sed ('but') has, among other things, the capacity to interrupt the current thematic chain and to shift the attention quite radically to a new discourse theme. Now a shift to a new discourse theme often goes hand in hand with a special type of parallel or contrastive focus construction. In this context 'contrastive' should be taken in the pragmatic and global (i.e. non-local) sense of 'alternate': a discourse theme evolves along one or more 'topical' chains; this implies that, when a theme shifts, one 'discourse topic' is replaced by (or alternates with) another 'discourse topic'. Especially when successive discourse topics belong to the same semantic class (e.g. both are individuals, both are locations, etc.) this may yield a parallel or alternate focus construction of the type 'about A ... ; as for B, on the other hand, ... '. The occurrence of the particle autem in such contexts is fully explainable, and not essentially different from its local use. In the following section I will deal quite extensively with this use of autem in 'global contrasts', that is, with autem in a text-organizing role.

23. Comparable examples of autem in a sed-clause are PI. Rud. 472, Ter. Ph. 601 and Verg. A. 2.101. As to PI. Rud. 472 KUhner-Stegmann (p. 78) comment: "sed bricht die Rede ab, autem drUckt die Verwunderung aus." In my opinion however the "Verwunderung"-element is to be considered a side-effect of autem (determined by the specific context), rather than a proper function or meaning of the particle. 24. It is noticeable that three of the six instances I gathered of quid-questions containing autem are introduced by transitional sed. The other two instances of quid-questions containing both sed and autem are Ter. Ph. 601 and Verg. A. 2.101. A comparable example of autem in a transitional quidquestion without sed is PJin. Ep. 9.29.2. Comparable to the combination of transitional sed and autem is, furthermore, the combination of ecce and autem. For the latter phenomenon see 10.3 below.

247

global use of autem

10.3 The global use of autem: autem as a marker of the organization of the text 10.3.1 parallel focus and thematic discontinuity

The second use of autem, which also emanates from the common discourse function of marking the discrete status of a piece of information, pertains to the thematic organization of the discourse. More specifically, autem may signal the discontinuity of one or more of the thematic strands along which the discourse evolves, thus bringing about a thematic reorientation. Very often such a discontinuity or reorientation is due to a shift of discourse topic. As noted earlier in this chapter (p. 228), 'contrast' or 'parallelism' may coincide with focus function as well as with topic function. The following example, which was discussed also in 10.2.1 above, clearly illustrates the closeness of the notions 'parallel/contrastive focus' and 'shifted discourse topic': (19) Nunc adeo hoc factust optumum ut nomine quemque appellem suo ... Advortite animum cunctae. Principio, Hedylium, ... Aeschrodora, tu ... Tu autem, Xytilis, Tu autem Phoenicium, (PI. Ps. 185-227) 00'

00'

00.

Keeping (19) at the back of one's mind, it is not difficult to see how one and the same particle (in this case autem) can function in some cases as a focus marking device with relatively local scope, and in other cases as a global marker of the organization of the text: in (19) the pieces of information accompanied by autem (viz. tu in vs. 209 and vs. 225) do not only count as parallel focus elements, but also serve as the alternate discourse topics along which the thematic segmentation of the discourse takes shape. There are other cases, however, in which it makes less sense to speak of a parallel or contrastive focus element, for instance when there is a considerable distance between the corresponding elements, or when the correspondence between the subsequent elements consists in no more than their being alternate discourse topics (i.e. no parallelism, no semantic contrast or relatedness). In such contexts autem has primarily a text-organizing role. Consider (36), where autem is used at the dividing line between two thematic sections: (36) URBANAE autem RES sic se habent (extensive description). Haec sunt, ut opinor, in re publica EGO autem (extensive description). Ac nostrae quidem rationis ac vitae quasi quandam formam, ut opinor, vides 00.

00.

248

§ 10.3.1 ('THE STATE OF AFFAIRS IN THE CITY (urbanae autem res) is as follows ... This, I think, is all the public news As for MYSELF (ego autem) ... There you have, I think, an outline sketch of my rule of life', Cic. Aft. 1.19.4-8)

Although text-organizing autem often coincides with a shift of discourse topic, I prefer to call autem a marker of thematic discontinuity rather than a marker of shifts of discourse topic. The following excursus on the notions thematic (dis )continuity and discourse topic will serve by way of explanation25 • Discourse is based on the exchange of information on some 'theme'. The theme can be announced explicitly, for instance in an introductory utterance. More often however the theme remains implicit. An extended mono logical discourse is commonly composed of several hierarchically ordered segments, which are each characterized by their internal thematic coherence. Thematic coherence, in turn, is based on the continuity or recurrence of certain pieces of information. Within a thematically coherent discourse segment various thematic strands (or chains) may be discerned, along which the discourse evolves. The most pervasive of these strands is referent continuity, that is, the recurrence of a particular nominal participant, most commonly the subject. In the tradition of text-based linguistics the term topic or discourse topic is reserved for these recurrent referents which play such an important role in the development of the discourse theme. Their important status is reflected in the fact that in most languages they get a grammatical coding, for instance in the form of subject marking. In addition one can often distinguish one or more other thematic strands which are operative at the same time, usually of minor importance. In narrative discourse, for instance, thematic coherence usually does not depend on referent continuity only, but also on location continuity, temporal continuity, action-event continuity, and the like26 • It is to be noted that the elements that make up these other, non-referential thematic strands are usually not referred to as 'topics' or 'discourse topics'. Sometimes we find the term 'setting' used for them. Each interruption of a thematic strand (whether it concerns a referential discontinuity or a discontinuity of one of the other possible thematic strands) in

25. The view presented here is largely in accordance with the view of Giv6n (1990). 26. In other text types (e.g. philosophical, argumentative or technical prose) one might likewise take logical or argumentative continuity into consideration.

parallel focus and thematic discontinuity

249

essence may count as a thematic discontinuity and hence - in terms of the organization and segmentation of the discourse - to a discourse boundary. When a relatively 'heavy' thematic chain is interrupted27 , or when several different chains are interrupted and shifted at the same time, this will usually result in a major discourse boundary. In written discourse this may be represented graphically by a paragraph mark or a new chapter. In other cases a shifted thematic chain may yield only a minor transition and discourse boundary, for instance when a thematic chain of minor (e.g. more local) importance is interrupted while other thematic strands are continued. As for Latin autem we can say that its text-organizational use is not confined to the signalling of shifts of discourse topic (i.e. the signalling of referential discontinuity) only28 , but that the particle may occur with any form of thematic discontinuity, and irrespective of whether it concerns a minor or a major thematic shift. In its text-organizing use autem to some extent resembles a number of other presentational particles, like for instance nam (see ch. 7) and igitur: it marks off units of the discourse and indicates their position within the overall discourse structure (cf. section 4.2.2.3). The units of discourse concerned, however, are in the case of autem to be defined in terms of thematic coherence and not (as appeared to be the case with nam, see chapter 7) in terms of the (linguistic) action structure. In other words, autem does not function primarily to link up communicative acts or configurations of acts on the level of the move. Its structuring role within the complex move is based on a different (viz. thematic) coherence principle. The effect of both types of presentational particle may ultimately however be the same. Both may for instance help to create an extended monologue with a relatively high degree of internal coherence. In the remainder of section 10.3 I will discuss the text-organizing use of autem on the basis of the various types of thematic discontinuity that may in general be distinguished.

27. A 'heavy' thematic chain involves the highly persistent recurrence of one and the same piece of information, mostly a nominal referent. 28. I use the term 'shift of discourse topic' for all forms of referential discontinuity, including 'shifts in perspective' (alternation of subjects).

250

§ 10.3.2

10.3.2 autem with shift of discourse topic As I noted above (see p. 248), thematic discontinuity is often due to referent discontinuity or, in other words, to discourse topic shifts in a narrow sense. We have also seen that the discourse function of autem is highly compatible with discourse topic shifts. In this section I will discuss the large group of instances of autem that coincide with such a shifted nominal referent. For the sake of presentation the material has been divided into two groups. First (10.3.2.1) I will deal with the large majority of cases in which the shifted discourse topic is given or accessible information. In such cases autem usually follows the shifted topic element, which - in accordance with its particular pragmatic function - occupies the first position in the clause (see below, pp. 257-261). The second, quite marginal group consists of cases in which the shifted topic element is new information and therefore tends to be placed later on in the clause (10.3.2.2). At several points in the discussion I will emphasize that there is no clear-cut distinction between instances of parallel focus (cf. 10.2 above) and instances of alternate or shifted topic. The distinction is rather a matter of degree: the determining factors (semantic and structural parallelism, as well as the relative textual distance between the corresponding elements) are to be taken as gradient phenomena. 10.3.2.1

autem with given and accessible discourse topics

An element counts as given information when it has already been explicitly referred to in the immediately preceding context. Likewise one might speak of accessible information when an element - instead of having been explicitly introduced - is inferrable from another element in the preceding discourse, or from the non-linguistic discourse situation. In Latin, as well as in many other languages, given information tends to be expressed by a pronoun or by zeropronominalisation29 , accessible information Gust like new information) by a full nominal phrase.

29. 'Zero-pronominalisation' of an element (most often the grammatical subject) means that the element is only verbalised to the extent that it is implicit in the personal ending of the finite verb.

autem with shift of discourse topic

251

shifted discourse topic introduced in preceding clause

The following two examples illustrate the use of autem in combination with a shifted given discourse topic. In the passage cited in (37) the status of discourse topic is to be assigned first to the protagonist Trimalchio. The focus of attention then shifts temporarily to the boy Croesus (puer autem), who has already been introduced as a focus element in the preceding clause (delicias suas ... quem Croesum appel/abat). After the interlude with the boy the focus of attention is again directed to Trimalchio (resumed discourse topic?o: (37) Nec non TRIMALCHIO ipse cum tibicines esset imitatus, ad delicias suas respexit, quem Croesum appellabat. PUER autem lippus, sordidissimis dentibus, catellam nigram atque indecenter pinguem prasina involvebat fascia panemque semesum ponebat supra torum [atque] ac nausea recusantem saginabat. Quo ... admonitus officio TRIMALCHIO Scylacem iussit adduci ... (,TRIMALCHIO himself also, after imitating a man with a trumpet, looked round for his favourite, whom he called Croesus. THE BOY (puer autem) had bleary eyes and very bad teeth, and was tying up an unnaturally obese black puppy in a green handkerchief, and then putting a broken piece of bread on a chair, and cramming it down the throat of the dog, who did not want it and felt sick. Because of this TRIMALCHIO remembered his duties, and he ordered them to bring in the dog Scylax', Petro 64.5-6)

We may say that the use of autem in (37) is one of the possible linguistic means of highlighting the organization of the passage cited: on account of its distinguishing and demarcating discourse function autem emphatically marks off the current thematic unit from the preceding one. Autem is an explicit signal for the reader that some thematic chain is being interrupted (viz. Trimalchio and his activities), and another one started (viz. the activities of the boy). Another linguistic clue to the organization of the passage in (37) can be deduced from the variation of the tenses used: Trimalchio's activities are represented in the perfect tense (respexit, iussit), the activities of the boy in the imperfect tense (involvebat, ponebat, saginabat)31. A comparable example is (38): 30. Note that the neutral puer ('the boy'; cf. homo, 'the man') in this context virtually equals the anaphorically used pronoun ille. Zero-pronominalisation would lead to a lack of clarity in the case of switch reference and is commonly confined to topic continuity. 31. Note that the imperfect appel/ahat is of a different order from the other imperfect verb forms: the state of affairs referred to is habitual and not part of the string of activities taking place in the discourse situation described (that is, it is not part of the narration proper). For the text- structuring function of the perfect/imperfect alternation in Latin narrative texts, cf. Chausserie-Lapn!e (1969), Bolkestein (1987), and Pinkster (1990: 254).

252

§ 10.3.2

(38) (One of Hegio's sons has been taken prisoner in a battle with the Eleans. In order to recover the captured boy his father buys up two Elean prisoners of war) Coepit captivos commercari hie Aleos, si quem reperire possit qui mutet suom, ilium captivom ... Reconciliare ut facilius posset domum, emit hose' de praeda ambos de quaestoribus. Hisce autem inter sese hunc confinxerunt dolum, ... (,HEGIO then began to buy up Elean captives, hoping to get hold of one that he could exchange for his son - the captive son, that is ... So in the hope of getting that son back home more readily he bought both of these prisoners from the commissioners who were disposing of the spoils. THESE SAME PRISONERS, however (autem), have got together and laid a scheme ... ', PI. Capt. 27-35)

Here autem again coincides with a shift of discourse topic (viz. from Hegio to the prisoners). The shifted discourse topic hisce ('these') has already been introduced in the preceding context (see e.g. hosce in the preceding clause). As to the information units marked off by (and within the scope of) autem in (37) and (38), we might say that they are larger than a single constituent (compare the use of autem as a marker of parallel focus), and that they may even amount to section-like segments. The presence of a pair of 'contrasted' nominal referents (e.g. Trimalchio vs. puer in ex. 37) is admittedly significant for the occurrence of autem, but only in so far as they are the pivotal elements of more extensive thematic units. The impact of autem in its text-organizational use clearly goes beyond the mere singling out of topical referents. This can be proved by passages in which autem does not cluster (or clusters less clearly) with a topical element, as will be illustrated later in this chapter (see p. 266-267). I will therefore speak of 'thematic units' rather than of 'information units' when dealing with autem used globally. announced alternation of discourse topics

Slightly different are examples in which the occurrence of alternate discourse topics has been previously announced. Such an announcement, as well as th~ use of autem, can be explained as an attempt by the author to make the thematic organization of a discourse as transparent as possible. The following three examples serve as illustration32 . In a way we can still speak of 'parallel'

32. Similar examples are Cic. Ac. 2.15; Ver. 5.7; Liv. 7.29.3; 33.14.12. A special instance of an 'announced' shift of discourse topic is Cic. Fin. 5.4, in which the preparatory particle quidem anticipates the occurrence of another (as yet unspecified) discourse topic: POMPONIUS quidem, inquam, noster iocari videtur, et fortasse suo iure. Ita enim se Athenis collocavit ut sit paene unus ex Atticis, ut id etiam cognomen videatur habiturus. EGO autem tibi,

autem with shift of discourse topic

253

text parts, considering that the descriptions of the activities of the subsequent discourse topics run largely in parallel (semantic parallelism). (39) Haec eodem tempore Caesari mandata referebantur, et LEGATI AB AEDUIS ET A TREVERIS veniebant: AEDUI questum quod Harudes ... finis eorum popularentur; sese ne obsidibus quidem datis pacem Ariovisti redimere potuisse: TREVERI autem pagos centum Sueborum ad ripam Rheni consedisse ... ; his praeesse Nasuam et Cimberium fratres. Quibus rebus Caesar vehementer commotus ... ('At the same time as this message was brought back to Caesar, DEPUTIES arrived FROM THE AEDUI AND THE TREVERI. The AEDUI came to complain that the Harudes ... were devastating their borders, and that they themselves had not been able to purchase peace from Ariovistus even by the delivery of hostages. The TREVERI, on the other hand (autem) , reported that one hundred cantons of the Suebi had settled on the banks of the Rhine ... ; that they were under the command of two brothers, Nasua and Cimberius. At this Caesar was exceedingly disquieted .. .', Caes. Gal. 1.37.3-4) (40) In exitu iam annus erat, et ambitio magis quam unquam alias exarserat consularibus comitiis. Multi et potentes petebant PATRICII PLEBEIQUE: P. Cornelius Cn. filius Scipio, qui ... et L. Quinctius Flaminius, qui ... et Cn. Manlius Vulso, hi PATRICII; PLEBEII autem C. Laelius, Cn. Domitius, C. Livius Salinator, M'. Acilius ('It was now the end of the year, and canvassing at the consular election was more spirited than ever before. The candidates were many and influential, PATRICIANS AND PLEBEIANS, P. Cornelius Scipio, the son of Gnaeus, who ... and L. Quinctius Flaminius, who ... and Cn. Manlius Vulso; these were the PATRICIANS. The PLEBEIANS now (autem) were C. Laelius, Cn. Domitius, C. Livius Salinator and M'. Acilius', Liv. 35.10.1-3) (41) (A father and a son are in love with the same girl) Nunc sibi uterque contra legiones parat, PATERQUE FILIUSQUE, clam alter alterum: PATER adlegavit vilicum qui posceret sibi istanc uxorem: is sperat, si ei sit data, sibi fore paratas clam uxorem excubias foris; FILIUS is autem armigerum adlegavit suom, qui sibi eam uxorem poscat: scit, si id impetret, futurum quod amat intra praesepis suas ('And now the pair of them, FATHER AND SON, are mustering their opposing legions, each without the other's knowledge: THE FATHER has commissioned his bailiff to ask the girl in marriage: he hopes that if the bailiff does get her, he himself will have waiting for him, unbeknown to his wife, a night watchman's berth away from home; THE SON, for his part (autem) , has commissioned his orderly to ask her in marriage: he knows that if the orderly should obtain her, he himself will have the object of his affections inside his own stall', PI. Cas. 50-57)

Piso, assentior usu hoc venire, ut acrius aliquanto et attentius de claris viris locorum admonitu cogitemus. This more global, text organizational use of the quidem ... autem construction (a whole sentence intervenes between quidem and autem) is quite exceptional, in contrast to the equivalent Ancient Greek /lip ... Oi construction. For the above example cf. also Solodow (1978: 35). For the text organizational use of the Ancient Greek /lip ... Oi construction cf. Bakker (1993).

254

§ 10.3.2

In a sense, examples like (42) below also belong in this category. Although the alternate subtopics postis ('the door-post'), aditus ('the entrance') and limen ('the doorway') have, strictly speaking, not been explicitly announced, they can be inferred rather easily from the preparatory 'supertopic' ianua ('the door'). In cases like this we speak of accessible rather than of given topics: (42) cum Agamemnone ad IANUAM pervenimus, in cuius POSTE libellus erat cum hac inscriptione fixus: "Quisquis servus sine dominico iussu foras exierit, accipiet plagas centum." In ADITU autem ipso stabat ostiarius prasinatus, cerasino succinctus cingulo, atque in lance argentea pisum purgabat. Super LIMEN autem cavea pendebat aurea, in qua pica varia intrantes salutabat ('We came with Agamemnon to THE DOOR. On THE DOORPOST a notice was fastened: "No slave to go out of doors except by the master's orders. Penalty, one hundred stripes." Just at THE ENTRANCE (in aditu autem) stood a porter in green clothes, with a cherry-coloured belt, shelling peas in a silver dish. In THE DOORWAY (super limen autem) hung a golden cage, and a black and white magpie in it greeted visitors', PetL 28.6-9)

series of alternate discourse topics Characteristic of the alternate discourse topics in (39) - (42) above is that they are, so to speak, subtopics of one covering 'supertopic', which explains why they can be announced together or referred to as a group. Alternations of subtopics may also involve more than two or three elements and as such may have the form of a list or catalogue. It is not surprising that autem is prone to occur in the environment of such series and lists: the explicit demarcation of the separate components (cf. the discourse function of autem) can be regarded as a feature highly characteristic of series and lists. Series and lists mayor may not be preceded by an announcement. I give only a few examples of the quite frequent use of autem in series. The first one is taken from Plautus: (43) OMNES festinant intus totis aedibus, SENEX in culina clamat, hortatur coquos: " ... ". VILICUS is autem cum corona, candide vestitus, lautus exornatusque ambulat. ILLAEC autem arrnigerum ilico exornant duae quem dent pro Casina nuptum nostro vilico ... ; digne autem COQUI nimi' lepide ei rei dant operam, ne cenet senex ... : illarum oratu faciunt; ILLAE autem senem cupiunt extrudere incenatum ex aedibus, ut ipsae solae ventris distendant suos (,EVERYONE is bustling about all over the house, THE OLD MAN (senex) is clamouring in the kitchen, urging on the cooks" ... ". As for THE BAILIFF (vilicus is autem), he is parading around with a garland and white clothes on, all spick and span. And THE TWO LADIES (illaec autem) - they're in a bedroom decking out the orderly to be our bailiff's wife in place of Casina ... ; and then THE COOKS

autem with shift of discourse topic

255

(autem coqui) , too, are doing their part, and, my! the lovely way they work to keep the old man from dining! ... They do anything the ladies ask. As for THEM (illae autem), they are bent on driving the old man out of the house without his dinner, so that they can swell their own stomachs all by themselves', PI. Cas. 763777)

With regard to autem in verse 775 (illae autem) , De la Harpe (1923: 68) comments that it shows the particle at a stage of development intermediate between an adversative and a causal particle: "IUae autem senem cupiunt extrudere explique illarum oratu faciunt du vers precedent; sans cette explication l'attitude des cuisiniers et l'ordre auquel ils obeissent seraient incomprehensible, parce que La raison nous en echapperait". However, I would again contend (see also p. 237) that the account of autem given in this chapter makes unnecessary De la Harpe's quite unlikely solution. There may indeed be an explanatory relation between the autem-clause and the preceding clause, but this relationship is not indicated by autem. Autem appears to be used in exactly the same way as in the preceding verses 767. 769 and 772, viz. as a marker of thematic discontinuity, due to a discourse topic shift. Its occurrence in 775 is fully in line with the catalogue-like nature of the whole preceding passage, and does not call for a special explanation. A comparable example from Cicero is (44)33. First there is an announcement of the 'supertopic' of the passage: the enormous area of the city of Syracuse, which appears to consist of four separate cities. Then the author elaborates on each of these four 'cities'. Two of the four subtopics are marked byautem: (44) Ea tanta est urbs ut ex QUATTUOR URBIBUS maximis constare dicatur; quarum UNA est ea, quam ... (extensive description). ALTERA autem est urbs Syracusis, cui nomen Achradina est ... TERTIA est urbs quae ... QUARTA autem est quae .. ('So large is the city that it is described as being FOUR GREAT CITIES joined together. ONE OF THESE (una) is the Island ... Then there is A SECOND TOWN (altera autem) in the city, called Achradina ... There is A THIRD TOWN (tertia est urbs), ... And there is A FOURTH TOWN (quarta autem est), which .. .', Cic. Ver. 4.118-119)

The impression of an enumeration or series is even stronger when an enumerative adverb such as tum ('then') is added. Consider (45): (45)

TUM PISTORES scrofipasci, qui alunt furfuribus sues ... , eorum si quoiiusquam scrofam in publico conspexero, ex ipsis dominis meis pugnis excu1cabo furfures ...

33. Cf. e.g. also Cic. Div. 1.93-94; Ver. 3.66.

256

§ 10.3.2 TUM PISCATORES, qui praebent populo piscis foetidos eis ego ora verberabo surpiculis piscariis ... TUM LANII autem, qui concinnant liberis orbas ovis ... reddam mortalis miserrumos (' And as for THE MILLERS (tum pistores) that feed sows on bran ... , if I spy a sow of anyone of them on the public highway, I'll up with my fists and stamp the stuffing out of those sows' owners ... THEN THE FISHMONGERS (tum piscatores) that offer folk stale fish ... , I'll whack their faces with their own fish baskets ... THEN THE BUTCHERS, TOO (tum lanii autem) , that bereave sheep of their little ones ... , I'll make them the saddest men alive!', PI. Capt. 807-822)

Discourse topic shifts of the type illustrated in (43) - (45) above (i.e. series of alternate discourse topics) usually involve minor thematic breaks, which obtain at a relatively local level of the text. A rare example of a series of subtopics which is to be described in terms of the macrostructure of the text is (48). In this example the successive discourse topic shifts (which have been explicitly announced by exponam enim vobis, Quirites, ex quibus generibus hominum istae copiae comparentur ('I shall show you, citizens, from what classes of men those troops are procured'» all seem to coincide with section boundaries (the textual distance between them is relatively large). The last member of the series is marked by autem: (46)

§ 17. Exponam enim vobis, Quirites, EX QUIBUS GENERIBUS hominum istae copiae comparentur ... § 18. UNUM GENUS est ... § 19. ALTERUM GENUS est ... § 20. TERTIUM GENUS est ... § 21. QUARTUM GENUS est ... § 22. QUINTUM GENUS est .,. POSTREMUM autem GENUS est ... ('I shall show you, citizens, from WHAT CLASSES OF MEN those troops are procured .. ,. § 18. THE FIRST CLASS ... § 19. THE SECOND CLASS .,. § 20. THE THIRD CLASS .. , § 21. THE FOURTH CLASS ... § 22. THE FIFTH CLASS ... THE LAST CLASS autem .. .', Cic. Catil. 2.17-22)

At this point in the discussion it is interesting to recall example (10), in which autem was described as a marker of parallel focus at clause level. (10) currenti, properanti hau quisquam dignum habet decedere. Ita TRES simitu RES agendae sunt, quando unam occeperis: et CURRENDUM et PUGNANDUM et autem IURIGANDUM est in via (PI. Mer. 117-119)

The resemblance to the more global, text-organizational use of autem (especially in series) is very apparent here, and can be taken as support for the assignment of a single discourse function to the particle autem. The difference between the two uses appears to be merely a matter of scope: in its local use autem marks off information units (i.e. constituents), in its global use thematic units.

autem with shift of discourse topic

257

Another instance of a local series in which autem is used to mark off one of the individual information units is (47)34: (47) Deglupta maena, sarrapis sementium, manstruca, halagora, sampsa, tum autem plenior ali ulpicique quam Romani remiges ('You excoriated sardine and semi-sarrapian, you pelt, saltsouk and olive pulp, and (tum autem) reeking of garlic and leeks worse than a bench of Roman rowers!', PI. Poen. 1312-1314)

shifted discourse topics and their position in the clause In the above examples we have seen quite a number of autem-clauses which consist of an initial nominal constituent followed by autem and then by the rest of the clause. The frequent combination of autem with a fronted noun is no coincidence. Rather it is an important distributional property of autem, which can be taken as co-indicative of the discourse function of the particle. It has been observed by De Jong (1989) and Jones (1991) that in Latin shifts of discourse topic are typically associated with clause-initial position of the discourse topics concerned, provided that the topical elements are 'given' or at least 'accessible' information35 • The correlation between sentence-initial position and given, topical information is not unique to Latin, but seems to hold for a large number of other languages as well. The common psycholinguistic explanation for this phenomenon is that the fronting of given information enhances the continuity of discourse and hence reduces the processing load for the hearer/reader. Given or accessible information functions, so to speak, as a familiar framework into which the new information can be integrated rather smoothly. Fronted discourse topics can thus be regarded as convenient 'headings' for what is to come. The discourse function of autem is highly compatible with this 'heading' function of fronted topics: both operate at transition points in a discourse and enhance its coherence in the case of a thematic discontinuity. The 'heading' function of fronted discourse topics may be made more explicit by for instance the addition of autem, by a certain syntactic independence pf the topical element from the rest of the clause, or by a combination of both. The investigated sample contains several instances of autem-clauses in

34. Cf. also PI. St. 213; Ter. Hau. 38; Cic. ruse. 4.5. 35. In Latin the fronting of given discourse topics is confined to shifted discourse topics, continuous discourse topics as a rule being zero-pronominalized.

258

§ 10.3.2

which a fronted topic has somehow been detached from the rest of the utterance 36 . I give three examples37 . In (48), which is the continuation of the passage quoted in (41), the discourse topic ille (which picks up senis from the preceding clause) alternates with the preceding discourse topic UXO?8. Its 'heading' function is emphasized by the insertion of the temporal subclause introduced by postquam, and by the postponement of pater, both of which features contribute to the impression that we are dealing with a 'loose' beginning of the sentence. (48) SENIS UXOR sensit virum amori operam dare, propterea una consentit cum filio. ILLE autem postquam filium sensit suom eandem illam amare et esse impedimento sibi, hinc adulescentem peregre ablegavit pater ('THE OLD MAN'S SPOUSE has discovered that her husband is engaged in a love affair, and therefore espouses her son's cause. HE, however (ille autem), discovering his son to be in love with that same girl and in his way, has sent the young fellow abroad, the father, that is', PI. Cas. 58-62)

Likewise in (49), the fronted discourse topic contagio ... belli is placed outside the curl-clause. Note, incidentally, the neat and well-balanced structure of the passage, the alternate discourse topics aditus and contagio having been announced in the preceding clause aditum jacilem hostibus an contagionem imitandi belli periculosam fuisse?: (49) Etenim propinquitas locorum ad utram partem hoc loco profertur? Utrum ADITUM facilem hostibus an CONT AGIONEM imitandi belli periculosam fuisse? ADITUS omnis ... (rejection of this suggestion). CONTAGIO autem ista servilis belli cur abs te potius quam ab iis omnibus qui ceteras provincias obtinuerunt praedicatur? (,What, indeed, is the nearness of the two countries (viz. Italy and Sicily) here adduced to prove? That the ENTRANCE to Sicily was an easy one for our enemies, or that there was a danger of THE INFECTION of revolt spreading to it? THE WAY into Sicily ... (rejection of this suggestion). And as for your INFECTION (contagia autem) of the slave revolt, why (cur) should that be brought up by you any more than by anyone of the governors of all our other provinces?', Cic. Ver. 5.6-7)

36. In Functional Grammar the pragmatic function Theme is reserved for constituents that have no syntactic relation with the subsequent clause they belong to semantically (Dik 1978; 1989: 135). For Latin Theme constructions cf. Hoffmann (1989), Pinkster (1990: 37; 171) and Somers (1994). The Theme-construction can be regarded as the most extreme fonn of a fronted topic. 37. Some other examples are: PI. Men. 34; Caes. Civ. 2.31.4; Cic. Aft. 2.16.4; Sen. Ep. 4.10. 38. For the combination ille autem see below on p. 274.

autem with shift of discourse topic

259

In (50) the constituent de Acutiliano autem negotio ('as to Acutilius' business') is emphatically placed in sentence-initial position, although it cannot be decided whether the constituent is syntactically displaced from the rest of the clause or not: (50) De Acutiliano autem negotio quod mihi mandaras, ut primum a tuo digressu Romam veni, confeceram (' As to Acutilius' business I settled things according to your directions, as soon as ever I got to Rome after your departure', Cic. Att. 1.5.4)

Example (50) is representative of a number of instances in Cicero's correspondence where the principle that fronted topics are typically 'given' or 'accessible' information might seem, at least at first, not to apply: Acutilius or 'his business' have not been mentioned earlier in the letter from which the passage has been taken, nor are they inferrable from the prior discourse topic or some other element in the preceding context. However, the topic may still count as 'accessible' in so far as it was brought up in an earlier letter by Atticus, to which this letter apparently forms a reaction (cf. the explicit quod mihi mandaras, 'what you instructed me to do'). As a consequence of the lack of direct semantic and pragmatic relatedness of the alternate discourse topics in such cases, the thematic breaks and ensuing discourse boundaries involved are relatively large. Hence we can say that in this particular context autem plays a role in indicating the macrostructure of the discourse: it functions more or less as an introducer of new sections or chapters. Example (50) is not an isolated instance of 'macrostructuring' autem in Cicero's correspondence39 • Another nicely-illustrative example is cited in (51), part of which has also been cited in (36). The structure of the letter is quite straightforward and has been made explicit by various means: an introduction with an announcement of the two main discourse topics; the addition of autem at important discourse boundaries; and the use of sectionclosing formulas.

39. In the first book of the Letters to Atticus II of the 41 instances of autem are of the type illustrated in (50) (de Acutiliano autem).

260

§ 10.3.2

(51) Overview of the structure of Cic. Aft. 1.19: introductory part

announcement of the two main discourse topics of the letter: (i) politics; (ii) Cicero's own affairs

part one

main body of the letter politics d-topic (i) a) external affairs (the imminent war in Gaul) b) internal affairs (discussions of the agrarian law) Cicero's own affairs d-topic (ii)

part two

additional issues: exchange of personal and practical information between Cicero and Atticus Atticus' business d-topic (iii) Atticus' opinion of Cicero's literary producd-topic (iv) tion Cicero's brother Quintus d-topic (v) d-topic (vi) Cossinius

It is noticeable that autem occurs at three important thematic boundaries of the letter: between discourse topic (i) a) and (i) b); at the transition from discourse topic (i) to discourse topic (ii); and at the transition from discourse topic (ii) to discourse topic (iii), which is also the transition from the main body of the letter to the heterogeneous collection of minor topics with which the letter ends. It is significant that these major breaks are explicitly prepared for also by means of section closing expressions: (51)' § 4-5 URBANAE autem RES se sic habent: ... (extensive description). Haec sunt, ut opinor, in re publica § 6-8 EGO autem ... (extensive description). Ac nostrae quidem rationis ac vitae quasi quandam formam, ut opinor, vides § 9 DE TUO autem NEGOTIO saepe ad me scribis ... ('(§ 4-5) THE STATE OF AFFAIRS IN THE CITY (autem) is as follows: ... This, I think, is all the public news. (§ 6-8) As for MYSELF (autem) ... There you have, I think, an outline sketch of my rule of life. (§ 9) You keep writing about THAT BUSINESS OF YOURS (autem) .. .', Cic. Att. 1.19.4-9)

The use of autem illustrated in (50) and (51) (i.e. autem with fronted topics which seem at first sight to convey new information) is rare outside Cicero's

autem with shift of discourse topic

261

correspondence40 • An instance from Petronius is (52), where a new stage in the description of a banquet at Trimalchio's house begins as follows 41 : (52)

Petauristarii autem tandem venerunt (,But at last the acrobats came in', Petf. 53.11)

Despite the fact that the discourse topic petauristarii ('acrobats') has in no way been anticipated in the preceding context, it may count as accessible information in so far as the appearance of acrobats is to be expected in the context of a banquet such as the one described here. The addition of tandem ('at last') seems to confirm this analysis. In the following section I will deal with the use of autem in clauses in which an indisputably new discourse topic is introduced. In such clauses the discourse topic involved tends to be placed non-initially. 10.3.2.2

autem with new discourse topics

Having described the use of autem with fronted discourse topics, I now turn to the quite marginal use of autem (in less than 1 % of the instances in my sample) in presentative sentences, that is, in sentences in which brand new entities are introduced into the discourse42 • Presentative sentences usually contain verbs like existential esse ('to be'), venire ('to come', 'to arrive'), apparere ('to appear'), verbs of perception, devices such as ecce ('look! ') which indicate sudden developments and unexpected arrivals, and the like. The newly introduced entity, which is to acquire discourse topic status in the ensuing discourse, is typically not sentence-initial, the first position in the

40. Note that the fronted topics in (51) (urbanae res, ego, de tuo negotio) are to be considered accessible rather than brand new topics: the political situation inside and outside Rome, Cicero's own situation and Atticus' personal affairs, are recurrent (and hence expected) themes in Cicero's letters to Atticus. 41. See e.g. also Cic. Ac. 1.39, where a preceding section is concluded by the closing formula haec Jere de moribus ('so much about the mores'), while the next section is introduced by De naturis autem sic sentiebat ('His views as to the natural substances were as follows'). The sentence-initial position of de naturis is no doubt to be explained by the correspondence with de moribus in the preceding clause. 42. The difference between given, accessible and new information is not always clear-cut. In the present section I am concerned only with brand new information (cf. Prince 1981). Presentative constructions are, on the whole, not very frequent in actual discourse. An explanation may be that people tend much more to talk about familiar subjects and persons (i.e. information that belongs to the 'shared knowledge' of the communication partners), than to bring up brand new topics which need an explicit introduction.

262

§ 10.3.2

clause usually being reserved for the presentative verb. I give two examples. Compare also example (18) in chapter 9, p. 219. Example (53) is fully comparable with (51) above (with autem at sectionboundaries), except that the shifted discourse topic C. Herennius is explicitly presented as a brand new element in the text. Fronting of the shifted topic is therefore not very likely. (53) Est autem C. Herennius quidam, tribunus pI., quem tu fortasse ne nosti quidem ... (,There is one C. Herennius, a tribune - you may not even know him ... ', Cic. Aft. 1.18.4)

Compare also example (54)43: (54) Quare de hoc quoque genere pastionis dicendum censui. (New section) Est autem id fere vel in villa, vel circa villam. In villa est quod appellant Graeci OPI'tOWI'OIt; KOIt npWTEpEWI'OIt; .... Rursus circa villam ... ('Therefore I have thought it fitting that I should speak also of the keeping of this kind of animal. (New paragraph) Autem the following kinds of pasturing of animals are generally carried on either at the farm or in its neighbourhood. At the farm there are what the Greeks call ornithonas (,poultry-houses') and peristereonas (,dovecotes') ... On the other hand, in the neighbourhood of the farm .. .', Col. 8.1.2-4)

Examples like these argue against a treatment of autem in terms of a specific semantic relationship between successive clauses: autem-clauses like these are characterized rather by a marked absence of any semantic relationship with the preceding context. If one chose to persist in calling autem a marker of adversative relationships one would be obliged, in view of examples such as (53) and (54), at least to provide this qualification with a clear discourse pragmatic definition.

43. I depart deliberately from the translation in the Loeb edition, which takes the autem-clause as contrastively related to the previous clause and starts a new section with the sentence In villa est .... It is in my opinion beyond doubt that the clause est autem Jere in vida vel circa villam functions as an introductory announcement of what is to be elaborated in the subsequent sentences In villa est ... and Rursus circa villam .... The pronoun id should accordingly be taken as a cataphoric rather than as an anaphoric pronoun. Regarding autem as a full-blown adversative conjunction marking a semantic relationship with the preceding clause (as the Loeb translation seems to do) would, moreover, be contrary to the common discourse function of autem as expounded in this chapter, and would call for a rather forced interpretation in the context concerned. Note that rursus in (54) functions as an equivalent of autem.

autem with other types of thematic discontinuity

263

10.3.3 autem with other types of thematic discontinuity 10.3.3.1

autem with shifted 'settings'

It was stated above (pp. 248-249) that the use of autem is not confined to the

type of thematic discontinuity that is based on shifts of discourse topic in a narrow sense, that is, on referent discontinuity. Thematic discontinuity may also be due to an interruption of one or more of the other thematic sub-strands along which a discourse commonly evolves. In narrative texts such thematic sub-strands may involve for instance location continuity, temporal continuity, circumstantial or situational continuity, and action-event continuity. In other text types (e.g. argumentative, technical or philosophical prose) we might also discern logical or argumentative continuity. Just as it appeared to be the case with shifted discourse topics, the discontinuous element is usually placed sentence-initial: just like shifted discourse topics they convey non-challengeable (i.e. given or accessible) information, thus providing a convenient base for the new information conveyed by the rest of the clause. In order to distinguish these elements from fronted discourse topics in a narrow sense we might term them settings. Shifted settings, like shifted discourse topics, may be indicative of the beginning of a new episode in a narrative, or of a new line of reasoning in other types of text. An example of autem after a shifted temporal setting is (55), in which the remarkable travelling habits of Verres during the winter and spring are being described44 . Note that the main discourse topic (Verres) is being continued: (55) Itinerum primurn laborem ... accipite quam facHem sibi iste et iucundum reddiderit. Primurn temporibus hibernis ... (extensive description). Cum autem ver esse coeperat ... ('To speak first of the laborious duty of making journeys, ... let me tell you how easy and agreeable he made it for himself. To begin with, during the winter he ... (extensive description). When spring began, on the other hand, .. .', Cic. Ver. 5.26-27)

An example of autem after a shifted locational setting is (56). In the previous context the author has described, in quite emotional words, the lamentable situation of the family of king Darius after his defeat at Issus. The focus of attention then returns from Darius' tent to the battlefield and the dry facts of warfare (viz. the casualty figures on both sides). The use of the organizational 44. Comparable examples are Cic. Au. 10.13.1; de Orat. 1.26-28; Ver.2.139.

264

§ 10.3.3

particle autem at this significant change of scene appears to be fully in place. In chapter 12 I will come back to this example. (56) In acie autem caesa sunt Persarum peditum C milia, decem equitum; at a parte Alexandri ad quattuor ... ('Now IN THE BATTLE 100,000 Persian foot-soldiers were killed and 10,000 horsemen. But on Alexander's side about 4500 were wounded .. .', Curt. 3.11.27)

In the travel report quoted in (57) the use of autem coincides with both a shifted temporal setting and a shifted locational setting. Autem helps to distinguish emphatically the various stages of the journey described: (57) Cum Brundisium ... vitavissem, ... veni Syracusas ... (elaboration). Cum autem me ex Sicilia ad Leucopetram ... venti detulissent, ab eo loco conscendi, ut transmitterem ('Having avoided Brundisium ... I came to Syracuse ... Now the wind having carried me from Sicily to Leucopetra ... I embarked from that place to cross over', Cic. Phil. 1.7)

In argumentative texts we find not uncommonly autem in a quite specific type of conditional clause, namely a fronted si-clause constituting the second member of a pair of alternative hypothetic situations. Such autem-containing si-clauses may be regarded as hypothetical situational settings. An example of this si ... si(n) autem-pattern is (58)45. Note that the si-clauses concerned do not differ essentially from the temporal cum-clauses quoted in (55) and (57): (58) Si honoris causa statuam dederunt, inimici non sunt; credamus testibus; tum enim honori tuo, nunc iam religioni suae consulunt. Sin autem metu coacti dederunt, ... Utrum tibi commodum est elige ('If they gave this statue to do you honour, they are not your enemies; let us, then, believe their evidence; they were thinking of your honour then, they are thinking of their consciences now. If, on the other hand, they were frightened into giving it, ... Choose the alternative that suits you!', Cic. Ver. 2.150)

For the sake of completeness it should be mentioned here that autem may also co-occur with 'rhetorical' settings. I use the term 'rhetorical settings' to refer to a number of discourse-structuring (notably enumerative) adverbs or particles, in particular nunc ('now'), deinde ('next', 'subsequently'), porro, post(e), praeterea, tum and tunc ('then'). Beside their use as temporal adverbs in narrative texts, in which case they indicate a new stage in the narrative 45. Other examples of the si ... si(n) autem-pattem are PI. Aul. 367; Cist. 67; Cic. Au. 2.9.1; Ver. 2.74; 3.182; 4.113. See also TLL, S. V. autem, p. 1596, 14-51. Examples of autem in comparable constructions (but without si) are Cic. Ver. 3.208; 3.227.

autem with other types of thematic discontinuity

265

('temporal setting', see above), these markers may also be employed in expository and argumentative prose to indicate a new step in an argument or in a sequence of ideas46 • I give two examples in which autem is attached to a rhetorical setting. Both of them illustrate moreover that the presence of some adversative or contrastive relationship (in a strictly semantic sense) is not an essential condition for the use of autem: the relations involved are enumerative or continuative rather than contrastive. (59) (The slave Pinacium is enumerating the goods that Epignomus has brought with him on his ship. The parasite Gelasimus is overhearing the scene) Argenti aurique advexit nimium ... Lanam purpuramque multam ... Lectos eburatos, auratos .. . TUM Babylonica et peristroma tons ilia et tappetia advexit, nimium bonae rei .. . POSTE, ut occepi narrare, fidicinas, tibicinas, sambucas advexit secum forma eximia ... POSTE unguenta multigenerum multa ... POSTE autem advexit parasitos secum (' An enormous amount of gold he has aboard ... And quantities of wool and cloth of purple ... And couches, inlaid with ivory and gold ... And then the Babylonian draperies and clipped carpeting and tapestry he has brought, all sorts of fine stuff ... Yes, and to continue, he has brought along lyre girls and flute girls and harp girls, perfect beauties ... Yes, and perfume galore of kinds galore ... And last but not least (paste autem), he has brought along some parasites', PI. St. 374-388)

Note that only the last member of the enumeration, which seems to contain a more strongly emphatic focus element than the other members47 , is marked by autem. We might perhaps conclude that in (59) both the text-organizational and the focalizing/highlighting use of autem playa role. The translation 'and last but not least' therefore seems to be a better rendering of poste autem than the rather flat translation of the Loeb edition ('yes, and besides'). An example of autem with a rhetorical setting in Cicero is (60): (60) De litterarum missione sine causa abs te aCCUSOL Numquam enim a Pomponia nostra certior sum factus esse cui dare litteras possem; PORRO autem neque mihi accidit ut haberem qui in Epirum proficisceretur nequedum te Athenis esse audiebamus ('You have no reason to complain of lack of letters from me, as Pomponia has never let me know when there was a messenger to give them to. Besides (porro autem) it has so happened that I have not had anyone starting for Epirus and have not yet heard of your arrival at Athens', Cic. Aft. 1.5.3)

46. cr. Hilton (1989: 173-84). 47. Note that the noun parasitos is placed emphatically after the verb.

266 10.3.3.2

§ 10.3.3

autem as a neutral transition particle

In a small number of cases the distinguishing and demarcating force of autem is not clearly associated with one particular element in the clause, such as a fronted discourse topic or setting. In these cases autem signals a quite neutral, and hardly translatable transition to a new stage or step in the narrative or argument. Examples (61) and (62) serve as illustration. In (62) the thematic reorientation is indicated not only by autem, but also by ecce ('look! ')48. (61) (In the preceding passage Seneca has expounded a theory of his school that he himself does not adhere to) Ne putes autem primum < me > ex nostris non ex praescripto loqui sed meae sententiae esse ('Now (autem) do not imagine that I am the first one of our school who does not speak from rules but has his own opinion', Sen. Ep. 113.23) (62) Ecce autem illis ipsis diebus, cum ... , nuntiatur mihi tantam isti gratulationem esse factam ut is domum quoque pueros mitteret qui uxori suae nuntiarent ('But to my surprise (ecce autem) , in the very same period, when ... , I received the news that Verres had been so warmly congratulated on this (i.e the election of Marcus Metellus as president of the Extortion Court, CK) that he even sent off slaves to his house to carry the news to his wife', Cic. Ver. 21)

Another instance of the use of autem with a neutral thematic reorientation is (63), in which the subsequent actions of the same group of people are described: (63) incredibili celeritate ad flumen decucurrerunt, ut paene uno tempore et ad silvas et in flumine et iam in manibus nostris hostes viderentur. Eadem autem celeritate adverso colle ad nostra castra atque eos, qui in opere occupati erant, contenderunt ('with incredible speed the enemy rushed down to the river, so that almost at the same moment they were seen at the edge of the woods, in the river, and then at close quarters. Then (autem) with the same speed they charged up the hill against our camp and the troops engaged in entrenching it', Caes. Gal. 2.19.7-8)

In this example there is no shift of discourse topic. The autem-clause can nonetheless be regarded as the next stage in the development of the narrative, in that the focus of attention is shifted from the events at the river to the subsequent events near Caesar's camp. The thematic reorganization that enhances the use of autem depends, in other words, on a shift in location (and inherently also in time), rather than of discourse topic. Since however the local

48. Some other examples of the combination ecce autem are: PI. Mer. 748; Mil. 207; 209; 1198; Mos. 382; 660; Per. 300; Cic. Au. 6.1.23; 14.12.1; 15.4a.l; eluent. 14; Ver. 17; 21.

autem with other types of thematic discontinuity

267

setting of the events has not been chosen as a major substrand of the thematic development (note that the shifted location is not placed sentence-initially), the transition signalled by autem appears to be relatively small and 'neutral'. As an illustration of the use of autem at the transition to a new idea or consideration we can also adduce instances in which autem occurs in the minor premiss of a syllogism49. This type of argumentative pattern is particularly well suited to a stepwise presentation in which the separate steps are explicitly demarcated from one another: (64) (On the question whether a woman who has received as a bequest "all the silver", is entitled also to the coin) Si pecunia signata argentum est, legata est mulieri. Est autem pecunia signata argentum. Legata igitur est ('If coined money is silver, it was bequeathed to the woman. But (autem) coined money is silver; therefore it was bequeathed', Cic. Top. 53)

10.3.3.3

autem with temporary thematic shifts

A last type of thematic discontinuity that may be signalled by autem is illustrated in (65), which has also been cited in 9.2.1, p. 220: (65)

§ 24 ... ipse cum equitatu antecedit ad castra exploranda Cornelia, quod is locus peridoneus castris habebatur. Id autem est iugum directum eminens in mare, utraque ex parte praeruptum atque asperum, sed tamen paulo leniore fastigio ab ea parte quae ad Uticam vergit ... (description continues for three more sentences). § 25 Hoc explorato loco Curio castra Vari conspicit ... ('Curio himself went on in front with his cavalry to explore the Cornelian camp, because that spot seemed particularly suitable for a camp. Now this (id autem) was a straight ridge, projecting into the sea, abrupt and rugged on either side, but with a somewhat gentler slope on the side facing Utica ... Having reconnoitred this place, Curio saw the camp of Varus .. .', Caes. Civ. 2.24-25)

In (65) a constituent from the preceding context with low topicality (viz. locus) is temporarily raised to discourse topic status: instead of a continuation of the description of the activities of the main discourse topic Curio, the narrative is temporarily interrupted for quite extensive background comment. The main line of the narrative and the main discourse topic Curio are regained in § 25

49. Some other examples of the use of autem in syllogisms are: Cic. Ac. 2.230; Inv. 66; Top. 56; Plin. Ep. 7.2.3; Sen. Ep. 113.20. For autem in somewhat similar argumentative patterns see e.g. Cic. Ver. 3.147; Sen. Ep. 4.6.

268

§ 10.3.3

with the resumptive ablative absolute construction hoe explorato loeo50 • The earlier cited example (37) from Petronius (see section 10.3.2.1 above) could perhaps be described in the same way: the activities of the boy Croesus appear to interrupt the main thematic string, an interpretation which is supported by the use of imperfect tenses. Temporary thematic breaks may sometimes have the form of a parenthesis. Two examples of a parenthesis containing autem are (66) and (67)51. (66) in ripa elephantos - quadraginta autem erant - disponit ('he posted the elephants, of which he had forty, along the bank', Liv. 21.5.10) (67)

"Sed iam confirrnata causa te, hominem amicissimum (me autem appellabat) et aliquot annis minorem natu, non dubitabo monere" (' "But now that my case is established, I will not hesitate to give some advice to you as a very dear friend" - autem he was addressing myself - "and a person some years my junior" " Cic. Ac. 2.61)

Note that in (67) the transition in the discourse indicated by autem involves a shift from the reported layer of the text to the 'here and now' (the communicative layer) of the author and his audience (cf. section 5.2.2.2). Although it is quite tempting to assume, in contexts like (65) - (67) above, an exceptional 'causal' or 'explicative' value of autem (as does e.g. De la Harpe (1923: 68) for 65), such an interpretation becomes unnecessary in an account in which autem is no longer regarded as an adversative conjunction, but as a marker of thematic discontinuity. The only difference from the examples cited earlier is that the thematic discontinuity involved concerns a shift to an embedded or secondary discourse topic, which admittedly may yield the impression of a 'causally' related text part. Assigning this quite specific causal or explicative value to autem would however be an unnecessary overburdening of its proper discourse function. I therefore do not agree with Latin grammarians and commentators who argue that autem in background

50. Example (65) is representative of quite a number of comparable autem-examples, most of them to be found in narrative texts, especially in Livy. Cf. e.g. also: Caes. Gal. 1.6.1; 7.23.1; Cels. 2.7.34; Cic. Aft. 1.14.1; 1.16.10; Div. 1.103 (cf. Val. Max. 1.5.3);yer. 2.171; 4.122; Liv. 21.12.5; 21.26.6; 33.14.12; Val. Max. 3.2.ext.6. Characteristic of these examples is that they usually contain a predicate in the imperfectum or plusquamperfectum, both tenses which are typically used in text parts which convey background information (see ch. 5, p. 122); besides we find, as expected, the habitual present. Another significant feature of the examples concerned is that after the embedded text part the main line of the narrative or exposition is often regained by a resumptive particle (e.g. sed; igitur) or a resumptive construction (cf. ex. 65). 51. Cf. TLL s.v. autem p. 1592, 58 ff. Other examples are for instance: Cic. Aft. 1.16.1; 5.3.2; Fam. 6.5.1; Phil. 2.16; Tusc. 2.45; Liv. 37.10.1; 37.18.12.

autem with other types of thematic discontinuity

269

comments is used instead of the more common enim or nam52 • The most that we can say is that the conditions of use of autem partially overlap with those of enim and nam. As for enim, I have argued earlier (ch. 8, e.g. p. 207) that it occurs typically in those parts of the text in which the interactional layer of writer and reader shines through, as is often the case in parenthetical comments for instance. The relatively rare use of autem in such contexts (cf. ex. 67)53 is, however, easily explained by autem's own, very different, discourse function. Comparable remarks can be made on the alleged resemblance between autem and nam: autem and nam can both occur in backgrounded text parts, but on the strength of different discourse functions, and with potentially different effects on the ultimate interpretation of the text. In section 10.4.4 below I will briefly return to the similarities and differences between autem and nam.

10.4 Conclusions

10.4.1 a discourse-pragmatic description of autem The discussion in this chapter has been arguing for the following conclusions: (i) that autem does not signal (weakly) adversative relationships between successive clauses, at least not in a strictly semantic or rhetorical sense; the contexts in which autem occurs are characterized rather by an absence of any specific semantic or rhetorical relationship between successive clauses or

discourse units. The most that we can say is that autem's proper discourse function is highly compatible with adversative relationships (in a strict, i.e. semantic or rhetorical sense), notably with those of the subtype 'semantic opposition' (see section 9.1; note that the semantic opposition relation usually does not obtain between propositions or clauses). The particle is, moreover, not excluded from clauses that maintain a causal, additive or consecutive relationship with the preceding clause54 • 52. Cf. e.g. De la Harpe (1923) and Petersmann (1977: 248). Szantyr (p. 490) refers to the equivalence of autem and enim in late Latin. Hand (1829-1845, vol. I, s.v. autem, p. 573) observes that autem has often been unjustly changed by editors into enim. 53. A comparable example of autem (though not in a parenthetical clause) is Petro 61.9, which is adduced by Petersmann (1977: 248) as an instance of autem in the role of enim. 54. For autem in causally related clauses see e.g. (24) and (65), for autem with additive relationships (60). An example of autem in a consecutively related clause is Cic. Ver. 3.47; see also Fin. 3.16 and 3.58, and the comment of Hand (vol. I, p. 571-2).

§ 10.4.1

270

(ii) that the various 'functions' or 'meanings' of autem that are assumed in the literature are all reducible to one common discourse function, including the alleged modal and causal uses. In accordance with the theoretical outlines given in chapters 4 and 5, the characterization of the particle autem may run as follows: stepwise analysis of the particle autem 1. basic meaning: 2. discourse function:

3. actual use:

4. side-effects:

'distinctiveness'55. indication of the discrete status of a text segment in relation to its preceding verbal or non-verbal context (presentational level of discourse) (i) local: marker of focus a) relational (parallel or contrastive) focus b) absolute focus (ii) global: marker of the organization of the text (more specifically: marker of thematic discontinuity) (i) reinforcement of the illocutionary intention of the speaker/writer, or indication of his specific

(ii)

emotional attitude adversativity, causality

From the above overview it is clear that we are dealing here, not with two different particles with two separate discourse junctions, but at most with two different uses which both emanate from one and the same discourse function. This discourse function is associated with the presentational level of discourse. The basic meaning of 'contrast' that is intuitively felt to be present is to be' associated with the discourse-pragmatic notions of distinctiveness and demarcation rather than with semantic adversativity. Depending on whether it is applied locally or with a more global scope, the particle is used as a focus marker (parallel or absolute), or as a marker of the organization of the text56 . 55. Mosts accounts in the literature assume an etymological connection with Greek OIJ, OIJT€, OIJOU; (lexical root au 'again', 'alternatively'). Cf. Hand (voL I, p. 559); Szantyr (p. 489); Janson (1979: 101). 56. Cf. the Greek particle Of, whose discourse function and uses are remarkably comparable with Latin autem. Cf. Ruijgh (1971) and Bakker (1993).

discourse-prag11U1tic description of autem

271

These two uses cannot always be kept strictly apart: we have seen for instance that parallel focus and alternate or shifted topic are clearly related (and partially overlapping) phenomena. There is no need to distinguish a separate, 'modal' particle autem, with a function on the interactional level of discourse: the specific illocutionary intentions or attitudes of the speaker that autem sometimes may seem to indicate (indignation, surprise, impatience, etc.) are to be regarded as side-effects of its use as a focus marker. The alleged uses of autem as an adversative or causal connective should also be described as sideeffects. As to the text-organizational use of autem we can say that the notion of 'contrast', or rather 'distinctiveness', that is basic to the particle, should be taken in the sense of 'discontinuity' of the thematic development of the discourse. As such autem is involved in the segmentation and organization of an extended monologue. In chapter 4 (p. 83) I mentioned autem among the indicators of the text-organizational strategy of Listing. The concept of Listing applies to situations in which a virtually completed move is followed by a new, 'coordinated' move by the same speaker, thus yielding an extended monologue. So far as autem is concerned, the connected moves are to be defined in the11U1tic terms rather than in communicative terms (as appeared to be the case with nam). On this point I would recall that, whereas the unit of discourse called 'act' is defined in strictly communicative terms, the unit of discourse called 'move' is defined in communicative as well as thematic terms (see chapter 4): a completed move is characterized on the one hand by its communicative coherence (i.e. coherence between the constituent communicative acts or configurations of acts), on the other by its thematic coherence (i.e. coherence between thematic units). Thematic units are in turn constituted by one or more thematic chains which are based on the recurrent mention of certain information units (lexical words that code either the subject or object noun, the verb, an adjective or an adverb). The description of autem as presented above in schematized form illustrates once more that we should be careful about distinguishing too strictly 'connective' particles on the one hand and 'situating' (e.g. 'modal') particles on the other (see ch. 1 and 4): the 'connective' and 'situating' uses of autem appear to merge into each other quite smoothly. In the same light we might also reconsider the alleged diachronic widening of the possibilities of use of autem, as has been suggested by Petersmann for instance in his study on the language of Petronius (1977: 247). A more plausible, synchronic explanation for the occurrence of 'modal' autem in questions and exclamations would be for

272

§ 10.4.1

instance that in accordance with the variety of text types represented in Petronius' Satyricon we find autem in all kinds of different contexts (both conversational and non-conversational) and, accordingly, with its whole range of possible uses and connected side-effects. Most other Latin texts lack the appropriate contextual conditions for the occurrence of 'non-connective', focalizing autem. 10.4.2 linguistic clues

In chapter 5, section 5.3 I discussed a number of possible linguistic clues that may help to determine the discourse function and uses of a particular particle. As to the distributional properties of autem that may count as linguistic clues to its proper function we can say that in general there are very few restrictions on the conditions of use of autem: almost any unit of a text can be emphatically demarcated or singled out, regardless of its meaning, size, function or syntactic status; accordingly autem's distinguishing and demarcating force may belong to a single word, to an entire clause, to two or more clauses, or even to a whole section. This fact is fully in line with the observation that there are no real restrictions as to the sentence type and illocutionary force of the sentence in which autem occurs 57: since the force of autem applies not to communicative discourse units (e.g. acts), but to infor11Ultion units (i.e. constituents of the clause) or the11Ultic units, the illocutionary force of the grammatical clause in which autem happens to occur is quite irrelevant for determining the particle's specific discourse function. Its indifference to its syntactic and illocutionary context is in accordance with the generally held, intuitive opinion that autem has a relatively 'neutral' load. However, there also appear to be a number of distributional properties of autem that do allow for a more refined and distinctive description of the particle. Most of these have already been mentioned in passing in the preceding sections. They have to do with phenomena such as word order, combinatory possibilities of the particle, and its preferred discourse type. The following contains an overview.

57. The use of autem in assertions and questions has been amply illustrated in the examples cited in this chapter. For the use of autem in a sentence with a directive illocutionary force see e.g. Cic. Tusc. 5.90: Quare ut ad quietum me licet venias. Munera autem ista, quibus es delectatus, vel civibus tuis vel dis immortalibus dona ('You may come to me therefore as to one at peace; but as for the gifts (munera autem) you delight in, present them to your fellow citizens or to the immortal gods').

linguistic clues

273

One important indication of the proper function of autem appeared to be the element in the first position of the clause autem belongs to. On pp. 257-258 we saw that the use of autem is highly compatible with fronted nominal phrases (NPs). Now NPs in sentence-initial position have been said to be indicative of discourse topic shifts and other forms of thematic discontinuity. The frequent occurrence of autem in the immediate environment of such fronted nominal phrases can hence be taken as a clue to a text-organizational use of the particle. Statistical research should make clear whether the frequent combination of autem with fronted NPs is indeed significant, and may hence be regarded as a distinctive property of the particle. A relevant observation in this respect is, for instance, that in Cicero's Letters 15 out of 350 investigated thematic de- and quod-constructions (which count as the most outspoken forms of fronted NPs) are accompanied by autem, whereas other discourse connectives in such an environment are rare or not attested at a1l58 • Another linguistic clue that points in the direction of the more specific textorganizational use of autem, may be derived from the combinatory possibilities of autem with other particles. On p. 264-265 above I mentioned the quite common collocation of autem with enumerative adverbs or particles (nunc ('now'), deinde (,next', 'subsequently'), porro, post (e) , praeterea, tum and tunc ('then'», which are typically used to indicate a new step in an argument or in a sequence of ideas. Apart from this rather specific type of collocation, there are very few other systematically used combinations of particles including autem. There are, admittedly, incidental combinations of autem with sed ('but'), but in all these cases sed has a text-organizational force (viz. as indicator of a sudden turn in the development of the discourse, see ex. (35) and note 23)59. For the combination ecce autem ('10 and behold!', 'look!'), which can be explained in the same way, see p. 26660 • For the combinations et autem, (ni)si autem and cum autem, I refer to 10.2.1 and 10.3.3.1 above. It is significant that these combinations are used almost exclusively in the second

58. lowe these data to Maartje Somers (unpubl. ms., University of Amsterdam, 1990). There are 3 instances of nam in Theme constructions, 2 of quidem, 3 of sed, 1 of tamen and 1 of vera. 59. The isolated instance of verum autem ('but on the other hand') in PI. Cas. 555 (verum autem altravarsum quam eam mecum ratianem puta) is to be regarded as artificial. The speaker's excessive cumulation of linguistic expressions for putting forward an alternative ('1. but 2. then, 3. on the other hand, 4. when I think it over ... ') can hardly be considered a common form of expression. 60. It is to be noted that ecce cannot be combined with any discourse connective other than autem or enim. For ecce enim see ch. 8, pp. 180-181.

274

§ 10.4.2

member of a parallel construction61 • Other text-organizational devices that are typically used in combination with autem appeared to be the use of metadiscursive expressions in which the thematic structure of a stretch of text is announced or recapitulated. This has been illustrated in (46) and (51) for instance. Another clue to the more specific discourse function of autem can be derived from the relatively frequent collocation of autem (compared with other discourse connectives) with the pronoun We ('he'). Examples can be found in (9), (43) and (48). For statistical data I refer to Bolkestein and Van de Grift (1994), who in their discussion of the discourse function of the third person pronouns hie, is and We pay attention also to certain combinations of discourse particles and pronouns. According to Bolkestein and Van de Grift62 the anaphoric pronoun We is typically used (instead of e.g. is or hie) in the case of a discourse topic shift, when there are more candidates available for the role of discourse topic. Ille has, in other words, the function of signalling referential ambiguity or thematic discontinuity. The high compatibility of autem with We (compared with e.g. the collocation of igitur with ille, which is relatively rare) is fully explainable when we assign a distinguishing and demarcating discourse function to autem63 • For the frequent collocation of autem with the personal pronouns ego and tu, and the implications for the specific function of autem, I refer to pp. 235-236. A characteristic distributional property of autem that is especially apparent in its local use is, furthermore, the immediate presence of some form of parallelism. We have seen in 10.2 that autem tends to be attached to the second member of a parallel construction in which two or more (semantically or pragmatically) corresponding notions are placed emphatically side by side, often (but not necessarily) due to some semantic contrast between the individual members. As to its preferred discourse type, we can say that autem is strongly associated with monological discourse (cf. section 5.2): even in Latin comedy,

61. I came across one instance of the comparable parallel constru.ction quoniam ... quoniam autem ('since ... since, on the other hand, ... '): Quint. Inst. 2.10.12. 62. Cf. also Pinkster (1987). 63. The low frequency of the collocation of ille and igitur observed by Bolkestein and Van de Grift can be explained by the discourse function of igitur. In Kroon (1989) I have argued that igitur introduces a central discourse unit after a preparatory unit leading up to it. It thus stands to reason that an igitur-unit does not provide the type of context in which a typical topic shift device such as ille is prone to occur, topic continuity being the 'default' situation in the case of an igiturunit.

pragmatic motivation for the use of autem

275

in which the dialogical discourse type prevails, the particle is usually found in typically monological (e.g. narrative or descriptive), continuous passages. Autem's preference for a monological discourse type is confirmed by the fact that the particle (in contrast with enim, and, as we will see later on, with autem's alleged synonyms vero and at) does not co-occur with 'interactional elements', the latter being indicative of a dialogical or diaphonic discourse type. A further significant distributional property of autem (interrelated with its monological nature) is its backward-orientation: as a rule autem-units belong to the continuation of a preceding, corresponding unit. Hence it is inconceivable (and in fact not attested, as far as I know) that a text should start with an autem-unit. 10.4.3 pragmatic motivation for the use of autem

The question as to the pragmatic motivation or communicative point of using a certain particle is especially relevant in the case of particles with a relatively broad range of applicability and a low degree of semantic specificity. It is particles of this type that run the risk of being described as 'meaningless fillers' which are scattered over a text quite randomly (see section 3.2). Although this description may seem at first sight to apply to autem, it appears to be possible to identify a number of rather specific contexts in which the use of this marker of distinctiveness and demarcation is particularly expedient. On a relatively local level of the text autem appears to be employed for instance to avoid referential competition or ambiguity (viz. in the case of 'switch reference'). Furthermore, it is used in situations where the unambiguous identification of one or more referents is at stake, as in (68), or in the case of an emphatic division of tasks, as in (69): (68) nunc internosse ut nos possitis facilius, ego has habebo usque in petaso pinnulas; tum meo patri autem torulus inerit aureus ('To make it easier for you to tell us apart I shall always wear this little plume on my hat; yes, and as for my father (tum meo patri autem) he will have a little gold tassel hanging from his', PI. Am. 142-144) (69) tu eum orato, ego autem orabo vilicum ('You plead with him, while I (ego autem) plead with the bailiff', PI. Cas. 273)

Especially series and lists call for a strictly structured presentation of their constituent elements. In contexts like these autem is a common alternative to (or corroboration of) enumerative adverbs such as tum or post ('then', 'next').

276

§ 10.4.3

On a more global text level autem characteristically appears to distinguish the various steps in an argument or exposition. Accordingly autem is especially in place in didactic and philosophical prose, that is, in text types in which a transparent and stepwise presentation of the material is an essential requirement for the success of the text64 • An explicitly structured presentation of the information is expedient also in letters, which are commonly characterized by the fact that they are composed of several more or less unrelated thematic units. The use of autem on the border between two such 'independent' units guarantees a smooth and uninterrupted continuation of the discourse. Narrative texts, by contrast, usually have a much higher degree of thematic continuity (referent continuity, action-event continuity, location and time continuity, etc.), which explains why autem is used more sparingly here. However, in narrative texts too, conditions may arise under which it is useful to underscore the distinctive status of separate thematic units. We might think for instance of situations in which various protagonists are involved in a number of different actions 65 • This can be illustrated with (70); compare also (43) on p. 254-255: (70)

. .. NAUT AE coacti fame radices palmarum agrestium ... colligebant et iis miseri perditique alebantur; CLEOMENES autem ... totos dies in litore tabernaculo posito perpotabat. ECCE autem repente ebrio Cleomene esurientibus ceteris nuntiatur piratarum esse navis in portu Odysseae; nam ita is locus nominatur; NOSTRA autem CLASSIS erat in portu Pachyni. CLEOMENES autem ... speravit ... ('THE SAILORS (nautae) , as food was so short, had to set about collecting the roots of wild palms ... and they tried to keep themselves alive on these. Now CLEOMENES (Cleomenes autem) ... spent whole days drinking in a tent pitched on the shore. AND NOW (ecce autem), while he was drunk and his men starving, the news suddenly arrived that there were pirate ships in the harbour of Odyssea (because that was what the place was called). Now OUR FLEET (nostra autem classis) was in the harbour of Pachynus. Now CLEOMENES (Cleomenes autem) ... counted on .. .', Cic. Ver. 5.87)

The text fragment in (70) contains a series of discourse topic shifts and one· instance of a sudden change in the action-event continuity. The attention is first on the Sicilian sailors, then on the commander of the fleet Cleomenes. Next there is a sudden change of circumstances (a message is brought that there are pirates in the neighbourhood), which is also marked by autem (ecce

64. It is of note that Vitruvius appears to be particularly fond of autem. See e.g. Vitro 5.4.5, 5.11.2 and 5.11.4, where autem is used four times within the same paragraph. 65. This description applies also to the passages in Latin Comedy in which one of the characters expounds the (usually quite complicated) plot of the play and the characters involved. Cf. the use of autem in e.g. PI. Capt. 27-36 and Cas. 50-65.

autem and other discourse connectives

277

autem). Then there is again a shift in attention to the fleet, and finally we are back with the commander Cleomenes. Narrative passages in which shifts in the focus of attention are as frequent as in (70) are quite exceptional, more persistent topic continuity being the unmarked situation in narrative texts. In contexts like (70) the use of an explicit discontinuity marker may considerably help to prevent the impression of an incoherent story. 10.4.4 autem and other discourse connectives It is now possible to compare autem with two categories of other discourse

connectives: (i) discourse connectives that may somehow be involved in the global organization of the text, such as nam, igitur and sed, in some of their uses; (ii) words that are conventionally termed 'adversative conjunctions', like sed, vero and at.

nam, igitur and sed As to the first category it has already been observed that there is a certain overlap in the conditions of use of autem and nam. Both particles are for instance found at section boundaries, in combination with a Theme/fronted topic construction, which implies that both can be employed for marking the structure of an extended monological stretch of text (cf. 4.2.2.3 and 7.3.3). There is, however, a difference between the two, which can be explained on the basis of their distinct discourse functions. Globally used autem is what I called in chapter 4 a Listing device: it serves mainly to link successive, functionally equivalent and independent thematic units (e.g. episodes) and hence to move the discourse in a forward direction. Globally-used nam, on the other hand, is a typical Dependency device: it not only helps to move the discourse in a forward direction (by broaching a new topic of discourse), but also and primarily indicates that from a communicative perspective the new thematic unit plays a subsidiary role: the nam-unit has the status of an 'afterthought', added for the sake of completeness. The difference between text-organizing autem and text-organizing nam can be illustrated with (71):

278

§ 10.4.4

(71) Quod quoniam tibi exposui, facilia sunt ea, quae a me de Vatinio et de Crasso requiris. Nam de Appio quod scribis sicuti de Caesare te non reprehendere, gaudeo tibi consilium probari meum. De Vatinio autem primum reditus intercesserat in gratiam per Pompeium ... ('And now that I have explained this to you, what you ask me about Vatinius and Crassus is easily answered. In Appius' case (Nam de Appio) you write that, just as in the case of Caesar, you have no fault to find, and I am glad that the line I took meets with your approval. Now as to Vatinius (De Vatinio autem), a reconciliation had been brought about between us, in the first place by the intervention of Pompey .. .', Cic. Fam. 1.9.19)

In the above example Cicero first announces the theme of the discussion that is to follow: the addressee's earlier questions about Vatinius and Crassus, respectively. However, before turning to an extensive treatment of the first sUbtopic of this theme (Vatinius), Cicero inserts a short remark about Appius: Appius obviously needs to be mentioned in the context of the present letter, but the topic needs no further elaboration (this is what nam implies). The passage on Vatinius, on the other hand, takes up the remainder of the paragraph and is concluded by the metadiscursive expression Babes de Vatinio; cognosce de Crasso (,So much about Vatinius; now learn about Crassus'), which forms the transition to the second announced subtopic. Thus, although nam and autem might seem to play similar roles in (71) (they both co-occur with discourse topic shifts; note the preposed de-constructions), they are not identical devices, and are certainly not interchangeable. The Latin discourse connective igitur ('so') is, like nam, a typical Dependency marker: it introduces a central discourse unit after a preparatory unit leading up to it66 • Although igitur most often signals a rhetorical relation (,conclusion') between successive acts (cf. 4.2.2.1), it has also a text-organizational use on a more global level of the text, which is more or less comparable to the text-organizational use of autem: like autem, igitur is apt to introduce a new section. However, igitur does not indicate a neutral transition from one independent thematic unit to another, but rather signals that after some preparatory (often discourse procedural) remarks the actual exposition is finally starting or being resumed. Compare French alors ('well then'), which has a comparable use. An example of this use of igitur is (72):

66. Cf. Kroon (1989). Note that nam, in contrast with igitur, signals the subsidiary and not the central discourse unit.

autem and other discourse connectives

279

(72) (Marcus in reply to Quintus) Non ita est, Quinte, ac potius ignoratio iuris litigiosa est quam scientia. Sed hoc posterius; nunc iuris principia videamus. (new section) Igitur doctissimis viris proficisci placuit a lege ... (,There you are mistaken, Quintus, for it is rather ignorance of the law than knowledge of it that leads to litigation. But that will come later; now let us investigate the origins of Justice. (new section) Well then, the most learned men have determined to begin with Law ... ', Cic. Leg. 1.18)

The last discourse connective that is capable (in one of its uses) of signalling a new thematic unit in the discourse is sed ('but'). Sed differs from autem in that its text-organizational use always involves a quite radical shift in the focus of attention: sed cuts off rather abruptly the current thematic chain, in order to report a new and unexpected event or fact (see e.g. ex. 35 above), or to resume an interrupted line of the discourse after the speaker/author has suddenly realised that he has been led too far away on a sidetrack. Although autem appeared to be compatible with this type of context too (cf. the combinations of autem with sed and ecce discussed earlier in this chapter), the particle is not interruptive per se67 •

sed, vero and at Autem appears to have little in common with sed, that is, with sed in its most common use as an adversative conjunction on a local text level. In contrast with sed, autem is for instance not used in substitution constructions of the type 'not X, but Y', or 'not only X, but also Y' (see 9.1). A comparison with the 'adversative connectives' vera and at might perhaps be more fruitful, considering that they are mentioned more often in the literature as near-equivalents of autem than is sed. In Kilhner-Stegmann (1912, vol. II), for instance, autem and vera are both described as adversative connectives as well as as modal adverbs, while Janson (1979: 101) and Marouzeau (1948) go so far as to state explicitly that there seems to be no great difference between autem and vera. However, a quick glance at the distributional properties of vera compared to those of autem shows us that we should be careful in assigning similar or comparable discourse functions to autem and vera. To mention only two differences: (i) whereas vera is prone to occur after a resumptive pronoun,

67. I would contend that text-organizational sed Gust like e.g. nam) is involved in the organization of communicative rather than thematic units.

280

§ 10.4.4

autem is most often combined with a full nominal phrase; and (ii) vero is combined significantly often with the negation particles non and especially nee, whereas the combinations nee autem and non autem are extremely rare. In 11.5.2 I will come back to the distributional differences between autem and vero. In the same way it is possible to draw up an extensive list of distributional differences between autem and at. The most important of these appears to be the fact that autem is typically associated with a monological, at with a dialogical or diaphonic discourse type, which may suggest that they operate on different levels of discourse (presentational and interactional, respectively). In the next two chapters I will attempt to show that the distinct distributional properties of vero (ch. 11) and at (ch. 12) are indicative of highly distinct discourse functions, both of which differ considerably from that of

autem.

11

VERO

The discourse-pragmatic description of vero given in the present chapter resembles earlier accounts (see 9.3) in that it confirms the view of vero as a remarkably versatile word with a range of quite different uses. The rather strict dichotomy between affirmative, 'modal' vero and adversative, 'connective' vero seems, however, not to do full justice to the proper character of the particle and to its use in actual discourse, nor is it immediately apparent how these two 'meanings' may be interrelated. The alternative description proposed in this chapter aims at providing a more nuanced and unitary picture of the particle, by distinguishing systematically between discourse functions, actual uses and incidental side-effects (11.1-11.4). By doing so it will become clear, among other things, that vero is not primarily a marker of sequential, notably adversative, relationships. Rather, vero is to be described as a situating particle with functions on the representational level of discourse (objective modality marker) and on the interactional level of discourse (subjective modality marker; conversation particle). This approach enables us moreover to compare vero with other discourse connectives (especially with its alleged synonym autem), and hence to give it a better-defined place in an overall system of Latin discourse particles (11.5).

11.1 The discourse functions of vero

Vero is a situating particle with a discourse function on both the representational and the interactional level of discourse. In its representational function (as an objective modality marker) vero designates the actuality or reality of a state of affairs as compared to a supposed, hypothetical or potential counterpart, the latter being implicit or explicit in the surrounding context. In its interactional function vero indicates (i) (as a subjective modality marker) a high degree of personal commitment on the part of the speaker/author to the truth or appropriateness of the message being communicated; or (ii) (as a

282

§ 11.1

conversation particle) the sincerity of the speaker's illocutionary intention or the actuality of the perlocutionary effect! (conversation particle). explanatory remarks Vera belongs to the 'situating particles', that is, to the group of particles that do not, on their own account, mark sequential relationships between explicitly expressed text units. In 3.1 and 4.1.2.1 I already stated that particles can all be viewed as relational devices in a broad sense, their common characteristic being that they fit their host unit into a wider perspective. When this wider perspective is provided explicitly in the surrounding verbal context we are usually dealing with a connective particle in a narrow sense. The wider perspective may however concern also an implicit idea or event which is evoked by the surrounding context, or some element of the extratextual communicative situation in which the text is embedded. For particles of this type, which are connective only in a broad sense, I have reserved the term situating particles. Situating particles may have a discourse function on the representational level of discourse (see ch. 4, p. 71), on the interactional level of discourse (see ch. 4, pp. 94-95), or on both. Consider the following proposal for a subclassification of this elusive and quite heterogeneous group of words: figure: classification of situating particles2

/ situating particles

~

. / focus particles representational/' '-...... objective modality markers

. . ______ subjective modality markers mteractlOnal - - - - - conversation particles

I. By 'illocutionary intention' is meant a speaker's communicative intention with respect to the content of his speech act, by 'perlocutionary effect' the effect the performance of a speech act has on the addressee. For the latter concept, cf. Austin (1962: 94-120). 2. In the German linguistic tradition focus particles are known under the name of Gradpartikel. Other terms referring to this group of items are scalar particles or logical particles. Subjective modality markers are sometimes called attitudinal particles, modal particles, or, in German, Ahtonungspartikel, although these terms sometimes apply to a larger category of words containing conversation particles as well. Conversation particles are referred to also as phatic particles. For a general discussion of the various ways in which this group of particles is treated in the literature I refer to ch. 3.

discourse junctions of vero

283

There are at least two types of situating particles with a function on the representational level of discourse: focus particles and objective modality markers. Focus (or 'scalar') particles evaluate (part of) the state of affairs that is represented in the host utterance in terms of some relevant scale of expectation. More specifically, focus particles evaluate (part of) a state of affairs in terms of the (implicit) norms and expectations that hold in the represented world. English examples are just, only and even3 • Objective modality markers (e.g. indeed, really, actually, maybe), on the other hand, evaluate the actuality-status of a state of affairs in the represented world as compared to a hypothetical or possible alternative. In 11.3 below we will see that vero, in some of its occurrences, behaves as an objective modality marker. The situating particles with a function on the interactional level of discourse differ from their representational counterparts in that they are somehow oriented to indicating the involvement of the discourse partners in the speech event (and their management of the conversation) rather than at conveying content. Among these interactional particles we may distinguish for instance subjective modality markers (or attitudinal particles) and conversation particles. Subjective modality markers reveal the speaker's personal evaluation of the (truth, tenability, appropriateness, expectedness, etc. of the) transmitted content. As such they can be said to belong to the expressive component of language, supplying the text, so to speak, with an emotional or affective added value. It is to be noted that subjective modality markers cannot always be clearly distinguished from objective modality markers, especially not in contexts in which the modality marker is combined with a first person subjective evaluation verb, such as I mean, I think, and the like. In this light it is not surprising that some situating particles can be used both objectively and subjectively, as is the case for instance with English really, actually or indeelf. The same holds for Latin vero, as we will see later on. An example of an unambiguously subjective modality marker in Latin is ne ('truly', 'indeed', 'assuredly')5. On the other hand subjective modality markers appear to approach the conversation particles in meaning, namely whenever the item concerned plays

3. For Latin examples, see ch. 4, section 4.2.1. 4. Compare expressions such as he actually bought the house (something he had considered doing for a long time; objective modality marker) and he actually bought the house! (can you believe it?; subjective modality marker). See also ex. (37) on p. 304. 5. See e.g. PI. Mil. 65: Ne illae sunt fortunatae quae cum isto cubant! (,Really! The girls that sleep with him are lucky! ').

284

§ 11.1

a clear role in the management of the communicative interaction between the discourse participants. 'Subjective' particles like English really or its Latin equivalent vero, for instance, seem sometimes to be used mainly to anticipate or elicit a certain reaction on the part of the addressee, rather than to express merely the speaker's personal emotions or attitude in relation to some transmitted content. Other situating interactional particles are less 'subjective' and more straightforwardly involved in the management of the conversation. We might for instance think of particles that solicit the commitment of the addressee (cf. English y'know, German ja or Latin enim), or particles that specify, modify or reinforce the illocutionary force of the utterance, or the interactional function of the move. Summarizing thus far we can say that it is possible to divide the situating particles into several subcategories (see figure 1 above), but that these subcategories are not sharply demarcated from one another. The adequacy of a representation in which the various subcategories blend into one another is demonstrated by the behaviour of vera to be discussed in the remainder of this chapter: the use of vera as a subjective modality marker on the interactional level of discourse appears to be very close on the one hand to its use as an objective modality marker on the representational level of discourse, and on the other to its use as a conversation particle. The linking factor that serves as an explanation for this phenomenon is to be found in the common basic meaning underlying the different functions and uses of the particle. In contrast to many other particles this basic meaning has a quite concrete, lexical basis, which is related to the adjective verus ('true', 'rea!'). With regard to the discourse function(s) of vera the further question needs to be raised, how the quite pervasive view of vera as an adversative conjunction fits in with the above description of vera as a basically non-connective, situating particle. The following considerations may serve as a starting-point for a more elaborate discussion in 11.4.2. The basic meaning of vero inherently implies a comparison between two elements which are contrasted in the matter of their relative veracity, truthfulness, actuality, sincerity, and the like (real vs. unreal, hypothetical, presupposed, false, etc.). The contrasted elements may be states of affairs in the represented world, but also for instance the beliefs, knowledge, intentions or attitudes of the discourse partners (i.e. elements of the communicative situation). In the latter case the use of vera may for instance imply that the speaker presupposes that the addressee might not be convinced of the truth or

discourse functions of vero

285

actuality of the transmitted content, or of the sincerity of the speaker's illocutionary intention; the element with which the vero-unit is contrasted is, in other words, a presumed false opinion or disbelief on the part of the addressee. Now, in a monological stretch of text, when the element to which the vero-unit is compared happens to be verbally expressed in the preceding context, the impression may arise of an 'adversative connective' use of vero ('but' instead of 'really'). In 11.4.2 I will however argue that this adversative connotation is, as a rule, only an incidental side-effect of the situating particle vero, to be explained in terms of the mere compatibility of vero with a certain type of 'adversative' context: it looks as if vero connects one sentence to another, but what it actually does is to express the speaker's judgement or position in the case of a contradiction between two statements. This explains why in the definition of the discourse functions of vero at the start of this section no mention was made of a connective, adversative use of vero. The basically non-connective nature of vero is indicated by the distributional properties of the particle, especially by the combinations of vero with unequivocal connective particles such as aut, at, sed, verum, et, neclneque, immo and a number of subordinating conjunctions. That vero is not a very likely candidate for a role as a marker of semantic or rhetorical relationships (see ch. 4) is suggested moreover by its preference for a dialogical or diaphonic discourse type (see ch. 5). In these particular distributional properties vero strongly resembles enim, as we will see later on. At various points in the following discussion I will come back to the distributional properties of vero.

11.2 Representational vero

11.2.1 between adverb and particle The lexical origin of the particle vero (a case form of verus: 'true', 'real') is, not surprisingly, most apparent in its representational manifestation, the representational level of discourse being concerned with the representation of content. In fact, however, it is often hard to make out in the case of apparently representational vero, whether we are dealing with a 'content' word with a lexical or referential meaning, or with a 'pragma' word, with a discoursepragmatic meaning. In the former case linguists tend to speak of an adverb,

286

§11.2

whereas in the latter case it is common practice to speak of a particle6 • An unambiguous instance of the 'adverb' vera is cited in (1), where vera ('justly', 'sincerely') occurs in parallel with its opposite falso ('wrongly', 'insincerely'): (1)

Eho, mavis vituperarier falso quam vera extolli? (,Oho! You'd rather be run down insincerely, than sincerely admired?', PI. Mos. 178)

A comparable example is (2), in which vera is used in a parallel construction with ioeo (lit. 'with a joke', 'in jest') andfalso: (2)

So: Sed age responde: iam vos rediistis in concordiam? Iu: Derides qui scis haec dudum me dixisse per iocum So: An id ioco dixisti? Equidem serio ac vera ratus (,But come sir, answer me. Have you made friends again now, eh? # Mocker! What I said was all in fun, and you know it # In fun, was it? Upon my soul, it was serious and the solemn truth', PI. Am. 962-964)

In both (1) and (2) the adverb status of vera is undeniable: in its role as a modifier of the predicate, vera can be said to be a constituent element of the propositional content of the utterance, rather than a 'pragma' word which evaluates (a part of) the propositional content, without being itself part of this content. One of the characteristics of the former type of adverb is that it can (and indeed often does) bear (part of) the focus of the clause, as is the case in (1) and (2f. Note that the pair falso ... vera in (1) and the pair serio ... vera in (2) (contrasting with ioeo in the preceding clause) cannot be omitted without a change in the meaning of the clause. The unambiguously adverbial (viz. predicate modifying) use of vera, illustrated in (1) and (2), is, however, extremely rare and, as far as I know, not attested in classical or later Latin. As a rule, we find the regular adverb form in -e (instead of in -0) used in this function, vero apparently being reserved for a more particle-like roles. Consider for instance (3), in which

6. The problematic distinction between 'normal' adverbs and particles (see also section 3.1) has been a matter of much dispute in the literature. For a discussion with respect to Latin see Pinkster (1972). I prefer to relate the difference solely to the criterion of whether or not the item concerned contributes to the propositional content of an utterance. 7. Sometimes the term 'adjunct' is used for those adverbial constituents which modify the predicate. 'Adjuncts' are opposed to so-called 'disjuncts'. Disjuncts do not modify the predicate, but evaluate the whole propositional content of the clause. For the notions 'adjunct' and 'disjunct', see e.g. Pinkster (1972; 1990: 25-38). 8. In Latin, adverbs may be productively derived from adjectives, in one of a number of ways. The most common are the forms in -e, -er, -fer or -iter. More rarely a case form in -0 of an adjective or substantive has become ossified as an adverb, cf. e.g. certo (alongside eerte), erebro,

representational function of vero

287

vere (and not vera) is on a par with the unambiguous adverb falso: (3)

Nam cum grave est vere accusari in amicitia, tum, etiam si falso accuseris, non est neglegendum ('It is a serious thing to be justly accused in a matter of friendship, and no one can overlook it even if he is falsely accused', Cic. Mur. 7)

It will be clear that the instances of vera in (1) and (2) fall outside the actual scope of the present study. This may hold to a certain extent also for (4) and (5), in which, too, vera seems to be an integral part of the propositional content of the clause in which it occurs. This is indicated by the fact that in (4) vera esse is contrasted to non esse, in (5) to simulari esse: (4)

Immo iste eum sese ait qui non est esse et qui vero est negat (,This fellow says he's the man he isn't, and says he isn't the man he really is', PI. Capt. 567)

(5)

(Charea is telling his friend Antipho how he took his chance and got off with the girl he was supposed to keep an eye on as a fake eunuch) an ego occasionem mi ostentam, tantam, tam brevem, tam optatam, tam insperatam amitterem? Tum pol ego is (sc. eunuchus) essem vero qui simulabar ('Was I to lose an opportunity so offered me, such a chance, so short, so much desired, so little expected? Jove! I should then have been in fact what I set up for', TeL Eu. 605-606)

However, in (4) and (5) (more than in 1 and 2) vera can be said to evaluate the actuality status of the represented state of affairs ('to be a person X' and 'to be a eunuch', respectively) for the present situation. That is, it indicates the degree to which the general state of affairs referred to is applicable to the represented world of the specific discourse context at hand. This description of the function of vera resembles the definition of objective modality markers given in 11.1. On account of this slight difference between vera in (1) and (2) on the one hand, and (4) and (5) on the other, it might perhaps be justified to assume different degrees of 'adverbiality': vera in (1) and (2) appears to be more strictly adverbial than vero in (4) and (5), the latter being more 'disjunctlike' and displaying certain features that are typical of an objective modality marker (i.e. of a specific type of situating particle). Despite their 'adverbial' character, examples of vera of the type illustrated in (4) and (5) are interesting for the present study in that they form an essential element in an integral description of the 'particle' vero. We will see later

modo and serio. The form vero also may be explained in this way.

288

§11.2

on that even the clearest examples of the 'pragma' word vero (e.g. vero in its role as a conversation particle) show a certain relationship with vero as used in (4) and (5). In between these extremes a number of intermediate and transitional uses can be discerned which may help towards a better understanding of this relationship. These will be dealt with in turn in the following sections. 11.2.2 vero as an objective modality marker Example (6) contains an instance from classical Latin of the use of vero as an objective modality marker. As in (4) and (5) above vero still seems to have traces of a concrete, lexical meaning, on account of which one might still consider calling it an adverb: (6)

Cur autem ea legatis nostris dabat, cum ad nos Cotylam mitteret, ornamentum atque arcem amicorum suorum, hominem aedilicium? Si vero tum fuit aedilis cum eum iussu Antoni in convivio servi publici loris ceciderunt (,Yet why did he give those commands to our envoys when he was sending us Cotyla, the ornament and bulwark of his friends, a man who had been an aedile? If indeed (vera) he was an aedile at the time when by Antonius' order public slaves lashed him with thongs at a banquet', Cic. Phil. 8.24)

The use of vero in the si-clause helps to make clear that an ironic reservation is expressed by Cicero with regard to the truth of part of the propositional content of the preceding clause, namely Cotyla's being an aedile at the time of the events described; in other words, in the si-clause the actuality status of a possible state of affairs in the presently represented world is ironically questioned. Vero is in this example intuitively felt to be less adverb-like (and accordingly more particle-like) than in (4) and (5). A possible explanation for this intuition is that in (4) and (5) the opposition between an actual state of affairs and its non-actual counterpart is more overtly presented as the focal information of the transmitted content than is the case in (6). Compare f9r instance the contrastive pair 'to be actually a eunuch' and 'to pretend to be a eunuch' in (5). Vero can be regarded as an objective modality marker also in its significantly frequent combination with quasi ('as ift. The relative frequency of this 9. I counted 72 instances of the combination quasi vera in the selected part of the Ibycus corpus (see ch. I); that is to say that in about 3% of all instances of quasi in the corpus concerned vera is added to it. Cicero appears to be especially fond of the combination (51 instances, of which 40 in his orations). Examples from Latin comedy are e.g. PI. True. 292 and Ter. Eu. 685. Combinations of quasi with other particles are much less frequent; there are e.g. only three attestations in the

representational junction of vero

289

combination can be explained quite easily, in that quasi typically introduces statements in which the actuality status of a state of affairs is expressly at issue; quasi and vero are, in other words, highly compatible. I give two examples. Again - as is the case with all 'representational' occurrences of vero - the lexical origin of vero can still be clearly felt: (7)

ipse hanc acturust luppiter comoediam. Quid? Admirati estis? Quasi vera novom nunc proferatur lovem facere histrioniam (,He himself, Jove, will take part in this comedy. What? Surprised? As if (quasi) it were actually (vero) a new departure, this, Jove's turning an actor!', PI. Am. 8890)

(8)

faba quidem Pythagorei utique abstinere (sc. iubent), quasi vero eo cibo mens, non venter infletur! ('The Pythagoreans make a point of prohibiting the use of beans, as if (quasi) actually (vero) thereby the soul and not the belly was filled with wind!', Cic. Div. 2.119)10

A last group of rather straightforward examples of objective modality marking vero is illustrated in (9) - (14) below. These instances have in common that vero underscores the actuality of a state of affairs as compared to a non- or less-actual (i.e. hypothetical, potential, imaginary, presumed, unreal, etc.) alternative, which is stated or implied in the preceding context. Thus, in (9), the assumption that the current master of the house Euclio might perhaps do more honour to the household god than his father, is followed by the statement that in fact (vero) Euclio is paying him less tribute. Note, incidentally, that on account of the use of the neutral coordinating particle atque ('and', 'but') vero cannot plausibly be assigned a connective, adversative role. (9)

Ubi is obiit mortem qui mihi id aurum credidit, coepi observare, ecqui maiorem filius mihi honorem haberet quam eius habuisset pater. Atque ille vero minus minusque impendio curare minusque me impertire honoribus (' After the death of him who had committed the gold to my keeping, I began to observe whether the son would hold me in greater honour than his father had. But in fact (vero), his neglect grew and grew apace, and he showed me less honour', PI. Aul. 15-19)

corpus of quasi enim (all of them in Cicero), and quasi autem is not attested at all. 10. Note that examples like (8) may give a clue as to how vero has come to be viewed as an adversative conjunction: a construction like a vero, non b ('a in reality, not b') is essentially equivalent to the construction non b, vero a ('not b, in reality a'). In the latter type of construction, in which the vero-member comes second, vero could be easily mistaken for an adversative conjunction, more or less equivalent to sed ('but').

290

§11.2

Likewise in (10), three hypothetical clauses with the pluperfect subjunctive - a construction which is used to express an unrealized possibility in the past - are followed by an emphatic announcement of the matching reality, which accordingly contains the particle vero: three different and justifiable policies that Verres could potentially have chosen are contrasted with the policy he actually (vero) follows. (10) Cum tibi senatus ex aerario pecuniam prompsisset ... quid facere debuisti? Si quod L. Piso ille Frugi ... quod superaret pecuniae rettulisses; sin, ut ambitiosi homines aut benigni, ... ex senatus aestimatione ... solvisses; sin ut plerique faciunt, ... sumpsisses id nummorum quod tibi senatus cellae nomine concesserat. Hoc vero quid est? ('When the senate drew that money from the treasury for you ... , what ought you to have done? Had you acted like L. Piso Frugi, you would have ... repaid the balance to the treasury. Had you acted like a kindly man or sought after popularity, ... you would have paid the farmers at the rate fixed by the senate. Had you done, as most men do, ... you would have exacted money to the amount allowed you by the senate for maintenance. But now, what is the situation actually (vera)?', Cic. Ver. 3.195)

In contexts like (10) it is not unusual to find vero combined with the 'actualizing' adverb nunc ('now'). Compare (11) and (12)11: (11)

... si non ad homines verum ad bestias, aut etiam si in aliqua desertissima solitudine ad saxa et ad scopulos haec conqueri ac deplorare vellem, tamen omnia muta atque inanima tanta et tam indigna rerum acerbitate commoverentur. Nunc vera cum loquar apud senatores populi Romani ... timere non debeo ... (' ... if I were minded to tell this tale of suffering and wrong to the stones and rocks of some lonely desert waste, cruelty and injustice so awful as this would rouse sympathy even in the world of mute and lifeless things. And since those who I am in fact (vero) addressing are senators of Rome, ... I may rest assured that ... ', Cic. Ver. 5.171)

(12) Percepit sapientiam, si quis tam securus moritur quam nascitur; nunc vero trepidamus cum periculum accessit ('A man has caught the message of wisdom, if he can die as free from care as he was at birth; but as it is in fact (vero), we are all a-flutter at the approach of the dreaded end', Sen. Ep. 22.16)

In (13) the non-actual counterpart with which the state of affairs of the vero-

II. The selected part of the Ibycus corpus contains 132 instances of the combination nunc vera, including a small number of cases in which nunc does not have the same 'actualizing' force (e.g. after a hypothetical statement) as in (\ I) and (\2). Comparable with (\ 1) and (\2) are e.g. Cic. Aft. 2.1.12; Liv. 45.8.4.

representational function of vero

291

unit is compared, is presented in the form of a false assumption on the part of the addressee Lucilius: Lucilius might expect Seneca 'to talk about the weather', but in actual fact (vera) Seneca will be occupied with much more serious matters: (13)

Putas me tibi scripturum quam humane nobiscum hiemps egerit, quae et remissa fuit et brevis, quam malignum ver sit ... Ego vero aliquid quod et mihi et tibi prodesse possit scribam ('Do you suppose that I shall write and tell you how kindly the winter season has dealt with us, - a short season and a mild one, - or what a nasty spring we are having, .,. No; I shall in fact (vera) communicate something which may help both you and myself', Sen. Ep. 23.1)

In conclusion we can say that the instances of vera cited in this section all show to varying degrees traces of its concrete lexical meaning 'in reality' or 'in truth'. They differ, however, from adverbs in a strict (i.e. non-modal) sense in that they appear to evaluate (part of) the propositional content, rather than being themselves a constituent element of the propositional content. Nonetheless they appear to belong, in one way or another, to the represented content of the host unit (rather than e.g. to some aspect of the communicative situation), and could thus be considered to have a function on the representationallevel of discourse. Note, moreover, that (despite their obviously similar behaviour) only some of the examples of vera in this section could be properly described as adversative conjunctions, the alternative description proposed here having more general applicability. This observation clearly challenges the correctness of calling vero an adversative connective in the examples (10) - (12) above, as most grammars and commentators would be inclined to do. 11.2.3 vero in reactive moves Almost all grammars and reference books mention the use of vero as an affirmative adverb/particle in (confirmatory) reactions 12 • An example is (14): (14)

Ms: Menaechmum, opinor, te vocari dixeras Mn: Ita vero (,You said your name was Menaechmus, I believe # I did indeed', PI. Men. 10951096)

12. See e.g. KUhner-Stegmann (1912: I, 798): "vera als nachdriick1ich bekriiftigendes Modaladverb".

292

§11.2

I would argue that this use of vero might be considered a specific instance of its objective modality-marking use discussed above in 11.2.2. The only difference from the examples cited there is that the non- or less-actual (hypothetical, presupposed, etc.) state of affairs and its actual counterpart are spread over two different moves (an initiating and a reactive one) by two different speakers. Thus in (15) speaker 2 confirms that the state of affairs proposed by speaker 1 as potentially applicable to the current represented world, really (vero) obtains. Depending on the specific interactional function (see chapter 4) of the initiating move, the reactive move containing vera may be interpreted as an endorsement, a confirmative answer, a consent, compliance, and so on. The reactive move can also count as a correction, namely in those cases where the state of affairs referred to by speaker 1 is evaluated by speaker 2 as not (or not fully) valid or applicable, and is therefore to be replaced by a more correct (true, actual) alternative. In the following I will illustrate both categories of instances (i.e. vera-reactions with a confirmative and those with a corrective import) with a number of examples. Confirmative reactions may be found anywhere in a dialogical sequence. However, they are especially prone to occur after an utterance in which a 'non-actual' (potential, etc.) state of affairs is expressed, that is, a state of affairs which may possibly obtain. Such utterances may take the form of a cautious statement or suggestion, as is the case in (14) above (cf. opinor, 'I believe'). The matching reactive move with vera counts as an endorsement. Other examples are (15) and (16)13. In (15) the aspect of non-actuality or doubted actuality is clear from the addition credo ('I imagine'), in (16) from the hypothetical si-clause: (15) M: Fuisti saepe, credo, cum Athenis esses, in scholis philosophorum A: Vero, ac libenter quidem (,You have, I imagine, since you have been in Athens, often attended philosophy lectures # Certainly, and I did so readily', Cic. Tuse. 2.26) (16) A:

Lucus quidem ilIe et haec Arpinatium quercus agnoscitur saepe a me lectus in Mario. Si manet ilIa quercus, haec est profecto; etenim est sane vetus. Q: Manet vero, Attice noster, et semper manebit ('Surely I recognize that grove yonder and this oak tree of Arpinum as those of which I have read so often in the "Marius"; if that famous oak still lives, this is

13. Comparable with (15) and (16) is PI. Cas. 402, where vera occurs in an endorsement after a wish: Quod bonum atque fortunatum mihi sit # Ita vera, et mihi (,Heaven be with me and bring me luck! # Yes, yes, and me!').

representational junction of vero

293

certainly the same; and in fact it is a very old tree # That oak lives indeed, my dear Atticus, and will live for ever', Cic. Leg. 1.1)

Other, though closely-related, types of utterance in which the state of affairs referred to is inherently 'non-actual', are the so-called yes-no questions (in which the actuality of the state of affairs is, so to speak, left open). It is not uncommon to find 'reactive' vera in the answers to such questions. Consider (17) - (19), which are examples of confirmative vera in an answer to a yes-no question introduced by the interrogation particles -ne and an 14 • Note that in (18) the question-answer pair is rendered indirectly: (17) M: Dasne aut manere animos post mortem aut morte ipsa interire? A: Do vero (,Do you grant that souls either survive after death, or else perish by the mere fact of death? # I do grant it indeed', Cic. Tusc. 1.25) (18)

'" eum interrogat velletne ad Masinissam reverti. Cum effusis gaudio lacrimis cupere vero diceret, tum ... (' ... he summoned him and asked whether he would like to return to Masinissa. When he shed tears of joy and said that he was indeed (vera) eager to do so, then ... ', Liv. 27.19.11-12)

(19)

A: Adeone me delirare censes, ut ista esse credam? M: An tu haec non credis? A: Minime vero ('Do you suppose me so crazy as to believe such tales? # You don't believe them true then? # Certainly not, indeed', Cic. Tusc. 1.10)

Other types of utterance which inherently harbour a non-actual, potential state of affairs and can accordingly prompt a confirmative or consentient reaction, are proposals and commissives. Examples (20) and (21) serve as illustration: (20)

"Ergo" inquit (sc. Trimalchio) " ... quare non vivamus? ... coniciamus nos in balneum ... " "Vero, vero" inquit Habinnas "de una die duas facere, nihil malo" (' "Well, well", he said" ... why should we not live? ... let us jump into a bath. ... " " Indeed, indeed, " said Habinnas " making two days out of one is my chief delight" " PetL 72.2-4)

(21)

Volurnnius in Sarnnio res prosperas esse ait, litteris eius accitum venisse; quae si falsae fuerint nec usus sui sit in Etruriam, extemplo conversis signis abiturum. "Tu vero abeas" inquit (sc. Appius), "neque te quisquam moratur"

14. Comparable examples of vera in a reaction to a ne-question are PI. Am. 755; As. 58; Men. 1108; for vera in an answer to an an-question, see e.g. also Ter. Ad. 468 and Plin. Ep. 7.11.2. Vera-reactions can also occur after num- and nanne-questions. For the former see e.g. Cic. Rep. 1.61 and de Orat. 2.259; for the latter Div. 2.114 and Fin. 2.10.

294

§ 11.2 (,Volumnius replied that affairs were prospering in Samnium, and that he had come at the behest of Appius' own letter, but that if this were a forgery and he were not needed in Etruria, he would immediately face about and march back. "Yes indeed, go!" cried Appius "No one hinders you" " Liv. 10.18.13)

We can explain in a similar way instances of vero such as that cited in (22), which differ from (20) and (21) only in that the initiating move (in 22 a proposal or promise made by Atticus in an earlier letter) remains implicitl5 : (22) (The opening of a letter to Atticus) ... Tu vero age, quod scribis, de Faberio (' ... Pray, do as you say about Faberius', Cic. Att. 13.32.1)

A last type of dialogical context in which 'reactive' vero may occur is the orderlrequest-compliance pair. Note that orders and requests (as subclasses of directive speech acts) inherently contain non-actual states of affairs. Examples are (23) - (25)16: (23) Ph: Sequere hac me, mi anime Ar: Ego vera sequor (,Come along with me, darling # Indeed I will', PI. As. 941) (24) To: ... amplectere sis Le: Ego vero ('please clasp me in your arms! # Indeed I will', PI. Per. 764) (25)

"Qua re, si tibi est commodum, ede illa quae coeperas et Bruto et mihi" A: C: "Ego vera, inquam, si potuero, faciam vobis satis" (' "So then, if it suits you, go on now and tell us what you then began, for the benefit of Brutus as well as for me" "Indeed I will" I replied "try to satisfy your curiosity if I can" " Cic. Brut. 20-21)

The following example too may be described in terms of a request-compliance pair, even though the request element is not explicitly expressed (cf. ex. 22 above)I?: (26) (Opening of a letter to Atticus) Ego vero Quinto epistulam ad sororem misi ('I indeed sent Quintus your letter for your sister', Cic. Att. 13.41.1)

More or less along the same lines one could describe the use of vero in corrective reactions. Corrective reactions with vero differ from the IS. Similar examples in Cicero's letters areAtt. 12.45.2, Fam. 16.10.1 and 16.23.1. I do not agree with Kuhner-Stegmann's remark (I, 798) that vera in Fam. 16.10.1 occurs "auJ3erhalb der Antwort". 16. See also e.g. PI. As. 645; Paen. 1418 (which is identical with example 23); Sen. Ep. 6.4. 17. The vera-clause in (26) could in principle be described also as an answer to a neutral yes-no question, but this is not very likely in the context concemed.

representational junction of vero

295

confirmative reactions in that speaker 2 challenges (instead of confirms) the actuality, validity or applicability of the state of affairs suggested or alluded to in the preceding move by speaker 1, and replaces it by what he considers to be more applicable or true in respect of the present situation. I give two examples 18. In (27) the corrective vero-reaction follows on a cautious statement (cf. forsitan, 'perhaps') in the preceding move. In (28) the vero-reaction corrects the suggestion implied in a yes-no question: (27)

"Sed de Graecis hactenus; etenim haec ipsa forsitan fuerint non necessaria" Tum Brutus: "Ista vero" inquit "quam necessaria fuerint non facile dixerim" (' "So much then for the Greeks, and even this perchance was superfluous" "How far, actually, from superfluous" rejoined Brutus "I cannot easily tell" " Cic. Brut. 52)

(28) Sc: "Quid? domi pluresne praesunt negotiis tuis?" Lae: "Immo vero unus" (' "How about your residence in the city? Are several persons in charge there?" # "On the contrary, in/act only one" " Cic. Rep. 1.61)

The last example nicely demonstrates that the signalling of corrective reactive moves is only a side-effect of the use of vero, the actual marker of corrective reactions being immo (,rather'). It is perhaps not immediately apparent how, taken in isolation, the use of vero

in reactive moves as illustrated in this section is to be integrated within a general framework for the description of particles, such as that outlined in the present study. At first sight vero might perhaps be considered to play a role on the interactional level of discourse, more specifically as a marker of interactional relations between the constituent moves of an exchange (see ch. 4, section 4.2.3.1): in the examples (14) - (28) cited above, vero seems in a sense to 'connect' two alternating moves within an interactional exchange. There is, however, an essential difference from 'real' markers of interactional relationships, such as at (see ch. 12) or immo, which are markers of objecting and corrective reactive moves, respectively. For one thing, in contrast to at or immo for instance, vero seems to be capable of occurring independently in

18. Some other examples are Cic. ruse. 1.112; 2.39; 5.22; Alt. 1.16.10; Sen. Ben. 1.14.2. Later in this chapter we will see that the use of vera in corrective reactive moves is on the border between the representational and the interactional use of the particle.

296

§11.2

one-word sentences, more or less as an equivalent of English yes 19 • Consider (15) above and (29)20: (29)

" ... tu orationes nobis veteres expJicabis?" "Vera" inquam "Brute" (' " ... won't you go over with us those early speeches?" "Certainly, Brutus" I said', Cic. Brut. 300)

Vera's capacity to function as a one-word sentence in the above-mentioned examples might prevent us from calling it a particle in a strict sense. Unlike 'real' particles, vero does not seem to indicate here how a certain (informational or communicative) unit of discourse fits with, or is related to, its textual or communicative environment. Rather, vera seems to function as a discourse unit itself, having more in common with 'content words' than with full-blown 'pragma words'. Another possible way of describing vera in instances such as (15) and (29) would be, then, to rank it with the interjections, as is the common solution for English yes and no and comparable expressions. In the examples involved vera seems to have in common with the interjections that from a syntactic point of view it can be relatively independent from the surrounding text ('real' particles, by contrast, always need to be associated with a host unit, being themselves devoid of any meaning in a lexical-semantic sense of the word). However, calling vera an interjection in (15) and (29) (and, by extrapolation, in all other examples in this section), has a number of drawbacks. For one thing, the concept of 'interjection' is itself not very well defined either: interjections usually figure in the grammars as a class of words which can in no way be classified with any other group of words, except perhaps for the equally mysterious particles21 •

19. In Latin various devices are used to compensate for the absence of real equivalences of English yes and no. See Kiihner-Stegmann (p. 531-532); Thesleff (1960) and Pinkster (1972: 138140). A good impression of the variety of Latin yes-substitutes is provided by the passage PI. Men. 1106-1109. The successive means of expression are: performative verb (jateor, 'I admit'); repetition of the predicate (est); resumptive adverb (ita, 'thus', 'so') in combination with vera; independently occurring adverb (certo, 'certainly'): Ubi lubet, raga: respondebo. nil reticebo quod sciam # Est tibi nomen Menaechmo? # FATEOR # Est itidem tibi ? # EST # Patrem fuisse Moschum tibi ais ? # ITA VERO ... # Esne tu Syracusanus ? # CERTO ('Ask (sc. the questions) when you like; I'll answer. Nothing that I know will I keep back # Is your name Menaechmus ? # That I admit # And you, is yours also? # It is # Your father was Moschus, you say? # I do, indeed ... # Are you a Syracusan ? # Certainly'). 20. See also e.g. Cic. Leg. 2.46; Mur. 65; PI. Mer. 685; True. 302. 21. Not surprisingly interjections are sometimes treated as a sub-class of the particles.

representational function of vero

297

What is more, the syntactic 'independence' of vero as assumed in (15) and (29) appears to be of an altogether different type from that of English yes and no: vero is most likely to be taken together with an implied predicate, which is to be derived from the preceding context. This is strongly indicated by the other examples cited in this section. In (16), (17), (18), (21), (23), and (25) the predicate of the preceding move is repeated in the vero-reaction, while in (14) the predicate is implied by the resumptive adverb ita ('thus', 'so'). This provides support for the assumption that the predicate is strongly implied in (15) and (29) as well. In this respect it is also instructive to set side by side (15) and the very similar example (30). The resemblance between (15) and (30) (the initiating moves of the exchanges in 15 and 30 are both cautious statements (cf. the hedging opinor and credo) and have a largely similar composition) more or less proves that vero in (15) is short for vero fui ('I've been indeed'): (30) M: Cadere, opinor, in sapientem aegritudinem tibi dixisti videri A: Et vero ita existimo ('You said, I think, that in your view the wise man is susceptible of distress # That is in fact my opinion', Cic. Tusc. 3.12) (15) M: A:

Fuisti saepe, credo, cum Athenis esses, in scholis philosophorum Vero, ac libenter quidem

A third argument against classifying vero as an interjection is provided by examples such as (18) above, in which vero is part of a reported speech: undisputed interjections such as English yes and no are excluded from indirect speech. All in all we may conclude that in the use of vero in reactive moves there are no arguments that force us to assume the existence of a reactive interactional particle vero, or of an interjection vero which is equivalent to English yes22 • As I stated already at the beginning of this section, there is more reason to assume that we are dealing here with a specific instance of the objective modality marking use of vero as discussed in the preceding section 11.2.2. The circumstance that vero occurs in the reactive move of a dialogical passage need not necessarily have implications for the basic discourse function to be assigned to vero. At most we can say that in those contexts in which the

22. Note that the interjection-solution (vera = yes) could in no way be applied to the use of vera in corrective reactive moves. This may count as another argument against the interjection-solution.

298

§11.2

represented world and the communicative situation coincide (as is the case in the examples adduced in this paragraph), the representational and interactional roles of vero may appear to merge. In support of the proposed view we might adduce again example (13), which I discussed in 11.2.2 among the rather straightforward instances of objective modality marking vero. On closer inspection it now appears that here too the vero-clause can be considered in a certain respect to be a reactive move, although strictly the passage is not conversational23 . The example thus nicely illustrates how closely the instances of vero cited in 11.2.2 are related to the instances of vero that occur in reactive moves: (13) Putas me tibi scripturum quam humane nobiscum hiemps egerit, quae et remissa fuit et brevis, quam malignum ver sit .... Ego vero aliquid quod et mihi et tibi prodesse possit scribam (Sen. Ep. 23.1)

11.2.4 summary of the representational uses of vero

In summary, then, we can say that there are a number of contexts in which vero shows, to varying degrees, the features of an objective modality marker with a function on the representational level of discourse. Characteristic of such markers is that they stand midway between undiluted 'content' words with a clear lexical or referential meaning, and so-called 'pragma' words with a more discourse-pragmatic meaning. I started this chapter by discussing a number of instances of vero that are near the 'content' pole of the continuum and would accordingly usually be termed 'adverbs' (11.2.1). Other instances of vero are intuitively taken to be more particle-like in that they are not themselves constituent parts of the state of affairs referred to in the clause, but rather evaluate the content of the clause 'from the outside'24 (11.2.2). That they still cannot be considered full-blown particles in the sense of undiluted 'pragma' words, is suggested by the fact that they clearly convey a lexical-referential meaning. This meaning ('in reality', 'actually', etc.) is explicitly counterbalanced by an opposite meaning in the preceding context, in the form of an expression of a state of affairs that is not

23. Example (13) is illustrative of the 'diaphonic' discourse type, i.e. of a monological discourse with overt dialogical 'traits'. See chapter 5, section 5.2.2. A similar instance is Sen. Ep. 57.7. 24. The grammar of KUhner & Stegmann (vol. I: 793) speaks of 'modal' adverbs as opposed to 'normal' adverbs. A syntactic correlate of the distinction between 'normal' and 'modal adverbs' is that usually only the former can by themselves constitute the answer to a wh-question, whereas only the latter can form the answer to a yes-no question.

representational junction of vero

299

(or less) actual/real. In 11.2.3 I discussed the use of vero in reactive moves, which appeared to be not really different from the objective modality marking use, although the relation with the interactional function of vero is perhaps more apparent here. In the next section I will discuss instances of vero that have a more manifest interactional nature, and the relation of these instances with vero's representational function.

11.3 Interactional vero

The characteristic feature of vero in all its occurrences is its capacity to indicate a high degree of actuality, validity, appropriateness, etc. of a certain element in a certain world. This common feature is captured by a basic meaning with a concrete lexical origin. In some of its occurrences, as we have seen above, the actuality-feature displayed by vero pertains to a certain state of affairs in the represented world, as compared to an (implicit or explicit) possible or non-actual alternative. In such cases vero was said to playa role on the representational level of discourse, as a so-called objective modality marker. Most often, however, the actuality-feature of vero appears to pertain predominantly or solely to the interactional level of discourse: not only acts and events in the represented world, but also communicative (speech) acts, speaker intentions, and the like can be evaluated in terms of their actuality, validity, appropriateness, etc. In this case the standard by which the actuality status is determined is not provided objectively by the represented, external world, but is vested in the subjective, internal world of speaker and addressee, that is, in the communicative situation in which the speech event takes place. Whenever vero seems somehow to be oriented towards indicating the involvement of the speaker in the speech event (or his management of the conversation), rather than at conveying content, I will assign to vero a function on the interactional level of discourse. More specifically we may say that vero in its interactional function indicates (i) (as a subjective modality marker, see 11. 3 .1) a high degree of personal commitment on the part of the speaker/author to the truth or appropriateness of the communicated content, or (ii) (as a conversation particle, see 11.3.2) the sincerity of the speaker's illocutionary intention or the actuality of the intended perlocutionary effect.

§ 11.3

300 11. 3.1 vero as a subjective modality marker

On p. 283 above 1 stated that subjective modality markers (or attitudinal particles) reveal the speaker's personal stance with regard to the truth, tenability, appropriateness, applicability, degree of expectedness, etc. of the transmitted content of an utterance. Some examples of subjective modality marking vero are (31) - (34): (31) nec enim habeo quid scribam nec hoc tempore quicquam difficilius facio; ad te vero et ad nostram Tulliolam non queo sine plurimis lacrimis scribere ('I don't know what to write, and just now there is nothing 1 find greater difficulty in doing. To you, for sure, and our darling Tullia 1 cannot write without a flood of tears', Cic. Fam. 14.2.1) (32)

... exclamat tua familiaris, "Hoc vero neminem umquam audivi!" Ego repressi (' ... your friend exclaimed: "That, for sure, 1 never heard anyone ... " 1 interrupted her', Cic. Aft. 15.11.2)

(33) (Cicero is relating Verres' habit of stealing statues; in the preceding section he has been concerned with the theft of a statue of Mercury) illud vero quid sit iam non queo dicere, quo nomine appellem nescio, quod in C. Marcelli statua ('Under what head to class, by what name to describe, his use of the statue of Marcellus for his crime, it is indeed beyond my power to tell', Cic. Ver. 4.89)

The subjective modality marker vero can be said to resemble the objective modality marker vero in that both are somehow concerned with evaluating the propositional content of their host unit. There is, however, also an essential difference between the two. Objective modality marking vero is used primarily to indicate, quite simply, that a possible state of affairs actually applies, or that in reality another state of affairs obtains than the one assumed. Subjective modality marking vero, on the other hand, displays the additional feature of indicating that the speaker personally takes full responsibility for the truthfulness of the content of the host utterance. For an example like (31) this means that it is not fully correct to render vero by 'it is indeed/ in facti in reality the case [viz. that 1 have to cry whenever 1 write to you]', as would be appropriate in the case of objective modality marking vero. Rather, paraphrases like 'you can trust me', 'you can count on that', 'I tell you', 'I assure you' and the like (which reveal the interactional character of vero) seem to be suitable in the context of (31 )25.

25. In the Loeb edition vera in (31) is wrongly rendered 'but'. The Bude edition does not translate vera at all.

interactional junction of vero

301

The same holds for (32) and (33): the immediate context in which vero occurs does not readily support an objective modality marking interpretation of the particle. We may further illustrate the difference between objective and subjective modality marking vero with example (34): (34) (A conversation about cattle-herding; Varro's discussion partner has just stated that sometimes in disolated districts provisional enclosures are made for the cattle) "... Longe enim et late in diversis locis pasci solent, ut multa milia absint saepe hibernae pastiones ab aestivis." "Ego vera scio" inquam "nam mihi greges in Apulia hibernabant, qui in Reatinis montibus aestivabant ... (' "... For they usually graze far and wide in all sorts of places, so that frequently the winter grazing grounds are many miles away from the summer." "I am well aware of that (ego vero scio)" said 1 "for 1 had flocks that wintered in Apulia and summered in the mountains around Reate ... " " Var. R. 2.2.9)

In this example vero seems at first sight to resemble the objective modality marking vero in confirmative reactions (e.g. in an endorsement after a cautious statement, see 11.2.3 above). However, in such a case we would expect the repetition of a state of affairs from the preceding context, in the form of a repeated predicate or the resumptive adverb ita. This is obviously not the case in (34?6. It thus seems more reasonable that vero in (34) plays a role on the interactional level of discourse: by adding vero the speaker does not so much emphasize the actuality of the state of affairs scio (e.g. in reaction to a prior remark of the discourse partner about the speaker's state of knowledge, cf. e.g. 'It is indeed true that I know this', 'I actually know this', etc.), but principally emphasizes the trustworthiness (and hence the acceptability) of the speaker's current utterance. As such the function of vero in (34) is of the same order as enim in the preceding move by the discussion partner, which also pertains to the interactional level of discourse (see ch. 8). Whereas the function of enim can be described as soliciting the commitment of the addressee to the speech event,

26. Note also that the initiating move in (34) cannot readily be characterized as a cautious statement.

302

§11.3

vera works as a way of highlighting the speaker's own commitmenf7. Both particles can accordingly be regarded as overt manifestations of the conversation management taking place. This amounts to saying that the subjective modality marker vero might perhaps be considered as in effect belonging to a special subcategory of the conversation particles: in its capacity of subjectively evaluating the (truth, appropriateness, etc. of the) content of an utterance it may at the same time serve a further-reaching conversational goal (stressing the speaker's personal responsibility for the truth of the transmitted content may for instance forestall a possible negative reaction on the part of the addressee and hence help to further the conversation in a smooth and uninterrupted way). As to the more concrete linguistic clues for assigning a subjective modality marking function to vero (instead of e.g. an objective modality marking function) we can point for instance to the frequent combination of vera with the personal pronoun ego28 • As we have seen before, subject pronouns in Latin (ego, tu, nos, vos: '1', 'you', 'we', 'you'-pl) are essentially optional elements, the person and number distinctions already being coded in the finite verb form of the clause. An explicit subject pronoun appears to be quite normal, however, when the subject referent receives a more than usual emphasis. For the first person singular this is the case for instance when a speaker stresses his own knowledge, experience, opinion and the like, in short when he wants to underline his personal involvement and the subjectivity of his jUdgmenf9. The Latin subject pronoun ego is accordingly often accompanied by a verb expressing an experience or opinion, and is highly compatible with attitudinal particles and other subjective or expressive elements. Since explicit ego appears to perform, as one of its functions, the underlining of the personal involvement and responsibility of the speaker, its occurrence in a vero-clause could be taken as a possible indication of a subjective modality marking (i.e. interactional) function of vera (compatibility criterion). Consider for instance (34) above and the following example, which contains also the expressive element herde:

27. Notice that the functions of subsequent enim and vera in (34) could in a sense also be described in terms of 'initiating' and 'reactive'. For a discussion on the functional similarities between enim and vera (on account of which they are frequently collocated in an utterance) I refer to 11.5.2. 28. The combination ega vera occurs 190 times in the selected part of the Ibycus corpus; 108 of these instances are in Cicero. 29. See Pinkster (1986; 1987) and my remarks in ch. 10, pp. 235-236.

interactional junction of vero

303

(35) Tr: Scio te bona esse voce, ne clarna nimis Da: Ego hercle vero clarno ('I know you have a good voice. Don't strain it, yelling # By Hercules, 1 will certainly yell!', PI. Mos. 576-577)

The occurrence of ego (and other subjective elements) in the same clause is however never alone decisive for assigning a subjective modality marking status to vero, as can be illustrated with example (36) in which an objective modality marking interpretation of vero seems not to be totally excluded: (36) Quod de domo scribis, hoc est de area, ego vero tum denique mihi videbor restitutus, si illa nobis erit restituta. Verum haec non sunt in nostra manu ('As to what you write about our house, or rather its site, 1 shall never, I assure you, feel myself fully restored until that has been restored to me', Cic. Fam. 14.2.3)

In (36) it is perfectly possible to take vero as merely indicating the actuality status of the state of affairs restitutus: in that case the interpretation should be that Cicero will not feel himself actually (vero) restored, until his estate has been restored to him. The presence of emphatic ego cannot form a real counterargument to such an interpretation, as it seems to be due to contrastive or parallel focus (ego vs. illa, note the fIrst position of ego in the clause) rather than to emphatic subjectivity. On the other hand, the opinion verb mihi videbor and the forceful expression tum denique ('then, and only then') may be taken to point to a more subjective and expressive value of vero in this clause. In such a subjective modality marking interpretation of vero the particle could be paraphrased with for instance 'you can take my word for that'. Perhaps we should conclude that in instances like (36) vero displays indistinguishably both functions at the same time 30 • The existence of such ambiguous instances once more illustrates the continuous gradient of the 'content' word vera at one extreme and the 'pragma' word (conversation particle) vera at the other. In the next section I will deal with a number of other interactional instances of vera which do not (or less clearly) display this ambiguity.

30. A comparable mixture of objectively and subjectively evaluating features of vera can be assumed for many instances of vera in reactive moves (see 11.2.3 above). When the non- or less actual alternative to the state of affairs evaluated by vera has the form of an assumption or belief of the (real or imaginary) addressee, it is often hard to decide whether we are dealing with representational (objective modality marking) or interactional (subjective modality marking) vera, especially when the vera-clause contains one or more 'subjective elements'.

304

§ 11.3

11.3.2 other interactional instances ofvero: vero as a conversation particle In most of the examples discussed hitherto vero has occurred in declarative sentences with an assertive illocutionary force. In such a context it is not always easy to decide whether vero plays a role on the representational or on the interactional level of discourse. This can be illustrated with the invented example (37), which has two possible English translations. Which one is to be preferred is strongly dependent on the context. Note that the first interpretation is more or less automatically implied by the second, but not vice versa: (37) frater vero heri advenit a) my brother did indeed arrive/ really arrived yesterday (as we had all expected for so long; as you have already suggested; etc.) (representational vero) b) really (i.e. I'm serious; you can take my word for that; I tell you; etc.), my brother arrived yesterday (interactional vero)

That it is useful and even necessary to distinguish an interactional function of vero is however demonstrated more clearly by the occurrence of vero in utterances with a non-assertive illocutionary force, especially in directives. In such communicative acts it is much less feasible that vero belongs wholly or predominantly to the represented content of the utterance. The following examples are all taken from Latin comedy3l: (38)

Si concordabi' cum illa, habebi' quae tuam senectutem oblectet: respice aetatem tuam # Te oblectet, tibi habe # Minue vero iram ('If you hit it off with her, you'll have somebody to be the charmer of your old age. Do have thought for your grey hairs # Let her be your charmer, keep her for yourself # Now, now, less temper! (lit.: really, restrict your anger)', Ter. Ph. 433435)

(39)

(Euclio urges Lyconides to return the pot of gold he assumes Lyconides to have stolen from him) Illam (sc. aulam) ... cedo. I, refer. Dimidiam tecum potius partem dividam. Tam etsi fur mihi es, molestus non ero. I vero, refer ('Give it me. Go, bring it back. You can have half of it, yes, yes, I'll divide. Even though you are such a thief, I won't make any trouble for you. Do go, I urge you, and bring it back!', PI. Aul. 766-768)

(40)

(Sagaristio is drunk) Nequeo, leno, quin tibi saltern staticulum olim quem Hegea faciebat. Vide vero, si tibi sati' placet

31. Other examples of vero in directive utterances in Latin comedy are PI. Am. 946; Bae. 1062; Cas. 727; Ep. 3; St. 186; Ter. Ad. 268. Vero regularly (but not necessarily) appears after a negative (verbal or non-verbal) reaction of the addressee to a prior directive.

interactional junction of vero

305

(,Pimp, I can't refrain from dancing for you that lovely little dance that Hegea used to render. Do see now, if it isn't rather winsome', PI. Per. 824-825)

In the above examples vero appears to indicate that the speaker is highly committed to the actual realization by the addressee of the intended action. In more general terms, vero can be said to emphasize the sincerity of the illocutionary intention of the speaker, or, viewed from a different perspective, the actuality of the perlocutionary effect intended by the speaker. By doing so it lends the directive a more urgent character. The more specific interactional effect of adding vero to the directive depends crucially on the context and the type of directive involved. The presence of vero can for instance reveal the irritation of the speaker in a repeated order; it can have a reassuring or encouraging effect in a granting of permission; it can make a request more urgent and picture the speaker in a supplicatory role, etc. Comparable examples of interactional vero in directive utterances outside the comedies are (41) and (42)32. In both cases it is hard to render the exact force of vero, although its conversation-managing role is quite clear: (41)

Age vero, nunc inserite oculos in curiam, introspicite penitus in onmis rei publicae partes (' And now, if you please, cast your eyes over the senate-house, and scan deeply every department of public life', Cic. Font. 43?3

(42) sin facinus ... detegitur, vos vero et liberos Germanici et nos parentes iustis solaciis adficite ('but if murder comes to light ... , then, I urge you, do you see to it that proper requital is made to the children of Germanicus and to us, his parents', Tac. Ann. 3.12)

For the sake of completeness I mention here also a group of vero-containing directives in which the particle might perhaps be better described in representational rather than in purely interactional terms. I am referring mainly to the use of confirmative vero in reactive moves as discussed in section 11. 2.3 above. Let us recall examples (21) and (22), the latter of which is repeated here 34 :

32. Some other examples are Cic. Aft. 1.11.3; Phil. 2.34; Liv. 2.12.14; Sen. Ep. 3.2. 33. There are 7 other instances in Cicero's orations of the exhortative formula age vero ('come!'),

potentially followed by a directive utterance. The formula has the function of calling for special attention or consideration. The ossified imperative form age behaves as an interjection. 34. See also e.g. Cic. Aft. 12.45.2; Fam. 16.23.1; Rep. 6.26.

306

§11.3

(22) (Opening of a letter to Atticus) ... Tu vera age, quod scribis, de Faberio (' ... Pray, do indeed as you say about Faberius', Cic. Aft. 13.32.1)

In examples like (22) vero does not merely strengthen the illocutionary intention or perlocutionary effect connected with the ongoing communicative act: rather than being just a 'metaeomment' about the communicated message (with a role in the conversation management; see exx. 38 - 42 above), vero appears to be itself a part of this message, given that it corresponds to a prior content element (viz. in a preceding letter). As such it is quite closely related to the state of affairs referred to by the predicate. Another instance of a directive utterance in which vero appears to have clear representational features is (43): (43) Nee ... praecipio tibi neglegentiam. Tu vero metuenda declina (,But I do not advise you to be indifferent. Rather do you turn aside from yourself whatever may cause you fear', Sen. Ep. 98.6-7)

The fragment could be read as follows: Lucilius might perhaps assume that Seneca wants him to adopt a passive attitude. This assumption is false (nee ... neg/egentiam). In actual fact (vero) Seneca counsels more active behaviour. Note that (43) is fully comparable with e.g (13), which was discussed in the section on objective modality marking vero (see p. 291 and p. 298). Examples like (21) and (43) thus warn us against assigning too readily to vero the interactional function of strengthening the directive illocutionary force of the ongoing communicative act, as most manuals are inclined to do; sometimes the particular conditions of use of vero point to a more specific function that is closely related to its original lexical meaning and accordingly can just as well be located on the representational level of discourse. Although the unadulterated interactional function of vero can be demonstrated best in the context of directive utterances (see exx. 38 - 42 above), it is not excluded with other illocutionary types. In (44) for instance vero is used in the context of a commission35 : (44) Sed tamen in his vel asperitatibus rerum vel angustiis temporis obsequar studiis nostris ... tibi vero, frater, neque hortanti deero neque roganti, nam neque auctoritate quisquam apud me plus valere te potest neque voluntate

35. Cf. also the subjective modality marking use of vero in exclamations; see e.g. Cic. Ver. 5.14; Au. 1.16.5; 15.11.2 (= ex. 32). For vero in an unfulfillable wish introduced by utinam see e.g. Vitro 3, praef. 1.

interactional function of vero

307

(,But none the less, though events are so harassing and my time so restricted, I will hearken to the call of our studies ... And (? For) really, when you, brother, exhort and request me, I will not fail you, for no man's authority or wish can have greater weight with me than yours', Cic. de Orat. 1.3-4)

Although it may be a matter of dispute what the precise function of vero in (44) is, it is clear at least that it cannot sensibly be assigned a representational function. Vero appears rather to underscore the sincerity of the speaker's illocutionary intention, that is, it indicates that the speaker's promise (neque ... deero) is to be considered absolutely sincere (vero: 'you can count on that', 'I guarantee'). Note that vero has the specific side-effect of laying stress on the immediately preceding word36 • We face here an issue to be explored more fully in section 11.4.3. To round off the discussion on the 'conversation particle' vero, a few remarks on the use of vero in interrogative clauses may also be in order. The most significant fact to note on this issue is that I have not found any straightforward examples of vero in prototypical (wh- or yes-no- ) questions; not surprisingly, most instances of vero-containing interrogative clauses appear to be rhetorical questions, which, however, are effectively assertions rather than questions. However, there appear to be quite a large number of instances of vero in what might be called 'verifying questions'. In such questions the speaker checks the actuality or correctness of (part of) a preceding communicative act by a different speaker. Such verifying (or echoing) questions may thus be regarded as a specific type of reactive move which, so to speak, temporarily hold up the progress of the discourse. The lexical origin of vero is quite clear, cf. 'Do you really mean X?', 'X indee&', 'Really so?', and so on. The verifying questions concerned are mostly fixed expressions of the type ain vero? (,You really say so?'), itane vero? ('Really so?'), or qui vero? ('X indeed? How/why so?'). They are illustrated in (45) - (47): (45)

Servo'sne es an Iiber? # Utcumque animo conlibitum est meo # Ain vero? # Aio enim vero (,Are you slave or free? # Whichever I please # So? In sooth? # Yes, so in sooth', PI. Am. 343-344)37

36. Cf. the commentary of Wilkins (1892) ad loc.: " ... vera here, as so often, has a climax force, which can only be reproduced by laying a stress upon the word it qualifies". 37. For the expression ain vera? see also PI. Am. 284; As. 722; Per. 184; Tru. 306; Ter. Eu. 803; Ad. 405.

308

§ 11.3

(46)

... equidem hercJe opus hoc facto existimo, ut illo intro eam # Itane vero, vervex? Intro eas? # Quid aliud faciam? (' ... Lord man! I certainly take it that what I ought to do is to go inside there! # Do you, really, you old wether? Go inside, eh? # What else should I do?', PI. Mer. 566-568)38

(47)

Eho an vidisti, pater? # Vidi. Verum non ex usu nostrost neque adeo placet # Qui < quia> non nostra formam habet dignam domo (,Oho! You saw her, father? # I saw her. But she won't do for us, she really isn't suitable # Indeed? Why not? # Because her looks are out of keeping with our household', PI. Mer. 393-395)39 vero? # Quia -

A further type of 'verifying question' are those introduced by the particle an. According to Bolkestein (1988) an-questions are typically used to ask for confirmation (i.e. verification) of an implication contained in the preceding words or behaviour of a (real or fictitious) discussion partner. In the light of this observation it does not come as a surprise to find an accompanied by vera, as is illustrated by (48)40: (48) Ad me adiri et supplicari egomet mihi aequom censeo. An vero nugas censeas, nihil esse quod ego nunc scio? ('I hold it proper that I be approached myself, that I be appealed to. Canst think, forsooth, that what I now know is but a trifling thing of no account?', PI. St. 293294)

Note that the use of vera in (45) - (48) is not essentially different from the use of vera discussed in section 11.2.3: in the latter case we are dealing with confirmative statements, while the examples (45) - (48) involve requests for confirmation. In both cases vera is to be seen as being itself part of the transmitted message rather than as a sort of metacomment which plays a role merely in the management of the interaction. Accordingly I am inclined to assign to vera in (45) - (48) a function on the representational level of discourse, although the particle seems to display some interactional features as well. Unambiguous instances of interactional vera in questions are, in fact, 38. For ita vera? see also e.g. PI. Cur. 725; Mil. 844; Rud. 971; Ter. An. 926. 39. The fonnula qui vera? appears to comprise two different aspects: speaker 2 is checking the truth or correctness of (part of) the preceding utterance by speaker 1, by asking at the same time for more explanatory (and hence supporting) infonnation. Other instances of qui vera? are PI. Aul. 339; Cas. 720; Mil. 296; Paen. 634. Comparable is the use of vera in Rud. 794: ... dabitur tibi magnum malum :: quam magnum vera? ('You shall get a good hard hiding :: Really now? How hard?'). 40. The selected part of the Ibycus corpus contains 88 instances of the combination an vera, 50 of which are from Cicero. It is noticeable that the interrogation particles num and nanne, which lack the specific 'verifYing' capacity of an, cannot be combined with vera.

interactional junction of vero

309

hard to find. The only instance in my sample that might possibly qualify for such a status is (49), in which vera appears to function as a 'metacomment' on the entire communicative act (vera could be rendered with 'the actual question rather is: .. .'). (49)

. .. mirabar quid hie negoti esset tibi ... # Die sodes, pater, tibi vero quid istie est rei? (' ... I wondered what business you could have here # Rather - please tell me father -, what brings you here?', Ter. Ad. 642-644)41

11.4 Side-effects At this point in the discussion the two main discourse functions of vera (representational and interactional) and the corresponding actual uses, as well as their mutual interrelatedness, have been amply introduced and illustrated. It is now possible to turn our attention to a number of further features that are traditionally assigned to vera, and especially to the widespread misconception that vera should be ranked among the adversative connectives. On the latter point we have already seen that in some of the alleged occurrences of the 'adversative connective' vera, vera is in fact better described as an objective modality marker, functioning on the representational level of discourse (see 11.2.2, p. 291). In such cases either the vera-clause is rather to be taken as asyndetically connected to the preceding clause, or the specific link between the two clauses is designated by an unambiguous conjunction. In the following (11.4.2) I will show that other instances of the alleged 'connective' vera fit very well into the picture of interactional vera sketched above in 11. 3. This will lead to the conclusion that the 'adversative connective' use of vera is in fact to be regarded as a side-effect of one of its two main discourse functions. Likewise I will consider two other side-effects of the use of vera (11.4.3) which might be easily mistaken (and actually have been so) for a proper function of the particle. The first has to do with an alleged text-organizational use of vera, viz. the use of vera to mark transitions in the thematic structure of the discourse (ostensibly comparable with a major use of autem, see ch. 10). A second, related side-effect, involves the use of vera as a focus particle. 41. I deliberately depart here from the translation in the Loeb edition. Note also that KiihnerStegmann (p. 80) cite this example to illustrate the development of the particle into an 'adversative' conjunction.

310

§ 11.4

The sections dealing with the side-effects of vero (11.4.2. and 11.4.3) will be prefaced in 11.4.1 by a discussion on the pragmatic motivations for the use of interactional vero in non-conversational texts, and, interrelated with this issue, on the typical (linguistic and other) features of the immediate verocontext.

11.4.1 preliminary remarks: pragmatic motivations and linguistic clues As I stated above, by using interactional vero the speaker may emphasize the truth or appropriateness of the message conveyed, or the sincerity of his illocutionary intention. Such a particle will be especially prone to occur with linguistic units (ranging from single items to entire utterances) that for some reason or other have a high degree of challengeability, that is, units that on account of their particularly salient (surprising, controversial, etc.) import run the risk of being challenged by the addressee. In such contexts the use of a 'speaker-authority' marker like vero forestalls a possible negative reaction on the part of the addressee (such as disbelief, lack of understanding, noncompliance, etc.), and thus guarantees an uninterrupted continuation of the discourse. It should be emphasized that, notwithstanding the essentially interactional basis of the notion 'challengeability', challengeable utterances are not confined to strictly conversational contexts. One of the major tenets of this study is, in fact, that monological texts may to various degrees reflect the fact that communication is taking place between a speaker (in this case the author) and an addressee (the reader). Moreover, monological texts may contain embedded voices of reported characters. For the conversational traits of monological texts I refer to section 5.2.2.2. Potentially challengeable contexts in monological texts are provided by for instance utterances with a polemic, controversial, unexpected, surprising, outrageous, unprecedented, or simply remarkable content. In the following I will show that the majority of the instances of vero in non-conversational texts (including the alleged 'adversative connective' instances) occur in the immediate presence of a content element with a relatively high degree of challengeability . The use of vero as a means of diminishing the degree of challengeability of (part of) an utterance usually seems to have a highlighting effect on the text unit concerned: the presence of an expressive particle indicating the subjective involvement and commitment of the speaker more or less automatically lends

side-effects of vero

311

the unit involved the status of prominent information in comparison with the surrounding text. As such vero may be said to play indirectly a role in indicating the 'informational relief of a text, that is, in distinguishing prominent from less prominent text parts (side-effect 2, see 11.4.3 below). The abstract notions of challengeability and saliency that appear to be characteristic of the prototypical vero-context (both in conversational and in non-conversational texts) can be 'translated' into a number of concrete linguistic clues. In a sense these can be taken as distributional properties of the particle, and as such may serve as a heuristic tool for co-determining the specific discourse function(s) of vero (see chapter 5). The following contains an overview of the most apparent of these clues. It is based on an investigation of about 400 instances of vero within their contexts, taken for the most part from Caesar, Celsus, Cicero, Livy, Petronius, Plautus, Sallust, Seneca, Tacitus and Terence. distributional properties of vero

(i)

(ii)

immediate presence of 'climax' elements such as: adjectives and adverbs in their superlative form; adjectives, adverbs and verbs with an inherently superlative meaning, including compound forms with intensifying prefixes such as prae- and per_42; adjectives and adverbs indicating an extreme on a gradable scale, such as totus (,whole', 'all', 'totar), orrmis ('all', 'everyone'), nullus ('none'), nenw ('no one'), nihil ('nothing'), numquam ('never'). co-occurrence with one of the following particles: (a) the 'focus' particles or adverbs iam ('now'), etiam ('even', 'also'), et ('even', 'also'), adeo or tam ('to such a high degree'), ne ... quidem ('not even'), and the like; (b) the interactional, agreement-soliciting particle enim (see also 11.5.2.2 below and chapter 8); (c) particles that introduce challenging reactive moves (objections, protests, etc.), such as at or imnw (challenging reactive moves are ideal contexts for particles that are capable of diminishing a rela-

42. For compounds with prae- in combination with vera see e.g. Cic. Ver. 3.35 and 3.223 (praeclara and praeclaras, 'outstandingly bright', resp.); with per- e.g. Ver. 1.153 (perfacile, 'very easily').

312

§ 11.4

tively high degree of challengeability, such as vero or enim)43; the negation particles nee and neque ('and not')44; utterances introduced by nee or neque are challengeable in about the same way as protests and objections introduced by e.g. at or immo: they commonly involve a rejection of a presupposition contained in the context or situation, or of a generally held opinion (see also ch. 8, pp. 196-197)45. (iii) the presence of a focus construction, such as the correlative patterns cum ... tum vero ... ('when ... , then, indeed, ... ') or cum vero ... tum ... ('when, indeed, ... then ... ') (i.e. vero in combination with a resumptive or preparatory adverb or pronoun). (d)

In addition to the above clues, which are indicative of the more specific interactional function of vero, the prototypical vero-context usually contains also a number of features that hold for interactional particles in general. These features are all somehow connected to the fact that interactional particles typically occur in a dialogical or diaphonic discourse type46 . They include: (iv) the presence of second- and, in the case of vero, especially first-person pronouns and verb forms; the use of present tense verb forms in a passage that is otherwise presented in the past; the presence of metacommunicative or performative expressions; the immediate presence of subjective expressions, such as puto and credo ('I think', 'I believe'); the use in the immediate context of questions and directives; occurrence in a rhetorical question; the presence of interactional elements such as interjections, swear words and vocatives. For a more elaborate discussion on this issue I refer to section 5.2.2.2, pp. 111-115.

43. There are 105 instances of the combination at vera in the selected part of the Ibycus corpus. The majority (74 instances) come from the works of Cicero, which also contain a high number of instances (59) of the combination imma vera. 44. See also KUhner-Stegmann (p. 81). 45. The combination nec/neque vera is extremely common, esp. in Cicero, in whose works there are 30 I instances (i.e. more than 10 percent of all the vera-instances in Cicero). When we also take into account the figures for at vera and imma vera in Cicero (cf. note 43) we find that about 20 percent of the vera-instances in Cicero occur in polemic utterances introduced by nec/neque, at or imma. 46. The two most common examples of diaphonic discourse are (i) the indirectly rendered, reported conversation between the internal characters of a narrative, and (ii) mono\ogical stretches of discourse that linguistically reflect the communicative interaction taking place between the author and his implied audience. See section 5.2.2.2.

side-effects of vero

313

Most vero-units in my sample (which come for the most part from mono logical texts) have one or more of the above mentioned characteristics, or some other clue indicating that the unit involved has the status of an informational climax with a relatively high degree of challengeability. This can be illustrated with a number of examples. I will first cite some random examples of vero in an absolute climax, that is, in a text unit that is climactic mainly on account of its own (surprising, remarkable, controversial, etc.) content (exx. 50 - 52). Next I will give a few examples of vero-units that form the last climactic element in a cumulative string (relational climax, exx. 53 - 58).

(a) absolute climax (50) Illud vera sine ulla dubitatione maxime nostrum fundavit imperium et populi Romani nomen auxit, quod ... Romulus ... (,What, for sure, without any doubt has done most to establish our Empire and to increase the renown of the Roman people, is that Romulus .. .', Cic. Balb. 31) (Indications: sine ulla dubitatione, ('without any doubt'); the superlative form maxime ('the most', 'preeminently'); the focus construction illud ... quod; cf. e.g. also ex. 32: neminem umquam ('no one ever'» (51) (On the necessity of avoiding crowds) Nihil vera tam damnosum bonis moribus quam in aliquo spectaculo desidere ('Nothing, really, is so damaging to good character as the habit of lounging at the games', Sen. Ep. 7.2) (Indications: nihil ('nothing'); tam) (52) Sed confecto proelio, tum vera cerneres quanta audacia ... fuisset in exercitu Catilinae ('When the battle was ended, then indeed (tum vera) you could have witnessed what boldness and resolution had pervaded Catiline's army', Sal. Cat. 61.1) (Indications: the presence of a focus construction of the cum .. . tum-type; the second person verb form cerneres ('you could have witnessed'), which signals a switch from the reported layer of the text to the communicative layer. Both devices may be indicative of a climactic passage within a narrative text. Compare, by way of contrast, a more 'neutral' alternative construction like confecto praelio magna audacia in exercitu Catilinae apparuit ('after the battle it became clear how brave Catilina's soldiers had been'). In 11.4.3 I will come back to the combination tum vera and to other features of climactic passages in narratives.)

(b) relational climax (vero in a cumulative string) (53) tres albi sues in triclinium adducti sunt ... quorum unum bimum nomenculator esse dicebat, alterum trimum, tertium vera iam sexennem

314

§11.4 (,Three white pigs were led into the dining-room ... ; one of them was two years old, the announcer said, the second three, and the other as much as (vera iam) six', Petr. 47.8) (Indication: content)

(54) arcessivit ad se, promisit intercessit dedit. lam vera (0 di boni, rem perditam!) etiam noctes certarum mulierum ... non nullis iudicibus pro mercedis cumulo fuerunt ('He sent for everybody, made promises, gave security, paid money down. Yes, believe it or not (good heavens, what a scandal!), even the favours of certain ladies ... were given to some of the jury to swell the bribe', Cic. Att. 1.16.5) (Indications: the cumulative string; the focus particles iam and etiam; the parenthetically added author's comment, making explicit the outrageous character of the content; the content itself) (55) (Perseus to his brother Demetrius) Regnare utique vis. Huic spei tuae obstat aetas mea, obstat gentium ius, obstat vetustus Macedoniae mos, obstat vera etiam patris iudicium ('You wish of course to be king. In the way of this ambition of yours stand my age, the law of nations, the ancient custom of Macedonia, and, last but not least, also the decision of our father', Liv. 40.9.8) (Indication: the cumulative string; etiam)

An interesting subgroup of examples of vera-units in a 'cumulative string' is illustrated by (56) - (58). In these examples the vera-unit is not only climactic (as in exx. 53 - 55), but also corrective (cf. the addition of imnw, 'rather'). Accordingly vera appears to have interactional as well as representational features (for the representational nature of vera in corrective reactions, see 11.2.3 above). The examples (56) - (58) thus illustrate once more that the representational and the interactional function of vero cannot always be clearly separated, but in fact shade into each other. (56) Interea quidem cum Musis nos delectabimus animo aequo, immo vera etiam gaudenti ac libenti (,Meantime I shall settle down to the enjoyment of the Muses with resignation, indeed even with hearty goodwill and delight', Cic. Aft. 2.4.2) (Indications of climactic status of vera-unit: content; etiam; immo) (57) Quod rogas ut in bonam partem accipiam si qua sint ... quae me mordeant, ego vera in optimam (' As to your request to take it in good part, if there is anything ... that wounds my feelings: I injact promise to take it in the best possible part', Cic. Att. 11.7.8) (Indication: the superlative form optimam, which replaces the positive degree

side-effects of vero

315

banam)47 (58)

Ecquid placent? # Ecquid placeant me rogas? Immo hercle vera perplacent ('So you're quite satisfied, sir? # Quite satisfied? Lord, Lord, man, I'm supersatisfied, and that's a jact', PI. Mas. 906-907) (Indications: PERplacent; hercle; imma)48

11.4.2 side-effect I: the 'adversative connective' vero We can now at last address the issue of how to deal with the 'adversative connective' use of vero that is familiar from the Latin reference books (see section 9.2.2)49. One thing that is clear is that it is only a restricted number of the vero-instances that might qualify for the label 'adversative connective'. An adversative connective interpretation is doubtful for instance in those cases in which there is no sequential discourse relationship between the vero-unit and a preceding matching conjunct, or in which vero is accompanied by some other, indisputably connective particle. We have seen plenty of such instances of vero in the preceding sections, most of which are from conversational (or conversation-like) texts. As to the other vero-instances (i.e. the ostensibly adversative connective instances, usually occurring in continuous, monological texts 50), a few observations may be of interest. First, granted for the moment that vero could indeed be regarded as an adversative connective, it would in any case not have the same range of use as the adversative connectives sed and verum, which can both occur in the structural type of relationship labelled 'substitution' in section 9.1. This means that we do not come across vero as a linking device in constructions of the type not A but B or not only A but also B51. In this

47. According to the OLD (s.v. vera, 5b) vera is used in this example to emphasize the preceding personal pronoun ego: 'for my part'. It will be clear by now that my explanation of the occurrence of vera in this example is quite different. 48. Comparable examples are PI. Cas. 38 and Cic. Rep. 1.66. 49. I take the notion of adversativity in the broad sense indicated in section 9.1. 50. In the section on vera in reactive moves (11.2.3, pp. 295-296) I argued that vera (in contrast to e.g. at or imma) cannot be considered a marker of sequential discourse relationships on the interactional level of discourse. 51. There are a few ostensible instances of vera in a 'substitution construction', but in these cases vera is preceded by sed. Cf. e.g. Lucr. 4.986: non homines solum sed vera animalia cuncta ('not only humans, but all and every living being') In my opinion we are dealing here with interactional vera in a climactic text unit. Other examples are Catu!. 115.8 and Cic. Ver.5.14.

316

§11.4

respect vero resembles rather the adversative connectives autem and at, which cannot occur in the 'substitution-construction' either. A second relevant point is that in continuous, monological passages verounits may well maintain sequential relations of a non-adversative type with the preceding text. This is the case in (59) for instance, where it is not easy to take the vero-clause as contrastively related to the preceding clause. The quite limited applicability of the notion of adversativity thus clearly challenges the appropriateness of distinguishing an 'adversative connective' function of vero. (59) (Concerning Marcellus' irreproachable behaviour compared to Verres' greed) Nihil in aedibus, nihil in hortis posuit, nihil in suburbano ... Syracusis autem permulta atque egregia reliquit; deum vero nullum violavit, nullum attigit ('He set up nothing in his mansion, in his garden, in his country-house near Rome .,. He left Syracuse a great number of very beautiful things. And (vero: last but not least), he did not profane or so much as touch a single one of its gods', Cic. Ver. 4.121)

The last, and most important observation, is that the alleged instances of 'adversative' vero are commonly found in the same quite specific type of context as outlined above in 11.4.1, that is, in salient, climactic text units with a high degree of challengeability. We can make this point clear by considering examples (60) - (63). Example (60) is adduced by the OLD (s. v. vero, 7a) as an instance of vero "with mild adversative force", to be rendered 'on the other hand' or 'at the same time'. This description implies a strong resemblance between vero and autem, the latter of which can indeed be considered a weak adversative connective (see chapter 10)52. On account of the clear climactic status of the vero-clause it is however not impossible, and in my opinion even better, to regard vero in (60) as an interactional evaluation particle of the same type as discussed in the preceding section, and essentially different from the adversative connective autem. (60) si puer parvus occidit, aequo animo ferendum putant, si vera in cunis, ne querendum quidem ('If a small child dies, the loss must be borne calmly; if (si vera) an infant in the cradle, there must not even be a lament', Cic. Tusc. 1.93)53

52. Cf. e.g. the OLD s. v. autem: " (expresses contrast ... without any pronounced adversative sense) On the ather hand, while". Example (60) is also adduced by Marouzeau (1948: 242) as an instance of 'adversative' vera, to be compared with autem. 53. The same argument may be applied to Apul. Met. 55, which is also mentioned in the OLD (s. v. vera, 7a) under the heading "vera with mild adversative force".

side-effects of vero

317

It should be emphasized that (60) is not an isolated case. In fact, the majority

of the vero-instances that might seem at first sight to have an adversative connective function occur in a text unit which conveys salient, climactic information. As a further example consider (61), which in the OLD (s. v. vero, 7b) is mentioned under the heading "with stronger adversative force", with the translation 'however' or 'yet'. In my opinion the occurrence of vero is not to be explained by some sort of adversative relationship with the preceding context, but rather by the highly remarkable and salient content of the verounit itself, which enhances the use of an interactional, 'speaker-authority' marker. Indications for such an interpretation are the presence of a cumulative string and the superlative form paucissimis. (61) ceteros (sc. consulatus) aut novem aut sex aut quattuor aut tribus mensibus (sc. gessit), secundum vero paucissimis horis (,The rest (viz. of the consulships) he held for nine, six, four, or three months, the second (vero) lasted only a few hours', Suet. Aug. 26.4)54

Also in (62) it is tempting to assume a strong adversative force of vero, as the translator of the Loeb edition seems indeed to do (witness the translation 'however'). However, the climactic content of the vero-unit, as well as the preparatory pronoun hoc ('this') point clearly towards vero's regular use as a subjective interactional particle: (62) Quem Orsines intuens "Audieram" inquit "in Asia olim regnasse feminas; hoc vero novum est, regnare castratum!" (,Orsines with a glance at him said: "I had heard that women once reigned in Asia; this however is something new, for a eunuch to reign!" ',Curt. 10.1.37)

As a last example I cite (63), which in Saur (1913: 85) is mentioned among the instances of 'adversative vero'. The translation in the Loeb edition cited here also seems to suggest an adversative interpretation of vero: (63) Delector iucundum tibi fuisse Tironis mei adventum; quod vero scribis ... plurimos manumissos, unice laetor ('I am charmed that the advent of my friend Tiro was agreeable to you. But I especially rejoice that (as your letter informs me) ... you manumitted a large number of slaves', Plin. Ep. 7.32.1)

54. A comparable example from the same section in the OLD is Cic. Brut. 112.

318

§ 11.4

In this example the vera-clause and the immediately preceding sentence may at first sight seem to behave as a pair of parallel utterances maintaining the weak type of adversative relationship that is usually signalled by autem. However, what in my opinion explains the use of vera (instead of e.g. autem), is the fact that we are dealing here with a cumulative sequence of units rather than with a pair of functionally equal units. There are at least two indications that the vera-clause is a rather forceful expression containing salient information: (i) the combination unice laetor ('I am particularly happy'), especially in comparison with the more neutral delector ('I am glad') in the preceding clause; and (ii) the use of preparatory quod ('what') instead of e.g. an accusative and infinitive construction, as in the preceding clause. A better translation of (63) would in my view be: 'as to your remark that you manumitted a large number of slaves, at that I especially rejoice'. Vera is hard to render in English. Summarizing we can say that there are admittedly quite a number of cases in which a unit containing vera maintains some sort of adversative relationship (usually of the type 'semantic opposition', see 9.1) with a preceding unit. However, there are strong indications that it is not vera that is to be held responsible for the specific adversative connotation involved, nor that it is the primary function of this particle to underline this adversative relation. A comparison with the use of vera outside the context of an adversative relationship suggests that the alleged 'adversative connective' vero behaves more as an interactional situating particle (or, in other terms, as a modal or phatic particle): the conditions of use of the 'adversative connective' vera appear to be basically the same as those of the 'modal adverb/particle' vera (see 11.3 and 11.4.1). This observation leads to the tentative conclusion that the 'adversative connective' use of vera may in fact be nothing more than an additional (though quite frequent) side-effect of one of the particle's two basic discourse functions 55 • To put it differently, vera is highly compatible with adversative. relationships, but does not directly signal them. The view of vera advocated here has the advantage that it has greater explanatory power than the assumption that vera is occasionally to be taken as an adversative conjunction.

55. In some cases the alleged adversative connective vera displays the features of a representational evaluation particle rather than of an interactional evaluation particle. See my remarks on p. 291 in 11.2.2, and examples (10) - (12).

side-effects of vero

319

11.4.3 side-effect 2: vero as an ostensible marker of the organization of the text or as a focus particle; the chiaroscuro effect Some studies and reference books hint at what I would call a text-organizational use of vero. DeWitt (1938: 453) for instance observes that "vero is also continuative at times ('moreover', 'furthermore'), denoting a new stage in the argument"S6. The qualification 'continuative' or 'transitional' is meant to capture those instances of vero which seem to occur between two units of the text which do not maintain a' specific (adversative, causal, etc.) semantic or rhetorical relationship. An example is (64): (64) (How Verres, the governor of Sicily, in successive years despoiled the community of Herbita) Primo anna ... Anno secundo ... Anno tertia vero ... consuetudine usus est regia ('In Verres' first year ... In the second year ... Now (vero) in the third year, he adopted a practice that may fairly be described as "royal" " Cic. Ver. 3.75-76)

If vero in cases like (64) can indeed be shown to signal merely a transition in the thematic development of the discourse (in a manner comparable with the discontinuity marker autem, see chapter 10), we will be forced to assume for vero a separate, connective discourse function on the presentational level of discourse, in addition to its situating functions on the representational and interactional level. In fact, however, the pragmatic motivation underlying the use of vero at thematic transition points may be shown to be quite different from the motivation for the use of the discontinuity marker autem. The decisive factor here is again, that also in the alleged 'continuative' cases of vero we usually find the climactic or cumulative context that is typical of interactional vero (see 11.4.1 above)57. This means that the vero-units concerned have the status of salient, central information. Such units, as we have seen earlier, highly favour the occurrence of a particle that emphasizes the authority of the speaker, his personal opinions with regard to the message conveyed, or the sincerity of his 56. Cf. also e.g. Saur (1913: 80): "Vera ist Adversativpartikel und driickt entweder ein gegensatzliches Verhaltnis aus ader es dient in abgeschwachter Bedeutung als Ubergangspartikel" (emphasis mine, CK), and Szantyr (p. 494): "Bei den Historikem erseheint es auBerhalb der Reden ziemlieh abgesehwaeht und nieht von autem versehieden, ebenso bei Petron., wo es geradezu als Ubergangspartikel auftritt". For the latter part of this statement see also Petersmann (1977: 251). 57. Comparable observations are made also by KUhner-Stegmann (p. 81), and by Menge (1960: 348): "Vera hat als Partikel ... die naehdriieklieh steigemde Bedeutung "erst gar, vollends", oder hebt dann, wenn dieser Affekt nieht so hervortritt, in der Bedeutung "aber, femer aber" den neuen Gedanken als einen Gewiehtigeren hervor".

320

§11.4

illocutionary intention. It might therefore be more fitting to regard the alleged transitional, text-organizational function of vero as an accidental side-effect of its regular interactional discourse function, which obtains only when the verounit happens to coincide with a thematic boundary in the discourse. This can be illustrated with a number of representative examples. In the example cited in (64) above Cicero describes how the governor Verres despoiled in successive years the Sicilian community of Herbita. The composition of the passage is quite straightforward (viz. it is built around a series of temporal settings), and might lead one to expect in the second or last member of the series a discontinuity marker, such as e.g. autem (see 10.4.2.1 and 10.4.3.1). However, the use of vero (instead of e.g. autem) in the last section of the description appears to be well-motivated in that the story that follows clearly outweighs the two preceding units in length and significance of content. The two preceding units form, so to speak, the overture to the climax that is to follow 58 . We find vero also at larger thematic breaks in the discourse, at the beginning of a new expository or narrative section. Again I would contend that the use of vero is due to the highly surprising or salient content of the vero-unit itself, rather than to the transition in the thematic structure of the discourse. It should be emphasized, moreover, that the occurrence of vero at paragraph boundaries appears to be quite coincidental, and is much less frequent than

autem in this position. Examples (65) - (67) serve as illustration.

58. The gradual build-up to a climax containing vera is, in fact, a major compositional device in the Verrines. Cf. e.g. also Ver. 1.60: Nam in isto, iudices, hoc novum reperietis. Audimus aliquem tabulas numquam confecisse ... Audimus alium non ab initio fecisse, sed ex tempore aliquo confecisse ... Hoc vera novum et ridiculum est, quod hic nobis respondit ... usque ad M. Terentium et C. Cassium consules con feci sse, postea destitisse ('You will find this novelty in Verres' case, gentlemen of the jury. We have heard of a man's never keeping any accounts ... We have also heard of a man's not keeping accounts to begin with, but doing so from a certain moment onwards ... But what we have here (vera) is a ridiculous novelty, that he told me that he had kept them duly up to the consulship of M. Terentius and C. Cassius, but stopped keeping them after that', Cic. Ver. 1.60) Other examples in the Verrines are 1.16; 2.185; 3.47; 4.37 and 4.51. An example from the Brutus is: Hoc autem studium non erat commune Graeciae, sed proprium Athenarum. Quis enim aut Argivum oratorem aut Corinthium aut Thebanum scit fuisse temporibus illis? ... Lacedaemoniurn vera usque ad hoc tempus audivi fuisse neminem ('The pursuit of oratory however was not shared in by Greece as a whole, but was peculiar to Athens. Who ever heard of an orator of Argos or Corinth or Thebes of that period ... ? And as for a Spartan orator (vera), I have never down to the present day heard of one', Cic. Brut. 4950)

side-effects of vero

321

In (65) vero is part of a rhetorical question with the status of a praeteritio, which inherently counts as a rather forceful and expressive kind of statement59 : (65) (New section) In sartis tectis vero quem ad modum se gesserit quid ego dicam? ('His conduct in dealing with the maintenance of public buildings, (vero) I need hardly describe!', Cic. Ver. 1.128)

Example (66) is part of an extensive exposition about lions. In the preceding section Pliny has dealt with the taming of lions. In the vera-clause cited here a new subtopic is broached within the main 'lions'-theme. The occurrence of vero can however be explained on account of the surprising content of the utterance: lions are usually not supposed to display the quality of mercifulness. (66) (New section) Sunt vero et fortuitae eorum quoque dementiae exempla ('Now there are also instances (vero: 'believe it or not') of occasional mercifulness even in lions', Plin. Nat. 8.56)

The vera-clause in (67), lastly, contains various signals that point in the direction of a climactic, relatively salient piece of information: the reinforcing adverb utique ('absolutely'); the alliterative combination of the 'heavy' adverbs imnwdice and imnwdeste ('excessively' and 'immoderately'); and the use of a lively infinitive of narration ('historic infinitive': gioriari, 'he boasted'), to which I will come back below. (67) Minucius vero cum iam ante vix tolerabilis fuisset rebus secundis ac favore volgi, tum utique immodice immodesteque non Hannibale magis victo ab se quam Q. Fabio gloriari ('As for Minucius, success and the favour of the crowd had already made him well-nigh insufferable. But now at all events, he cast away all modesty and moderation and boasted of his triumph not only over Hannibal but over Quintus Fabius as well', Liv. 22.27.1)

The examples (64) - (67) discussed above make clear also that the observed text-organizational side-effect of the use of vera in monological texts can be specified further as a chiaroscuro effect. By this I mean that vera is highly compatible with units of the text that count as communicatively more prominent or relevant than the neighbouring units. Another way of saying this is that vera is one of several different devices that together may be somehow indicative of the 'informational relief of a text. 59. Closely comparable are Ver. 1.49 and 1.95; some other examples in the Verrines of vera at paragraph boundaries are 4.58 and 4.89, the latter of which has ,been discussed on p. 300 as (33).

322

§ 11.4

In argumentative or expository prose, for instance, we often find vero combined with the particle iam ('now') at the transition to a new thematic unit that is 'heavier', more to the point, more startling, etc. than the preceding unit(s)60. lam in this combination calls attention to a new important point in the exposition, whereas vero displays its regular interactional function of emphasizing the speaker's personal commitment to the message conveyed. In narrative prose vero may in the same way be indirectly involved III revealing the structure of a narration. It has more than once been pointed out in the literature that episode structure in Latin narrative texts depends partly on the alternation of two different narrative modes or styles: a neutral (historical, objective) mode and an affective (experiential, subjective) mode 61 . The neutral mode is reserved for the unmarked relating of facts and events, whereas in the affective mode the author reveals his own attitude towards his narrative account, or presents the narrated facts and events as the experiences of a person or group present in the events. The affective mode is hence considered to be livelier than the neutral mode. The alternation of the two modes is one of the devices available for distinguishing episodes from one another, in that sections with a significant status within the narrative (such as decisive turning points, sudden, dramatic changes in the narrated situation, new actions with far-reaching consequences, new, highly dramatic situations evoking strong emotions, etc.) tend to be expressed in the affective mode. The alternation of the two narrative modes is, in other words, a particular chiaroscuro technique 62 . Now the important observation for the present discussion is that the use of vero in narratives is confined largely to climactic passages in the affective mode. This is proved by the fact that the particle usually co-occurs with linguistic devices associated with the affective narrative mode, such as the historic infinitive and present (and the interplay with other tense forms)63, the narrative 'incidence marker' tum ('then'; tum can be considered in a sense the narrative counterpart of transitional iam in expository or argumentative texts,

60. Cf. e.g. the comment of Wilkins ad Cic. de Drat. 1.8: "iam vera calls attention to the fact that the next point is one of even more importance". 61. See e.g. Chausserie-Lapree (1969), who speaks of 'recit soutenu' versus 'recit dramatique' or 'anime', and Viljamaa (1983) who, referring to Weinrich (1964: 55-72) and Lyons (1977: 688-9), distinguishes a historical and an experiential narrative mode. 62. For a more extensive discussion of this phenomenon, cf. also Kroon (J994a). 63. For the occurrence and function of the historic infinitive see especially Chausserie-Lapree (1969) and Viljamaa (J983); for the historic present Chausserie-Lapree (e.g. on p. 371). Sznajder (1993) discusses both phenomena.

side-effects of vero

323

see above )64, or typically interactional particles such as enim (see also exx. 54 and 55 in section 8.4.1)65. An instance of vero in a passage containing a historic infinitive is (67) above. A further, very telling example of a chiaroscuro structure in a narrative passage is (68), in which the chiaroscuro effect is clear not only from the collocation tum vero, but also from the alternation of the 'backgrounding' imperfect tense (adpropinquabant) with the 'foregrounding' perfect tense (tulerunt). See also my remarks in chapter 7, e.g. p. 157 and n. 21. (68)

lam signa quoque legionum adpropinquabant. Tum vero Masaesulii non modo primum impetum sed ne conspectum quidem signorum atque armorum tulerunt ('And now the units of legionaries also were approaching. Then indeed (tum vero) the Masaesulians failed to sustain not only the first attack but even the sight of the standards and arms', Liv. 30.11.11)

The examples (69) and (70) further illustrate the collocation tum vero in the immediate presence of a historic infinitive66 , while in (71) tum vero occurs in a clause with a historic present67 : (69)

Tum vero ingenti sono caelum strepere et inter horrendos fragores micare ignes;

metu ornnes torpere ('And now (tum vero) the heavens resounded with a frightful tumult, and between the terrific crashes the lightning flashed; .. , they were all stunned with fear', Liv. 21.58.5) (70)

Ita cons ilium Britannorum in ipsos versum, transvectaeque praecepto ducis a fronte pugnantium alae aversam hostium aciem invasere. Tum vero patentibus locis grande et atrox spectaculum: sequi vulnerare capere, atque eosdem oblatis aliis trucidare ('Thus the tactics of the Britons recoiled on themselves, for our squadrons passed over by the general's order from the front of the battle and attacked the enemy's line from behind; then indeed (tum vero), wherever the open ground permitted, began a grand and gory drama of pursuit, wounding, capture, and then - as other fugitives crossed our men's path - of butchery of the captives', Tac. Ag. 37.2)

64. For the combination tum vera (less often tunc vera or ibi vera) see Chausserie-Lapree (1969: 520-31). He describes the combination as a mise en scene technique employed in passages with a strong affective or dramatic colouring. As to the contribution of the vera-element in this collocation however he is not very specific. It is noticeable that the conditions of use of tum vera appear to be much the same as the conditions of use of the historic infinitive as outlined by Viljamaa (1983) for Livy. As to the frequency of occurrence of tum vera Chausserie-Lapree observes that there are 14 occurrences in Vergil's Aeneid (bk 1-6), 6 in Tacitus, 6 in Sallust, 13 in Curtius, 8 in Caesar, and 39 in Livy. 65. For enim see ch. 8. For the combination enim vera in narrative texts see also ChausserieLapree (1969: 532-7). 66. Some other examples are Liv. 5.7.8-11; Sal. Jug. 84.1; 94.3. 67. Cf. e.g. also Caes. Gal. 5.37.3; Cic. eluent. 14; Liv. 22.17.6; 40.40.10.

324

§ 11.4

(71) (The Persian enemy is waltmg for Alexander's army on the summit of a rock) Quod ubi contemptu sui pergere vident, tum vera ingentis magnitudinis saxa per montium prona devolvunt (,But when they saw that in scorn of them it (i.e. Alexander's army) continued to advance, then indeed (tum vera) they rolled stones of huge size down the slopes of the mountains', Curt. 5.3.18)

The chiaroscuro effect associated with interactional vero may also obtain at a more local level of the text than is the case in the examples (64) - (71) above. In (72), for instance, we come across vero in the most prominent (i.e. informationally salient) part of a narrative structure which has the highly condensed form of only one (complex) temporal clause68 : (72) Ubi per exploratores Caesar certior factus est tres iam partis copiarum Helvetios id flumen traduxisse, quartam vera partem citra flumen Ararim reJiquam esse, ... (,When Caesar's scouts informed him that three quarters of the Helvetian forces had already crossed the river, but that a quarter (quartam vera partem) remained on the near side of the river Saone, .. .', Caes. Gal. 1.12.2)

A further (and the most local) type of chiaroscuro effect of vero is the focalizing effect that the particle sometimes seems to have on a single word or constituent of the clause69 • Consider for instance (73), in which vero appears to lend focus status to the word insignis: (73) Accedit eodem etiam noster Hortensius, multi praeterea boni; insignis vera opera Favoni fuit (,Our friend Hortensius joined him too, and many other loyalists, while a more than usual energy (insignis vera opera) was displayed by Favonius', Cic. Aft. 1.14.5)

Vero may seem to behave as a focus particle also in those cases where it follows a personal pronoun (ego, tu), as is the case in the earlier cited examples (44) and (57) (cf. pp. 306-307 and 314-315). The same holds for

68. I do not agree with the comment of Riemann (1886: 572) with regard to this example, which implies a full equivalency between vera and autem: " ... on les emploie surtout IA OU Ie grec employerait men ... de ... , pour etablir une relation soit entre deux termes iso\es, soit entre deux propositions, qu'on place, pour ainsi dire, en regard l'une de I' autre". In my opinion the use of vera is justified by the fact that its host unit (the subclause quartam ... reliquam esse) contains the most important information for the continuation of the narrative (in which it is described how Caesar attacks the remaining part of the Helvetii). It is not clear to me why the Loeb edition reads Jere ('almost') instead of vera. 69. In Saur (1913) the focalizing force of vera that is intended here is presented as the first of three main categories of use of vera: (i) zur Hervarhebung; (ii) Adversativpartikel and Obergangspartikel; (iii) zur Steigerung. Cf. also the OLD, s. v. vera, 5b and 5c, and Roby (1874, II: 461-463): " ... [vera) gives special emphasis to the word preceding it".

side-effects of vero

325

vero in combination with a preparatory or resumptive pronoun or adverb, as in (74), where vero may be rendered most adequately by laying a stress upon tum ('then'): (74) Cum per omnem orationem satis frequenti adsensu succ1amatum esset, tum vera ea vociferatio simul indignantium minitantiumque . .. exorta est ut finem dicendi faceret (' Although throughout all the speech there had been often enough outbursts of applause, at that moment indeed (tum vero) such an outcry arose from those who were at once enraged and threatening ... that he put an end to his speech', Liv. 42.53.1)

It should however be borne in mind that vero, although it readily clusters with

salient information units, is not a focus particle in the sense of being used primarily for conferring focus status on one particular constituent of the clause (viz. on the immediately preceding word?o. As a matter of fact it is not always easy to pinpoint the exact element of saliency that might explain the occurrence of vero: usually the presence of the particle can be explained by some general idea of saliency which is dependent on various factors (one of which may admittedly be a neighbouring constituent with strong focus). Likewise it is often difficult to determine the precise host unit of the particle and the full range of its proper force 71 • Thus in (74) a description of vero as a focus particle highlighting the element tum does not seem to capture fully the contribution of vero to the passage. The conclusion is therefore that also the 'focalizing' force of vero is to be regarded as one of the side-effects of the particle's quite specific interactional function as described in section 11.3.

70. It is to be noted that vero - at least in classical Latin prose - is placed quite mechanically in the second position of its host unit (clause, subclause, etc.). Combined with the fact that the first position in a Latin clause is often reserved for a relatively salient piece of information (such as the adjective insignis in ex. 73) this may easily give rise to the unwarranted conclusion that vero may be used as a focus particle in a strict sense. Compare e.g. (63), where it is most unlikely that vero is indicative of focus function of the preceding element. Vero pertains in this example to the entire utterance rather than to one element in particular. 71. I would recall that the concept of 'unit' or 'host unit' is not coextensive with the concept of 'clause'. Cf. ch. 5.

326

§ 11.4

11.5 Conclusions 11. 5.1 a discourse-pragmatic description of vero The accounts in the literature of vero, undoubtedly one of the most elusive Latin particles, are scanty and not very precise. In this chapter I have investigated the properties of vero in the light of the received view that the particle should be ranked partly among the adversative connectives. Although the present analysis does not pretend to be an exhaustive study of vero, at least the following conclusions may be drawn: (i) The common treatment of vero as being both a 'modal' and a 'connective' (notably adversative) particle appears to be unwarranted. The alleged connective (viz. adversative) force actually turns out to be a contextual exploitation (a side-effect) of the particle's more general 'modal' function. Accordingly there is no reason to assume for vero a diachronic development from a 'modal' to a 'connective' function. That we are dealing with only one particle vero is proved by the fact that 'adversative' vero and 'modal' vero have the same, quite specific conditions of use. (ii) More precisely vero is a situating particle with a primary function on the interactional level of discourse and a secondary function on the representational level72. In both cases it is concerned with indicating the actuality or reality of an act or event, viz. of a communicative event or an event in the represented world, respectively. In its interactional function vero draws special attention to the actuality of (part of) the ongoing communicative act. In a way it could be considered to be a metastatement about what the speaker is saying. In this function vero may help to make the communicative act being performed less challengeable (i.e. more acceptable to the addressee), and can hence be said to play a role in the management of the interaction. In its representational function vero is itself part of the communicated message and approaches an adverb in value. The two functions of vero cannot always be kept strictly apart, as the particle in some cases seems to display representational as well as interactional features (see also e.g. pp. 298-299; 303 and n. 30; 308).

72. For the notions of primary and secondary discourse function see section 5.1.

discourse-pragmatic description of vero

327

Making use of the analytical framework proposed in chapters 4 and 5, I come to the following schematized picture of vero, which may count as a closer specification and adjustment of the accounts to be found in Latin reference books 73 : stepwise analysis of the particle vero 1. basic meaning: 2. disc. function:

3. actual use:

4. side-effects:

actuality; reality (i) primary function (interactional level of discourse): indicating the speaker's personal commitment with regard to (part of) the communicative act. (ii) secondary function (representational level of discourse): indicating the actuality of a state of affairs in the represented world. (i) related to the primary discourse function: precluding the potential challengeability. of (part of) a surprising, controversial, climactic, etc. utterance by emphasizing: (a) (as a subjective modality marker) the actuality (i.e. truth) of the transmitted content (b) (as a conversation particle) the actuality (i.e. sincerity) of the speaker's illocutionary aims, which may e.g. lead to a reinforcement of the illocutionary force (see p. 305). (ii) related to the secondary discourse function: (as an objective modality marker) highlighting the actuality of a represented state of affairs as compared to a possible, less actual alternative; in reactive moves representational vero may moreover be indicative of the affirmative or corrective status of the move. (i) adversative connection (ii) chiaroscuro effect: indicating the organization of the text; indicating emphatic focus

73. I leave out of account here the quite exceptional 'adverbial' use of vero discussed in 11.2.1.

328

§ 11.5.1

Some of the aspects of the above analysis of vero can also be found, in some form or another, in the concise treatment of the particle in Kuhner-Stegmann (I, p. 798-799; II, p. 80-81). The main difference in Kuhner-Stegmann's account, however, is that they do not differentiate between properties of the particle and properties of the surrounding context, as appears from expressions such as "[vero] driickt einen Gegensatz bekraftigend und versichernd aus" (II, p. 80), and "seiner urspriinglicher Bedeutung entsprechend driickt vero sehr oft eine Steigerung (sogar, vollends) aus" (II, p. 81): Kuhner-Stegmann apparently ascribe to vero the function of expressing (among other things) a contrast or a climax, whereas in the present account contrast and climax figure rather as properties of the context in which vero is prone to occur. As to the pragmatic motivation for the use of interactional vero (which is by far the commonest use of the particle) it is important to note that vero characteristically occurs in the immediate presence of a content element with a relatively high degree of challengeability, that is, with surprising, controversial, climactic, etc. information units. The pragmatic motivation for using the actuality marker vero can accordingly be described as precluding the potential challengeability of the utterance, namely by emphasizing the actuality of (part of) the communicative act. A number of linguistic clues indicative of a challengeable context have been listed on p. 311-312. These include, among other things, the presence of a superlative adjective, of a focus particle, or of a negation particle. It turns out that vero in virtually all of its occurrences clusters with one or more of these elements. The alleged adversative, transitional and focal uses of vero are, accordingly, to be regarded as contextually determined side-effects of the particle's proper discourse function. The conclusion that vero's discourse function is to be located primarily on the interactional level of discourse is based furthermore on the more general observation that the particle tends to occur in dialogical or diaphonic discourse (so-called dia-discourse, cf. 5.2.2.2). On this point it is significant for instance that in narrative texts the use of vero is largely confined to directly or indirectly rendered conversation, or to those passages in which the subjective involvement of the author is displayed74 • In Curtius' History of Alexander the Great, for instance, 34 of the 65 instances of vero occur in (in)direct speech, while the remaining instances occur for the most part in a narrative climax. On the 74. Note that this is in accordance with the observation that vera in narrative texts is much more compatible with the 'affective mode' than with the 'neutral mode' (see p. 322-324 above).

discourse-pragmatic description of vero

329

latter point it is noticeable that the text contains 15 instances of the climaxmarking combination tum vero (cf. pp. 322-324). More or less the same picture is found also in Livy. The proposed description of vero may, by extrapolation, be applied also to the relatively few cases in which the context does not reveal a clear motivation for the use of vero. In these cases vero can be said to provide the text, on its own account, with an added emotional or expressive value. 11. 5.2 vero and other discourse connectives As to the question where to locate vero in an overall discourse-pragmatic system of Latin particles, it is now clear that there is no good reason to group vero together with undisputed adversative connectives such as sed or verum ('but'). In contrast to sed and verum, vero does not signal structural relationships (of the type not only X, but also Y) or sequential discourse relationships, while the alleged adversative value of vero actually turns out to be a property of the context. Vero is also markedly different from the broadly adversative connectives at and autem. Although vero resembles at in having its primary function on the interactional level of discourse, it differs from at in that it does not signal sequential discourse relationships, but rather has a situating nature75 • This functional dissimilarity is apparent for instance in the fact that vero, in contrast to at, may combine with all kinds of connective particles. Likewise, the assumption of many latinists that vero is more or less interchangeable with autem can be disproved by comparing the distributional properties of the two particles. Such a comparison reveals that the only thing vero and autem have in common is their preference for the second position in the host unit. Consider the following - not necessarily exhaustive - overview: distributional differences between vero and autem

(i)

autem tends to cluster with full nominal phrases which usually have the status of shifted discourse topics. This tendency cannot be observed for vero, which follows more or less equally all kinds of elements, with a

75. In ch. 12 I will describe at as a reactive interactional particle, which indicates that the language user in some respect challenges the preceding move of the discourse partner. At is, in other words, a marker of interactional relations (see also section 4.2.3.1).

330

§ 11.5.2

slight preference for collocation with (resumptive) pronouns. The combination of autem with a resumptive pronoun is rare; (ii) interrelated with (i): in contrast to autem, vera is not systematically used at major thematic transition points (paragraph boundaries); (iii) in contrast to autem, vera is not systematically used in the minor premiss of a syllogism; (iv) vero and autem differ significantly in their possibilities of combining with other particles. Autem is practically limited to combinations with enumerative particles or adverbs (see 10.4.2, p. 273), and to the combination with ecce76 • Vera, on the other hand, tends to cluster with 'focus' particles or adverbs (iam, etiam, adeo, tam, etc.), with negation particles, with protesting reaction particles such as at and immo, and with interactional particles like enim, age and sane. These combinations are excluded or extremely rare with autem; (v) vera has a preference for dialogical or diaphonic discourse, autem for monological discourse. The quite different sets of distributional characteristics point to the different discourse functions of autem and vera. Even when used in a typical autemcontext (i.e. in a monological passage at the introduction of a new thematic segment), vera is not to be considered interchangeable with autem: whereas autem would merely signal, in a neutral way, the occurrence of a new element in a string of thematic units, vera in such a context rather provides an indication of the centrality or informational saliency of the particular unit within the overall thematic structure. In other words, in the case of autem the discontinuity of the passage is focused on, whereas in the case of vera the emphasis is on the aspect of cumulation. Instead of including vera in discussion of adversative connectives, it might therefore be more insightful to take it, together with e.g. enim, in a separate functional group of situating interactional particles. Both particles have to do with the involvement of the discourse participants in the communicative process: vera indicates the commitment of the speaker, whereas enim solicits the commitment of the addressee. Both particles thus help to make the com76. In the investigated selection of the Ibycus corpus 63 instances can be found of the combination ecce autem, but not a single instance of the combination ecce vera. The non-occurrence of the combination ecce vera is quite explainable: ecce characteristically announces an unexpected event which takes place before the eyes of the speaker and the addressee at the very moment of speaking. A context like this is not compatible with a speaker-authority marker such as vera.

vero and other discourse connectives

331

municative act being performed less challengeable, and hence can be said to play a role in the management of the interaction. The view that vera and enim belong to the same functional group is confirmed by the fact that they share a considerable number of distributional properties. The most important are:

distributional similarities between vero and enim (i)

vero and enim both combine readily with unequivocal connectives. This may imply that they do not themselves mark sequential discourse relationships, i.e. that they are not connective in a strict, textual sense; (ii) vera and enim both occur typically in dialogical or diaphonic discourse. This is indicated, among other things, by the fact that they readily cluster with other interactional elements such as interjections, subjective evaluation verbs and first and second person pronouns; both are, moreover, quite common in rhetorical questions. In narrative texts vera and enim typically occur in (in)direct speech, or in passages with a subjective or emotional colouring (narrative climaxes, contemplation by the author, and the like); (iii) vera and enim both tend to occur in challengeable (i.e. surprising, controversial, polemic, etc.) contexts. This is indicated e.g. by the frequent combination of vera and enim with at (which will be described in the next chapter as a typical introducer of challenging reactive moves), and with negation particles. In view of the above similarities in the conditions of use of vera and enim (and the fact that their discourse functions turn out to be complementary) it is not surprising that in Latin texts they are often found in each other's vicinity77, or even in juxtaposition78 . Note, incidentally, that in the selection of the Ibycus corpus here investigated, there is no example of the combination of vera with enim's alleged synonym nam. This can be explained quite nicely by taking into account the specific discourse functions of nam and vera, which

77. A nice instance is Sil. 10.519-59 (concerning the great losses of Rome in the Punic war) nec socium numero pariter leviore perempto. I Sed vera, sed enim reliqui pia turba senatus I munera sortito invadunt ('and of the allies also an equal number had fallen at the same time. But really, but of course, the surviving senators did their duty and entered upon the functions prescribed to them by lot'). Cf. also Sil. 12.332. 78. The collocation enim vera was so common that it came to be felt as one single word. In the investigated selection of the Ibycus corpus I counted 168 instances.

332

§ 11.5.2

more or less predict their incompatibility: whereas nam typically introduces subsidiary discourse units (see ch. 7), vero usually occurs with central discourse units with a highly salient content. Furthermore the fact that vero is highly compatible with enim, but incompatible with nam, provides additional evidence for my contention (see 8.5.2) that there is a considerable difference between nam and enim.

12 AT

This chapter deals with the Latin but-equivalent at. After an introduction (12.1) it discusses in successive sections at's primary discourse function on the interactional level of discourse (as a marker of challenging reactive moves: 12.2), and its secondary function on the presentational level of discourse (viz. as a marker of new, surprising thematic units: 12.3). In 12.4 it will be demonstrated that also the alleged 'pathetic' instances of at can be subsumed under one of these two functions. In 12.5 I will summarize the main results of the investigation, including a schematized overview of the way in which the various uses of the particle are interrelated. In this overview the concept 'frustration of expectation' will play a central role. Attention will also be given to the other Latin but-equivalents sed, autem and vera.

12.1 Introduction When we attempt to describe, in terms of the theoretical framework outlined in the first part of this study, the primary discourse function of the Latin butequivalent at, the situation at first sight appears to be somewhat complicated. Judging by the information in Latin manuals (see section 9.2.3), which base their statements largely on Hand's study of Latin particles, there is no single discourse function of at. At appears to have an interactional use, a representational use, and a presentational use as well, as may be illustrated with (1) (3), respectively. (1)

eloquere # at pudet ('Go on, tell me # But I'm ashamed to', PI. Cas. 911)

(2)

tempore illi praecepto, at hi numero avium regnum trahebant (' ... the one party laid claim to the kingship from priority, but the other from the number of the birds', Liv. 1.7.2)

(3)

Postquam Caesar dicundi finem fecit, ceteri verbo alius alii varie adsentiebantur. At M. Porcius Cato rogatus sententiam huiusce modi orationem habuit: " .... "

334

§ 12.1 (' After Caesar had finished speaking, the rest briefly expressed their adherence to one or another of the various proposals. But M. Porcius Cato ... spoke to the following purport: " ... " " Sal. Cat. 52.1)

Example (1) is representative of the widespread use of interactional at in conversational texts, notably in Latin comedy. In this use at links two moves in an interactional exchange (as a marker of interactional relations, cf. section 4.2.3.1). More specifically it signals the introduction of a challenging reactive move (an objection, protest, disagreement, and the like). Examples (2) and (3) are illustrative of the use of at in typically monological text-types, such as narrative and expository prose. In these types of text one does not expect to find very many interactional particles, and certainly not reactive particles with a challenging character. As is wellrecognized, however, at is fairly common in these types of text as well. The common view is that at in non-conversational text types more or less acts as a stronger variant of the adversative coordinator sed ('but'), as seems to be the case in (2) (cf. also ex. 34 in 9.2.3). In addition to this seemingly representational use the handbooks also hint at a usage of at that is roughly comparable to the basic function of autem (cf. ch. 10), and would thus qualify (in terms of the framework proposed in ch. 4) for the label 'presentational'. This observation is based on instances like (3), where at seems to signal a change of discourse topic. All in all this yields the picture of a chameleonic particle at, a passe-partout which - in addition to its rather specific interactional use - can also be used for the representational job of sed, or the presentational, organizational job of autem. For a few isolated authors, such as Celsus, this view might hold good (see section 12.3 beloW). In general, however, the interactional flavour of at, which is not shared by autem and sed, turns out to be more pervasive in Latin literature than we might at first be inclined to assume. What is essential to realize, in this respect, is that in monological texts the interactional nature of at may not always be as evident as in other types of text. In what follows I will attempt to demonstrate how the various uses of at are interrelated. I will start with a discussion of the interactional features of at, which may also be present, as I will contend, in essentially monological contexts (section 12.2). In section 12.3 I will then address the issue of how a primarily interactional particle such as at may come to be used (as a secondary function) on the presentational level of discourse as well. In 12.4 I will give some attention to the so-called pathetic use of at in exclamations, supplications

335

introduction

and the like (the so-called usus patheticus, cf. 9.2.3). In 12.5 I draw some conclusions, and compare at with other discourse connectives.

12.2 The interactional discourse function of at In order to substantiate the claim that at is predominantly an interactional particle signalling the connection of moves within an exchange, I must first briefly recall some of the remarks made in chapter 5 with regard to the concept of 'discourse type'. In section 5.2.2.2 I argued that there are at least three discourse types that could in essence be called 'interactional', that is, discourse types which are capable of hosting interactional elements. They differ only in the degree of 'embeddedness' or 'overtness' of the conversation taking place within them. In figure 1 (repeated from 5.2.2.2) these discourse types are italicized. figure: discourse type and diapbony DISCOURSE TYPE diaiogai

,/"

dialogical (1)

"-

monoiogical (2)

monoiogai

/

dialogical

"-

monoiogicai (4)

(3)

diaphonic (5)

""

monophonic

Discourse type 1 (dialogical dialogal) is conventionally known as 'dialogue' or 'conversation', and is obviously the most straightforward interactional discourse type. For discourse type 2, which is very rare, see chapter 5, section 5.2.2.1. The term dialogical monologal discourse (discourse type 3) is reserved for those texts which are overtly phrased by a central reporter (for instance the author/narrator), but otherwise have the formal characteristics of a conversational exchange. In Latin literature (especially in Cicero) we come across this discourse type in cases where an author enters into a discussion with an implied (sometimes fictitious) interlocutor, or in the case of directly rendered speech/conversation. I use the term diaphonic discourse (discourse . type 5) to refer to the most deeply 'embedded' (or least overt) forms of

336

§ 12.2

conversational interaction. The term applies to monological monologal stretches of discourse (i.e. to conventional monologues) which somehow allude to the occurrence of a conversational interaction, but without a quotation of the actual words, and without all the formal properties of an interactional exchange being present. Now given the above, the important observation concerning at is that many 'monological' instances of at turn out to occur in one of these three interactional discourse types, that is, in a dialogical or at least a diaphonic environment. The examples cited in subsections 12.2.1-12.2.3 below serve to illustrate this point. The material is divided into three categories, in accordance with the three 'interactional' discourse types distinguished in figure 1. It is to be noted that the boundaries between the three different 'interactional' discourse types hosting at are not very clear-cut: we should rather assume a continuous gradient, ranging from discourse with a high degree of 'interactionality' to discourse with a relatively low degree of 'interactionality'. I will start my exposition with the clearest examples of interactional at (such as ex. 1 above), and will then attempt to demonstrate that this interactional discourse function can be discerned also in less clear cases, such as (2) and (3), which both involve an essentially monological discourse type. 12.2.1 discourse type 1: dialogical dialogal (direct dialogue without an overt central reporter)

Examples (4) - (6) illustrate the interactional use of at to signal challenging or dissonant reactive moves in dialogical discourse. (4)

Do: Ubi tu nata es? Vi: Ut mihi mater dixit, in culina ( ... ) Do: At ego patriam te rogo quae sit tua (,Where were you born? # In the kitchen, so my mother told me ( ... ) # But I'm asking you what your country is', PI. Per. 630-635)

(5)

Pe: Heia vero! age dice Ep: At deridebitis Ap: Non edepol faciemus (,Come on, out with it! # But you two will make fun of me # No, no, upon my word', PI. Epid. 262-263)

(6)

Da: Confracta navis in mari est illis Sc: Ita est. at hercIe nobis villa in terra et tegulae ('They've been shipwrecked # That's right. But we've been housewrecked', PI. Rud. 152-153)

interactional junction of at

337

From the above examples it appears that at-units do not necessarily count as full cancellations or rejections of the prior move. Usually at is used in a situation in which part of the prior move is (implicitly or explicitly) conceded or acknowledged, while another aspect is being challenged. The aspect that is being challenged need not be (part ot) the verbally expressed content of the prior move. In (4), for instance, it is not the truth or validity of the propositional content of the utterance 'ut mihi mater dixit, in culina' that is being challenged in the following move, but rather its relevance in view of the particular communicative intentions of the interlocutor. In (5), Epidicus' response to the prior directive move of Periphanes likewise does not count as a downright rejection of the proposed course of action, but only as a motivation for a virtual refusal: the at-unit indicates that Epidicus is not convinced that he can comply in a satisfactory way with Periphanes' (metadirective) adhortation to give an advice. The example is comparable with (1) above. Example (6) contains, like (4), an at-unit in which the relevance of a prior move to the discourse situation at hand is being challenged (rather than e.g. the truth of its content). The example is interesting in that the conceded part of the prior move (viz. the truth of its content) is made explicit by means of ita est. As a matter of fact, the presence of a concession (i.e. the part of the reactive at-move in which the speaker resumes and hence acknowledges the content of the prior move) is a very characteristic part of the environment of at, which will recur several times in my discussion below. This 'concession' element may take various forms. In (7) for instance it is contained in a subordinate clause introduced by quamquam ('although'); in (8) the concession interpretation can be derived from a literal repetition of (part ot) the preceding move. (7)

Ly: Qui, malum, homini scutigerulo dare lubet? Cl: Quia enim filio nos oportet opitulari unico Ly: At quamquam unicust, nihilo magis ille unicust mihi filius quam ego illi pater (,But, dash it! how can you want to give her to that shield-porter fellow? # Why, because both of us ought to assist our son, our only son # Well, no matter if he is our only son, he's no more my only son than I am his only father', PI. Cas. 262264)

(8)

Da: Tange dum La: Tangam hercle vero Da: Tanges, at scin quo modo? ('Well, touch them # Indeed I will touch them, by gad! # You will, eh? Know the results though?', PI. Rud. 796-797)

338

§ 12.2

Instead of a literal repetltlOn, one can also make use of anaphoric ita ('so'; 'thus'), as is demonstrated by (6) above. In (9) both devices (repetition and anaphoric ita) are combined in an expression in which also the concession marker quidem ('admittedly') is added: (9)

Ph:

An, ut nequid turpe civis in se admitteret propter egestatem, proxumo iussast dari, ut cum uno aetatem degeret? quod tu vetas De: Ita, proxumo quidem; at nos unde? aut quam ob rem? ('Wasn't it the intention of the law that no Athenian gentlewoman should be driven to shame by her poverty and so it is enjoined that she be married to the next of kin to live out her life with him? But you are above the law # Yes, to the next of kin, but where do we come in? Why be married to us?' Ter. Phorm. 415-418)

In (10) it is the element credo that is indicative of the fact that the speaker acknowledges the truth or validity of the prior utterance. (10)

Gn: Num quid nam hie quod nolis vides? Pa: Te Gn: Credo; at numquid aliud? ('You don't see anything here, do you, that you'd rather not? # You # Quite so, but not anything else?', Ter. Eun. 272)

The dialogical examples (6) - (10) above are interesting in that they may help to understand how an interactional particle such as at may come to be used in monological discourse types as well: in (6) - (10) at does not occur at the beginning of a move, but ostensibly functions as a move-internal connective. I will come back to this point in 12.2.3, when discussing the use of at in pseudo-apodotic clauses. Generalizing from the examples (4) - (10) we may say that at in these instances is a marker of the non-acceptability (for whatever reason) of a prior move by a different speaker. In terms of discourse coherence this means that in the at-units concerned certain expectations about the unmarked continuation of the discourse are being frustrated: instead of for instance an acceptance of an offer, a compliance with an order or an approbation of a statement, the 'problemizing' reactive move introduced by at postpones the proper closing of the exchange. It is to be noted that the 'problemizing' move introduced by at may, in turn, be challenged itself. This may lead to a complex exchange structure consisting of several lower-order exchanges. Consider for instance (11), in which the stacking of 'problemizing' moves has a comical effect: (11) La: Ha: La:

Est quidam homo qui illam ait se scire ubi sit At pol ille a quadam muliere, si eam monstret, gratiam ineat At sibi ille quidam volt dari mercedem

interactional function of at

339

Ha:

At pol ilia quaedam, quae illam cistellam perdidit, quoidam negat esse quod det. La: At enim ille quidam operam bonam magis expetit quam argentum Ha: At pol illi quoidam mulieri nulla opera gratuita est ('There's a certain man who says he knows where it (viz. a certain casket) is # Well, goodness me, there's a certain woman who would be grateful to him, if he'd show it to her # Well, that certain man wants a reward given him # Well, goodness me, that certain woman who lost the casket says she hasn't anything to give the certain man # Well, you see, that certain man is keener for a kind favour than for money # Well, goodness me, in the case of that certain woman, no favour is done gratis', PI. Cist. 735-740)

Thus far we have seen examples in which a speaker 2 challenges (by means of an at-unit) a prior move by speaker 11. Slightly different are examples such as (12) and (13). Here the at-unit apparently occurs after the closing of an exchange, as an afterthought-like elaboration of the initiating move of this exchange2 • (12) Sa: Leno ego sum Ae: Scio Sa: At ita ut usquam fuit fide quisquam optuma ('I am a slave-dealer # 1 know it # But as honest a man at that as ever man was anywhere', Ter. Ad. 161) (13)

Ph: Pa: Ph: Pa: Ph: Pa: ('Do will,

Fac, ita ut iussi, deducantur isti Faciam At diligenter Fiet At mature Fiet what 1 told you, have those people brought across # Yes sir # Attend to it # 1 sir # Make haste about it # 1 will sir', Ter. Eun. 207-208)

The effect of the at-units in (12) and (13) is however the same as in (4) - (11): they expand the exchange in place, delaying its closure, and thus frustrating the expected continuation of the discourse (which would have been for instance the beginning of a new exchange). Note, moreover, that it is perfectly possible to describe the at-units in (12) and (13) as reactions to supposedly incomplete

I. In the sequences concerned the move introduced by at can have various interactional functions (for the term 'interactional function' cf. section 4.2.2.1), although most of the at-units tum out to be assertions. For an at-question compare however (10), for a directive at-unit see e.g. PI. As. 464. For the moves preceding an at-unit there appear to be no restrictions whatsoever on their interactional function. 2. In more technical terms we could speak of a discontinuous move.

340

§ 12.2

uptakes on the part of the addressee. Such a description comes close to the description proposed for the examples (4) - (11), in which the at-units form a reaction to an explicitly expressed challengeable utterance (cf. esp. ex. 8). The examples in this section have all been taken from Latin comedy, which is the main source for dialogical dialogal discourse. Other examples of the use of at in a dialogical dialogal discourse type can however be found in for instance the dialogal parts of Apuleius' Metamorphoses, in Cicero's philosophical and rhetorical works (which for the most part have the form of intellectual conversations), and in Petronius. In these cases it is sometimes hard to determine whether we are dealing with a dialogical dialogal or a dialogical monologal discourse type, namely when the author who mediates the conversation retreats entirely into the background. 12.2.2 discourse type 3: dialogical monologal

As I stated above, dialogical discourse is characterized by alternating moves by different speakers, which are related by their respective interactional functions. In monologal text types (which make up the bulk of Latin literature) we come across this discourse type in roughly two situations: in direct speech, and in what is usually called the occupatio. The use of at in directly rendered conversation needs no further comment: it is of course no different from the examples from Latin comedy cited in 12.2.1 above. More interesting for the argument of this chapter is the use of at in the occupatio. Occupatio is the technical term for a much used device in which a central reporter (usually the author) enters into a discussion with an implied (often fictitious) opponent, in order to preclude possible objections from his audience. The 'discussions' concerned are close to real dialogue, and the function of at in such contexts is the same as described in section 12.2.1 (viz. the introduction of a challenging or dissonant move). Often, at co-occurs with the interactional particle enim, for which see chapter 8. Other formal indications that may help to create the impression of an actual conversation, are for instance the use of reporting phrases like aliquis dicat ('someone might say'), dices ('you will say'), or inquis ('you say"). An example is (14), in which the implied interlocutor coincides with the addressee of the letter this fragment is taken from. Comparable instances are (15) and (16).

interactional function of at (14)

(15)

'at' inquis ' 'HpaKAELOELOP aliquod'. Non recuso id quidem (' "But", you say, "write something in the style of HeracIeides." refuse', Cic. Att. 15.4.3)

341

That I don't

'At enim minora di neglegunt .. .': sic enim dicitis (' "But" (it may be objected) the gods disregard smaller matters". This is how you argue', Cic. ND 3.86)

(16) sunt hic inter se quos nunc credo dicere: " ... novum attulerunt, quod fit nusquam gentium." at ego aiio id fieri in Graecia ... (,There are some here who, I suppose, are now saying to each other: " ... something new this - something that happens nowhere on earth!" But I say it does happen in Greece', PI. Cas. 67-71)

The occupatio technique is especially common in Cicero's orations and in the speeches in Livy. These embedded, essentially monologal dialogues may be quite extensive, as is shown by (17). Note the similarity with the dialogal example (11) in 12.2.1 (with a series of at-clauses): (17) dices: 'Quid postea, si Romae adsiduus fui?' Respondebo: 'At ego omnino non fui' 'fateor me sectorem esse, verum et alii multi' 'At ego, ut tute arguis, agricola et rusticus' 'Non continuo, si me in gregem sicariorum contuli, sum sicarius' 'At ego profecto, qui ne novi quidem quemquam sicarium longe absum ab eius modi crimine' ('You will say, "If I was constantly in Rome, what follows from that?" I shall reply, "But I was never there at all." "I confess that I am a broker, but so are many others." "But I, as you yourself reproach me with being, am a farmer and a rustic." "If I have mixed with a crowd of assassins, it does not follow at once that I am an assassin." "But most certainly I, who do not even know any assassin, am far beyond the reach of such an accusation." " Cic. S.Rosc. 94)

Examples (14) - (17) all contain metacommunicative expressions (e.g. reporting clauses) which explicitly announce an embedded conversation. In the following two examples there are no such metacommunicative phrases. However, in (18) the use of the vocative Romani, and of vos ('you') instead of nos ('we'), contribute to the impression of a real exchange between two (groups ot) discourse participants, in which the actual words of the implied interlocutor(s) seem to be quoted in fu1l 3 • In (19) the context, and especially ita opinor (which is a typical turn-initiator in reactive speech acts: 'I indeed think so'), point to a real (i.e. formal) exchange structure4 •

3. A comparable example is Liv. 39.37.1. 4. For more examples, cf. Kiihner-Stegmann (p. 85-86). See also ex. (26) in ch. 9.

342

§ 12.2

(18) (From a speech of the Roman commander Titus Quinctius, directed to the Spartan king Nabis) At enim, ut iam ita sint haec, quid ad vos, Romani? Hoc tu dicas Iiberantibus Graeciam? (' "But granted now that all is so, how does this concern you, Romans?" Is this what you would say to the deliverers of Greece?', Liv. 34.32.13) (19)

At enim magnificum est Iiberare civitates servas. Ita opinor, si nihil hostile

adversos vos fecerunt; sin autem ... ('But there will be protests because it is a fine thing to liberate enslaved cities. Such is my opinion, if they have committed no hostile acts against you; but if .. .', Liv. 37.53.28)

It is noticeable that the use of at in (14) - (19) is close to its observed use in the minor premisses of syllogisms and in other argumentative patterns. Consider (20) and (21), which could in principle be regarded as altercations between two interlocutors, but in which any formal features of an exchange structure are lacking5 : (20) aut metuamus Carthaginiensis oportet, si incolumes eos reliquerimus, aut eorum urbem diruamus. at metuere quidem non oportet. restat igitur, ut urbem diruamus ('We must either live in fear of the Carthaginians if we leave them with their power undiminished, or we must destroy their city. But we certainly should not live in fear. The alternative is, then, to destroy their city', Cic. Inv. 1.72) (21) domus tibi deerat? at habebas. pecunia superabat? at egebas ('You had no house, perhaps? But you had one. You had plenty of money, perhaps? But you were in need', Cic. Seaur. 45)

With examples like these we enter into the realm of monological discourse. In section 12.2.3 I will discuss further instances in which a strictly dialogical form has been abandoned, as the alternating moves all appear to be presented from the deictic orientation of the author/narrator.

12.2.3 discourse type 5: diaphonic monoiogicai discourse The next stage down the scale of 'embeddedness/overtness of conversation' is diaphonic discourse. I use 'diaphonic discourse' as a cover term for all kinds of monological discourses that can be said to have dialogical 'traits', without displaying a full-fledged dialogical form. It is significant that the majority of the instances of at that are adduced in the literature to illustrate its usus

5. For more examples of the use of at in argumentative patterns, cf. e.g. KUhner-Stegmann (p. 8587), and TLL (s. v. at, p. 998, 37ff; p. 999, 27ft).

interactional junction of at

343

grammaticus (i.e. the monological, non-conversational use of at, allegedly comparable to sed or autem), turned out to occur in such a diaphonic environment6 • In 12.2.3.1-12.2.3.4 I will discuss a number of these and comparable instances, arranged according to the explicitness of the dialogical traits. 12.2.3.1

indirectly rendered conversation in oratio obliqua

The clearest instance of diaphonic discourse is indirectly rendered conversation in oratio obliqua. The only difference from directly rendered conversation is that in oratio obliqua the deictic orientation of the interlocutors is replaced by the deictic orientation of a mediating reporter or narrator (usually the author). An instance of at in indirectly rendered conversation is (22). The function of at is evidently the same as in examples (4) - (21) above (viz. to introduce a dissonant reaction). Note that third person plurals are used instead of second person plurals (as would have been used in the direct variant), the speeches being presented from the deictic orientation of the narrator Livy. The clause si ... mittere is a repetition of the preceding, initiating move, and as such counts (in combination with saltem) as the concession element that is often found in the environment of at (see above. p. 337). (22) (The leading men of the Sicilian city of Henna try to persuade the Roman commander Pinarius to give up the city. The latter advises them to send legates to the consul Marcellus, who has the right to decide the matter) Se vero negare illi missuros testarique, si verbis nihil agerent, vindictam aliquam libertatis suae quaesituros. Tum Pinarius: at illi si ad consulem gravarentur mittere, sibi saltern darent populi concilium ... ('But they said that they would not send them, and asserted that if they accomplished nothing by words, they would seek some means of recovering their freedom. Upon that Pinarius said that, if they objected to sending to the conSUl, very well, let them at least give him an assembly of the people .. .', Liv. 24.37.1011)

12.2.3.2

implied interlocutor with the status of an 'embedded voice'

Relatively unproblematic are, furthermore, instances like (23), in which a speaker/author apparently reacts to an opinion of an implied interlocutor (cf. also 21 above):

6. For the term usus grammatic us, see 9.2.3.

344 (23)

§ 12.2 enim ego earn violavi et everti, quod Argivorum civitatem teneo. Quomodo hoc tuear? (,But (in your eyes) I violated it (sc. the treaty), because I hold the city of Argos. How shall I defend myself against this charge?', Liv. 34.31.6/

At

It is clear that the function of at in (23) does not substantially differ from the regular interactional function of at sketched above in 12.2.1. That the at-clause counts as an objection or a reproach against a prior move (not quoted here) of a discussion partner, is proved by tuear ('I will defend'), which retrospectively qualifies the preceding utterance as a challenging move. It is, moreover, also suggested by the combination of at with the interactional (agreement-soliciting) particle enim. That the discourse type is monological rather than dialogical appears however from the fact that in the at-clause a first person ego is used instead of a second person tu. In more general terms we could say that in cases such as (23), the at-clause conveys someone else's opinion, the deictic orientation of the reporter being however maintained throughout: we are dealing, not with an opposing interlocutor, but with an opposing 'embedded voice'.

12.2.3.3

pseudo-apodotic clauses containing at

In (24) and (25) the conversational exchange that in my opinion explains the use of at is still less overt: (24) si tu in legione bellator clues, at ego in culina clueo ('You may be a well known hero in the army, but that's what I am in the kitchen', PI. True. 615)

(25) "Si vos urbis, Quirites, si vestri nulla cura tangit, at vos veremini deos vestros ab hostibus captos. . .. " ('If you feel no concern, Quirites, for your city, or for yourselves, yet fear your gods, whom the enemy hold captive ... ', Liv. 3.17.3) In both (24) and (25) at occurs between a subordinate clause and a main clause, which is obviously not a very likely environment for a reactive marker in an interactional exchange. Handbooks usually speak of 'at at the beginning

7. A comparable example is Cic. Phil. 2.12. Consider also instances like Plin. Ep. 4.14.1: Tu fortasse orationem, ut soles, et flagitas et expectas. At ego quasi ex aliqua peregrina delicataque merce lusos meos tibi prodo. ('You expect and demand, perhaps, as usual, an oration; but I am going to put into your hands, as if they were some choice bits of foreign merchandise, some of my poetical amusements').

interactional junction of at

345

of an apodosis'8, and discuss the instances concerned in a separate section or among the typically 'monological' uses of at. There may be more to say than that, however. For one thing, the subordinate clause in this quite frequent si ... at construction is of a very special type, viz. a quotative or semi-factual conditional. Quotative conditionals, which are always in the indicative mood9 , record contextually or situationally given information, which stems from a source other than than the reporter himselflO. This amounts to saying that quotative conditionals characteristically contain 'embedded voices' (see above), against which the reporter may feel free to raise an objection, which may be introduced by at. Such an analysis of the si ... at constellation is supported by the fact that the subordinate clause is relatively loosely related to the main clause, as appears from the fact that the si-clause falls outside the illocutionary scope of the main clause. Consider in this respect (25) above, in which the si-clause falls outside the scope of the directive illocutionary force of the main clausell. We are thus not dealing here with a common conditional constellation in which si and at function as a correlative ('if ... then') pair - as is for instance the case in 'true' (i.e. representational) condition relations, which may for instance be marked by the correlative pair si ... tum. The relation between the si-clause and the at-clause is to be described as concessive rather than as conditional. This is significant in the light of the observation made earlier that the presence of a concession element is characteristic of the environment of at in general, and that concession elements often count as a resumption of the words of an interlocutor. The concession-like character of the si ... at-construction is indicated, among other things, by the occurrence in the at-clause of typically 'concessive particles' like tamen, sane, saltern and certe. Example (26) offers an illustration.

8. Rosen (1989: 398) calls at a 'superordinator' in the construction concerned. 9. An exception is formed of course by quotative conditionals in oratio obliqua, which inherently are in the subjunctive mood. See e.g. Caes. Gal. 1.43.9, Liv. 3.31.7, and example (27) below. 10. The semi-factual (rather than e.g. hypothetical) character of the si-clause is proved not only by the fact that it is always in the indicative mood, but also by the fact that in this specific construction type si non is used instead of nisi. Semi-factuaVquotative conditionals to some extent resemble pseudo-conditionals. For the former category see Dik (1990); for the latter see van de Griend (1989: 452) and Pinkster (1990: 35-6). 11. A comparable example is Verg. A. 1.542.

346

§ 12.2

(26) in quo si non praesens periculum, at certe longinqua obsidione fames esset timenda (,There might be no present danger in it, but there was certainly famine to fear from a prolonged siege', Caes. Gal. 5.29.7)12

Moreover, the at-clause may also be preceded by an unambiguous concessive clause, as is the case in (27) and (28), where the concessive subordinators etsi and quamquam ('although') are used13. Example (27) is repeated from 9.2.3. (27) etsi scelestus est, at mi infidelis non est (,Rascal though he is, he's a loyal one to me', PI. Trin. 527-528) (28) quamquam ego vinum bibo, at mandata non consuevi simul bibere una ('I do drink wine, but 1 never had the habit of drinking down my orders along with it', PI. Pers. 170)

That there is no real structural correlation between concessive si (etsi, quamquam, etc.) and at is furthermore suggested by paratactic parallels of the si ... at constellation. Compare (29) - (31), which are fully comparable with (24) and (25) (cf. si ita vis in ex. 30), except for the fact that the embedded voice is carried by an independent clause: (29) sit fur, sit sacrilegus ... ; at est bonus imperator ... ('Granted that Verres is a thief, that he is a sacrilegious thief.. ; yet he is a great commander .. .', Cic. Ver. 5.4)14 (30) satis, si ita vis, fortasse naturae (sc. vixisti) ... ; at, quod maximum est, patriae certe parum ('Long enough, if you will have it so, have you lived for nature; but, what is more than all this, for your country all too brief a span', Cic. Marc. 25) (31) nihil ad rem! - Ne sit sane; at certe gravius ('That is no matter # Granted by all means; but it is certainly more impressive', Cic. Fin. 4.73)

In conclusion we can say that the (at first sight) curious, 'structural' use of at in 'apodotic' main clauses is fully explainable when we take differences in discourse type into account: si ... at constructions turn out to be monological counterparts of purely dialogical examples such as (7) above, which I repeat

12. Further examples are Gal. 1.43.9; Quint. 12.11.31 (si non ... at certe); Cels. 2.15 (idem). 13. For more examples of etsi- and quamquam (,although')-clauses followed by an at-clause, cf. TLL s. v. at, p. 1006-1007. 14. Some other random examples are Ter. Haut. 572 (with the concessive element esto) and Cic. Dom.77.

interactional function of at

347

here for convenience 15 • The only difference is that in (7) the initiating move of the interlocutor (which is resumed and acknowledged in the concessive quamquam-clause) is explicit. (7)

Ly:

Qui, malum, homini scutigerulo dare lubet? Quia enim filio nos oportet opitulari unico Ly: At quamquam unicust, nihilo magis ille unicust mihi filius quam ego illi pater (PI. Cas. 262-264)

Cl:

Interesting in this respect is also example (22) above, which takes up an intermediate position between the clearly dialogical instances, and the clearly monological, 'apodotic' ones. Note, incidentally, that because of the position of the si-clause in (22) (after at), an apodosis interpretation is excluded. The resemblance of this example to the si ... at examples provides a further argument against treating at in the si ... at constellation as an apodosis marker which is strongly correlated with a protasis marker. More generally we can now conclude that at may display its primary, interactional function (viz. indication of a challenging reaction) in all kinds of syntactic environments, irrespective of the grammatical concept of clause. The constellation of an opinion put forward and a dissonant reaction to it can receive various formal expressions, for instance: (i) two constituent moves of an interactional exchange, each potentially consisting of more than one clause, cf. section 12.2.1; (ii) only one move, consisting of two main clauses, as in example (29); (iii) as in (ii), but containing a subordinate clause and a main clause, as for instance in (24) and (25); or (iv) two phrases or word groups, as in (26). From this it follows that placing discussion of at in the sections on clause combining (as is the habit of Latin grammarians) may not be very felicitous. A further example of intraclausally used at (syntactic environment iv) is (32): (32) Verrem amatis! ita credo: si non virtute, non industria, non innocentia, non pudore, non pudicitia, at sermone, at litteris, at humanitate eius delectamini (Cic. Ver.3.8)

15. Additional support for a description of the apodotic instances of at as markers of a conversational contrast in an embedded dialogue is supplied by Livy, since all the apodotic instances he uses occur in direct or indirect speech. It is also noteworthy that in the 'apodosis' -constellation at cannot be substituted by sed, autem or vero.

348

§ 12.2

The embedded dialogue in this example can be made explicit as follows: Cicero: You are ardent supporters of Verres Opponents: [You cannot say that, as] he lacks virtue and industry and integrity, a sense of honour and of decency; But at least you enjoy his conversation, his culture, his good Cicero: taste [so my conclusion is still that you are ardent supporters of his] Consider also (33), repeated from 9.2.3, which, although not an instance of the si ... at constellation, appears to behave in a fully comparable way (note especially the phrase ut aliquis fortasse dixerit) 16. This implies that I do not consider (33) an instance of a substitution relation, as one might at first sight be inclined to do (cf. section 9.2.3). (33) non honestum cons ilium, at utile, ut aJiquis fortasse dixerit ... ('The trick (viz. Ulysses' ruse) was not morally right, but, as someone may perhaps say, it was expedient .. .', Cic. Off. 3.97)

12.2.3.4

at in sentences of narration proper

The last (and least overt) group of diaphonic at-instances, but probably the most interesting one, is exemplified by (2) (which I repeat here with more context), (34) and (35). They all involve sentences of narration proper, that is, they are not meant as imitations of the speech of narrated characters. (2)

Priori Remo augurium venisse fertur, sex voltures; iamque nuntiato augurio cum duplex numerus Romulo se ostendisset, utrumque regem sua multitudo consalutaverat: tempore illi praecepto, at hi numero avium regnum trahebant (,Remus is said to have been the first to receive an augury, from the flight of six vultures. The omen had been already reported when twice that number appeared to Romulus. Thereupon each was saluted king by his own followers: the one party laid claim to the kingship on the ground of priority, but the other (at hi) on the ground of the number of the birds', Liv. 1.7.1-2)

(34) est enim inter magnos homines summa dissensio. Princeps Thales ... ex aqua dixit constare omnia. At hoc Anaximandro ... non persuasit ('It is a subject extremely debated among the great. Thales ... said that all things are made of water. But in this (at hoc) he did not carry conviction with A.', Cic. Ac.2.118)

16. By extrapolation we can use the same explanation for instances like PI. St. 160: illa me in alvo menses gestavit decem, at ego illam in alvo gesto plus annos decem.

interactional function of at (35)

349

Sed Allobroges diu in incerto habuere quidnarn consili caperent. In altera parte erat ... ; at in altera ... ('The Allobroges for a long time were in doubt what course to pursue. On the one hand was ... , but on the other .. .', Sal. Cat. 41.1-2)

In (2), (34) and (35) a narrator relates the fact that there has been a conflict of opinion between characters of the narrated world. One explanation why at is used in (2) (instead of, for instance, sed) may be that Livy wants to evoke the image of an actual, verbal dispute between two parties, without going so far as to suspend the progress of the narrative by an embedded speech. At is, so to speak, the last trace of a conversational mode. The other but-equivalents sed and autem lack this interactional flavour which gives the narrative a more lively character. The same explanation may be given for (34) and (35)17, and for a considerable number of other (though related) instances of monological at in my sample, which are represented by (36) below. (36)

quem (sc. Posidonium) ut vidisset (sc. Pompeius) ... molesteque se dixisset ferre, quod eum non posset audire, at ille: 'tu vero' inquit 'potes .. .' ('When he had seen him ... and had said that he regretted that he was not able to hear him, Posidonius said: "But you can hear me" " Cic. Tusc. 2.61)

The subtle difference between "at" inquit " ... " ("but, he said, and at inquit: "... " (but he said: may be described as a mere difference in the discourse type chosen. In the former expression at is part of the direct speech of a narrated character (dialogical discourse), whereas in the latter at belongs to the narrative proper (monological discourse). In both, the use of at has, in my view, more or less the same effect: it enhances the interpretation of a dissonant reaction. On this point it would also be interesting to compare pairs like at ille: "... " and ille autem: "... "18. Although more research is needed here, my expectation is that ille autem is used preferably at the introduction of a new, initiating and non-challenging move (i.e. at the introduction of a shifted discourse topic), whereas at ille will be reserved for the introduction of reactive and challenging moves. Such a difference would explain why the interactional particle at is used instead of the presentational topic shifter autem in instances like (3): II

II

II

••• " ) ,

••• " )

17. Indicative of this interpretation are the elements dissensio and in altera parte ... in altera ... , respectively. Some further random examples are Ter. An. 563; Cic. Ac. 2.74; Caes. Civ. 2.42; 3.51; Liv. 5.9.3; Tac. Hist. 4.69; Apul. Met. 5.28. 18. The former expression may be paraphrased 'but the other one replied', the latter 'now the other one said the following'. Both are frequent combinations, cf. Bolkestein and Van de Grift (1994).

350 (3)

§ 12.2 Postquam Caesar dicundi finem fecit, ceteri verba alius alii varie adsentiebatur. At M.Porcius Cato rogatus sententiam huiusce modi orationem habuit: " .... " (Sal. Cat. 52.1)

All in all I would contend that in the instances cited in this section at is not a mere equivalent of autem. Although the context in which at occurs may seem indeed to point to a presentational use of the particle (as e.g. a marker of thematic discontinuity, cf. ch. 10), a comparison with the majority of other atinstances reveals that also in (2), (3) and (34) - (36) clear interactional overtones are present.

12.3 The presentational discourse function of at

In the preceding sections I discussed the ways in which interactional at may be used in monological discourse, and how in specific contexts it may seem to display certain presentational features as well. On this point it is useful to realize that in monological discourse a 'turn of speaking' (Le. an announcement that another of the narrated characters takes the floor) inherently involves a shift of discourse topic 19 • We may now attempt to explain a number of atinstances that are more genuinely presentational, and seem to approach more closely the main discourse function of autem (viz. as a marker of thematic discontinuity), since no overt reference is made to a verbal dispute between narrated characters or between the author and his audience20 • The instances of presentational at, with which I am concerned in this section, occur for the most part in narrative discourse. On account of the markedness of this type of discourse, and of the relatively low number of the instances involved, I am inclined to consider this presentational function as secondary to the particle's more prominent interactional function discussed above. The two functions are moreover not wholly unrelated: in fact, as I already observed at the end of the preceding section, it is impossible to draw a sharp line between the interactional and the presentational instances of at. In

19. For the term 'shift of discourse topic' see chapter 10. I use it to refer to all kinds of referential discontinuity, including 'shifts in perspective'. 20. A full overlap with the function of Qutem is observable in a few isolated authors only, as I will argue below.

presentational junction of at

351

what follows I will discuss a number of predominantly presentational atinstances and make some tentative remarks about their relationship with the interactional ones. In monological discourse at sometimes seems to be used as a mere equivalent of the discontinuity marker autem. Consider the following two examples, which can be roughly compared with examples such as (5) and (38) in chapter 10 (see p. 229 and p. 252). In (37) below at occurs in the second member of a parallel construction21 , while in (38) the particle coincides with a shift of discourse topic (in this case a shift in perspective). (37) (The ever-hungry parasite Gelasimus jestingly assumes his mother to be Hunger personified) nam illa me in alvo menses gestavit decem, at ego illam in alvo gesto plus annos decem ('she carried me in her belly a mere ten months, while I've carried her in mine ten years and more', PI. St. 159-160) (38)

... Ligures ... dediderunt sese, nihil quidem illi pacti; speraverant tamen non atrocius quam superiores imperatores consulem in se saeviturum. At ille arma omnibus ademit, oppidum diruit, ipsos bonaque eo rum vendidit ... (' ... the Ligurians surrendered, without, indeed, making any stipulations; nevertheless they had hoped that the consul would treat them with no greater severity than former commanders had shown. But he (at ille) disarmed them all, demolished their town, and sold them and all their property .. .', Liv. 42.8.1-3)

It is important to note, however, that in the case of presentational at (as opposed to autem) , shifts in the development of the discourse almost always coincide with a strong frustration of expectation: the at-units involved typically

convey surprising or outrageous facts and events, poignant contrasts, sudden or unexpected turns of events, dramatic incidents or tableaux, in short, any kind of information that counts as particularly striking, on its own account or in view of discourse expectations built up in the preceding context. In (37) above, for instance, the climactic at-clause counts as a quite unexpected turn of phrase, forming a comical contrast with the preceding clause. Likewise in (38), the at-clause involves not merely a neutral discourse topic shift, but also a frustration of the expectations built up in the prior discourse: although the Ligurians are hoping for a merciful treatment by the Roman consul, what they actually receive is quite the opposite. At lends the passage a more 'affective' character than it would have had with for instance autem: it signals that the

21. A comparable example is PI. Most. 781.

352

§ 12.3

text segment following at has the status of a dissonant reaction to anticipated (false) expectations on the part of the addressee. Such a description reveals the link with the interactional discourse function of at discussed in section 12.2. The impression of 'interaction' that is evoked by the use of at in this particular case does not, however, pertain to the narrated characters, but rather to the narrator and his addressee. That is, the use of at implies a reference to the world of the narrator and his addressee at the time of the delivery and reception of the narrative. Examples (39) - (41) below are fully comparable with (38)22. They have in common that in the clause preceding at certain expectations have been built up that are subsequently frustrated by the information conveyed in the at-clause. Example (39) contains what is traditionally called a de conatu use of the imperfect (prohibebant): a certain action of prevention is performed, which however (as becomes clear from the at-clause) is not brought to a successful end: (39) Ipsi ex silvis rari propugnabant nostrosque intra munitiones ingredi prohibebant. At milites legionis septimae, testudine facta et aggere ad munitiones adiecto, locum ceperunt ('The enemy came out of the woods to fight in small groups, and sought to prevent our troops from entering the fortifications, but the soldiers of the seventh legion formed a "tortoise" and threw up a ramp against the fortifications, and so took the position', Caes. Gal. 5.9.5-6)

Likewise in (40) the discourse topic shift brought about by the at-clause involves a strong frustration of the expectations raised in the preceding context. Note that this frustration of expectation has already been prepared for by the use of the interactional concession marker sane in the preceding clause23: (40) Cum inde Romam proficiscens ad Aquinum accederet, obviam ei processit, ut est frequens municipium, magna sane multitudo. At iste operta lectica latus per oppidum est ut mortuus ('When, setting out for Rome from that place, he was approaching Aquinum, as the borough is a populous one, quite a large crowd came to meet him. But he was carried through the town in a closed litter like a corpse', Cic. Phil. 2.106)

22. Some further examples are Caes. Civ. 1.13; 2.39; Cic. Au. 5.1.3; Liv. 1.10.1; 2.23.10-11; 26.38.7; Apul. Met. 5.24. Slightly different is Liv. 4.52.8, in which a striking tum of events is described which does not coincide with a shift of discourse topic in a strict sense (i.e. referent discontinuity, cf. 10.3.1). 23. For sane, cf. Risselada (\994).

presentational junction of at

353

In (41) the at-clause functions more or less as an anticlimax. The Roman army, armed to the teeth, marches against the enemy, in order to recapture their camp. Instead of fighting the expected fierce battle, however, the drunken enemy surrenders without striking a blow: (41) Et parte alia T. et C. Aelii ... cum equitatu adveniunt. Confestim et quos binos oneraria in iumenta imposuerant secuti, et consul cum toto agmine. At Histrorum pauci, qui modice vino usi erant, memores fuerant fugae; aliis somno mors continuata est, integraque sua omnia Romani, praeterquam quod vini cibique absumptum erat, receperunt ('And on another side T. and C. Aelius ... were coming up with the cavalry. Straightway also came the soldiers whom they had mounted in pairs on the packanimals, and the consul with the main body. But a few of the Histrians, who had indulged only moderately in the wine, bethought them of flight, for the rest, death was a prolongation of sleep; and the Romans recovered their possessions in their entirety, except for the wine and food that had been consumed', Liv. 41.4.3-4)

In (42) as well, the use of at coincides with a shift of discourse topic: (42) Receperat (sc. coniunx) in sinum filium nondum sextum annum aetatis egressum, in spem tantae fortunae, quantam pater eius paulo ante amiserat, genitum. At in gremio anus aviae iacebant adultae duae virgines non suo tantum, sed etiam illius maerore confectae ('She had taken into her arms a son, who had not yet passed his sixth year, born to the hope of as great a fortune as his father had lost a short time before. But in the lap of their aged grandmother lay two grown-up maidens, her granddaughters, overwhelmed with grief, not for themselves merely, but also for her', Curt. 3.11.24-25)

In this example the connotation of frustrated expectation that is associated with at (but usually lacking with autem) does not so much involve a frustration of

expectations built up in the immediately preceding context, but is rather based on the striking content of the at-unit itself: the author sketches an unusually harrowing scene, which is part of a more extensive, rather emotional description of the lamentable situation of Darius' family after his defeat at Issus. The use of at (instead of e.g. autem) adds considerably, in my opinion, to the dramatic impact of the passage24 • It is significant, in this light, that when

24. A comparable explanation can be given for the use of at in the first line of the fourth book of Vergil's Aeneid, where Dido's strong emotions are described, leading up to the description of her suicide: At regina gravi iamdudum saucia cura / volnus alit venis et caeco carpitur igni ('But the queen, long since smitten with a grievous love-pang, feeds the wound with her life-blood, and is wasted with fire unseen', Verg. A. 4.1-2). In this example and in (42) at can be described as a signal that something unexpected (viz. particJuarly striking and dramatic) is coming up.

354

§ 12.3

after this scene the attention finally returns to the battlefield and the dry facts of warfare (see ex. 47 below and ch. 10, pp. 264), the more neutral particle autem is used instead of at. Examples (43) and (44) are of a slightly different nature, in as far as they do not involve a shift of discourse topic in a strict sense (i.e. a referential discontinuity). Example (43) contains an eyewitness report of a shipwreck. At in this example signals a sudden turn of events: whereas the narrator (an eyewitness in this case) expects from what he sees that one of the girls is going to be drowned, a miraculous incident seems to prevent this. It is to be noted that this instance comes close to the 'pathetic' use of at in exclamations, to be discussed in 12.4 below. The expressive, affective overtones in this example illustrate once more how much the interactional and presentational uses of at are interrelated: (43) Salvae sunt si illos fluctus devitaverint. Nunc, nunc pericIumst. Unda eiecit alteram. At in vadost, iam facile enabit. ('They're all right if they only dodge those breakers! Now! Now there is danger. One's overboard! Where it's shallow, though!', PI. Rud. 168-170)

Example (44) contains a poignant contrast between a situation in the past and a present situation. The act of unheard-of impudence referred to in the at-clause can be regarded as going against what could generally be expected: (44) Alii ... nunc sunt mores # Id equidem certo scio. nam olim populi prius honorem capiebat suffragio, quam magistro desinebat esse dicto oboediens; at nunc, priu' quam septuennis est, si attingas eum manu, extemplo puer paedagogo tabula dirrumpit caput (,The customs of today are different ... # Indeed they are! I realize the truth of that. Why, in the old days a young man would be holding office, by popular vote, before he had ceased to hearken to his teacher's precepts. But nowadays, before a youngster is seven years old, if you lay a finger on him, he promptly takes his writing tablet and smashes his tutor's head with it', PI. Bac. 437-440)

From examples such as (44) it appears that frustrations of expectation are highly compatible with strong contrasts (see also Sal. Cat. 12, cited below as 59). This may be further illustrated with (45) and (46). (45) (In the preceding section Caesar has described at length how his men were starving because of problems with the food supply) At exercitus Afrani omnium rerum abundabat copia (,But the army of Afranius had abundance of provisions of every kind', Caes. Civ. 1.49.1)

presentational function of at

355

(46) (In the preceding section Caesar has described the miserable situation in Pompeius' camp: starvation, disease, and shortage of water) At Caesaris exercitus optima valetudine summaque aquae copia utebatur, tum commeatus omni genere praeter frumentum abundabat ('On the other hand, Caesar's army enjoyed excellent health and an abundant supply of water, and abounded with every kind of provision except grain', Caes. Civ. 3.49.5)

Examples (45) and (46) also illustrate the remarkable tendency of Roman historiographers (who for obvious reasons are preoccupied with warfare), to use at when a particular discourse topic shift involves a shift to a hostile camp25. Consider also example (47): (47) In acie autem caesa sunt Persarum peditum C milia, decem equitum. At a parte Alexandri quattuor milia quingenti saucii fuere, ex peditibus CCC omnino et duo desiderati sunt, equitum centum quinquaginta interfecti. ('Now in the battle 100,000 Persian foot-soldiers were killed and 10,000 horsemen. But on Alexander's side about 4500 were wounded, of the infantry in all 302 were missing, 150 were killed', Curt. 3.11.27)

Even when there is no strong contrast (based on e.g. a striking discrepancy) in the content of the opposed text units, shifts to a hostile camp are apparently felt to be significant enough to host an expressive particle like at (note that a shift of attention to a hostile party almost always involves a marked turn of events?6. Consider for instance (48). In the preceding context it is described how Caesar is marching against Apollonia and the neighbouring communities, and is admitted into these towns without striking a blow. The attention is then shifted to the hostile camp of Pompeius: (48) At Pompeius cognitis his rebus, quae erant Orici atque Apolloniae gestae, ...

Dyrrachio timens diurnis eo nocturnisque itineribus contendit. ('But Pompeius, when he learnt of what had happened at Oricum and Apollonia, fearing for Dyrrachium, hurried there, marching night and day', Caes. Civ. 3.13.1)

The existence of examples such as (48), in which the force of at can hardly be distinguished from the force of autem, helps to explain how at can eventually be used as a mere marker of thematic discontinuity ('marker of weak con25. Some further random examples are Caes. Civ. 1.42; 2.14; 3.11; Gal. 1.52; 5.32; Liv. 41.4.4 (= ex. 41 above); Curt. 3.11.27; Sal. Jug. 50.\-3. 26. Shifts to a hostile camp may also be signalled by autem (cf. e.g. Caes. Civ. 3.55). Most commonly however they get no marking at all, except for fronting of the shifted discourse topic. What is significant here, is that at almost always occurs with a highly specific type of shifts of discourse topic, whereas autem can be used more freely (viz. also - and even preferably - with more neutral shifts of discourse topic).

356

§ 12.3

trasts', in more traditional terms). According to Saur (1913: 8) this usage of the particle was not fully developed until Tacitus, although earlier instances can be found in for instance Caesar, and especially Sallust, as well. Example (49) illustrates the use of at at a major thematic boundary, which is based on a discourse topic shift in combination with a shift in the local setting of the story. In the preceding context Tacitus has described Titus' preparations for storming Jerusalem. He then resumes the story of the revolt of Civilis which he dropped at 4.79: (49) Hanc adversus urbem gentemque Caesar Titus . . . aggeribus vmelsque certare statuit ( ... ). At Civilis post malam in Treveris pugnam reparato per Germaniam exercitu apud Vetera castra consedit, 00. (' Such was the city and people against which Titus Caesar now decided to use earthworks and mantlets (00')' But meanwhile Civilis, after his reverse against the Treveri, recruited his army in Germany and encamped at Vetera, 00.', Tac. Hist. 5.13-14)

Likewise Tacitus uses at a number of times to shift the attention from Rome to other parts of the empire, or vice versa27 • Summarizing we can say that in narrative texts at can function, just like autem, as a marker of thematic discontinuity, thus giving an indication of the segmentation of a text. In most cases, however, the relationship with the more common interactional discourse function is quite clear. For the sake of completeness I mention here one more group of presentational at-instances, which have in common that they occur in 'instructional' texts at the transition from one topic or instruction to another. The cases concerned remarkably often involve a constellation of a hypothetical clause (introduced by si, ubi, cum or the relative pronoun qui), followed by the focal information, which may take the form of an instruction28 • In this quite specific environment at appears to be fully interchangeable with autem and vero, which seem to alternate with at for reasons of variatio. An example is (50): (50) (In the preceding section Celsus has dealt with infections of the spleen) At renes, ubi adfecti sunt, diu male habent ('As to the kidneys, these when affected, continue diseased for a long while', Cels. 4.17.1)

27. The tum At Romae (,But meanwhile in Rome') is used ten times altogether in the Histories and Annals: Hist. 2.55; 4.68; Ann. 1.7; 46; 2.82; 3.22; 44; 4.52; 6.29; 15.18. 28. In Celsus' medical handbook 132 out of 210 at-instances are part of a hypothetical constellation of the type referred to; in Columella's On Agriculture this is the case in about 50 percent of the at-instances. About the same proportion is found in Gaius' Institutiones.

presentational junction of at

357

In examples such as (50) at appears to be devoid of any specific 'added value' which would distinguish it from autem. The frequent occurrence of at in what in other text types counts as a typical autem-environment can perhaps be explained by the inherently interactional character of instructional texts29 •

12.4 So-called pathetic use of at In his discussion of the particle at, Hand (1829-1845, vol. I) distinguishes between a grammatical use, a rhetorical use, and a pathetic use of at (see also 9.2.3). Hand seems to apply the term usus patheticus to all those instances of at which occur in utterances (usually non-assertive) with a strongly emotional character, that is, in lively orders, indignant questions, emotional exclamations, amazed reactions, ardent prayers, curses, threats, strong oaths, and the like. An example quoted by Hand is the exclamation in (51). (51) (In an aside) at ut scelesta sola secum murmurat! ('How the old criminal is mumbling away to herself, though!', PI. Aul. 52)

Hand's category 'pathetic at' is taken over by most Latin grammars, lexicons and dictionaries. Usually these manuals are more outspoken than Hand in what they believe to be the specific function of 'pathetic at': according to them at lends the host utterance a particular expressive or emotional force. Consider for instance the description of 'pathetic at' given by the TLL (s. v. at, p. 995): "emphatica vel pathetica quae inest in particula vis aptam facit ad exprimendum indignationem, minas, imprecationes, preces, sim." ('the emphatic or pathetic force that is present in the particle makes it fit for expressing indignation, threats, curses, entreaties, and the like')30. From descriptions like this we might perhaps conclude that 'pathetic at' differs quite radically from the other functions of at, discussed in 12.2 and 12.3: instead of signalling some sort of relation with the preceding verbal context (cf. Hand's usus grammaticus and usus rhetoricus, i.e. 'connective'

29. We come across at with the conditions of use of autem also in Pliny's Natural History, which is not an instructional text in a strict sense, but can at least be said to display some of the features of the didactic genre. For an overview of interactional elements in Pliny's Natural History I refer to Melzani's contribution in Cova et al. (1986), pp. 227-233. It is noticeable that the combination at si (in the sense of sin autem) also occurs frequently in argumentative texts, which tend to have strong diaphonic characteristics. 30. Note that the TLL does not make a distinction between speaker attitudes (e.g. indignation) and speech acts (such as threats, curses and entreaties).

358

§ 12.4

at), the particle is believed to express various kinds of emotion ('pathetic', i.e. 'situating' at), and as such seems to be entitled to individual treatment in a separate category. A reconsideration of the pathetic at-instances cited in Hand, Saur (1913) and a number of manuals, indicates however that in most cases a more integrated (and hence more insightful) description is possible. In the first place quite a number of the instances quoted turn out in fact not to differ substantially from those discussed in 12.2 and 12.3. Consider for instance (52), which Saur (1913) adduces as an instance of pathetic at in an exclamation. In my opinion the example can be explained perfectly well as an instance of the use of interactional at in an occupatio (see 12.2.2). The alleged pathetic force of at should rather be considered a side-effect of the use of at, due mainly to the exclamatory nature of the utterance (cf. quantus) and the ellipse of the predicate3 !. (52) (Cicero quotes Demosthenes as an example of someone who loved being popular) Quid hoc levius? At quantus orator! (,What could be more petty? Ah, but how consummate an orator!', Cic. Tusc. 5.103)

Likewise in (53) at can be explained better as a marker of a challenging reactive move (comparable to the instances discussed in 12.2.1), than as expressing a strong emotion in an exclamatory utterance, as Hand (I, p. 440)

seems to do. (53) (Hegio has Tyndarus arrested) He: Abducite Ty: At unum hoc quaeso '" (,Off with him # But I do ask this one thing of you, sir: .. .', PI. Capt. 746-747)

Not surprisingly, Latin grammarians tend to detect instances of 'pathetic' at not only in exclamations, but also in directive utterances. An instance that is quoted by Hand (I, p. 440) is (54). (54) (New section) At videte hominis intolerabilem audaciam ... ! (,But observe the man's intolerable effrontery ... !', Cic. Dam. 115)

Again I would contend that the use of at in this example is not essentially different from the more common 'connective' use, more specifically from its

31. Comparable accounts (i.e. interactional at in challenging reactive moves) can be given for e.g. Cic. Tusc. 3.63; 3.83; Ver. 4.22; eluent. 199; Alt. 13.33a; PI. Am. 793.

so-called pathetic use of at

359

use as a marker of anew, surpnsmg thematic unit. Compare for instance example (42) in 12.3. The particularly emotional character of the utterance is to be ascribed to the choice of words (intolerabilis; audacia) and the metadirective appeal to the audience, rather than to at. A third type of utterance in which 'pathetic' at is allegedly prone to occur, is the indignant or irritated question, an example of which is quoted in (55) below. Contrary to Kuhner-Stegmann (vol. II, p. 85) I prefer to describe the function of at in this example as marking a challenging reaction in a conversational dialogue (see 12.2.1): speaker 1 asks a question, speaker 2, instead of answering the question, challenges the appropriateness of its having been asked in the first place. (55) (A husband is blaming his wife for not having informed him about his daughter's pregnancy) ... non sic ludibrio tuis factis habitus essem # Quibus? # At rogitas? Peperit filia: hem, taces? Ex qui? (' . .. you would never have played upon me with these doings # What doings? # You ask? (at rogitas?). My daughter has just given birth to a child. What, not a word? Who is the father?', Ter. Hee. 526-527)

Summarizing thus far we can say that a number of alleged instances of 'pathetic' at can in fact be regarded as regular cases of the interactional and presentational uses of at discussed in 12.2. and 12.3. In other cases the resemblance is less clear, but usually nonetheless demonstrable. On this point it is important to realize that the concept 'frustration of expectation' - which I assume to be the basic meaning of at and to underlie all its occurrences - can have various manifestations in actual discourse: as a 'frustration of expectation marker' at indicates that the upcoming unit may cause surprise to the addressee or audience in view of: (i) a frustration of expectations built up in the preceding verbal context (i.e. expectations about the unmarked continuation of the discourse; most of the interactional and presentational instances of at discussed in 12.2 and 12.3 belong to this type); or (ii) a frustration of expectations built up in the immediate discourse situation, or based on general knowledge and norms (this type involves e.g. reactions to, or accounts of, surprising, striking, or extraordinary events that are being observed in the immediate discourse situation; most of the so-called pathetic instances belong to this type). Whereas the alleged 'pathetic' instances of at in (52) - (55) can be explained as frustration of expectation markers of the first type, most other 'pathetic' instances quoted in Latin manuals involve a frustration of expectation of the second (less transparent) type. Consider for instance (43), repeated from 12.3:

360

§ 12.4

(43) Salvae sunt si illos fluctus devitaverint. Nunc, nunc periclumst. Unda eiecit alteram. At in vadost, iam facile enabit. (PI. Rud. 168-170)

In this example, cited by Hand in his section on the pathetic use of at in exclamations, at signals a frustration of the expectations built up in the immediate discourse situation: a shipwrecked girl miraculously and unexpectedly escapes almost-certain death by drowning. I consider it more appropriate to discuss this example on a par with the 'presentational' instances of at (viz. as a marker of a surprising shift in the thematic development of the discourse, see 12.3), than to assume that at merely reinforces the exclamatory character of the utterance32 • More or less the same holds for the aside in (51) above, in which the function of at can be described as marking a reaction to a most striking event in the immediate discourse situation (viz. the particularly irritating behaviour of the old woman on the stage). A description of at in terms of the expression of indignation or contempt on the part of the speaker fails to make clear the similiarities with the 'connective' use of at, and is, in my opinion, an overburdening of the proper function of at. Examples (56) and (57) are illustrative of the use of at in irritated or indignant questions 33 • Instead of expressing irritation or indignation (see the quotation from the TLL on p. 357 above), at can better be taken as marking a reaction to the observed fact that someone is (still) not behaving as he is expected to do. This reaction takes the form of a request for confirmation with a challenging character: (56) (Beginning of a scene) Exi foras, sceleste. At etiam restitas, fugitive? 'Out you come, scoundrel! Still struggling, good-for-nothing?', Ter. Eu. 668) (57) Pugnum in os impinge # Iniqua haec patior cum pretio tuo # At etiam minitatur audax? (,Let him have one on the jaw! # I'll make you smart for my being put upon so! # What? Threatening me, the brazen scoundrel?', PI. Rud. 710-711)

In (56) the expected behaviour that fails to be displayed is non restitare ('not to tarry'), in (57) non minitari ('not to use threats'). Note that the impression of irritation or indignation is due mainly to the presence of the particle etiam,

32. Cf. Hand's remark "gaudium exprimitur", and the quite cryptic addition "sed potest etiam alio modo intelligi" (I, p. 441). 33. Although Hand does not incorporate this category into his discussion of 'pathetic at', most of the other studies consulted do. Cf. e.g. KUhner-Stegmann (p. 85), who speak of at "in leidenschaftlichen Fragen". Other instances are e.g. PI. Cap. 563 and As. 923.

so-called pathetic use of at

361

and to the type of question involved (requests for confirmation with a challenging character, which may be interpreted as threats). Similar descriptions are possible for most other instances quoted in handbooks in the sections on 'pathetic' at. The most difficult to explain are the few instances of at in prayers, oaths and wishes34 • It is only in this particular environment that the particle may seem to approach the value of a mere attention-getting device ('pay attention, for I have something urgent and important to say'). An example is (58), also mentioned by Hand (I, p. 441): (58) To: Do:

Dic bona fide: iam liberast? < Pol aio> i ad forum ad praetorem, exquire, siquidem credere mihi non vis. libera, inquamst: ecquid audis? To: At tibi di bene faciant orones! ('Tell me honestly - is she free now? # She is. Go on, go to the forum to the praetor. Ask, in case you're unwilling to trust me. She's free, I tell you - Can't you hear? # Well, well, may all the powers of heaven bless you!', PI. Per. 485488)

In conclusion we can say that there is little support for distinguishing, alongside an interactional and presentational discourse function of at, a third discourse function involving the expression of quite specific illocutionary forces (such as threats, curses and entreaties), and of particular speaker moods Goy, indignation, contempt, etc.). It is more insightful to describe the alleged 'pathetic' instances of at in terms of one of the various actual uses (see 5.1) that are connected with the interactional (or, more sporadically, presentational) discourse function of at. The more specific speaker moods that may be observed in at-clauses are to be considered as side-effects of the use of at, depending on other elements in the clause, such as the presence of the indignation marker etiam, or the amazement-marker hercle. The most we can say is that a high degree of emotionality is apparently very compatible with the 'frustration of expectation' connotation of at. Note, finally, that the instances discussed in the present section supply additional support for the view that the difference between 'connective' and 'modal' (or situating) particles is to be regarded as a continuous gradient rather than as a strict dichotomy.

34. Most instances of 'pathetic' at in prayers, oaths, wishes, and the like, can however be explained as regular cases of the introduction by at of a challenging reactive move (cf. 12.2.1), or of a shift to a surprising new thematic unit. Cf. the TLL, s.v. at, p. 995, 4-70, and KUhnerStegmann (p. 85).

362

§ 12.4

12.5 Conclusions

12.5.1 a discourse-pragmatic description of at From the conventional accounts of at in Latin handbooks one can easily get the impression that at covers almost the entire spectrum of adversativity (cf. also 9.2.3): at first sight all three basic types of adversative relationship (i.e. substitution, opposition and concession, cf. 9.1) seem to apply to this particle. In the present chapter I have attempted to adjust this view, making use of the theoretical framework outlined in chapters 4 and 5. This has led to the following picture of at. stepwise analysis of the particle at 1. 2.

basic meaning: discourse function:

3.

actual use:

frustration of expectation (i) primary function (interactional level of discourse): indicating that in the upcoming communicative unit presumed expectations on the part of the addressee/audience will be frustrated (ii) secondary function (presentational level of discourse): indication that the attention is emphatically directed to a new thematic unit (i) (most frequent and prototypical use): marker of rebuttals, i.e. of challenging or problemizing reactive moves (so-called connective use of at in dialogical or diaphonic contexts); cf. 12.2 (ii) marker of surprised reactions to unexpected facts or events that are being observed in the immediate discourse situation (,situating' or 'modal' use of at in interactional situations); cf. esp. exx. (56) and (57) in 12.4 (iii) marker of unexpected turns in the line of the story in monological contexts; cf. 12.3 and 12.4 (iv) marker of strongly opposed discourse topics in monological contexts; cf. 12.3

discourse-pragmatic description of at

4.

side-effects:

363

(i)

expression of 'speaker moods' such as indignation, joy and contempt (ii) expression of the more specific illocutionary force of an utterance (e.g. threat, curse, entreaty) (iii) reinforcement of the illocutionary force of an utterance

explanatory remarks All instances of the particle at have in common that they signal, in one way or another, that in the upcoming unit certain expectations about the unmarked continuation of the discourse are being frustrated. A good candidate for the basic meaning of at would therefore be the concept 'frustration of expectation'. For various reasons I prefer this term to, for instance, a more conventional term like 'concession': (i) because of the connotation of interactionality that is inherently connected with the concept 'frustration of expectation'; (ii) because the term 'concession' is usually associated with a more grammatical, interclausal type of relationship than we are dealing with here; and (iii), because at pertains to the unit that commonly follows a concession, rather than to the concession itself5 • In 9.1 I observed that concession/frustration of expectation may play a role on various discourse levels. Latin at turns out to have functions on both interactional and presentational levels of discourse. The interactional function of at is clearly primary, and more pervasive in Latin literature than is commonly assumed. The most frequent use of at on the interactional level of discourse involves the introduction of a challenging reactive move (see actual use i). In this particular use we might speak, in terms of the theoretical framework of chapter 4, of a marker of interactional relations between the constituent moves of an exchange (see 4.2.3.1). Instead, however, of constituting a satisfactory reaction which properly rounds off the initiated exchange (which would be the expected, 'default' situation), moves introduced by at

35. It is notable that Latin linguists who venture to assume a common denominator for the various uses of at, choose for this role the concept 'opposition' rather than 'concession'. See e.g. Hand (I, p. 418) and Szantyr (p. 488-489).

364

§ 12.5.1

rather expand the discourse in place, for instance by challenging the relevance or appropriateness of the initiating move, or by criticizing its specific wording. As such, at-moves count as initiating and reactive at the same time. The use of at on the interactional level of discourse is not confined to conversational texts in a strict sense. As a matter of fact, it turns out to be quite common in monological texts as well (see 12.2.3), a fact which has commonly been overlooked by earlier studies. An important observation on this point is that monological texts, too, may to various degrees display interactional features (see the discussion in 5.2.2.2). In addition to the primary interactional discourse function of at, we may also distinguish a function on the presentational level of discourse. In this function at more or less resembles autem (see the end of section 12.3). I regard this presentational function as secondary, because it concerns a relatively small group of instances in a well-defined textual environment and corpus. What is more, there appears to be a clear affinity between these presentational instances of at and the more common interactional instances, indicated by the presence of a quite large group of intermediate cases. This intermediate group consists of at-instances that signal new thematic units with a quite unexpected content, that is, instances that display both presentational and interactional features. The inherently interactional feature of unexpectedness (i.e. frustration of expectation) distinguishes at in these instances from the more purely presentational particle autem. The list of actual uses connected with both discourse functions of at is not exhaustive. I have left out, for instance, more specific instantiations of the quite general actual use (i), such as 'indicating the minor premiss of a syllogism in case of mUltiperspective argumentation '36 , or 'introducing a pseudoapodotic clause'. Since the presentational discourse function of at follows naturally from its interactional function, and since in many cases both functions appear to play a role, I do not, moreover, explicitly distinguish between actual uses that are related to the interactional level of discourse, and actual uses belonging to the presentational level of discourse. It is to be noted finally that the actual uses of at may have a 'connective' as well as a 'situating'

36. By 'multiperspective argumentation' I mean the kind of argumentation in which the various argumentative steps are presented from different perspectives (diaphonic discourse type). 'MuItiperspective argumentation' contrasts with 'monoperspective argumentation'. In the latter the argumentative steps are all presented from the perspective of the speaker/author (monological discourse).

discourse-pragmatic description of at

365

character, depending on whether the at-unit reacts to the preceding verbal context, or rather frustrates a presumed hearer-expectation which is not based on the preceding verbal context; see also 12.4. The side-effects mentioned under 4 in the above overview have been treated quite extensively in section 12.4 on the so-called pathetic use of at. linguistic clues

The linguistic clues on which the above conclusions are based are varied. 1 confine myself here to the most systematic and significant clues:

discourse type: The fact that at occurs predominantly in dialogical or diaphonic discourse strongly suggests a discourse function on the interactional level of discourse.

collocations with other particles: At appears to be combinable with four, quite specific, groups of particles only. These combinatory peculiarities of at provide insight into the proper function of the particle, especially when compared to the combinatory features of other discourse connectives, especially of the so-called adversative connectives. (i) at is frequently combined with the concession markers sane, quidem, tamen, certe and saltern; this provides strong evidence for the 'frustration of expectation' meaning that 1 assume to be basic to all instances of at; (ii) at can be combined with contra, e contrario and potius (cf. actual use iv); (iii) at can be combined with the interactional particles enim, vero and enimvero, and the particle-like 'subjective evaluation verbs' credo and puto ('I believe', 'I think'); a common characteristic of these elements is that they all, in different ways, help to 'tone down' the challengeability of an utterance (see e.g. chapters 8 and 11 on enim and vero). The observation that at is frequently combined with one of these elements, supports the view that at-units have a challengeable nature 37 ;

37. In the selected part of the Ibycus corpus (see chapter I) I found e.g. 105 instances of the combination at vera and 190 instances of at enim (from a total of 5536 instances of at). It is significnt that the combination at enim is much more frequent than the combination sed enim (see ch. 8, note 2).

366

§ 12.5.1

(iv) at can be combined with certain markers of 'speaker mood': we find collocations of at with the amazement markers pol, hercle and ecastor, and with the indignationlirritation marker etiam (see 12.4).

occurrence of at in pseudo-apodotic clauses: The use of at in pseudo-apodotic clauses (and the highly specific characteristics of such clauses, see 12.2.3.3) appeared to be of essential importance for understanding the similarity between the use of at in conversational texts, and the use of at in monological texts. Apart from this, the remarkable si ... atconstellation might offer support for the view that the communicative structure of a discourse may be quite independent of its grammatical structure: in terms of communicative structure 'apodotic' at can be said to connect two alternating moves within an interactional exchange; in terms of grammatical structure, however, we are dealing with a connection between a subordinate and a main clause uttered by one single person. In other words, the use of at in pseudoapodotic clauses seems to demonstrate that grammatical clauses may not only serve as 'input' to wider communicative structures (e.g. as communicative acts), but can also have an internal communicative structure. This observation would form a major complication for discourse grammars that take the grammatical clause as a starting point on the basis of which larger communicative structures (such as moves and exchanges) may be built. 12.5.2 at and other discouse connectives Finally I will compare at with the other but-equivalents autem, vero and sed, and with immo and ergo.

autem, vero and sed On account of its 'frustration of expectation' connotation, at bears a certain resemblance to both autem and vero. As to at's resemblance to autem it has already been observed that in monological texts both particles can be used on the presentational level of discourse, as markers of thematic discontinuity (cf. 12.3). In the case of autem this is to be considered a systematic use, following naturally from the connotation of 'distinctiveness' that is characteristic of all instances of autem (see ch. 10). In the case of at however we are dealing with a less systematic use,

at and other discourse connectives

367

which appears to be closely connected with the particle's 'frustration of expectation' connotation. In accordance with this difference in basic meaning, autem commonly signals a neutral continuation of the discourse, whereas we find 'text-organizational' at almost always at transitions to surprising information, introductions of strong oppositions, and the like: in contrast with autem, at apparently has an interactional 'added value', which can be described as signalling that the speaker/author deems the upcoming unit surprising in view of (what he assumes to be) hearerlreader-expectations. That at and autem are quite different devices, and that autem is a more prototypical text-organizational particle than at, appears also from a comparison of their distributional properties: (i) unlike at, autem is remarkably often combined with a shifted discourse topic (in the form of a fronted NP); (ii) unlike at, autem occurs remarkably often in series and lists; (iii) unlike autem, at cannot be used at the introduction of a thematic unit with the status of background information; (iv) autem is combined in a systematic way with one group of particles only, viz. those with the meaning 'now', 'next', or 'then', which indicate new steps in an argument or sequence of ideas; we do not find systematic collocations of at with these particles (at usually combines with interactional particles, see above); (v) unlike at, autem is often preceded by a metadiscursive expression which explicitly announces the thematic structure of the passage; (vi) autem is typically associated with a monological discourse type, at with a dialogical or diaphonic discourse type. The distributional properties of at are in fact more comparable with those of the alleged but-equivalent vero (which I characterized in ch. 11 as a speakerauthority marker rather than an adversative connective). Both particles have a main discourse function on the interactional level of discourse, and are accordingly usually associated with a dialogical or diaphonic discourse type. The two particles frequently combine with the interactional, agreementsoliciting particle enim, and can also be combined with each other. In monological texts, moreover, both are often used in text units with a highly surprising or salient content. This can be illustrated with example (59), in which at is used in what could be considered a typical vero-environment as well (see 11.4.3; note especially the use of the superlative forms and of the historic infinitive adimere)38:

38. A comparable example is Liv. 2.24.3, which, too, contains an historic infinitive, and in which at is moreover combined with vera.

368

§ 12.5.2

(59) Verum illi delubra deorum pietate, domos suas gloria decorabant ... At hi contra, ignavissumi homines, per summum scelus omnia ea sociis adimere quae fortissumi viri victores reliquerant (,But they adorned the shrines of the gods with piety, their own homes with glory, ... The men of today, on the contrary, basest of creatures, with supreme wickedness are robbing our allies of all that those heroes in the hour of victory had left them', Sal. Cat. 12, 4-5)

My conclusion would be that at and vera are strongly compatible devices: while at signals that the upcoming unit has an unexpected, and therefore challenging nature, vero, as a speaker-authority marker, can be described as attempting to diminish the challengeability of a text unie 9 • More than with autem and vera, at has been compared with sed in Latin manuals. At and sed are usually looked upon as stylistic variants, at being considered the more lively, sed the more colourless marker of adversative relations. Although I have not made a thorough study of sed, I have the impression that the assumed interchangeability of at and sed is in fact rather limited. For one thing, at cannot signal substitution relations (see 9.1), which is one of the main functions of sed. And although sed appears to be the only other Latin but-equivalent that is capable of marking 'adversative' relations on the interactional level of discourse (i.e. contrast relations between the constituent moves of an exchange), this use of sed is relatively rare and confined to a quite specific type of rebuttal (see also 9.1, note 12). In monological, argumentative contexts at and sed seem indeed to be interchangeable. Both appear for instance to signal rhetorical contrast relations on the presentational level of discourse, of the type 'Marco is fast, but he has some knee problems' (cf. 9.1, ex. 6). My hypothesis would be, however, that in the case of at the author chooses a multiperspective presentation of the argumentation, in the case of sed a monoperspective presentation (see note 36 above). In the multiperspective variant the first of the connected units (which comprises the view of a person other than the speaker) is conceded, the second asserted. In the monoperspective variant both units are asserted40 • Finally, it can be observed

39. There is a small distributional difference between at and vera in monological texts: vera usually occurs in an absolute climax or in the last member of a cumulative string, at preferably occurs with pairs of strongly opposed units. Some other instances of at in a typical vera-environment, but with a strong opposition, are Cic. Off. 1.35; PI. Mos. 781; Bac. 473-440; Liv. 2.24.3 (= ex. 44 above). 40. This hypothesis would explain e.g. why in a Latin equivalent of the clause '(please) look up that phone number - but don't bother if it will take you more than a few minutes' (see 9.1, ex. 9), at would be excluded: the second clause is an elaboration (by the same speaker) of the directive

at and other discourse connectives

369

that in narrative texts sed is usually associated with thematic continuity rather than with thematic discontinuity (as appears to be the case with at).

ergo and immo From the above it appears that - from a discourse point of view - there may be not much reason for ranking at in one and the same category as the other Latin but-equivalents autem, sed and, to a lesser degree, vero. Although they may all involve, in one way or another, some aspect of adversativity, their functions in the discourse are widely different. It may be more interesting instead to discuss at on a par with particles like immo and, especially, ergo, which have much more comparable discourse functions. Immo (,rather') has in common with at that it may signal a challenging reactive move, and as such can be considered a marker of interactional relations. It differs from at in that it corrects (and therefore eliminates) the content of the preceding move, rather than acknowledging it.· An instance of the characteristic use of immo can be found in 4.2.3.1, in example (18). In 4.2.3.1 I also mentioned the 'consecutive connective' ergo ('so') among the possible markers of interactional relations between subsequent moves. Ergo has in common with at that in conversational texts it occurs in reactive moves that expand the discourse in place, rather than carrying it forward. It is found, for instance, in requests for confirmation of inferences that are based on preceding moves, in repetitions of earlier moves, in metadirective expressions, in short, in all kinds of units that initiate a side-sequence whenever there is a chance of 'communication breakdown', thus 'recycling' or 'halting' the topic discussed. More specifically ergo functions as a signal that essentially given (or inferrable) information is being (re)activated, in order to avoid any possible misunderstanding between the discussion partners. As such the particle plays a role in the management of the conversation. Example (60) forms an illustration. (60) Philolaches hie habitat, quoiius est pater Theopropides. qui ... hie tibicinam liberavit # Philolaches ergo? # ita (,Philolaches lives here. His father's Theopropides. And he ... set free a music girl # Philoloaches? Really? # Just so', PI. Mos. 970-972)

first clause.

370

§ 12.5.2

This interactional discourse function of ergo can be observed also in monological types of text. Consider (61), in which ergo is used in a multiperspective argumentation, at the introduction of a new argument. Note the similarity with at-instances cited in for instance (14) - (17) above: (61)

Ergo, inquiet aliquis, donavit populo Syracusano illam hereditatem

('Well, well, someone may tell me, he gave the estate to the citizens of Syracuse', Cic. Ver. 2.45)

All in all we might say that in conversational contexts ergo and at are, in a sense, complementary devices. They both introduce interactional moves that expand the discourse in place. But while at signals that the upcoming unit counts as a challenge (e.g. of an implication contained in a preceding move), ergo, conversely, somehow entails a (re)confirmation or a request for confirmation. Other distributional similarities between at and ergo in conversational texts concern the fact that both particles may occur in asides (so apparently with a more 'situating' than 'connective' force)41, and that both can be combined with expletives like edepol and (m)ecastor4 2 • Furthermore, just like at, ergo appears to have functions on both the representational and presentational levels of discourse, which, too, are clearly connected with the particle's interactional function43 . For a more extensive discussion of ergo, I refer to Kroon (1989).

41. For the use of ergo in an aside, see e.g. PI. Mos. 74. Latin grammars tend to speak of an 'affirmative' use of ergo in contexts like this. 42. For the combination of ergo and edepo/, see e.g. PI. Mere. 371, for the combination of ergo and meeastor, PI. Mil. 74. 43. In Kroon (1989) I use the terms 'semantic implication' and 'pragmatic implication' with respect to the representational and presentational functions of ergo.

13 CONCLUSIONS

This study contains an investigation of the role of so-called coordinating conjunctions in Latin. From antiquity onwards, Latin linguists have described particles such as nam, enim, autem, vero and at in terms of the expression of semantic relations between successive main clauses. In the course of this study it has become clear that this is an oversimplified account of the matter. What Latin grammars usually do, is to describe the context in which these particles tend to occur, rather than their proper functions. These functions turn out in fact to be strongly discourse-bound and particularly resistant to conventional syntactic and semantic analysis. The units of text to which the particles apply, moreover, do not necessarily have the form of grammatical clauses, but may be larger or smaller than a single clause (paragraphs or even larger segments at one extreme, constituents or single clause elements at the other). For these reasons I prefer the term 'discourse particles' (or, more specifically, 'discourse connectives') to the term 'coordinating conjunctions'. In the first, theoretical part of this study a discourse-pragmatic framework for the description of 'connective' particles is developed. This framework is based on an evaluation of earlier attempts in this direction, and on the results of an empirical study of a number of Latin connective particles, which are discussed in the second part of the study. Central to this framework is the recent insight that connective particles somehow play a role in signalling or maintaining the discourse coherence, and, even more important, that discourse coherence obtains on various 'levels' or has various 'dimensions'. Coherence relations obtain on at least three levels of discourse, for which I use the terms representational, presentational and interactional level of discourse. The representational level of discourse is concerned with the representation of some real or imagined world outside the language itself. Connective particles with a function on this level usually signal relations between the states of affairs that make up the represented world. They usually take the form of subordinating conjunctions; therefore this type of particle is left out of account, except for a few occasional remarks in chapter 4. The presentational level of discourse is concerned with the particular way in which a speaker or author 'stages' and organizes the represented content. Particles with a role on this level indicate for instance how a particular discourse unit is

372

chapter 13

functionally related to other discourse units. Moreover they may highlight the macrostructure of a discourse. The interactional level of discourse, fmally, is concerned with the coherence of a unit of discourse in the light of the conversational interaction taking place between two or more interlocutors in a particular communicative situation. Particles with a function on the interactional level of discourse may signal how a communicative move fits into an interactional exchange, or how a. certain discourse unit is evaluated in terms of the communicative situation. The Latin particles investigated turn out to have functions on different discourse levels: nam and autem have primary functions on the presentational level of discourse, whereas enim, vero and at are primarily interactional particles. An important criterion for distinguishing between typically 'presentational' and typically 'interactional' particles appears to be the factor discourse type: presentational particles tend to occur in monological discourse types, interactional particles in dialogical discourse types. Of crucial importance, however, is the observation that not only typically dialogical Latin texts, but also essentially monological texts are apt to host interactional particles. To explain this phenomenon the concept of diaphony (as developed by the ParisGeneva school of linguistics; see ch. 5) proves to be of special use. A further criterion for the description and subcategorization of Latin discourse connectives appears to be the type of relationship indicated by the particles, which may be strictly textual (i.e. between two verbally expressed units of text), or rather extratextual (i.e. between a unit of text and some nonverbal, extratextual element). Nam, autem and at primarily appear to indicate textual relationships (as 'connective particles in a strict sense'), while enim and vero appear to be less strictly connective devices (I use the term 'situating particles' for the latter). Application of this discourse-pragmatic descriptive framework to the Latin so-called coordinating conjunctions thus yields at least three different functiongroups of particles: (i) a group with e.g. nam (see ch. 7), autem(see ch. 10) and igitur (see Kroon 1989), which all have a primary function on the presentational level of discourse, and usually behave as connective particles in a strict, textual sense; (ii) a group with e.g. enim (see ch. 8) and vero (see ch. 10), which both are primarily 'situating' particles with a function on the interactional level of discourse (the connective connotation that these particles undeniably have is to be described in terms of a regularly occurring side-effect of the particle when used in a particular context); (iii) a group with e.g. at (see ch. 12) and ergo (see Kroon 1989), which have a primary function on the

conclusions

373

interactional level of discourse, and usually behave as connective particles in a strict sense. These function-groups differ considerably from the semantic categories of connective particles that are distinguished in Latin grammars. In these traditional accounts nam, enim, igitur and ergo are usually grouped together under the heading of 'causal coordination,l, autem, vero and at under the heading of 'adversative coordination'. The results of the two contrasted lines of approach are graphically represented in the following figure: figure: discourse pragmatic versus semantic approach presentational 'connective'

,.'"

- - ... - - - _... - causal/connam; igitur ,t_ _ _ _ _ _ ... _ _ _ _ _ secutive

interactional 'situating'

interactional 'connective'

enim

ergo

---

1---------------------- --

adversative

I

autem

I I

,- -

-

-

-

-

-

-

-------

vero -

-

-

-

-

-

-

-

-

-

-

-

-

-

. -

-

-

-

--I

i

at

1

-

-

--I

The horizontal rows marked off by the dotted lines indicate the traditional subcategorization of Latin connective particles into the semantic categories 'causal/consecutive' and 'adversative'. The vertical rows that make up the square boxes represent a subcategorization in terms of primary discourse function. Classification within the same function-group does not, of course, mean that the particles concerned have identical discourse functions. The more specific discourse function of a particle is determined by an abstract basic meaning, which differs from particle to particle. For instance, enim and vero both are markers of the involvement of the discourse participants in the communication

1. Nam and enim are considered to be equivalents of the English conjunction 'for', igitur and ergo of English 'so'.

374

chapter 13

process, and are therefore grouped together in the category 'situating particles with a function on the interactional level of discourse'. However, whereas vera indicates the commitment of the speaker, enim is used to solicit the commitment of the addressee. This difference in discourse function is connected with a difference in their basic meanings: 'actuality' or 'reality' in the case of vero, 'consensus' in the case of enim. In consequence, the individual particles within each function-group are distinct items to the extent that no other particle within the same group has an equivalent set of actual uses (although some of these actual uses might in essence be expressed also by some other particle from the same or a different-function group). In other words, there are no real functional equivalents. Secondly, it should be observed that the above figure is a rather rough abstraction of what actually turns out to be a very refined and complex system. In addition to their primary discourse functions, particles may also have secondary (or derived) discourse functions. These secondary functions are related to the primary discourse function through a common basic meaning. Furthermore, the interplay of the discourse function of a particle (which is the pivotal element in the analysis and description of a connective particle in that it determines its more distinctive features vis a vis other particles) with properties of its context may lead to a number of different actual uses. This system, in which a distinction is made between basic meaning, discourse function(s), actual uses and side-effects (see chapter 5), may count as a more sophisticated explanation of a situation that has often been described as mere chaos or an extreme form of polysemy. A further point to be noted with regard to the diagram above is that the difference between connective and situating particles should be considered a continuous gradient rather than a strictly binary opposition: the Latin connective particles investigated appear to range from strongly connective to mainly situating, with all possible shades in between. At the 'situating' end of the scale, where we find for instance enim and vera, the contrast with so-called modal particles appears to be small. I therefore expect that the descriptive framework developed here for 'connective particles' may (with a few adaptations) be a useful tool for the description of other (e.g. 'modal') particles as well. The discourse-pragmatic approach proposed in this study has the advantage over traditional, semantic-syntactic accounts of Latin connective particles, that it provides a more unified picture of the particles involved, in which the interrelatedness between the various uses of a particular particle is made

conclusions

375

explicit, and in which ostensibly deviant instances can be explained rather easily. As such it makes unnecessary the assumption of unlikely diachronic developments. Moreover, the individual descriptions (see the summaries given at the end of each descriptive chapter) provide a better insight into the effects that a particular Latin particle can have in a particular context. As such they demonstrate that Latin particles are seldom meaningless 'fillers' distributed over a text at random, a view that is sometimes found in commentaries on Latin texts (cf. the discussion in Kroon 1992). Furthermore, the discoursepragmatic approach enables us to distinguish more clearly between nearsynonyms. The necessity for such a distinction is strongly suggested by observed differences in the distributional properties of the alleged synonyms. Since the argumentation is based firmly on objective linguistic criteria (as is necessary in a study of a language for which one cannot rely on the intuitions of native speakers), the present study may be useful to particle research in general and to the development of its analytical tools. Given its strongly empirical basis it might appear, moreover, to be a useful intermediate stage in the development of a more encompassing model of discourse coherence. It is my belief that an undertaking as vast and complicated as the development of such a model, should start 'bottom-up' rather than 'top-down'.

BIBLIOGRAPHY

Abbreviations Kiihner-Stegmann:

Szantyr: TLL:

OLD:

Kuhner, R. & C. Stegmann, AusfUhrliche Grammatik der lateinischen Sprache II: Satzlehre (vol. 2). Hannover: Hahnsche Buchhandlung 1912-14 Hofmann, J.B. & A. Szantyr, Lateinische Syntax und Stilistik. Munich: Beck 1965 Thesaurus Linguae Latinae editus auctoritate et consilio academiarum quinque Germanicarum Berolinensis Gottingensis Lipsiensis Monacensis Vindobonensis. Leipzig: Teubner 1900ff. OxjordLatinDictionary, edited by P.G.W. Glare. Oxford: Clarendon Press 1982

References Abraham, W. (ed.) 1991 Discourse particles. Amsterdam: Benjamins Altenberg, B. 1984 Causal linking in spoken and written English. Studia Linguistica 38, 20-69 Anscombre, J.C. 1984 La representation de la notion de cause dans 1a langue. Cahiers de Grammaire de Toulouse-Le Mirail8, 1-53 Anscombre, J.C. & O. Ducrot 1977 Deux mais en fran!;ais? Lingua 43, 23-40 1983 L'argumentation dans la langue. Bruxelles: P. Mardaga Austin, J.L. 1962 How to do things with words. Cambridge, Mass.: MIT-Press Austin, R.G. 1977 P. Vergili Maronis, Aeneidos tiber sextus (comm.). Oxford: Clarendon Press Axelson, B. 1945 Unpoetische Warter. Lund: Gleerup Bakker, E.J. 1988 Linguistics and formulas in Homer. Scalarity and the description of the particle per. Amsterdam: Benjamins 1993 Boundaries, topics and the structure of discourse. An investigation of the Ancient Greek particle de. Studies in Language 17, 275-311 Baratin, M. 1989 La naissance de la syntaxe a Rome. Paris: Minuit Barendt, P.O. 1902 Ciceronian use of nam and enim. Classical Review 16, 203-209

378

bibliography

Bazzanella, C. 1990 Phatic connectives as interactional cues in contemporary spoken Italian. Journal of Pragmatics 14, 629-647 Beaugrande, R.-A. de & W. Dressler 1981 Introduction to text linguistics. London: Longman Berry, M. 1981 Towards layers of exchange structure for directive exchanges. Network 2, 23-32 Blakemore, D. 1987 Semantic constraints on relevance. Oxford: Blackwell Blum-Kulka, S. 1983 The dynamics of political interviews. Text 3,131-153 1985 Modifiers as indicating devices: the case of requests. Theoretical Linguistics 12, 213-229 Bodelot, C. 1987 L'interrogation indirecte en latin. Syntaxe - valeur illocutoire - formes. Paris: Societe pour l'Information Grammaticale Bolkestein, A.M. 1983 The role of discourse in syntax: evidence from the Latin Nominativus cum Infinitivo. In: K. Ehlich & H. Van Riemsdijk (eds) Connectivity in Discourse and Syntax. Tilburg: University Press, p. 111-140 1985 Discourse and case-marking; three-place predicates in Latin. In: C. Touratier (ed.) Syntaxe et latin. Aix-en-Provence: Universite de Provence, p. 191-225 1987 Discourse functions of predications: the background/foreground distinction and tense and voice in Latin main and subordinate clauses. In: J. Nuyts & G. de Schutter (eds) Getting one's words into line. On word order and Functional Grammar. Dordrecht: Foris, p. 163-78 1988 Nurnne an an non? In: D. den Hengst & J.J.L. Smolenaars (eds) Propemptikon W.J.H.F. Kegel. Amsterdam: University of Amsterdam, p. 94-102 1989 Parameters in the expression of embedded predications in Latin. In: G. Calboli (ed.) Subordination and other topics in Latin. Amsterdam: Benjamins, p. 3-35 1991 Causally related predications and the choice between parataxis and hypotaxis in Latin. In: R. Coleman (ed.) New studies in Latin linguistics. Amsterdam: Benjamins, p. 427~451 1992 Anaforische subjecten en tekststructuur in het Latijn. GrammalTTT, Tijdschrijt voor Taalkunde 1, 111-124 Bolkestein, A.M. & M. v.d. Grift 1994 Participant tracking in Latin discourse. In: J. Herman (ed.) Linguistic studies on Latin. Amsterdam: Benjamins, p. 283-302 Bolkestein, A.M. & R. Risselada 1985 De tekstuele funktie van valentie: Latijnse drieplaatsige predikaten in kontekst. Tijdschrijt voor Tekst- en Taalwetenschap 5, 161-76 1987 The pragmatic motivation of semantic and syntactic perspective. In: M. BertucelliPapi & J. Verschueren (eds) The pragmatic perspective. Amsterdam: Benjamins, p. 497-512

bibliography

379

BrauBe. U. 1983 Bedeutung und Funktion einiger Konjunktionen und Konjunktional-adverbien: Aber, nur, immerhin, allerdings, dafi1r, dagegen, jedoch. In: Untersuchungen zu Funktionswortern (Adverbien, Konjunktionen, Partikeln) (= Linguistische Studien, Reihe A Heft 104). Berlin: Akademie, p. 1-40 Brix, l. & M. Niemeyer 1866-1901 Ausgewiihlte Komodien des T. Maccius Plautus. Bnd. I: Trinumnus. Leipzig: Teubner (6th revised ed., Leipzig 1931) Buhler, K. 1934 Sprachtheorie. lena: Fischer Burton, D. 1980 Dialogue and discourse: a sociolinguistic approach to modern drama dialogue and naturally occurring conversation. London: Routledge & Kegan Paul Butler, C.S. 1982 The directive junction of the English modals. Unpublished Ph.D. Thesis, University of Nottingham Chafe, W.L. 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In: C.N. Li & S.A. Thompson (eds) Subject and Topic. New York: Academic Press, p. 25-56 Chausserie-Lapree, l.R. 1969 L'expression narrative chez les historiens latins. Paris: De Boccard Clement, W.K. 1897 The use of enim in Plautus and Terence. American Journal of Philology 18, 402416 Cova, P.V. et al. 1986 Studi sulla lingua di Plinio il Vecchio. Milano: Vita e Pensiero Crothers, E.l. 1979 Paragraph structure inference. Norwood, NY: Ab1ex Davies, E.C. 1979 On the semantics of syntax. Mood and condition in English. London: Croom Helm Denniston, l. D. 1954 The Greek particles. London: Oxford University Press (1st ed. 1934) Dewitt, N.W. 1938 The semantics of Latin particles. Classical Journal 33, 450-456 Dijk, T.A. van 1977 Text and context. Explorations in the semantics and pragmatics of discourse. London: Longman 1979 Pragmatic connectives. Journal of Pragmatics 3, 447-456 1981 Studies in the pragmatics of discourse. The Hague: Mouton Dik, S.C. 1968 Coordination: its implications for the theory of general linguistics. Amsterdam: North Holland 1989 The theory of Functional Grammar I: the structure of the clause. Dordrecht: Foris

380

bibliography

1990 On the semantics of conditionals. In: J. Nuyts, A.M. Bolkestein & c. Vet (eds) Layers and levels of representation in language theory. Amsterdam: Benjamins, p. 233-261 Doherty, M. 1987 Epistemic meaning. Berlin: Springer Verlag Dougan, Th.W. & R.M. Henry 1905-1934 M. Tulli Ciceronis Tusculanorum Disputationum Libri Quinque (2 vols). Cambridge: Cambridge University Press Draeger, A. 1878 Historische Syntax der lateinischen Sprache II. Leipzig: Teubner Drescher, M. & Th. Kotschi 1988 Das "Genfer Modell". Diskussion eines Ansatzes zur Diskursanalyse am Beispiel der Analyse eines Beratungsgespriichs. Sprache und Pragmatik 8, 1-42 (Arbeitsberichte Lund) Ducrot, O. 1982 Note sur l' argumentation et l' acte d' argumenter. Cahiers de Linguistique Franfaise 4, 143-163 1983 Operateurs argumentatifs et visee argumentative. Cahiers de Linguistique Franfaise 5, 7-36 Ducrot, O. et al. 1975 Car, parce que, puisque. Revue Romane X 2,248-480 1980 Mais occupe-toi d'Arnelie. In: Ducrot et al. (eds) Les mots du discours. Paris: Minuit, p. 93-130 Ducrot, O. et al. (eds) 1980 Les mots du discours. Paris: Minuit Dunbar, R. W. 1985 Context and syntax: the effect of discourse pragmatics on clause structure in German. In: J.R. Wirth (ed.) Beyond the sentence. Ann Arbor: Karoma, p. 21-31 Edmondson, W.J. 1981 Spoken discourse: aA model for analysis. London: Longman Eemeren, F.H. van & R. Grootendorst 1984 Speech acts in argumentative discussions. Dordrecht: Foris Ellis, J. 1987 The logical and textual functions. In: Halliday, M.A.K. & R.P. Fawcett (eds) New developments in Systemic Linguistics. London: Pinter, p. 107-129 Enk, P.J. 1932 T. Maccius Plautus, Mercator (2 vols.). Leyden: Sijthoff (8th. ed.; repro New York: Arno Press, 1979) Enkvist, N.E. 1989 From text to interpretability: a contribution to the discussion of basic terms in text linguistics. In: Heydrich, W. et al. (eds) Connexity and coherence: analysis of text and discourse. Berlin: De Gruyter, p. 369-382 Ernout, A. & F. Thomas 1951 Syntaxe latine. Paris: Klincksieck

bibliography

381

Fawcett, R.P. 1980 Cognitive linguistics and social interaction: towards an integrated model of a Systemic Functional Grammar and the other components of a communicating mind. Heidelberg: Groos Verlag Ferrara, A. 1985 Pragmatics. In: T.A. van Dijk (ed.) Handbook of Discourse Analysis, II: Dimensions of discourse. New York: Academic Press, p. 137-157 Fleischmann, S. 1991 Toward a theory of tense-aspect in narrative discourse. In: 1. Gvozdanovic & Th.A.J .M. Janssen (eds) The junction of tense in text. Amsterdam: North Holland, p.75-97 Foley, W.A. & R.D. van Valin 1984 Functional syntax and universal grammar. Cambridge: Cambridge University Press Fontenrose, J. 1944 The meaning and use of sed enim. Transactions of the American Philological Association 75, 168-195 Foolen, A. 1989 Beschreibungsebenen fur Partikelbedeutungen. In: H. Weydt (ed.) Sprechen mit Partikeln. Berlin: De Gruyter, p. 305-317 1993 De betekenis van partikels. Een dokumentatie van de stand van het onderzoek met bijzondere aandacht voor maar. Diss. Katholieke Universiteit Nijmegen Franck, D. 1980 Grammatik und Konversation. KonigsteinlTs: Scriptor Fraser, B. 1990 An approach to discourse markers. Journal of Pragmatics 14, 383-395 Friedlander, L. 1891 Petronii Cena Trimalchionis. Leipzig: Hirzel (2nd ed. 1906) Fugier, H. 1987 Les connecteurs "de cause" en latin. Modeles Linguistiques 9, 9-17 1989 Quod, quia, quoniam et leurs effets textuels chez Ciceron. In: G. Calboli (ed.) Subordination and other topics in Latin. Amsterdam: Benjamins, p. 91-119 Giv6n, T. 1990 Syntax: A junctional-typological introduction II. Amsterdam: Benjamins Giv6n, T. (ed.) 1983 Topic continuity in discourse. A quantitative cross-language study. Amsterdam: Benjamins Goffman, E. 1981 Footing. Forms of talk. Philadelphia: University of Pennsylvania Goldberg, J.A. 1983 A move toward describing conversational coherence. In: Craig, R.T. & K. Tracy (eds) Conversational coherence; form, structure, and strategy. New Delhi: Sage Publications, p. 25-45 Greenbaum, S. 1969 Studies in English adverbial usage. London: Longman

382

bibliography

Griend, M.E. van de 1989 Pseudo-conditionals in Latin. In: M. Lavency & D. Longree (eds) Actes du cinquieme Colloque de Linguistique Latine. Cahiers de l'Institut de Linguistique de Louvain 15,1-4. Louvain-Ia-Neuve: Peeters, p. 447-456 Grimes, J.E. 1975 The thread of discourse. The Hague: Mouton Haiman, J. & S.A. Thompson 1984 'Subordination' in universal grammar. In: C. Brugmann & M. Macaulay (eds) Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, p. 510-523 Hale, W.G. 1887 The cum-constructions, their history andfunctions. New York: Cornell University Press Hale, W.G. & C.D. Buck 1903 A Latin Grammar. Boston: Ginn Halliday, M.A.K. 1973 Explorations in the functions of language. London: Arnold 1985 An introduction to Functional Grammar. London: Arnold Halliday, M.A.K. & R. Hasan 1976 Cohesion in English. London: Longman Hand, F. 1829-1845 Tursellinus seu de particulis Latinis commentarii (4 vols). Leipzig: Weidemann (repr. Amsterdam: Hakkert, 1969) Harpe, 1. De la 1923 Etude sur tamen conjonction adversative et son passage au sens causal avec remarques comparatives sur les particules sed, autem, nam, enim. Diss. Universite de Lausanne Harras, G. 1984 Was erkHiren Kausalsatze? In: G. Stickel (ed.) Pragmatik in der Grammatik. Dusseldorf: Schwann, p. 175-194 Harweg, R. 1968 Pronomina und Textkonstitution. Munich: Fink (2nd. revised ed., 1979) Hentschel, E. 1986 Funktion und Geschichte deutscher Partikeln: 'ja', 'doch', 'halt' und 'eben '. Tubingen: Niemeyer Heringer, H.J. 1988 Ja, ja, die Partikeln! Konnen wir Partikelbedeutungen prototypisch erfassen? Zeitschrijt Jilr Phonetik, Sprachwissenschajt und Kommunikationsforschung 41, 730-754 Hilton, J. 1989 Temporal markers in the Letters of Cicero. In: M. Lavency & D. Longn!e (eds) Actes du cinquieme Colloque de Linguistique Latine. Cahiers de l'Institut de Linguistique de Louvain 15,1-4. Louvain-Ia-Neuve: Peeters, p. 173-184 Hinds, J. et al. 1987 Perspectives on topicalization. The case of Japanese wa. Amsterdam: Benjamins

bibliography

383

Hobbs, J.R. 1985 On the coherence and structure of discourse. Center for the Study of Language and Information (CSLI). Stanford University, Technical Report 85-37 Hofmann, J.B. & A. Szantyr 1965 Lateinische Syntax und Stilistik. Munich: Beck Hoffmann, M.E. 1989 A typology of Latin Theme constituents. In: M. Lavency & D. Longn!e (eds) Actes du cinquieme Colloque de Linguistique Latine. Cahiers de l'Institut de Linguistique de Louvain 15,1-4. Louvain-Ia-Neuve: Peeters, p. 185-196 Janson, T. 1979 Mechanisms of language change in Latin. Stockholm: Almquist & Wiksell Jones, F. 1991 Subject, topic, given, and salient: sentence-beginnings in Latin. Proceedings of the Cambridge Philological Society 37, 81-105 Keil, H. 1961 GrammaticiLatini (8 vols). Hildesheim: Olms (repr. of 1st edition, Leipzig, 18551880) Klein, J. 1987 Die konklusiven Sprechhandlungen: Studien zur Pragmatik, Semantik, Syntax und Lexik von Bergriinden, Erklaren warum, Folgern und Rechtfertigen. Tubingen: Niemeyer Konig, E. 1986 Conditionals, concessive conditionals and concessives: Areas of contrast, overlap and neutralization. In: E.C. Traugott et a!. (eds) On conditionals. Cambridge: Cambridge University Press, p. 228-246 The meaning of focus particles. A comparative perspective. London: 1991a Routledge 1991b Identical values in conflicting roles: the use of German ausgerechnet, eben, genau and gerade as focus particles. In: W. Abraham (ed.) Discourse Particles. Amsterdam: Benjamins, p. 11-36 Konig, E. & J. Van der Auwera 1988 Clause integration in German and Dutch conditionals, concessive conditionals and concessives. In: J. Haiman & S.A. Thompson (eds) Clause combining in grammar and discourse. Amsterdam: Benjamins, p. 101-133 Konig, E. & P. Eisenberg 1984 Zur Pragmatik von Konzessivsatzen. In: Stickel, G. (ed.) Pragmatik in der Grammatik. Dusseldorf: Schwann, p. 313-332 Kroon, C.H.M. 1989 Causal connectors in Latin: the discourse function of nam, enim, igitur and ergo. In: M. Lavency & D. Longn!e (eds) Actes du cinquieme Colloque de Linguistique Latine. Cahiers de l'Institut de Linguistique de Louvain, 15,1-4. Louvain-IaNeuve: Peeters, p. 231-243 1992 Particula perplexa. In: G. Bakkum et a!. (eds) Pentecostalia. Bundel ter gelegenheid van de vijftigste verjaardag van Harm Pinkster. Pub!. University of Amsterdam, p. 53-63

384 1994a

1994b

1994c

bibliography Latijnse maar-equivalenten in narratief proza. In: R. Boogaart & J. Noordegraaf (eds) Nauwe betrekkingen. Voor Theo Janssen bij zijn vijftigste verjaardag. Stichting neerlandistiek VU, Amsterdam. Munster: Nodus Publikationen, p. 129-136 Discourse markers, discourse structure and Functional Grammar. Paper read at the VIth International Conference on Functional Granunar, York, August 2226, 1994. Discourse connectives and discourse type: the case of Latin at. In: 1. Herman (ed.) Linguistic studies on Latin. Amsterdam: Benjamins, p. 303-317

Kuno, S. 1976 Subject, theme, and the speaker's empathy. A reexamination of relativization phenomena. In: Ch.N. Li (ed.) Subject and topic. New York: Academic Press, p. 417-444 Kuhner, R. & C. Stegmann 1912-1914 Ausfilhrliche Grammatik der lateinischen Sprache II: Satzlehre (2 vols). Hannover: Hahnsche Buchhandlung Langen, P. 1880 Beitriige zur Kritik und Erkliirung des Plautus. Leipzig: Teubner (repr. Hildesheim: Dims, 1973) Lavency, M. 1975-6 La valeur de la conjonction cum en latin. Les etudes classiques 43,365-386 and 44,45-59. 1985 Problemes du classement des propositions en cum. In: C. Touratier (ed.) Syntaxe et Latin. Aix-en-Provence: Universite de Provence, p. 279-286 1989 Pour une description syntaxique de la phrase latine: complements conjoints et complements adjoints. In: G. Calboli (ed.) Subordination and other topics in Latin. Amsterdam: Benjamins, p. 153-179 Lehmann, C. 1988 Towards a typology of clause linkage. In: J. Hairnan & S. A. Thompson (eds) Clause combining in grammar and discourse. Amsterdam: Benjamins, p. 181-225 1989 Latin subordination in typological perspective. In: G. Calboli (ed.) Subordination and other topics in Latin. Amsterdam: Benjamins, p. 153-179 Levinsohn, S.H. 1987 Textual connections in Acts. Atlanta, Georgia: Scholars Press Levinson, S.C. 1983 Pragmatics. Cambridge: Cambridge University Press Lindner, K. 1991 'Wir sind ja doch alte Bekannte': the use of German ja and doch as modal particles. In: W. Abraham (ed.) Discourse particles. Amsterdam: Benjamins, p. 162-201 Lodge, G. 1904-1933 Lexicon Plautinum (2 vols). Leipzig: Teubner (repr. Hildesheim: Dims, 1971) Longacre, R.E. 1979 The paragraph as a granunatical unit. In: T. Givon (ed.) Syntax & Semantics 12: Discourse and syntax, p. 115-134

385

bibliography

1983 The grammar of discourse. New York: Plenum Luke, K.K. 1989 The cantonese utterance particle la and the accomplishment of common understandings in conversation. Papers in pragmatics 3, 39-87 Lundquist, L. 1980 La coherence textuelle. Copenhagen: Nyt Nordisk Forlag Lyons, 1. 1977 Semantics. Cambridge: Cambridge University Press Mann, W.C. et al. 1992 Rhetorical Structure Theory and text analysis. In: W.C. Mann & S.A. Thompson (eds) Discourse description. Diverse linguistic analyses of a fund-raising text. Amsterdam: Benjamins, p. 39-78 Mann, W.C. & S.A. Thompson 1987 Rhetorical Structure Theory: a framework for the analysis of texts. Papers in Pragmatics 1, 79-105 1988 Rhetorical Structure Theory: toward a functional theory of text organization. Text 8, 243-281 1989 Rhetorical Structure Theory: a theory of text organization. ISIIRS-87-190 from Document Center, USC/lSI, 4676 Admiralty Way, Marina del Rey, CA 90292 Marouzeau, 1. 1948 La construction des particules de liaison. Revue des Etudes Latines 26, 235-267 1949 L'ordre des mots dans la phrase latine III. Paris: Les Belles Lettres Martin,l.R. 1983 Conjunction: the logic of English text. In: 1.S. Petofi & E. Sozer (eds) Micro and macro connexity of texts. Hamburg: Buske Verlag, p. 1-72 Matthiessen, Chr. & S.A. Thompson 1988 The structure of discourse and 'subordination'. In: 1. Haiman & S.A. Thompson (eds) Clause combining in grammar and discourse. Amsterdam: Benjamins, p. 275-329 Maynard, S.K. 1989 Functions of the discourse marker dakara in lapanese conversation. Text 9, 389414 1992 Speech act declaration in conversation: functions of the japanese connective datte. Studies in Language 16, 63-89 Mellet, S. 1988 L'imparjait de l'indicatif en latin. Paris: Societe pour l'Information Grammaticale 1994 Elements pour une etude de la synonymie syntaxique; I' exemple des conjonctions de cause. In: C. Moussy (ed.) Les problemes de la synonymie en latin. Paris: Presses de I'Universite de Paris-Sorbonne, p. 203-221 1995 Quando, quia, quod, quoniam: analyse enonciative et syntaxique des conjonctions de cause en latin. In: D. Longree (ed.) De USU. Etudes de syntaxe latine offertes en hommage Marius Lavency. Cahiers de l'Institut de Linguistique de Louvain 70. Louvain-Ia-Neuve: Peeters, p. 211-228 Menge, H. 1960 Repetitorium der lateinischen Syntax und Stilistik. Munich: Hueber (13th ed., revised by A. Thierfelder)

a

386

bibliography

McGlynn, P. 1963-1967 Lexicon Terentianum. London: Blackie & Son Moeschler, I. 1985 Argumentation et conversation. Elements pour une analyse pragmatique du discours. Paris: Hatier 1987 Trois emplois de parce que en conversation. Cahiers de Linguistique Franraise 8, 97-110 Moeschler, I. & N. de Spengler 1981 Quandmeme: de la concession it la refutation. Cahiers de Linguistique Franraise 2, 93-112 1982 La concession ou la refutation interdite, approche argumentative et conversationnelle. Cahiers de Linguistique Franraise 4, 7-36 Nagelsbach, K.F. 1905 Lateinische Stilistik. Neurenberg (repr. Darmstadt: Wissenschaftliche Buchgesellschaft, 1967) Norden, E. 1916 Aeneis, P. Vergilius Maro; Buch VI. Leipzig: Teubner Ostman, 1.-0. 1981 You know: a discourse-junctional approach. Amsterdam: Benjamins Owen, M. 1989 Book notice D. Schiffrin Discourse markers (1987). Journal of Linguistics 25, 255-257 Pease, A.S. 1955-1958 M. Tulli Ciceronis De Natura Deorum. Cambridge MA: Harvard University Press (repr. Darmstadt: Wissenschaftliche Buchgesellschaft, 1968) Petersmann, H. 1977 Petrons urbane Prosa. Untersuchungen zu Sprache und Text (Syntax). Vienna: Akademie Verlag Pike, K.L. 1967 Language in relation to a unified theory of the structure of human behaviour (2nd rev. ed.). The Hague: Mouton Pike, K.L. & E.G. Pike 1983 Text and Tagmeme. London: Pinter Pinkster, H. 1972 On Latin adverbs. Amsterdam: North-Holland 1986 Ego, tu, nos. Opmerkingen over het gebruik van subjektpronomina, in het bijzonder in Cicero de Oratore II. Lampas 19, 309-322 1987 The pragmatic motivation for the use of subject pronouns: the case of Petronius. In: S. Mellet (ed.) Etudes de linguistique generale et de linguistique latine offertes a G. Serbat. Paris: Societe pour I'Information Grammaticale, p. 369-379 1990 Latin syntax and semantics. London: Routledge Polanyi, L. & R.I.H. Scha 1983 The syntax of discourse. Text 3,261-70

bibliography

387

Posner, R. 1980 Semantics and pragmatics of sentence connectives in natural languages. In: J. Searle et al. (eds) Speech act theory and pragmatics. Dordrecht: Reidel, p. 169203 Poyser, G.H. 1952 A usage of nam. Classical Review 66, 8-10 Prince, E.F. 1981 Towards a taxonomy of given-new information. In: P. Cole (ed.) Radical pragmatics. New York: Academic Press, p. 223-255 Quirk, R. et al. 1985 A comprehensive grammar of the English language. London: Longman Rathmayr, R. 1985 Die russischen Partikeln als Pragmalexeme. Munich: Verlag Otto Sagner Rebert, H.F. 1929 The origin and meaning of Latin at. Classical Philology 24, 169-175 Redeker, G. 1990 Ideational and pragmatic markers of discourse. Journal of Pragmatics 14, 367-381 1991 Linguistic markers of discourse structure: review article of Discourse markers, by D. Schiffrin. Linguistics 29, 1139-1172 Riemann, O. 1886 Syntaxe latine. Paris: Klincksieck (7th revised ed., 1942) Rijksbaron, A. 1976 Temporal and causal conjunctions in Ancient Greek. Amsterdam: Hakkert Risselada, R. 1993 Imperatives and other directive expressions in Latin. A study in the pragmatics of a dead language. Amsterdam: Gieben 1994 Modo and sane, or what to do with particles in Latin directives. In: J. Herman (ed.) Linguistic studies on Latin. Amsterdam: Benjamins, p. 319-342 Roby, H.J. 1874 A grammar of the Latin language from Plautus to Suetonius, part II. London: Macmillan & Co Rosen, H. 1989 On the use and function of sentential particles in Classical Latin. In: M. Lavency & D. Longree (eds) Actes du cinquieme Colloque de Linguistique Latine. Cahiers de L'Institut de Linguistique de Louvain 15,1-4. Louvain-la-Neuve: Peeters, p. 391-402 Rosengren, I. 1987 Begriindungen und Folgerungen als kommunikative Handlungen. In: W. Motsch (ed.) Satz, Text, sprachliche Handlung. Berlin: Akademie, p. 179-197 Roulet, E. 1984 Speech acts, discourse structure and pragmatic connectives. Journal of Pragmatics 8, 31-47 1991 Le modele genevois d'analyse de discours; evolution et perspectives. Pragmatics 1, 243-248 Roulet, E. et al. 1985 L 'articulation du discours en Jrant;ais contemporain. Bern: Lang

388

bibliography

Rubattel, C. 1988 Structure syntaxique et forme logique des unites discursives monologiques. Cahiers . de Linguistique Franraise 9, 7-25 Rudolph, E. 1979 Zur Klassifizierung von Partikeln. In: H. Weydt (ed.) Die Partikeln der deutschen Sprache. Tubingen: Niemeyer, p. 139-151 1981 Zur Problematik der Konnektive des kausalen Bereichs. In: J. Fritsche (ed.) Konnektivausdrucke, Konnektiveinheiten. Grundelementeder semantischen Struktur von Texten I. Hamburg: Helmut Buske Verlag, p. 146-244 Rumelhart, D.E. 1975 Note on a schema for stories. In: D. Bobrow & A. Collins (eds) Representation and understanding. New York: Academic Press, p. 211-236 Ruijgh, C.J. 1971 Autour de te epique. Amsterdam: Hakkert Sanders, T. 1992 Discourse structure and coherence. Aspects of a cognitive theory of discourse representation. Diss. Katholieke Universiteit Brabant Saur, H. 1913 Die Adversativpartikeln bei Lateinischen Prosaikern. Diss. Tubingen Schenkeveld, D.M. 1986 From particula to particle - the genesis of a class of words. In: I. Rosier (ed.) L 'heritage des grammairiens latins de I 'antiquite aux lumieres. Louvain-la-Neuve: Peeters, p. 81-93 Schiffrin, D. 1987 Discourse markers. Cambridge: Cambridge University Press Schiwy, P. 1932 Die syntaktischen Funktionen der Partikel nam. Diss. Breslau Schleppegrell, M.l. 1991 Paratactic because. Journal of Pragmatics 16, 323-337 Schourup, L.C. 1985 Common discourse particles in English conversation: like, well, y'know. New York: Garland 1988 Review of D. Schiffrin Discourse markers (1987). Language 64, 633-637 Schrijvers, P.H. 1970 Horror ac divina voluptas: etudes sur la poetique et la poesie de Lucrece. Amsterdam: Hakkert Searle, J.R. 1969 Speech acts. Cambridge: Cambridge University Press Shackleton Bailey, D.R. 1965-1970 Cicero's letters to Atticus (7 vols). Cambridge: Cambridge University Press Sicking, C.M.J. 1986 Griekse partikels: definitie en classificatie. Lampas 19, 125-141 Sinclair, 1.M. & R.M. Coulthard 1975 Towards an analysis of discourse: the English used by teachers and pupils. London: Oxford University Press

bibliography

389

Solodow, J.B. 1978 The Latin particle quidem. Boulder Colo.: American Philological Association Somers, M.H. 1994 Theme and topic. The relation between discourse and constituent fronting in Latin. In: J. Herman (ed.) Linguistic studies on Latin. Amsterdam: Benjamins, p. 151163 Sperber, D. & D. Wilson 1986 Relevance. Communication and cognition. Oxford: Blackwell Spooren, W.P.M.S. 1989 Some aspects of the form and interpretation of global contrastive coherence relations. Diss. Katholieke Universiteit Nijmegen Suzuki, R. 1990 The role of particles in Japanese gossip. Berkeley Linguistics Society 16, 315-324 Sweetser, E. 1990 From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press Szantyr: see Hofmann & Szantyr Sznajder, L. 1993 Sur la concurrence entre presents historiques et infinitifs de narration chez les historiens latins. Paper read at the 7th Colloquium on Latin Linguistics, Jerusalem, April 1993 Thesleff, H. 1960 Yes and no in Plautus and Terence. Helsinki: Societas Scientiarum Fennica Thompson, S.A. & W.C. Mann 1986 A discourse view of concession in written English. In: S. Delancey & R. Tomlin (eds) Proceedings of the second annual meeting of the Pacific Linguistic Conference. 1987 Antithesis: a study in clause combining and discourse structure. In: R. Steele & T. Treadgold (eds) Language topics. Essays in honour of Michael Halliday. Amsterdam: Benjamins, p. 359-381 Toolan, M.J. 1988 Narrative. A critical linguistic introduction. London: Routledge Traugott, E.C. 1982 From propositional to textual and expressive meanings. Some semantic-pragmatic aspects of grammaticalization. In: W.P. Lehmann & Y. Malkiel (eds) Perspectives on historical linguistics. Amsterdam: Benjamins, p. 245-271 Vandepitte, S. 1993 A pragmatic study of the expression and the interpretation of causality. Conjuncts and conjunctions in modern spoken British English. Brussel: PaIeis der Academien Van der Auwera, 1. & W. Vandeweghe 1984 Inleiding. In: J. Van der Auwera & W. Vandeweghe (eds) Studies over Nederlandse partikels. Antwerp Papers in Linguistics ill 35, p. 7-15 Van Valin, R. 1984 A typology of syntactic relations in clause linkage. In: C. Brugmann & M. Macaulay (eds) Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society

390

bibliography

Viljamaa, T. 1983 Infinitive of narration in Livy. A study in narrative technique. Turku: Annales Universitatis Turkuensis Wakker, G.C. 1994 Conditions and conditionals. An investigation of Ancient Greek. Amsterdam: Gieben Warner, R. 1985 Discourse connectives in English. New York: Garland Weinrich, H. 1964 Tempus. Besprochene und erziihlte Welt. Stuttgart: Kohlhammer (2nd. ed. 1971) Weydt, H. 1969 Abtonungspartikel: die deutschen Modalworter und ihre jranzosischen Entsprechungen. Bad Homburg: Gehlen Weydt, H. (ed.) 1977 Aspekte der Modalpanikeln. Tubingen: Niemeyer 1979 Die Partikeln der deutschen Sprache. Tubingen: Niemeyer 1983 Partikeln und Interaktion. Tubingen: Niemeyer 1989 Sprechen mit Partikeln. Berlin: De Gruyter Wierzbicka, A. 1986 Introduction to special issue on particles. Journal of Pragmatics 10, 519-534 Wilkins, A.S. 1892 M. Tulli Ciceronis De Oratore libri tres. London: Oxford University Press (repr. Hildesheim: Olms, 1965) Wolski, W. 1986 Partikeliexicographie. Ein Beitrag zur praktischen Lexikologie. Tubingen: Niemeyer

INDEX OF TERMS

The index contains references to passages in which the terms listed are defined (or redefined) accessible information act actual use adjacency pair Adjunct adverb as contrasted to particle adverb in a strict sense modal adverb adversative conjunction adversativity afterthought

250 50, 65-66, 65 n. 12, 78, 271 98, 100-101 66,90 286 n. 7

basic meaning bottom-up approach

see sub meaning 8

causality conversational causality epistemic causality sociophysical causality causal relation central challengeability chiaroscuro classifications of relations discourse-pragmatic classifications formal and structural classifications (logical-)semantic classifications pragmatic classifications clause clause complex clause combining coherence cohesion in the sense of Halliday communicative frame communicative situation communicative units

129-131 131 131 131 129-132 65-66,76 196-198, 310 321 9-10 20-24 24-30 10-11 12-24 25 n. 39 25 n. 39 25,67 30 30 12, 59-60 112-113 63 see sub unit (of communication)

285-288 (esp. n. 6) 38 35 210 210-217 155

392

index of terms

concession direct concession indirect concession interactional concession logical concession rhetorical concession conjunct conjunction connective connector consensus coordinating conjunction coordination coordinator

211 211-212 212 216 216 216 7 40 see discourse connective 40 184 69 25 40

deictic orientation Dependence dia-discourse dialogical discourse dialogical traits dialogical discourse type diaphonic discourse diaphony discourse discourse act discourse boundary discourse connective discourse contrast discourse function primary discourse function secondary discourse function discourse marker discourse particle discourse topic fronted discourse topic new discourse topic shift of discourse topic discourse type dialogical discourse type dialogical dialogal discourse type dialogical monologal discourse type dialogal discourse type diaphonic discourse type mono logical discourse type monological dialogal discourse type monological monologal discourse type

113 80-88 114 see discourse type 112, 114-115 see sub discourse type see sub discourse type 52,111-115 30 n. 50 see act 249 36, 40, 56 211,213 98-100

100 100 36 see sub particle 228 n. 3,248 257 261-262 249 (esp. n. 28) 52, 109 110 110 109 52,111-115,178 52, 109-110 110 110

index of terms

393

monologal discourse type discourse unit Disjunct

109 see sub unit 286 n. 7

embedded voice empathy enonciateur

111,344 199 (esp. n. 38) 110 n. 22 245 16, 131 50, 65-66 see sub type of relationship

epanorthosis evidence exchange extratextual relationship

absolute focus contrastive focus parallel focus focus marking focusing device frustration frustation of expectation function-group Geneva discourse model given information

227 (esp. n. 1) 239-240 227 (esp. n. 2), 239 227 n. 2, 239 121 see focus see (frustrated) relation 351, 359 372-373 50-54 250

higher order speech acts homonymy host unit

17-19 see maximalism 35

illocutionary act illocutionary force directive illocutionary force interactional function interactional level of discourse interaction management interjection intratextual relationship

18, 65 n. 12 (as opposed to discourse act), 78 (idem) 77-78, 118, 189-195 190-191 76-78,90-92 see sub level of discourse 71 296-297 see sub type of relationship

justification

16, 131

layer of the text communicative layer reportive (= narrative) layer level of discourse interactional level of discourse presentational level of discourse

80 112 80, 112 58-62 61-62, 89-95 61-62, 73-89

focus

index of terms

394 representational level of discourse Listing locuteur

61-62, 69-73 80-88 110 n. 22

maximalism meaning basic meaning functional meaning logical meaning pragmatic meaning procedural meaning referential meaning metacommunicative expressions metadirective minimalism moderate minimalism monological discourse type monologal discourse type monoperspective argumentation monophonic discourse monosemy move counter-move initiating move problemizing move reactive move multiperspective argumentation

43-44, 97-98 98-99 41 41 see functional meaning 46 41 114-115 192-194 43-44, 97-98 44,98 see sub discourse type see sub discourse type 364 n. 36 114 see minimalism 50, 65-66, 271 see reactive move 90-92 338 90-92 364 (esp. n. 36)

narrative mode affective mode neutral mode new information

322 322 322 261 n. 42

objective modality marker organization of the text

283 157 (esp. n. 22), 158, 340 67, 75, 80-89

particle affirmative particle connective particle conversation particle discourse particle focus particle functional definition modal particle morphological definition

136 39-40 283-284 36, 56 36,42, 71, 283 35 35-36,42 34

occupatio

index of terms

395

negative definition scalar particle situating particle perlocutionary polysemy pragmatic motivation pragmatics praeteritio presentational level of discourse pseudo-apodotic clause

34 see focus particle 63,95,282 192 see (moderate) minimalism

rebuttal Recursion referent continuity relator relation adversative relation argumentative relation external relation frustrated relation global relation interactional relation internal relation local relation pragmatic relation rhetorical relation semantic relation semantic-pragmatic relation speech act relation representational level of discourse rhetorical contrast rhetorical function Rhetorical Structure Theory

211, 214 80, 83-88 248 7, 34 7 210-217 78-79 12 15, 212 24, 81-83, 161,226 67-68, 90-93 12 24, 81-83, 161,226 15 67-68, 67 n. 15, 75 (esp. n. 27), 76-80 67-68 12-17 17-19 see sub level of discourse see discourse contrast 76-78 20-24

semantic opposition semantic-functional relationship semantics sentence sequential discourse relationship setting rhetorical setting side-effects situating particle speech act state of affairs

211,212-213 see sub type of relationship 10 25 n. 39 175 248,263 264 98, 101 see sub particle see illocutionary act 69

100 10 158 (esp. n. 23) see sub level of discourse 344-348

396

index of terms

structural relationship structure conversational structure formal structure hierarchical (discourse) structure linear structure text structure thematic structure subjective evaluation verb subjective modality marker subordinating conjunction subordination subsidiary substitution subtopic syntax

see sub type of relationship

taxonomies of relations text Text Grammar text relation text structure text type thematic chain thematic coherence thematic (dis )continuity thematic strand thematic unit Theme top-down approach topic topicality type of relationship extratextual relationship intratextual relationship semantic-functional relationship structural relationship

see classifications of relations 30 n. 50 24-26, 28 n. 44 see sub relation see sub structure 109 248 248 248,249 see thematic chain see sub unit 120 (esp. n. 30), 258 n. 36 7-8 228 n. 3 see topic 62 63-64 63-64 67-68 64-67

67,94 28 50,66-67 66-67 26-30 28-29 115 283-284 69 25 65-66,76 211 254 10

unit discourse unit thematic unit unit of communication unit of content unit of form verifying question

50,64 64 28-29, 64-67 28-29 28 307-308

INDEX OF CRITICALLY DISCUSSED PASSAGES

Apuleius Met. 8.17: 180, 181 n. 10 Caesar Civ. 2.24-25: 220, 267 Gal. 1.12.2: 324; 3.12-13:

156;

5.9.5-6: 352 Cicero Ae. 2.118: 348

Aft. 1.17.5: 231; 1.19: 259; 3.7.1:

160; 5.12.1: 149 and n. 8 Div. 2.115: 233 Dam. 115: 358 Off. 3.97: 348 N.D. 1.27-28: 154; 1.27-28: 159 Phil. 2.106: 352 Sui. 33: 236 Tuse. 1.93: 316; 2.61: 349; 5.103:

Sallustius Cat. 41.1-2: 349; 52.1: 223, 334, 350; 52.33: 165, n. 36 Jug. 95-96: 83, 181

Columella 8.1.2-4: 262

3.11.27:

264;

Livius 1.7.2: 133, 223; 3.17.3: 344; 24.37.10-11: 343, 347; 34.31.6: 344; 41.4.3-4: 353 Petronius 38.4: 134 n. 8, 155, 163; 54.1-3: 157; 64.5-6: 251 Plautus Am. 536-539: 243; 553-555: 166 As. 712-716: 238, 241

Bae. 1059-1062: 221

Plinius Maior Nat. 28.30-31: 153, 159 Plinius Minor Ep. 7.32.1: 317

358 Ver. 4.126: 153, 160; 3.8: 347

Curtius Rufius 3.11.24-25: 353; 10.1.37: 317

Capt. 602-607: 151, 168 n. 43; 746747: 358 Cas. 763-777: 255; 887-889: 204; 982: 146, 150 Ep. 25: 229, 238, 245 Mer. 248-251: 201 Mil. 95-97: 153, 159; 1225-1226: 93, 151, 168 n. 43 Paen. 1056-1057: 237 Ps. 185-227: 234, 247 Rud. 168-170: 354, 359; 686: 166; 710-711: 360 Trin. 23-26: 137, 148; 59-61: 173 n. 2,209 True. 615: 344; 838: 234, 236

Seneca Ep. 11.7: 138, 185 Suetonius Aug. 26.4: 317 Terentius Ad. 190: 150 n. 11; 193: 146, 150 Eu. 474-475: 242; 668: 360 Hee. 526-527: 359 Vergilius A. 4.1-2: 353 n. 24; 6.315-318: 201; 8.81-85: 183, 204 G. 3.69-71: 191,204

INDEX OF LATIN WORDS

an at autem

308 ch. 12 and passim ch. 10 and passim

quaeso quasi quia

certe ceterum cum

186, 365 83,88 72-73

quidem quod

demum

71

quoniam

ecastor ecce edepoI enim ergo

sane sed

92,352,365 69-71, 86, 88, 216217 and n. 12, 218, 246, 279 and n. 67, 315, 329, 368

etiam

187, 365 266, 273 and n. 60 187, 365 ch. 8 and passim 92-94, 99, 119, 122, 369-370, 373 63, 69-71, 73, 106107 71 and n. 19, 95, 366

tandem tum

95 117, 322-323

fortasse

187

utique

71

hercIe

187, 365

iam igitur

ch. 11 and passim 216, 315, 329 186 and n. 16

immo

322 63 and n. 9, 86, 88, 100-101, 114, 117, 119, 121-122, 167, 181, 274 and n. 63, 278, 373 92,295,369

vero verum videlicet

modo

71 and n. 19, 95

nam nimirum

ch. 7 and passim 186 and n. 16

obsecro

93

profecto

186

et

93 117, 288-289 16, 19, 69, 79, 105106, 106 n. 16, 131132, 176 120, 235, 338, 365 16, 79, 106 n. 16, 131-132 19, 105-106, 106 n. 17, 131-132

DISCOURSE PARTIKELS IN HET LATIJN. DE FUNCTIES VAN NAM, ENIM, AUTEM, VERO EN AT

Samenvatting

Dit proefschrift handelt over een kategorie van discourse partikels in het Latijn die in de handboeken worden samengenomen onder de noemer 'coordinerende conjuncties'. Van deze groep van partikels worden nam, enim, autem, vero en at in detail besproken. De functie van bovengenoemde discourse partikels wordt traditioneel beschreven in termen van het uitdrukken van semantische relaties tussen opeenvolgende hoofdzinnen: nam en enim worden beschouwd als min of meer synonieme markeerders van causale relaties tussen opeenvolgende zinnen, autem, vero en at als onderling concurrerende middelen voor de uitdrukking van adversatieve relaties. Deze weergave is een oversimplificatie van de werkelijke situatie en roept op een aantal punten problemen op. Zo komen bij ieder van de genoemde partikels gebruikswijzen voor die zich niet, of slechts met zeer grote moeite, laten onderbrengen in een beschrijving in termen van causale of adversatieve zinsverbinding: in de bedoelde gevallen is bijvoorbeeld geen sprake van 'causaliteit' of 'adversativiteit' (in de ruimste zin van het woord), of de verbonden elementen hebben niet de vorm van grammaticale zinnen. Gebruikelijke oplossingen voor dit soort problemen zijn het aannemen van homonymie, van (vaak nogal onwaarschijnlijke) diachronische ontwikkelingen, of van een extreme vorm van polysemie. Een verder probleem waaraan nauwelijks aandacht is besteed, is het verschil in 'distributionele eigenschappen' tussen vermeende synoniemen: de lingulstische context waarin bijvoorbeeld nam wordt aangetroffen in het Latijn blijkt aanzienlijk te verschillen van die van enim en ook tussen de gebruikscondities van autem, vero en at bestaan grote verschillen. Een strikt semantisch-syntactische benadering kan geen goede verkiaring geven voor bovengenoemde problemen. In deze studie wordt gebruik gemaakt van moderne inzichten afkomstig uit met name de pragmatiek en discourse analyse om tot een meer inzichtelijke beschrijving en classificatie van Latijnse 'coordinerende conjuncties' te komen. Hiertoe wordt een analy-

400

samenvatting

tisch raamwerk ontwikkeld aan de hand waarvan deze, en vergelijkbare partikels in andere talen, in kaart kunnen worden gebracht. Het eerste, theoretische deel van het boek (hoofdstuk 1-5) is geheel gewijd aan de stapsgewijze ontwikkeling van dit raamwerk. Na een korte inleiding (hoofdstuk 1) en een systematisch overzicht van de stand van zaken in onderzoek naar tekstrelaties (hoofdstuk 2) en in partikelonderzoek (hoofdstuk 3) worden in hoofdstuk 4 de contouren geschetst van het eigenlijke beschrijvingsmodel, dat zowel op theoretische overwegingen als op empirisch onderzoek is gebaseerd. Centrale uitgangspunten van dit model zijn twee recente inzichten uit de op discourse toegespitste taalwetenschap: (i) connectieve partikels (waaronder de zogenaamde 'coordinerende conjuncties') zijn op een of andere manier betrokken bij het signaleren van de coherentie van een bepaalde discourse; (ii) discourse-coherentie vindt plaats langs verschillende dimensies of op verschillende niveaus. In het hier voorgestelde model wordt er vanuit gegaan dat er drie relevante niveaus van discourse-coherentie zijn. Deze niveaus worden het representationele, presentationele en interactionele discourse-niveau genoemd. Het representationele discourse-niveau betreft de representatie van een bepaalde wereld buiten de taal zelf. Coherentie-relaties op dit niveau worden met name gevormd door semantische relaties tussen de gerepresenteerde standen van zaken in een bepaalde wereld. Het presentationele discourseniveau heeft betrekking op de manier waarop de taalgebruiker binnen een bepaalde monologische teksteenheid de gerepresenteerde inhoud organiseert en presenteert. Coherentie-relaties op dit niveau zijn functioneel van aard en betreffen met name de thematische structuur van de tekst en de zogenaamde retorische relaties tussen communicatieve handelingen en clusters van handelingen. Coherentie op het interactionele discourse niveau, tenslotte, heeft te maken met het inpassen van taaluitingen in een interactionele context. Relaties op dit discourse-niveau betreffen bijvoorbeeld de onderlinge relatie tussen met elkaar corresponderende communicatieve 'zetten' van verschillende gesprekspartners binnen een communicatieve 'uitwisseling'. Op aIle drie niveaus kunnen coherentie-relaties expliciet gemarkeerd worden, bijvoorbeeld door bepaalde soorten van partikels. Omgekeerd verschillen dergelijke partikels onderling onder meer wat betreft het discourseniveau waarop ze primair opereren. Naast discourse-niveau is voor de beschrijving en classificatie van coherentie-markerende partikels verder nog van belang of deze de relatie tussen twee expliciet uitgedrukte teksteenheden

samenvatting

401

signaleren (connectieve partikels in strikte zin), of een teksteenheid plaatsen binnen zijn ruimere, niet-expliciet uitgedrukte context, bijvoorbeeld binnen de niet-talige communicatieve context waarin de tekst is ingebed (connectieve partikels in ruime zin). In hoofdstuk 5 worden de theoretische noties van het in hoofdstuk 4 gepresenteerde model geoperationaliseerd voor de feitelijke beschrijving van Latijnse partikels in het tweede deel van het boek. In hoofdstuk 5 wordt allereerst ingegaan op de verschillende componenten waaruit een volledige 'definitie' van een discourse partikel zou moeten bestaan en op de vraag wat de relatie tussen deze componenten is. Vervolgens wordt aandacht besteed aan de lingulstische relevantie van (en evidentie voor) het onderscheiden van de drie voorgestelde discourse-niveaus. Dit vormt tevens de aanleiding voor een uitweiding over het belangrijke begrip discourse type en, in samenhang hiermee, het verschijnsel diafonie (dat wil zeggen, dialogische verschijnselen die optreden binnen een monologische context). In het laatste deel van hoofdstuk 5 wordt een opsomming gegeven van (met name lingulstische) criteria die gebruikt kunnen worden bij het vaststellen van de functie van een bepaald partikel. Het tweede deel van het boek bevat systematische, synchronische beschrijvingen van de Latijnse partikels nam, enim, autem, vero en at, in overeenstemming met de theoretische lijnen die in het eerste deel zijn uitgezet. De hoofdstukken over nam (hoofdstuk 7) en enim (hoofdstuk 8) worden voorafgegaan door een algemene inleiding op het omstreden begrip 'causale relatie' en de manier waarop dit begrip gebruikt is in eerdere beschrijvingen van deze partikels. Op dezelfde manier worden de hoofdstukken over autem (hoofdstuk 10), vero (hoofdstuk 11) en at (hoofdstuk 12) voorafgegaan door een bespreking van het begrip 'adversatieve relatie'. De besprekingen van de afzonderlijke partikels worden besloten met een samenvattende analyse van het betreffende partikel en een contrastieve vergelijking met (echte of vermeende) synoniemen. Toepassing van het theoretisch model op deze Latijnse partikels levert uiteindelijk een onderverdeling op in drie functie-groepen (hoofdstuk 13): (i) een groep met onder andere nam en autem, die beide een primaire functie vervullen op het presentationele discourse-niveau en beschouwd kunnen worden als connectieve partikels in strikte zin; (ii) een groep met onder andere at, waarbij het gaat om strikt connectieve partikels met een primaire functie op het interactionele discourseniveau;

402

samenvatting

een groep waartoe onder andere enim en vero behoren, die eveneens beide primair een rol vervullen op het interactionele discourse-niveau, maar geen connectieve partikels in strikte zin zijn. Deze onderverdeling in functie-groepen wijkt aanzienlijk af van de traditionele classificatie in semantisch-syntactische kategorieen. De gekozen discourse-pragmatische benadering blijkt een aantal duidelijke voordelen te hebben ten opzichte van de gebruikelijke semantisch-syntactische benadering. Allereerst leidt ze tot meer gei"ntegreerde en homogene beschrijvingen van afzonderlijke partikels, waarin de relaties tussen de verschillende aangetroffen gebruikswijzen expliciet worden gemaakt en ogenschijnlijk afwijkende gebruikswijzen kunnen worden verklaard. De functionele beschrijvingen geven bovendien een beter inzicht in de effecten die de partikels in een bepaalde context kunnen hebben, waardoor het duidelijk wordt dat Latijnse partikels niet zomaar beschouwd kunnen worden als betekenisloze elementen die naar willekeur over een tekst worden verspreid - een visie die nogal eens wordt aangetroffen in commentaren op Latijnse teksten. Verder blijkt door de gekozen benadering een beter onderscheid gemaakt te kunnen worden tussen vermeende synonieme of concurrerende uitdrukkingsmogelijkheden. Naast de bijdrage die deze studie levertaan de beschrijving van het Latijn, levert zij ook inzichten op die bruikbaar zijn bij de beschrijving van vergelijkbare verschijnselen in andere talen. Omdat bij de argumentatie zoveel mogelijk gebruik wordt gemaakt van objectieve lingui"stische criteria, draagt deze studie bij aan de ontwikkeling van het analytisch instrumentarium van partikelonderzoek in het algemeen. Tevens kan zij gezien worden als een empirisch gefundeerde stap in de richting van meer algemene modellen van discourse coherentie. (iii)