Information Structure and its Interfaces 9783110213973, 9783110208771

The volume presents recent results in the field of Information Structure based on research on Italian and Italian dialec

231 110 5MB

English Pages 435 [440] Year 2009

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Frontmatter
Contents
Acknowledgements
Introduction
Topic accent and prosodic structure
The prosodic annotation of C-ORAL-ROM and the structure of information in spoken language
Universals of information structure
Constructions with preposed infinitive: Typological and pragmatic notes
Negation and «Focus Clash» in Sardinian
Do-support in a Sicilian variety, an Italian pseudo-cleft, and the packaging of information
Clitics and anaphoric relations in informational patterning: A corpus-driven research on spontaneous spoken Italian (C-ORAL-ROM)
The informational structure and the scope of lexical modality in spoken Italian (C-ORAL-ROM)
On two functions of verb-subject order in Old Italian
Between thematicity and grammaticalisation: The diachronic rearrangement of information structure and the position of clitic pronouns in Italian
Information structure in Slavic languages
Position, function and interpretation of topics in Somali
Acquiring the grammar of topicality in L2 Italian: A comparative approach
Appendix” or “postposed Topic”: Where does the difference lie?
Backmatter
Recommend Papers

Information Structure and its Interfaces
 9783110213973, 9783110208771

  • 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

Information Structure and its Interfaces



Interface Explorations 19

Editors

Artemis Alexiadou T. Alan Hall

Mouton de Gruyter Berlin · New York

Information Structure and its Interfaces edited by

Lunella Mereu

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Information structure and its interfaces / edited by Lunella Mereu. p. cm. ⫺ (Interface explorations ; 19) Includes bibliographical references and index. ISBN 978-3-11-020877-1 (cloth : alk. paper) 1. Grammar, Comparative and general ⫺ Topic and comment. 2. Grammar, Comparative and general ⫺ Syntax. 3. Italian language ⫺ Topic and comment. 4. Italian language ⫺ Dialects ⫺ Topic and comment. I. Mereu, Lunella. P298.I545 2008 415⫺dc22 2008048798

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-020877-1 ISSN 1861-4167 ” Copyright 2009 by Walter de Gruyter GmbH & Co. KG, 10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Christopher Schneider, Laufen. Printed in Germany.

Contents

Aknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Section 1 Interfaces of information structure Topic accent and prosodic structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Claudia Crocco The prosodic annotation of C-ORAL-ROM and the structure of information in spoken language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Antonietta Scarano Universals of information structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Lunella Mereu Section 2 Grammar and information structure in Italian and some of its dialects Constructions with preposed infinitive: Typological and pragmatic notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Giuliano Bernini Negation and “Focus Clash” in Sardinian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Franck Floricic Do-support in a Sicilian variety, an Italian pseudo-cleft, and the packaging of information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Ignazio Mauro Mirto Clitics and anaphoric relations in informational patterning: A corpusdriven research on spontaneous spoken Italian (C-ORAL-ROM) . . . . . . . 169 Emanuela Cresti The informational structure and the scope of lexical modality in spoken Italian (C-ORAL-ROM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 Ida Tucci

vi

Contents

Section 3 Diachronic aspects in Italian On two functions of verb-subject order in Old Italian . . . . . . . . . . . . . . . . . . 229 Ludwig Fesenmeier Between thematicity and grammaticalisation: The diachronic rearrangement of information structure and the position of clitic pronouns in Italian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Shingo Suzuki Section 4 Pragmatically oriented languages Information structure in Slavic languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 Lucyna Gebert Position, function and interpretation of topics in Somali . . . . . . . . . . . . . . . 325 Mara Frascarelli and Annarita Puglielli Section 5 Some more aspects of information structure Acquiring the grammar of topicality in L2 Italian: A comparative approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 Marina Chini “Appendix” or “postposed Topic”: Where does the difference lie? . . . . . 387 Edoardo Lombardi Vallauri

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413 Index of subjects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415 Index of languages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421 Index of persons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423

Acknowledgements

This volume originates from the “Giornate su La struttura dell’informazione” which were held at the Università Roma Tre in Rome in the period 23–24 October 2006 as a joint project between Lunella Mereu (Università Roma Tre, Rome) and Giorgio Banti (Università di Napoli «L’Orientale», Naples). The Department of Linguistics of the Università Roma Tre (Raffaele Simone, head of Department) and the Department of the Studi e Ricerche su Africa e Paesi Arabi of the Università di Napoli «L’Orientale» (Giorgio Banti, head of Department) provided financial support for the Giornate which were attended by linguists from all over Italy, Europe as well as Japan. I wish to express my gratitude to my colleague and friend Giorgio Banti for having shared with me the organization of the Giornate and for his suggestions in the very first steps of the preparation of the present volume. A word of thanks to all the speakers who greatly helped to make the event a success and to the numerous participants whose contributions kept the discussion lively and interesting. Thanks are also due to Michele Abrusci, Dean of my Faculty at Roma Tre, for introducing the Giornate. A thank you for their help during the various steps of the preparation of the Giornate also goes to our friends on the administrative and technical staff, in particular to Giuseppina Vecchioni (Secretary of the Department of Linguistics at Roma Tre) whose superior organizational skills helped to make the Giornate a success, together with Paola Sentinelli, Tiziana Pierdomenici, Silvia Avella and Marco Pagliai. A very special thanks to Raffaele Simone for his useful suggestions and his constant friendly encouragement during the various stages of the preparation of the event and of this volume. Finally I wish to express my gratitude to Ursula Kleinhenz for her constant help and suggestions during the preparation of the manuscript for publication and to Frank Benno Junghanns for his competent help in formatting the papers for the present volume.

Rome, October 31, 2008

Lunella Mereu

Introduction

This volume features a selection of papers presented at the Conference “Giornate su La Struttura dell’Informazione” held at the Università Roma Tre, Rome, 23–24 June 2006. The conference, organized by Giorgio Banti (Dipartimento di Studi e Ricerche su Africa e e Paesi Arabi of the Università di Napoli “L’Orientale”) and by Lunella Mereu (Dipartimento di Linguistica dell’Università Roma Tre), aimed at discussing recent results in the field of Information Structure as emerging from research on Italian and Italian dialects, and from Italian studies on typologically different languages including Italian. Ever since a pragmatic organization of the sentence has been recognized, this has been one of the main topics of study in Linguistics. The number of studies produced in almost a century of work is huge and rich both in terms of theory and descriptions of information structure systems represented by different languages. This has of course also created some confusion as to the choice of the theories and the terminology used, making it often difficult to draw a common denominator from all the material produced. For this reason we will not discuss the state of the art reached up to now, as it would take a good part of the publication.1 However, some background about the foundational work done from the 1930s onwards will be useful to characterize the theoretical and methodological choices included in the present volume. It goes without saying that the systematic study of the pragmatic organization of discourse at the sentence level started with the work by the Prague School and its theory on Functional Sentence Perspective (Mathesius 1929; Firbas 1964)2 and continued with the studies by Halliday (1967) to whom we owe the term ‘information structure’. Next, Chafe (1976) made a number of semantic distinctions and introduced the term ‘information packaging’ which was also adopted by Vallduvì (1992) in a formal context.3 What soon became clear in these studies is that intonation, morphosyntax and discourse interact in determining the identification of the pragmatic notions of information structure. Empirical evidence in this direction has come from the important work by typologists who, in exploring the morpho-syntactic structure of the languages of the world, have identified a series of languages that grammaticalize sentence-related discourse functions such as Topic and Focus. We are referring to Li and Thompson’s distinction

2 Introduction (1976, 1981) between subject-oriented and topic-oriented languages in their studies on Mandarin Chinese, or Givón’s concept of word order variation as “controlled by discourse-pragmatic considerations pertaining to new vs. old, topical vs. non-topical, discontinuous vs. disruptive information” (Givón 1984: 204). This has led all linguistic approaches, both formal and non-formal, to overlap a topic – comment/focus structure with the NP – VP syntactic structure. In addition, in the 1980s, this has created much interest in focus structure under the assumption that there are many focus-prominent languages (Abraham and De Meij 1986). Consequently there were a number of descriptions of languages in terms of the single different focus structures they realize (Puglielli 1981; Horvath 1986; Kiss 1987; Lecarme 1991) and theoretical works, primarily but not exclusively formal, on focus (Dik et al. 1981; Rooth 1985; Sasse 1987; Rochemont 1986; Brody 1990; Horvath 1995). After concentrating on focus in the 1980s, interest shifted to topic structure as well (Reinhart 1982; Lambrecht 1994; Kiss 1995; Dik 1997; Büring 1997, 1999) and, finally, complete descriptions of the topic – focus architecture in different languages (Aissen 1992; Payne 1987; King 1995; Choi 1999; Lombardi Vallauri 1996, 2001; Frascarelli 2000; Mereu 2004; Bernini and Schwartz 2006) appeared. Important results have come from all these works. Let us summarize the most significant ones. First, information structure is not an exclusive matter of syntax as generative grammar seems to imply with the architecture of the left periphery in the recent ‘cartographic’ literature (Rizzi 1997, 2004). Rather it is an interface issue including “the interaction between the syntactic and phonological component … and the syntactic and semantic component” although “the question as to the actual division of labor between the components and their interaction is still controversial” (Molnàr and Winkler 2006: 1–2). Secondly, topic and focus are not “complementary” or “mutually exclusive” (Molnàr and Winkler 2006: 5; Erteschik-Shir 2007: 27 f.), as most literature in the 1980s seemed to claim. As a matter of fact, languages such as Korean (Choi 1997, 1999) or Wambon (Dik 1997) use the same morphological means to express topic and contrastive focus. There are also languages which use the same phonological means, that is pitch accent, to mark topics and contrastive focus (Mereu and Trecci 2004; Mereu and Frascarelli 2006; Molnàr 2006). Although the importance of prosody has been associated with the study of information structure since the beginning, only recently have we started

Introduction

3

to understand that this means finding ways to correctly measure prosodic data investigating the intonational contours of utterances. This leads us to focus on two more issues which have recently emerged: I) information structure cannot be studied in isolated sentences, and II) we need samples of spoken language to extract chunks to be analysed phonologically, syntactically and pragmatically. Both of these issues were not clear in most works in the past century, at least up until the 1980s, as these were based more on the analysis of single sentences, or even questionanswer pairs, rather than on a systematic study of actual language use. And of course the work in formal syntax, based on the grammaticality judgements of single ‘laboratory’ sentences, did not encourage a change of perspective4. Chafe (1980) with his ‘pear stories’ was among the first to encourage typologists to have informants tell stories in order to draw data out of the discourse. To analyze data of this kind is important because they are contextualized. Context allows linguistic behaviour which can be considered ‘out’ on the basis of grammaticality judgements, but which is acceptable for the speakers or writers of utterances in discourse as this responds to communicative needs and efficiency. Therefore, it is no longer possible to trust the speaker’s intuitions about single isolated sentences; instead we need to use data from spontaneous or semi-spontaneous production of any kind, either conversations or monologues, formal or informal, written or spoken, even if our aim is to study the information structure of single utterances. One final point we would like to note is the importance of cross-linguistic studies, as it is only through the comparison of different systems that we can detect the actual range of phenomena connected with information structure and hope for generalizations in terms of universals. This means that we have both to provide deep analyses of information structure coding in single languages and to look carefully at variation registering the different ways to express topic and focus in languages. We also have to abandon “… the accepted principle that topics tend to precede foci” (Erteschik-Shir 2007: 214) in the unmarked information structures of all languages, as different, even opposite, pragmatic strategies might be at work in single languages. This has often been observed in typological studies and it is a point that needs to be reinforced. We have already hinted at this when we mentioned the distinction by Li and Thompson (1976) between subject-oriented and topicoriented languages; more in general we might speak of syntactically-based and pragmatically-based languages as does Mithun (1987) and of different pragmatic principles operating at the sentence level in different languages.

4 Introduction After this short overview of the current state of information structure, we can now address the papers collected in the present volume. The papers are primarily based on actual speech or written documents. Most of them concentrate on Italian or Italian dialects, mainly, but not exclusively, from a synchronic perspective. However, a cross-linguistic perspective is also taken into account, as there are also papers on a few pragmatically-oriented languages. Some papers are specifically oriented towards aspects of information structure interfaces, while others examine the relationship between syntactic and semantic aspects of Italian grammar and information structure under the hypothesis that the different coding of the individual phenomena investigated is conditioned by information structure. The papers are grouped into five sections: I. Interfaces of information structure; II. On the interaction between grammar and information structure in Italian and in some of its dialects; III. Diachronic aspects of information structure in Italian; IV. Information structure in pragmatically-oriented languages; V. Some more aspects of information structure. The first section is primarily concerned with the interface between information structure and prosody but also the search for universals of information structure in terms of its interfaces. The first paper by Crocco deals with topic accent and its realization in Neapolitan Italian. After providing a summary of both theoretical and empirical results on topic accents in Italian and English, the paper concentrates on data from Neapolitan showing how variation in the prosodic topic marking depends on the syntactic position of this discourse function. While Crocco’s paper is based on a hypothesis of the interaction between prosody, syntax and pragmatics, Scarano’s is meant to demonstrate that syntax plays no role in the segmentation of the utterances from a pragmatic point of view. An isomorphism between the information structure of utterances and intonation is the premise on which data on Italian, drawn from the C-ORAL-ROM corpus (Cresti and Moneglia 2005), are analysed. However, rather than describing tonal events, the paper is based on the parsing of speech in prosodic units and aims to show that it is possible to identify utterances from a semantic and pragmatic point of view on the basis of their prosodic structure. Finally Mereu’s paper analyses morpho-syntactic variation in the languages of the world in terms of two opposite pragmatic principles and shows prosodic and morpho-syntactic evidence in favour of a distinction between syntactically-oriented and pragmatically-oriented languages. The second section concentrates on the interaction between grammar and discourse and specifically shows how grammatical behaviour, syntactic or semantic, is conditioned by the communicative efficiency or the discourse

Introduction

5

patterning of utterances. The paper by Bernini is concerned with constructions with an initial infinitive followed by an inflected form of the same verb. After presenting a typological survey of these constructions in many languages, Bernini shows that the split between the lexical-semantic component, the infinitive predicate, and the grammatical component, the inflected finite form of the verb, encodes the topic – comment structure. He argues that the construction is used to focus semantic functions such as assertion and epistemic modality as they are expressed by inflection. The next two papers are about two dialects of Italian, Sardinian and Sicilian. The paper by Franck is about the Logudorese variety of Sardinian, and it deals with the incompatibility between negation and the interrogative marker a, and between this marker and fronting in yes/no questions. The reason for this conflict is given by the informational and semantic properties of negation and total questions: the two modalities clash because of the focalization value of both. Cross-linguistic data showing the incompatibility of both modalities in similar constructions in other languages is provided as further evidence that confirms their pragmatic nature. Mirto’s paper examines a construction with the verb fari ‘do’ which is used in an area of Sicily. Formally the construction includes a do-verb and an uninflected verbal form that is an infinitive, and is compared to a pseudo-cleft with a do-verb in standard Italian. In this construction the information contained in the lexical verb is broken down into two distinct components, the existence of an unspecified event and its denomination, the former corresponding to the utterance theme while the latter to the rheme or the unit with higher prominence. The two final papers in this section, like Scarano’s in the preceding one, share data drawn from the Italian C-ORAL-ROM corpus and are based on the same theory, known as Language in Act Theory and Informational Patterning Hypothesis (Cresti 2000). Cresti’s paper deals with anaphoric clitics, their distribution in different information units to be distinguished and the constraints they show. After summarizing the theory and showing the information units the utterance may contain, Cresti describes the clitics and their head of reference in relation to the different information units in which they may occur. In this way she demonstrates how the relation between clitics and their antecedent is constrained by illocutionary and informational factors. Tucci’s paper, on the other hand, is concerned with the relation between modality and the information structure of spoken language. From the corpus analysis it emerges that lexical modality cannot be only a semantic matter, as only specific types of information units can bear modal lexical indexes and specific modal values are associated with different types of information unit.

6 Introduction The third section is devoted to analysing the articulation of information structure in old Italian texts and therefore it is exclusively concerned with written documents from the 13th and 14th centuries Tuscan vernacular as in both Fesenmeier’s and Suzuki’s paper, although in the latter they are compared to modern Italian. Fesenmeier deals with the occurrence of VS order in Old Italian showing how postverbal definite subjects can receive different pragmatic interpretations. While a definite preverbal subject is by default a topic, a postverbal subject can be interpreted as a topic or not in relation to a number of parameters, in particular the state of affairs the utterance expresses and its relation to the context, textual structure or extralinguistic reality. Interestingly, the author shows that even in old texts the preverbal subject is a stressed topic or a cognitively salient topic, while the postverbal one, if a topic, is unstressed or unemphasized. While Fesenmeier is exclusively interested in written language use, Suzuki uses these documents to draw data on spoken language too, being interested in clitics, their position before or after the verb, and their function. He observes clitic behaviour in the diachronic evolution of Italian in relation to the development of the SVO structure from the old XVS. He concludes that I. the process towards the pre-verbal position of the thematic S and the process of the disappearance of the enclisis in declarative clauses occurred over a long period of time, II. both processes influenced each other, and III. gradually the clitic has undergone a grammaticalisation process determining its final grammatical agreement function. The fourth section contains two papers on typologically different but pragmatically-oriented languages. Gebert’s paper provides an interesting contrast between two groups of Slavic languages in their realization of information structure contents. In particular, she focuses on some morphosyntactic properties associated with marked word orders, ascribing the differences to the fact that Russian and Polish express Case morphologically, while Bulgarian and Macedonian have lost this type of noun inflection. Frascarelli and Puglielli deal with the formal and pragmatic properties of topics in Somali from a generative perspective. They analyse the intonational, syntactic and discourse realization of topics in a cartographic approach drawing from Rizzi’s (2004) recursive structure of the left periphery, a recursive array of CPs each with distinct functional projections such as Topic or Focus Phrase. They show that topics in Somali are associated with different projections of Topic Phrase in distinct positions in the left periphery, each with its own formal and discourse properties. The last section contains two papers dealing with two issues of a different nature, but both relevant for information structure. The first, by Chini,

Introduction

7

is concerned with acquisitional data, specifically with topic treatment in narratives by Italian L2 learners and by native speakers of Italian of the same age and culture. Interestingly, it turns out that information structure is language dependent as it implies mastering the morphosyntactic, phonological and discourse strategies of a language. L2 learners behaviour shows that learners tend to rely on L1 grammatical patterns and textual preferences even at an advanced level of L2 acquisition. Lombardi Vallauri’s paper deals with a specific category, which in various ways is proposed by supporters of a tripartite organization of information structure. He concentrates on models of this kind present in Italian literature, specifically Cresti’s (2000, this volume) model of information structure, in which a distinction is made between the preposed and postposed topic and the latter is called Appendix. The author shows, also on the basis of cross-linguistic evidence, that topics and Appendices are equivalent in terms of information value and that their difference is only a syntactic matter. Lunella Mereu

Notes 1. There are a number of works recently published which provide a survey of the main schools on information structure; see, amongst the most recent ones, Steube (2004) for the Prague school, Erteschik-Shir (2007) for Generative Grammar, Gómez-Gonzalez (2000) and Molnár and Winkler (2006) for reference to various approaches. 2. It has to be pointed out that we will simply mention some of the relevant bibliographical material on information structure without aiming at being exhaustive. 3. Vallduvì will propose a tripartite structure for information structure on the basis of data on right dislocations in Catatalan; see Cresti (2000), Lombardi Vallauri (this volume) and Mereu (this volume) for different interpretations of right dislocations or topics in Italian, and consequent proposals of bipartite or tripartite structures. 4. We must mention here as an exception the work by Jackendoff who in the early 1970s already spoke about A and B accents to differentiate the marking of the different pragmatic functions in sentence grammar. This indicates an early interest of his towards the data about spoken language. See Crocco (this volume) for an insight into Jackendoff’s (1972) phonological data about information structure.

8 Introduction References Abraham, Werner and Sjaak De Meij (eds.) 1986 Topic, Focus and Configurationality. Papers from the 6th Groningen Grammar Talks. Groningen 1984. Amsterdam/Philadelphia: John Benjamins. Aissen, Judith 1992 Topic and Focus in Mayan. Language 68: 43–80. Bernini, Giuliano and Marcia L. Schwarts (eds.) 2006 Pragmatic Organization of discourse in the Languages of Europe (Eurotyp 20-8). Berlin /New York: Mouton de Gruyter. Brody, Michael 1990 Remarks on the Order of Elements in Hungarian Focus Field. In Approaches to Hungarian. Structure and Arguments, Istvan Kenesei (ed.), 95–121. Jate: Szeged. Büring, Daniel 1997 The Meaning of Topic and Focus. The 59th Street Bridge Accent. London: Routledge. 1999 Topic. In Focus: Linguistic, Cognitive and Computational Perspectives, P. Bosch and R. van der Sandt (eds.), 142–165. Cambridge: Cambridge University Press. Chafe, Wallace 1976 Givenness, Contrastiveness, Definiteness, Subjects, Topics and Point of View. In Charles N. Li (ed.), 25–56. Chafe, Wallace (ed.) 1980 The Pear Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Production. Norwood: Ablex. Choi, Hye-Won 1997 Information Structure, Phrase Structure, and their Interface. In Proceedings of the LFG 97 Conference (on line), Miriam Butt and Tracy Halloway King (eds.). Stanford: CSLI Publications. 1999 Optimizing Structure in Context: Scrambling and Information Structure. Stanford: CSLI Publications. Cresti, Emanuela 2000 Corpus di italiano parlato. Vol. I /II, CD-ROM. Firenze: Accademia della Crusca. Cresti, Emanuela and Massimo Moneglia 2005 C-ORAL-ROM. Reference Corpora for Spoken Romance Languages. Amsterdam/Philadelphia: John Benjamins. Dik, Simon 1997 The Theory of Functional Grammar, Part 1: The Structure of the Clause. Berlin /New York: Mouton de Gruyter.

Introduction

9

Dik, Simon, Maria E. Hoffman, Jan R. de Jong, Sie Ing Djiang, Harry Stroomer, and Lourens de Vries 1981 On the Typology of Focus Phenomena. In Perspectives on Functional Grammar, T. Hoekstra, H. Van der Hulst and M. Mortgat (eds.), 41– 74, Dordrecht: Foris. Erteschik-Shir, Nomi 2007 Information Structure.The Syntax-Discourse Interface. Oxford: Oxford University Press. Firbas, Jan 1964 On Defining the Theme in Functional Sentence Perspective. Traveaux Linguistiques de Prague 1: 267–289. Paris: Klincksieck. Frascarelli, Mara 2000 The Syntax-Phonology Interface in Focus and Topic Constructions in Italian. Dordrecht: Kluwer. Givón, Talmy 1984 Syntax: a Functional-typological Introduction. Vol. I, Amsterdam / Philadelphia: John Benjamins. Gómez-Gonzalez, María A. 1985 [2000] The Theme-Topic Interface. Evidence from English. Amsterdam /Philadelphia: John Benjamins. Halliday, Michael A. K. 1967 Notes on Transitivity and Theme in English. Journal of Linguistics 3 (part 1): 37–81, (part 2): 199–244. Horvarth, Julia 1986 FOCUS in the Theory of Grammar and the Syntax of Hungarian, Dordrecht: Foris. 1995 Structural Focus, Structural Case and the Notion of Feature Assignment. In Katalin É. Kiss (ed.), 176–206. Jackendoff, Ray 1972 Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. King, Tracy H. 1995 Configuring Topic and Focus in Russian. Stanford: CSLI Publications. Kiss, Katalin É. 1987 Configurationality in Hungarian. Dordrecht: Springer. Kiss, Katalin É. (ed.) 1995 Discourse Configurational Languages. Oxford: Oxford University Press. Lambrecht, Knud 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press.

10 Introduction Lecarme, Jacqueline 1991 Focus en Somali: syntaxe et interprétation. Linguistique Africaine 7: 33–65. Li, Charles N (ed.) 1976 Subject and Topic. New York: Academic Press. Li, Charles N. and Sandra Thompson 1976 Subject and Topic: a New Tipology of Language. In Charles N. Li (ed.), 457–490. 1981 Mandarine Chinese: a Functional Reference Grammar. Berkeley: University of California. Lombardi Vallauri, Edoardo 1996 La sintassi dell’informazione. Roma: Bulzoni. 2001 La teoria come separatrice di fatti di livello diverso. L’esempio della struttura informativa dell’enunciato. In Atti del XXXIII Congresso Internazionale della Società di Linguistica Italiana. Dati empirici e teorie linguistiche, Napoli, 28–30 ottobre 1999, F. A. Leoni, E. S. Krosbakken, R. Sornicola and C. Stromboli (eds.), 151–173. Roma: Bulzoni. Mathesius, Vilem 1929 [1983] Functional Linguistics. In Praguiana, Josef Vachek (ed.), 121–142, Amsterdam /Philadelphia: John Benjamins. Mereu, Lunella 2004 La sintassi delle lingue del mondo. Roma: Laterza. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. In Il Parlato Italiano. Atti del Convegno Nazionale (13–15 febbraio 2003), Federico Albano Leoni, Franco Cutugno, Massimo Pettorino, and Renata Savy (eds.), Napoli: M. D’Auria Editore – CIRASS (CD-ROM). Mereu, Lunella and Mara Frascarelli 2006 L’interfaccia sintassi-fonologia. Interpretazione e implicazioni teoriche. In “Analisi prosodica”. Teorie, modelli e sistemi di annotazione, Atti del 2° Convegno Nazionale AIS (Associazione italiana Scienze della Voce) Università di Salerno, 30 novembre – 2 dicembre 2005, Renata Savy and Claudia Crocco (eds.), Università di Salerno, CDROM. Mithun, Marianne 1987 Is Basic Word Order Universal? In Coherence and Grounding in Discourse, Russell S. Tomlin (ed.), 281–328. Amsterdam/Philadelphia: John Benjamins. Molnár, Valéria 2006 On Different Kinds of Contrast. In Valéria Molnár and Susanne Winkler (eds.), 197–234. Molnár, Valéria and Susanne Winkler (eds.) 2006 The Architecture of Focus. Berlin /New York: Mouton de Gruyter.

Introduction

11

Payne, Doris 1987 Information Structuring in Papago Narrative Discourse. Language 63: 783–804. Puglielli, Annarita 1981 Frase dichiarativa semplice. In Sintassi della lingua somala 2, Annarita Puglielli (ed.), 1– 44. Roma: Ministero degli Affari Esteri. Reinhart, Tanya 1982 Pragmatics and Linguistics: an Analysis of Sentence Topics. Philosophica 27 (1): 53–94. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. In Elements of Grammar. Handbook of Generative Syntax, Liliane Haegemann (ed.), 281–337. Dordrecht: Kluwer. Rizzi, Luigi (ed.) 2004 The cartography of Syntactic Structure. The Structure of CP and IP, Vol. 2. Oxford: Oxford University Press. Rochemont, Michael 1986 Focus in Generative Grammar. Amsterdam /Philadelphia: Benjamins. Rooth, Mats 1985 Association with Focus. Ph.D. diss., University of Massachusetts, Amherst. Sasse, Hans-Jürgen 1987 The Thetic / Cathegorical Distinvtion Revisited. Linguistics 25: 511– 80. Steube, Anita (ed.) 2004 Information Structure. Theoretical and Empirical Aspects. Berlin / New York: Walter de Gruyter. Vallduvì, Enric 1992 The Informational Component. New York: Garland.

Section 1 Interfaces of information structure

Topic accent and prosodic structure Claudia Crocco

1. Introduction Since the appearance of Jackendoff’s work in 1972, several studies have approached the problem of the prosodic correlates of topicality. The central concern of many of these investigations is the individuation of a topic accent or accents, i.e. of melodic prominences that can be associated with topics. This paper aims to contribute to the debate concerning the prominence of the topic by focussing on the general properties of prosodic structure. The perspective I will adopt combines the findings concerning topic accents with general research on prosody. Topic accents have been studied in several European languages, such as English, German, French, Greek and Italian among others. This paper deals with a regional variety of Italian, Neapolitan Italian, although a number of general considerations could be equally applied to other languages in principle. I will take the preceding findings on the prosody of topics and on the phonology of Neapolitan Italian as a starting point and I will argue that topic is not associated with a unique accent, but rather with a bundle of prosodic phenomena. Furthermore I will claim that prosodic marking of topic may be better understood in the light of the studies on prosodic structure, and that the actual realisation of the prosodic marking of topicality may vary according to the different position occupied by the topic with respect to the prosodic nucleus of the utterance. Finally, I will interpret the results concerning the prosody of topics from a functional perspective. I will claim that, on the prosodic level, the pragmatic function of topic corresponds to a form, and that this form has different realisations, according to the actual position of the topic in the utterance. The paper is composed of two main parts. The first (§§ 2, 3) presents the theoretical background of this research; the second (§§ 4, 5) analyses and discuss some data from Italian task-oriented speech. In the first sections a number of studies on topic accent in English and Italian are reviewed (§ 2). The following section (§ 3) sums up several studies providing the theoretical background for the analysis of Neapolitan Italian

16 Claudia Crocco intonation. Section § 4 is devoted to the analysis of the tonal contour of a number of utterances with and without sentence topics. Finally, in the last sections (§§ 4.4, 5) the results are summarised and discussed and the general conclusions are drawn.

2.

The prosody of topic

2.1. Terminological premise The following review sums up results obtained by scholars working in different frameworks and adopting unequal terminologies. Information structural facts have been analysed from several different perspectives (cf. Vallduví and Enghdal 1996 for an overview). Although in my exposition I will maintain the original terminology used by the different authors, I follow Gundel’s definitions of topic and comment (1988: 210): “Topic: An entity, E, is the topic of a sentence, S, iff, in using S, the speaker intends to increase the addressee’s knowledge about, request information about or otherwise get the addressee to act with respect to E.” “Comment: A predication, P, is the comment of a sentence, S, iff, in using S the speaker intends P to be assessed relative to the topic of S.”

Furthermore, in this paper I will refer to the main pitch accent of the utterance using the expressions prosodic focus or nucleus.

2.2. The prosody of topic. A short overview Apart from a number of studies explicitly and specifically devoted to the accentuation of topic (e.g. Büring 1997; Gundel, Hedberg and Zacharski 1997; Hedberg and Sosa 2007; Frascarelli and Hinterhölzl 2007; Mereu and Trecci 2004; Cresti and Firenzuoli 2002) most information on the prosody of topic can be found in studies concerning related issues, such as intonation meaning, information structure and prosody, accentuation of given and new referents1, word order, information structure and syntax. I will review a number of prosodic, syntactic and pragmatic studies which have contributed to the field.

Topic accent and prosodic structure

17

2.2.1. Semantics, phonology and phonetics of the “B Accent” A first reference to the existence of what may be called a “topic accent” can be found in Jackendoff 1972. Jackendoff bases his analysis on the distinction, made in Bolinger (1965), between two different types of “pitch accents”, namely accent “A” (fall) and “B” (fall-rise). Jackendoff borrows Bolinger’s terminology2 and applies the accentual distinction to the analysis of the intonation pattern of sentences articulated in “presupposition” and “focus”. He proposes relating the two accents to the independent (B accent) and to the dependent (A accent) variables composing the sentence, in other words to the units traditionally called “topic” and “comment”. Jackendoff (1972: 262) claims that “in many simple sentences, the stress and intonation contour divide the sentence exhaustively into two foci, the subject and the predicate, with a B pitch accent on the former and an A accent on the latter”. The semantic interpretation of the B accent originally proposed by Jackendoff (1972) has been strongly criticised by Ladd (1980) and Gussenhoven (1983) in their work on intonation and meaning. According to Ladd (1980: 161), the fall-rise tone conveys “a single abstract general meaning of ‘focus within a given set’”, i.e. of picking up an element from a set of other possibilities which is in the context (Ladd 1980: 153). Moreover, the B accent cannot be treated as a “variable”. As Ladd explains (1980: 158), “by treating fall-rise independently of fall […] we can readily describe sentences with only a fall-rise, and we are able to describe the discourse relationship it expresses far better than a system which treats both tones [i.e., A and B accent] as two variables of a single function”. The analysis of fall and fall-rise tones proposed by Jackendoff (1972) is also considered inadequate by Gussenhoven (1983). In his study on the semantics of nuclear tones in English, Gussenhoven (1983: 79) argues that although Jackendoff’s proposal “is the first attempt to assign meaning to nuclear tones that went beyond a vague association of falls with finality and rises with non-finality”, the analysis does not fit the data, because “the interpretations given [for the examples discussed in Jackendoff 1972] are by no means obligatory”. The author argues instead that the fall-rise expresses the selection of a variable from the background. In a paper on the meaning of the fall-rise tune in English, Hirschberg and Ward (1985) analyse the accent from a phonetic and phonological point of view. The analysis shows that the generic label of fall-rise covers more than one type of pitch accent. In particular, the authors claim that two different fall-rise accents exist: one of them could be analysed, in Pierrehumbert (1980) terms, as a L–+H* L– H% tune, and the other, which would corre-

18 Claudia Crocco spond to Jackendoff’s B accent, as a L*+H – L– H% tune3. Given this distinction, it is inaccurate to use the label “fall-rise” for both contours, since it mixes up tunes which have nuclear accents differently aligned to the stressed syllable (L+H* vs. L*+H). The authors, who also suggest that the two fall-rises could differ in other phonetic details, argue that the L*+H – L– H% contour conveys “uncertainty”. In an unpublished experiment (quoted in Pierrehumbert and Hirschberg 1990) Liberman and Pierrehumbert reanalyse Jackendoff’s B accent as a L+H* L– H% tune. Pierrehumbert and Hirschberg (1990: 296) claim that “the meaning of the L+H* pitch accent is closely related to that of L*+H”, investigated in depth by Hirschberg and Ward (1985). The authors add that “the most common use of “L+H* […] is to mark a correction or a contrast” (Pierrehumbert and Hirschberg 1990: 296), and that it is also the typical tonal pattern of background information4.

2.2.2. Topic accent(s) The hypothesis that the utterance may contain prominences related to information units other than the comment has been further developed by several authors working on information structure, prosody and on the relationship between the two. The approach of Vallduví and Zacharski (1994) and Vallduví and Enghdal (1996) to information packaging conflates the two dichotomy topic-comment and focus-ground in a new focus-ground partition, where the ground is further articulated in “link” and “tail”. The link, which corresponds to the topic of the topic-comment articulation, is associated with the B accent, L+H*. In more detail, nonsubject links, topicalised links and contrastive links are always B-accented, while “the accentual marking of subject links […] is optional, if the link in question is noncontrastive” (Vallduví and Enghdal 1996: 473). Similarly, Steedman (1991: 275) claims that “at least one function of the tune L+H* L– H% is […] marking what the utterance is about”, i.e. marking what in “older traditions” was called the “theme” or “topic”. Following the model proposed in Vallduví (1992), Büring (1997) also analyses the sentence in three information units, i.e. “topic”, “focus” and “background”. Topic is understood as “‘what the sentence is about’5 or ‘the entity anchoring the sentence to the previous discourse’” (Büring 1997: 55), while the background is “that part of the sentence which is neither Topic nor Focus” (Büring 1997: 54). The author bases his analyses of topic accent

Topic accent and prosodic structure

19

in English and German respectively on the works of Jackendoff (1972) and Féry (1993). Both in English and German, topic and focus are said to bear phonologically distinct pitch accents. However, in English, topical accent corresponds to the B accent (Jackendoff 1972), while in German sentence topics are marked by a Rise accent, analysed as L*+H. In addition, Jackendoff’s analysis of B accents is refined by Gundel (1978), who points out that activated topics are not B-accented. Accentuation is limited to new (unactivated) and contrastive topics (Gundel 1988). The phonological view on topic accent developed since Jackendoff (1972), has been recently tested on real speech with interesting results. Gundel, Hedberg and Zacharski (1997) applied the findings of previous studies on topic accent (Pierrehumbert 1980; Pierrehumbert and Hirschberg 1990; Steedman 1991; Vallduvi and Zacharski 1994; Vallduví and Enghdal 1996) to analysing a corpus of radio news stories in English. The analysis shows that topics do not seem to have a constant prosodic realisation. The examination of the correspondence between topics and tunes, pitch accents and boundary tones shows that there is no one-to-one relationship between them. Topics can be realised as a L+H* L– H% tune (the B accent) or as a H* L– H% tune, and can be followed by both L% and H%. However, H% boundary tones occur more frequently on topics. The traditional view of the accentuation of topic is also questioned by the results of Hedberg and Sosa’s work on broadcasted speech (2007). On the basis of their observations, the authors “deny that there is any prosodic category as distinctive as a ‘topic accent’ as opposed to a ‘focus accent’” (Hedberg and Sosa 2007: 101). In fact the results suggest that the relationship between informational categories and tonal events is less straightforward than the literature implies. Following the categorisation elaborated in Lambrecht and Michaelis (1998), the authors distinguish between Ratified, Unratified and Contrastive topics. The data show that “except for Ratified Topics, which tend to be unaccented, most phrases in each information structure category are marked by H*”. Moreover, the L+H* accent appears to be significantly related to topical or focal contrastiveness. However, this pitch accent can sporadically mark Unratified topics as well. Given this picture, Hedberg and Sosa (2007; see also Hedberg 2006) suggest that rather than topicality, this pitch accent could mark the category of “kontrast” (Vallduví and Vilkuna 1998) or that of “Contrastive Focus” (Gundel 1998). Further analysis on questions evidences that topic is mostly “marked with some kind of H* Accent” (Hedberg and Sosa 2002: 378).

20 Claudia Crocco 2.2.3. The prosody of topic in Italian: prominence and phrasing A complex relationship between information structure and prosody appears implicitly in other studies on read and spontaneous speech. Let us summarise a number of them concerning Italian. In a recent paper Frascarelli and Hinterhölzl (2007) examine syntactic and tonal features of topic expression in German and Italian. The analysis is illustrated by a number of examples of broadcasted speech (German) and spontaneous conversation (Rome Italian). According to the authors, in both languages topic expressions are realised by three different tonal events. The phonological form of the pitch accent, however, varies across the two languages. While in Italian topics can be realised as a bitonal accent L*+H or by single tones, i.e. H* and L*, in German topic expression may correspond to L+H*, L*+H or to the single tone L*. In both languages different pitch accents mark different types of topic. Namely, the introduction of a new “aboutness topic” (Lambrecht 1994) is marked by a L*+H tune in Italian and by a L+H* tune in German; contrastive topics (Büring 1997) correspond to H* tones in Italian and to a L*+H tune in German; finally, familiar topics (Givón 1983), i.e. topics that are already part of the background information, are realised by L* tones in both languages.6 In addition Frascarelli (20007) investigates the phonological domain of topic. In this work the author analyses a corpus of read and spontaneous speech collected in Florence, Milan and Rome following the approach of Prosodic Phonology (Selkirk 1984; Nespor 1993). Frascarelli claims that the phonological domain of the topic is the Intonational Phrase (I), composed by one or more minor constituents, i.e. Phonological Phrases (). According to the author, however, non branching-topics8, especially in fast speaking styles, may restructure into adjacent Intonational Phrases. This can happen also when the adjacent Intonational Phrase is a second topic (Frascarelli 2000: 48–53). Example (1a, b; adapted from Frascarelli 2000: 48) shows a case of nonbranching topic restructuring: (1)

a. [[ in America]]I [[ ci sono andato] [ tanti anni fa]]I In America there be-1SG go-PP many years ago ‘I went to America many years ago’ [tanti anni fa]]I b. [[in America] [ci sono andato]

Other analyses on phrasing, however, do not entirely confirm the findings summarised above. Gili Fivela (1999) investigates experimentally the pros-

Topic accent and prosodic structure

21

ody of the left periphery in Pisa Italian9. This study provides a phonetically and phonologically detailed comparison between focalised and topicalised left detached elements10. The data consist in read utterances where the desired reading is forced by the context. According to the author, although sequences of two topics can merge in one prosodic unit (as argued in Frascarelli 1997, 2000), this option can not be related to branchingness. Furthermore, topic separation and merge do not seem to be necessarily related to the speech rate or to the length of the constituents. Rather, the speakers seem to keep the two topics separated when they occur in marked order.11 Sentences with two left-detached topic constituents in unmarked vs. marked order are exemplified in (2a, b; example adapted form Gili Fivela 1999: 533): (2)

a. Il nascondiglio a Giovanni glielo svelava ieri The hiding place to Giovanni to him-it reveal-PAST-3SG yesterday pomeriggio afternoon ‘the hiding place, he revealed it to Giovanni yesterday afternoon’ b. a Giovanni il nascondiglio glielo svelava ieri to Giovanni the hiding place to him-it reveal-PAST-3SG yesterday pomeriggio afternoon ‘to Giovanni he revealed the hiding place yesterday afternoon’

Moreover, Gili Fivela’s results show that left-detached topical constituents are always separated from the matrix by a low tonal target and may carry different types of pitch accents. While single left-detached topics are consistently realised with an H* followed by a low boundary tone, sequences of two topics show different realisations, bearing usually H* or H+L* accents. In a study devoted to the comparison of broad and narrow focus accents, Avesani & Vayra (2000) analyse marked word order with left-detached topic constituents. They argue that marked word order such as left-dislocations and antepositions influence only prosodic phrasing, while they leave the accentual structure of the utterance unaffected12. The example (3) is taken from Avesani & Vayra (2000: 2; accentual and boundary tones are indicated as “T” plus diacritic; “ip” stands for “Intermediate Phrase”, Pierrehumbert 1980):

22 Claudia Crocco (3)

T* T– T*T–T% | | [[ Di porte]ip [ ne poteva aprire solo una] ip] ip Of doors of them can-PAST-3SG open only one ‘The doors, he could open only one of them’

In these constructions the main accent corresponds to the rightmost lexical stress of the utterance (in the example (3), it is located on the penultimate syllable “u” of “una”), just as in sentences with canonical word order (SVO)13: (4)

Giovanni poteva aprire una porta Giovanni can-PAST-3SG open a door ‘Giovanni could open a door’

However, the internal phrasing of the utterance is different in the two cases. On the one hand, in the unmarked case with canonical word order there are no internal prosodic boundaries that separate the various constituents. On the other hand, the left-detached constituent, i.e. the topic of the utterance, which also bears a pre-nuclear tone, is isolated in a separated Intermediate Phrase, as indicated by the presence of an edge tone. The following constituent, i.e. the comment, carries the nuclear pitch accent, followed by boundary tones, as in the unmarked constructions. Interesting data on intonation and phrasing of right-detached topical subject in a corpus of spontaneous speech are provided in Bonvino (2004), on Rome Italian. In Italian, subjects in VS construction can be either pragmatically neutral or marked as right-detached constituents (Sornicola 1999). In the former case the subject is part of the comment, while in the latter the subject is topical. Compared to right-detached subjects, neutral postverbal subject show a higher degree of cohesion with the VP. Based on f0 and duration measurements, Bonvino claims that the degree of coherence is marked also by prosodic means: while neutral postverbal subject are grouped in the same tone unit of the VP, right detached topical subject tend to be realised in a separated tone group. Moreover, topical postverbal subject are intonationally flat. Other works on spontaneous speech investigate the accentual phenomena related to topicality (Mereu 2004; Mereu and Trecci 2004; Mereu and Frascarelli 2006; Mereu 2008). According to the authors, utterance-initial topic constituents are pragmatically and, therefore, prosodically prominent, while postverbal topic, which are not salient, are flat and “follow the declination”

Topic accent and prosodic structure

23

(Mereu and Trecci 2004: 17). Moreover, according to Mereu and Trecci (2004), both contrastive and non-contrastive topics in utterance-initial position are prosodically marked by a pitch rise on the last stressed syllable of the constituent. A different theoretical approach to the analysis of information structure is adopted in Cresti (2000) and Cresti and Firenzuoli (2002)14, on Florentine Italian. According to Cresti (2000) the topic-comment structure expresses the illocutionary force of the utterance. Around the illocutionary nucleus, i.e. the comment, a number of other units may satellite optionally. Among these units, the topic and the appendix play a relevant role in structuring the information. On the one hand, the topic always precedes the comment limiting the field of application of its illocutionary force. On the other hand, the appendix may follow the comment, integrating its content with further information.15 Crucially, the actual topic/comment/appendix articulation of the utterance relies entirely on the prosodic realisation.16 The authors adopt the approach to prosodic analysis proposed by ‘t Hart, Collier and Cohen (1990), and analyse the melodic movement associated with topics as prefix contours. Topic contours are said to be composed of two tonal events, namely a rise-fall (“hat” pattern), located on the stressed syllable of the topic, and a fall-rise located on the last syllable of the topic unit. Other investigations on Rome Italian (Giordano and Crocco 2005) and on other varieties (Crocco and Savy 2007) also suggest that the topic bears a peak-shaped accent which can be compared to the pattern described in Cresti and Firenzuoli (2002) and Mereu and Trecci (2004). Finally, interesting remarks on the prosody of topics can be found in studies mainly concerning the pragmatic value of right-dislocations and other marked word orders. According to Berruto (1986), antitopics, i.e. postverbal given topics, are said to be separated from the matrix by means of an audible break. 2.3. A few explanatory remarks Before moving to the next section, a few explanatory remarks are needed on why the studies summarised above have been presented together here. A relevant and problematic issue concerning the study of “topic accent” is the well-known lack of agreement on what should be actually defined as “topic”. This makes any comparison between the different prosodic analyses uncertain and potentially unreliable. Moreover, several analyses of the prosody of topicality are not supported by experimental investigations of data.

24 Claudia Crocco These studies, which often adopt a syntax-centred perspective, provide phonological analyses but few phonetic details (cf. Gili Fivela’s criticism at fn. 9). I decided to review these studies together, despite the fact that they adopt a range of different and sometimes conflicting analytical approaches, because they agree on a number of general tenets, which allow us to consider them as composing a global picture. First they agree on the existence of structural interactions between the prosodic and information level. Furthermore, the presence of prominences, or boundaries, or both, marking certain types of topic (the theme, the link, or the ground according to the different terminologies) is generally acknowledged for several languages including Italian. I take for granted these general claims, together with the phonological regularities concerning the focus-related pitch range variations, which will be described in the following section (§ 3). On this basis, I will argue (§ 4) that the actual realisation of the prosodic marking of topicality may vary according to the different position occupied by the topic in the utterance. 3.

Prosodic structure of Neapolitan Italian. Pre- and post-nuclear accents

In this section I will summarise a number of relevant findings concerning the phonology of Neapolitan Italian, while focussing on the difference between nuclear and post-nuclear pitch accents. The theoretical framework adopted in most of the studies to which I will refer is the Autosegmental Metrical phonology (Pierrehumbert 1980; Ladd 1996). 3.1. Regional variability of Italian It is a well-known fact that Italian intonation varies considerably across the country (Canepari 1980; Bertinetto and Magno Caldognetto 1993; Endo and Bertinetto 1997). The great geographical variability of Italian is confirmed by investigations on several regional varieties of Italian.17 Overall, these studies show that, although the regional varieties share a number of features, several differences remain. The choice of the nuclear accent in questions and statements in Florentine, Neapolitan, Bari and Palermo Italian (Grice et al. 2005), can be taken as example of the differences and the similarities which can be found in regional Italian. In fact, the nuclear pitch accent of the statement can be analysed as H+L* in all observed dialects, while question tunes vary conspicuously across the four varieties.

Topic accent and prosodic structure

25

3.2. Prosody of Neapolitan Italian In this section I will describe the accentual phonology of only one variety, i.e. Neapolitan Italian (henceforth NI), which has been the object of several investigations (Caputo 1993, 1996; D’Imperio 1999, 2001a, 2001b, 2002; 2003; Grice et al. 2005). The description presented in this section is based largely on D’Imperio’s findings. According to the current accounts, (D’Imperio 2001a, 2002; Grice et al. 2005) in NI four pitch accents can be found in nuclear position. These accents differ both in form and function, since together with the edge tones, they mark various types of utterances. As the phonological status of the intermediate phrase in NI remains problematic,18 I will briefly review only the boundary tones typical of this variety. In NI the right side of the intonation phrase can be marked by a high or a low boundary tone (L% and H%). In more detail, H% is attested in utterances expressing continuation (cf. Avesani 1995; Gili Fivela 2002, 2003), while L% can mark the right edge of both declarative utterances and polar questions. The phonology of NI allows the presence of pitch accent both in prenuclear and in post-nuclear position, i.e. before and after the main pitch accent of the utterance. Regrettably, however, the prosody of the pre-nuclear region has not yet been investigated in depth. As far as it is known, only a reduced number of tones can occur in this position “with a predominance of H* and L+H*” (Grice et al. 2005: 363). More is known about the prosodic realisation of post-nuclear pitch accents. These prominences may occur in narrow focus utterances, where the prosodic focus encompasses only one constituent instead of the whole utterance (Ladd 1980, 1996). In narrow focus utterances the prosodic nucleus can be located somewhere before the last lexical stress, creating a “post-nuclear region”, which in (5) corresponds to the word “solo” (the word carrying the nuclear accent is boldfaced): (5)

Mario dorme solo Mario sleep-3SG alone ‘Mario sleeps alone (he doesn’t eat alone)’

In NI, as well as in other geographical varieties (Grice et al. 2005), the postnuclear region can host pitch accents. Post-nuclear accents appear to be different from the pitch accents occurring in pre-nuclear and in nuclear position, because they are phonetically reduced, or, more accurately, downstepped.19 This means that when there is a pitch accent after the prosodic nucleus, it usually appears compressed and subordinated to the nuclear accent.

26 Claudia Crocco The tonal inventory of NI listed below contains nuclear, pre-nuclear and post-nuclear accents attested in NI (Grice et al. 2005; D’Imperio 2003): 20 – – – – – – –

H+L*: L+H*: L*+H: L*: H*: !H*: !H+L*:

nuclear, declarative broad focus; nuclear, declarative contrastive focus; nuclear, yes/no question (broad focus and contrastive focus); nuclear, continuation; pre-nuclear, statements and questions; post-nuclear, yes/no questions post-nuclear, early focus statements

The existence of pre- and post-nuclear prominences in NI is crucial for the discussion concerning the prosody of topic. As it was seen in § 2, several linguists agree that the topic is prosodically marked by pitch accents, edge tones or both. It is also generally assumed, however, that the prosodic nucleus, i.e. the most relevant pitch accent in an utterance, is located on the informationally most relevant part of the utterance itself, i.e. the comment in an utterance articulated in topic and comment.21 Therefore topical prominences have to be located before or after the prosodic nucleus, i.e. they have to be either pre-nuclear or post-nuclear prominences. As just said above, pre-nuclear and post-nuclear pitch accents display different prosodic features. In particular post-nuclear accents are said to be downstepped. Interestingly, post-nuclear melodic compression does not seem to be a peculiarity of NI. Rather, it is attested in several other languages, such as French (Di Cristo and Jankowski 1999; Delais-Roussarie et al. 2002), English (Xu and Xu 2005), and Mandarin Chinese (Xu, Xu, and Sun 2004). The presence of different “prosodic regions” has been recently interpreted as a phenomenon connected to the focus (Xu 2005; Xu and Xu 2005; Xu 2006 a, b). According to this hypothesis, in several languages, postnuclear range suppression is part of the realisation of the prosodic focus, rather than an independent prosodic event: “in a declarative sentence, focus is realised not only by expanding the pitch range of the focused item, but also by compressing the pitch range of post-focus items, and possibly requiring that the pitch range of pre-focus items remain neutral” (Xu, Xu and Sun 2004: 81). Accordingly, since topics may occur in different positions of the utterance, i.e. before or after the nucleus, their position with respect to the prosodic focus is relevant for their realisation. In the following section I will discuss a number of examples of utterances where the topic occurs in different linear, or syntagmatic, positions.

Topic accent and prosodic structure

27

The discussion is focussed on the accentual phenomena related to the topic. I will argue that according to the different position occupied by the topic in the utterance, the actual realisation of the prosodic marking of topicality may vary. This means that the phonological regularities described in this section, i.e. the focus-related range variations, affect the realisation of the prosodic marking of the topic as well. 4.

Prosodic realisation of topical NP and PP

The example and the discussion below are designed to show that topic may undergo different prosodic realization according to its position with respect to the nucleus. Moreover, I will show that the different realizations of the topic are coherent with the general prosodic characters of the pre-nuclear and the post-nuclear regions. In order to show this, the prosodic realisation of the utterances containing a topic is contrasted with other utterances where the topic does not occur. I will only discuss cases in which the topic is realised as full NP or PP, occurring at the beginning or at the end of the utterance, and containing at least one lexical accent. In fact in Italian the pronouns tend to be prosodically weak and dependent to the adjacent lexical word, except when they are contrastive (Nespor 1993). The examples consist of questions and statements taken from dialogues of the API corpus22 which collects conversations mostly elicited using the Map Task technique (Anderson et al. 1992) .23 4.1. Pre-nuclear region Examples (6) and (7) show declarative utterances where a topical leftdetached NP occurs at the beginning of the sentence. In (6) the NP is the subject of the utterance, while in (7) it is the direct object 24. In both cases the NP is a given topic, restated for the sake of clarity. The utterance (6) is an autonomous conversational turn, while (7) opens a new conversational game (Carletta et al. 1996), after the abrupt interruption of the preceding one (Ferrari, Soria and Milos 2001). In both examples the prosodic nucleus is located on the last stressed syllable of the utterance: (6)

il viale dei lillà a me sta da un’altra parte The alley of the lilacs to me be-3SG from an other place ‘The lilac alley, in my map, is somewhere else’

28 Claudia Crocco

Figure 1. “Il viale dei lillà a me sta da un’altra parte”, uttered by a female speaker. F0 settings: 100 – 400 Hz. Maximum pitch value of the utterance: 269 Hz. The melodic peak corresponds to the NP “il viale dei lillà”. The nucleus is on the last word “parte”.

Figure 2. “Il bar Mimì non ce l’hai”, uttered by a female speaker. F0 settings: 100 – 400 Hz. Maximum pitch value of the utterance: 309 Hz. The melodic peak corresponds to the NP “il bar Mimì”. The nucleus is located on the word “hai”.

Topic accent and prosodic structure

(7)

29

il bar Mimì non ce l’ hai the bar Mimì not there it have-2SG ‘“Mimì bar”, you don’t have that’

Figures 1 and 2 show the pitch tracks and the waveforms of the utterances (6) and (7). In both examples, the topical NP carries a peak-shaped pitch accent. The rise of the peak is located on the syntactic head of the NPs, i.e. “viale” and “bar”, and the highest pitch value of the utterance is reached just after the end of the first word, i.e. the syntactic head of the NP. High utterance-initial peaks can occur also on locative PPs, as in (8)figure 3. These PPs may occur before presentative sentences. They act as “scene-setting” topics (Maslova and Bernini 2006) and provide spatial information relevant for the accomplishment of the task and for the understanding of the sentence. Example (8) is a nonfinal utterance: (8)

in alto c’ è arrivo in high there be-3SG endpoint ‘on the top there is the endpoint’

Figure 3. “In alto c’è arrivo”, uttered by a male speaker. F0 settings: 100–400 Hz. Nonfinal utterance. The melodic peak corresponds to the PP “in alto”. Maximum pitch value: 230 Hz. The nucleus is located on the word “arrivo”.

30 Claudia Crocco The interpretation of the utterance-initial peaks is controversial. On the one hand, they can be interpreted as a product of the f0 declination (cf. Sluijter and Terken 1993; Swerts, Strangert and Heldner 1996; Chilin 1997; Gussenhoven 2004); on the other hand, they can be related to the beginning of a new paratone (Brown et al. 1980; Yule 1980; Umeda 1982; Swerts and Geluykens 1993; Nakajima and Allen 1993; Xu and Xu 2005) or, as suggested by the examples shown above, to the presence of a sentence topic. Furthermore, physiological and linguistic factors might interact, as argued, among others, by Gussenhoven (2004: 121): “The natural tendency of the f0 to decline during the utterance is in part brought under speaker control, behaviour which may in turn lead to grammaticalized versions of this natural phenomenon.” However, similar peaks can be found also in association with sentential non-topical elements. The utterance (9)-figure 4, begins with two imperative VPs marked by rather high melodic peaks. The utterance is split in two different tone units: “prosegui e arriva fino” and “dove sta scritto portico”. For the sake of clarity I report in (9) the text of the entire utterance. The boundaries of the two tone units are indicated with square brackets: (9)

[prosegui e arriva fino]tu [a dove sta scritto go-IMP-2SG and come-IMP-2SG until to where be-3SG write portico]tu porch ‘go and come where “porch” is written’

On the basis of a superficial comparison it is clearly impossible to determine whether or not these peaks share properties. In fact, these prosodic events could differ in a number of phonologically and phonetically relevant aspects, such as the scaling and the temporal alignment. Therefore, further investigations are needed to shed light on the phonology and the phonetics of these pre-nuclear peaks. Nevertheless, it is apparent that at the beginning of the utterance, i.e. in pre-nuclear position, the pitch is rather dynamic. Figures 1, 2, 3 and 4, show that large melodic movements can occur on utterance-initial elements, which may, or may not, be topical.

Topic accent and prosodic structure

31

Figure 4. “Prosegui e arriva fino”, uttered by a male speaker. F0 settings: 80–400 Hz. Nonfinal utterance. Melodic peaks are visible on the first two words of the utterance (max. pitch values on the two peaks: 282 Hz; 248 Hz). The nucleus is located on the word “fino”.

4.2. Post-nuclear region The examples (10)-figure 5, and (11)-figure 6, show two declarative utterances with narrow prosodic focus. In both cases the nucleus is followed by another word: in (10) the post-focal word is a given topic, while in (11) it is a (non-topical) PP. (10) io non ce l’ ho il bar Mimì I not there it have-1SG the bar Mimì ‘I don’t have it, the “bar Mimì”’ (11) verso sinistra in pratica toward left in practice ‘In practice you must go to your left’

32 Claudia Crocco

Figure 5. “Io non ce l’ho, il bar Mimì”, uttered by a male speaker. F0 settings: 50– 250 Hz. After the nucleus, located on the word “ho”, the utterance appears melodically compressed.

Figure 6. ”Verso sinistra, in pratica”, uttered by a male speaker. F0 settings: 80–280 Hz. After the nucleus, located on the word “sinistra”, the utterance appears melodically compressed.

Topic accent and prosodic structure

33

In both cases the rhythmic word prominence, i.e. the metrical stress, is preserved and is clearly perceivable, although the pitch movements appear to be strongly reduced and compressed. The post-nuclear compression seems to be present both in declarative utterances and questions; however, it is more evident in declarative utterances. Example (12)-figure 7 shows a polar question with a narrow focus on the VP. The nucleus is followed by a right-detached topic. The pitch accent on the right-detached NP appears to be downstepped. (12) ce l’ hai la via splendida? There it have-2SG the street Splendid ‘Have you got the “Wonderful street”’?

Figure 7. “Ce l’hai la via splendida?”, uttered by a male speaker. F0 settings: 50 – 250 Hz. The pitch accent on the word “splendida” appears to be clearly downstepped.

As seen above (§ 3), pitch range suppression or compression is typical of the post-nuclear region in many languages. The examples presented above show that this phenomenon affects the realisation of topical elements as well as that of non-topical elements. Hence, similar to what happens in the prenuclear region, local pitch range features influence the actual realisation of the topical accentual prominence.

34 Claudia Crocco 4.3. Conclusive remarks The examples in the previous section are presented to show the differences between topic located on the left and on the right of the prosodic focus, and to compare their realisation to the realisation of non-topical units in similar positions. In pre-nuclear position, pitch appears to be dynamic and large melodic movement can be found on utterance-initial elements, while in post-nuclear position, just as observed in other studies such as Grice et al. (2005), pitch accents are strongly reduced. Melodic dynamism vs. compression appears to be a general prosodic feature affecting both topical and non-topical elements: therefore, topical expressions occurring in different syntagmatic positions in the utterance have different prosodic realisations. Overall, the prosodic realisation of topic in pre- and post-nuclear positions seems to be coherent with the general phonological regularities described in §3 (D’Imperio 2001a, Xu and Xu 2005). In my initial hypothesis I suggested that topics are associated with a bundle of prosodic phenomena rather than with a single pitch accent. This hypothesis shares similarities with those proposed by Frascarelli and Hinterhölzl’s (2007) for Italian, and by Hedberg and Sosa (2007) for English (cf. § 2). The examination of the data, together with a range of evidence provided by preceding studies (cf. § 2), show that topic is related not only to melodic prominence, but also to other prosodic phenomena, such as metrical stress.25 In cases of severe range reduction in the post-nuclear region, such as in (10)-figure 6, where the pitch accent is undistinguishable (or even absent),26 the metrical position is preserved and the lexical stress is still audible. Moreover, the data on left and right detachment of topical elements summarised in § 2 suggest that, at least in some cases, topics are separated from the rest of the utterance by a weak break, i.e. a melodic and/or rhythmic boundary. In the light of these findings, it seems possible to claim that topicality can be prosodically marked both by culminative and delimitative means, i.e. by a bundle of prosodic events, rather than by a single pitch accent.

4.4. A functional perspective The data discussed in the preceding sections also support the hypothesis that the realisation of topical accentual prominences is coherent with the structural prosodic regularities summarised in § 3. It is worth repeating,

Topic accent and prosodic structure

35

however, that the global prosodic features of the pre-nuclear and the postnuclear region affect topical and non-topical prominences in similar ways (fig. 1, 2, 4).27 Topical prominences behave just as other accents with respect to the pitch range variation in pre- and post-nuclear position. In fact, any differences between topical and non-topical pitch accents occurring in the same linear position are not directly apparent. The prominences examined in this paper share the function rather than the formal features: they have been grouped on an implicit functional basis as being part of the linguistic expression of topicality. However, I believe that adopting an explicit functional perspective could be more fruitful, for the following reasons. In the analysis of the interaction between two linguistic levels it is essential to keep in mind that the two levels are independent from one another (Lombardi Vallauri 2001). Topic and prosodic events, such as pitch accents, pertain to different levels of the linguistic organisation: the former is an informational/pragmatic category, while the latter are elements of the phonological structure. Moreover, according to the structure-based theories of intonations (Gussenhoven 1983; Ladd 1996), prosody has its own structure, which is independent from, though related to, other levels of the linguistic organisation, including information structure.28 As it has been shown in the previous section, the prosodic realisation of topic-related prominences displays different feature according to the syntagmatic position occupied by the topic in the utterance.29 These variations depend in part on the structural factors examined in this paper, which affect both topical and non-topical prominences. In other words, as far as prosodic realisations are concerned, structural prosodic factors come first. Prosody does not reflect directly the topicality, although it provides an important means for the linguistic expression of information structure and, hence, of topicality. Rather, the realisations of the prosodic form of topic are mediated by the general properties and characters of the prosodic structure: intonation and information structure are related, but independent. Furthermore, it seems possible to say that the relation between topic function and prosodic realisations is not biunivocal, because the function of topic does not correspond to a single prosodic event, just as in principle the same pitch accent can correspond to more than one function (cf Xu and Xu 2005). Accordingly, prosodic events can be analysed either as elements of the prosodic structure or as realisations of the information structure. On the one hand, several materially different prosodic realisations can be linked to the same abstract form and, ultimately, to the same function. On the other hand, the same prosodic event, such as a pitch accent, can be analysed from a

36 Claudia Crocco phonetic or phonological point of view per se, or can be put in relation to one or more linguistic functions.30 Therefore, I suggest considering the pragmatic function of topic corresponding, on the prosodic level, to an abstract prosodic form, which may have different realisations (see figure 8). These realisations may involve different rhythmic and melodic means such as pitch accents, stress, and boundary phenomena. In this perspective, the prosodic realisations of topical phrases discussed above can be seen tas actual manifestations of the prosodic form of topic. Topic function Abstract prosodic form Actual prosodic realisations pre-nuclear- --- -- -- --- -- -- --- --post-nuclear Figure 8.

5. Conclusions In this paper I suggest that the study of topic-related prosodic phenomena should be integrated with the results of the general investigation of prosody, in order to develop an integrated approach. Taking the preceding findings on the prosody of topics and on the phonology of Neapolitan Italian as a starting point, it is argued that topic is not simply associated with a unique accent, rather with a bundle of melodic and rhythmic phenomena. Furthermore, it is claimed that topical accentual prominences can be seen as part of the general phenomenology of the prosodic structure. In fact, the actual realisation of the prosodic marking is affected by syntagmatic factors: topic accentuation appears to vary according to the different position occupied by the topic with respect to the prosodic nucleus (D’Imperio 2001a; Xu and Xu 2005). Finally, a functional perspective is proposed for the interpretation of the topic-related prosodic events. This perspective is meant to provide a general frame for the analysis of the relationship between the prosodic level and topicality. I propose considering the pragmatic function of topic corre-

Topic accent and prosodic structure

37

sponding, to an abstract prosodic form, which may have several actual realisations according to the actual linear position of the topic in the utterance. By adopting a functional perspective it is possible to take for granted the structural autonomy of prosody and information organisation while analysing the interactions between the two linguistic levels. The phonetic and phonological qualities of the pitch accents occurring on topic phrases are not analysed in this paper. Indeed the examples presented in § 3 have been used merely to establish an initial comparison. However, investigation into the details of pre-and post-nuclear accents would be highly desirable. A comparison between topic-related accents and other prominences would be an invaluable contribution to the deeper understanding of the interactions between prosodic structure and information organisation.

Notes 1. The accentuation of new, accessible and given referents is a particularly relevant issue as far as the accentuation of topical element is concerned. The existence of a close relationship connecting the accentuation of topicality and the accentuation of the activation state of the linguistic expression of referents is an established fact (see, among others, Chafe 1976; Prince 1981; Brown 1983; Pierrehumbert and Hirschberg 1990; Lambrecht 1994; Swerts, Avesani and Krahmer 1999; Baumann and Grice 2004, 2006; Baumann 2006; Avesani and Vayra 2005; Cruttenden 1993, 2006). 2. As Ladd 1980 points out (Ladd 1980: 216, fn.4), Jackendoff (1972) makes use of Bolinger’s terminology incorrectly, since in the original system both “fall” and “fall-rise” are considered to be “A accents”. 3. For the conventions of the “Tones and Breaks Indices” (ToBI) prosodic annotation in American English and in Italian the reader is referred respectively to the work of Silverman et al. (1992) and Avesani (1995). 4. According to the authors however, both meanings express the commitment of the speaker to “an item chosen from a salient scale” (Pierrehumbert and Hirschberg 1990: 297). Moreover, “the S[peaker] employs the L+H* accent to convey that the accented item – and not some alternative related item – should be mutually believed.” (Pierrehumbert and Hirschberg 1990: 297). 5. However S(entence)-Topic may have also other uses (Büring 1997: 56), i.e. it could be “contrastive” (topic moving “the conversation form an entity given in the previous discourse”, “partial” (topic used to “‘narrow down’ a given D(iscourse)-topic”, or “implicational” (topic “used to indicate that the speaker would like to discuss alternative issues”).

38 Claudia Crocco 6. The authors underline that although in Italian, familiar topics are likely to occupy the rightmost position in the utterance, i.e. to occur after the prosodic focus of the utterance, the presence of the L* tone must not be related to the melodic compression usually found in post-nuclear sequences (see D’Imperio 2003; Di Cristo and Jankowski 1999). According to the authors, in fact, this tone marks familiar topics occurring in other positions. 7. See also Frascarelli 1997. The hypothesis proposed by the author is discussed and tested on experimental data in D’Imperio and Gili Fivela (2003). 8. A branching topic is composed by more than one . 9. It is worth noting that Gili Fivela’s investigation of the prosody of left periphery is motivated by the fact that “most of the time, in works on syntax [of left periphery], claims about […] intonation are not really substantiated; other times, the effort is directed to systematically relate the syntactic structure to the prosodic structure. This is the case for analysis realised in the Prosodic Phonology framework, in which researchers look at phonological rules to define prosodic constituents: topic and syntactic focus, then, have been defined as different prosodic constituents because they are domains of application of different phonological rules.” 10. Here topic is defined in terms of contextual givenness, i.e. as being previously mentioned, and aboutness. For a general picture concerning marked word orders in Italian see Benincà, Frison and Salvi (2001). 11. Left-detached constituents occur in marked order when the indirect object is placed before the direct object, as in “[a Giovanni]indir. obj. [il nascondiglio]dir. obj. glielo svelava ieri pomeriggio” vs. “[il nascondiglio]dir. obj. [a Giovanni] indir. obj. glielo svelava ieri pomeriggio”. 12. Cf. also Grice et al. (2005: 376): “[left] dislocation can only be made possible if a weak prosodic break is inserted between the dislocated element and the rest of the sentence”. 13. The syllable which carries the nuclear stress is boldfaced in the examples. 14. For an exhaustive account of this theoretical approach, see Cresti (2000). 15. The positional criterion is crucial for the identification of the topic. Since the topic can never follow the comment, in several cases post-comment appendices may correspond to the topic as defined by authors such as Lambrecht (1994) or Gundel (1988). 16. In this approach, different intonation may lead to different information interpretation of the same utterance. For instance, according to the prosodic realisation, a sentence such as “Ieri pioveva a Venezia”, “Yesterday it was raining in Venice” can be intonated as composed of a single comment [ieri pioveva a Venezia]comment; of topic and comment [ieri]topic [pioveva a Venezia]comment; of comment and appendix [ieri pioveva]comment [a Venezia]appendix, or as topic and comment and appendix [ieri]topic [pioveva]comment [a Venezia]appendix. 17. Among others, Caputo 1993, 1996; Grice 1995; Grice and Savino 1995a, b, 1997; Grice, Savino and Refice. 1997; D’Imperio 1999, 2001a, 2001b, 2002, 2003;

Topic accent and prosodic structure

18. 19. 20.

21.

22.

23. 24.

25. 26. 27.

28. 29.

39

D’Imperio and Gili Fivela 2003; Marotta 2002; Marotta and Sorianello 1999, 2001; Savino 2001; Avesani 1995; Avesani and Vayra 2000, 2003, 2004, 2005; Gili Fivela 1999, 2002, 2003; Gili Fivela and Savino 2003; Giordano 2006; Interlandi 2004. See D’Imperio (2001b, 2002) for an extensive discussion on the question tune and on the HL- phrase accent in NI. “Downstep is generally defined as compression of the pitch range as a consequence of some phonological regularity” (Grice et al. 2005: 372). Nuclear and post-nuclear accent are listed separately here. From a theoretical point of view, however, it is not necessary to keep the downstepped pitch accents apart from their non-downstepped counterparts. In fact, downstepped pitch accents can be considered as the product of the application of a phonological process, i.e. downstep, to normal pitch accents. This is the hypothesis of Ladd (1996), followed by Grice et al. (2005: 372), who claim that “downstep involve[s] an orthogonal variable independent of pitch accent type”. It is worth noting however, that every prosodically well-formed utterance, regardless to its inner informational articulation, must have a nucleus followed by edge tones: in fact, nucleus is “the obligatory accent of the contour” (Ladd 1996: 213). API (“Archivio del Parlato Italiano”, Crocco, Savy, and Cutugno 2003) is the continuation of a previous project, called AVIP (“Archivio delle Varietà dell’ Italiano Parlato”, Bertinetto 2001). Both projects aimed to construct corpus of spoken Italian and were carried out by several Italian academic institutions (Albano Leoni 2003). The Map Tasks were performed by male and female speakers, who had university education and were in their twenties. The subject NP in (6) can be also considered as left-dislocated, since (Benincà, Frison and Salvi (2001: 144) “[i]l soggetto è dislocato quando è separato dal verbo da altri costituenti, che a loro volta possono essere dislocati a sinistra […].” [the subject is dislocated when it is separated from the verb by other constituents, which can be left-dislocated too]. Metrically stressed syllables can bear a pitch accent (Ladd 1996). Deaccentuation of given information is in Italian a rare although not impossible phenomenon (Avesani and Vayra 2005). It is worth noting that, although the actual realisation of the accentual phenomena is affected by syntagmatic factors, this does not contradict the hypothesis that topic accents could be phonologically distinguishable from pitch accents with other functions in principle. However, in order to establish such a phonological distinction, punctual analyses of the prosodic features of topic-related prominences are required. Cf. Ladd 1996; see in particular the debate on prominence and focus summarised in Ladd (1996: 160 –167). It goes without saying, however, that the actual prosodic realisation of the topic does not depend only on phonological and phonetic factors, but that it is also

40 Claudia Crocco influenced by syntactic as well as cognitive factors (cf. fn. 1). In languages, such as Italian, where morphological marking of topicality is lacking, topicality is primarily expressed by morphosyntactic means. The actual prosodic realisation of a topic expression, hence, is relevantly affected by both morphosyntactic and structural phonological factors. Possible examples of the interaction between morphosyntactic and prosodic factors are, on the one hand, the accentuation of full NP and PP, which varies according to the syntagmatic position occupied by the topic in the utterance; and on the other hand, the lack of prominence of pronominalised topics, which, in normal conditions, receive neither stress or accent. 30. Cf. Xu and Xu (2005).

References Albano Leoni, Federico 2003 Tre progetti per l’italiano parlato. In Italia linguistica anno Mille. Italia linguistica anno Duemila: Atti del XXXIV Congresso Internazionale di Studi della Società di Linguistica Italiana (SLI), Nicoletta Maraschio, and Teresa Poggi Salani (eds.), 675–683, Roma: Bulzoni. Anderson, Anne H., Miles Bader, Ellen Gurman Bard, Elizabeth H. Boyle, Gwyneth Doherty, Simon C. Garrod, Stephen Isard, Jacqueline Kowtko, Jan McAllister, Jim Miller, Catherine Sotillo, Henry Thompson, and Regina Weinert 1992 The HCRC Map Task Corpus. Research Paper of the Human Research Communication Centre (HCRC / RP) 29, University of Edinburgh. Avesani, Cinzia 1995 ToBIt. Un sistema di trascrizione per l’intonazione italiana. Atti delle V giornate di studio del Gruppo di Fonetica Sperimentale (GFS), 1994, Trento, Italy: 85–98. Avesani, Cinzia and Mario Vayra 2000 Costruzioni marcate e non marcate in italiano: Il ruolo dell’intonazione. Atti delle X giornate di studio del Gruppo di Fonetica Sperimentale (GFS), 1999, Napoli, Italy: 1–14. 2003 Broad, narrow and contrastive focus in Florentine Italian. Proceedings of the 15th International Congress of Phonetic Sciences (ICPhS), 2003, Barcelona, Spain: 1803–1806. 2004 Focus ristretto e focus contrastivo in italiano. Il parlato italiano: Atti del Convegno Nazionale, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.), CD-Rom: F01.pdf, Napoli: D’Auria Editore.

Topic accent and prosodic structure 2005

41

Accenting, deaccenting and information structure in Italian dialogues. Proceedings of the 6th International Speech Communication Association Special Interest Group on dialogue (ISCA-SIGdial), Workshop on Discourse and Dialogue, 2005, Lisbon, Portugal: 19–24. Baumann, Stefan 2006 Information structure and prosody: Linguistic categories for spoken language annotation. In Methods in empirical prosody research, Stefan Sudhoff, Denisa Lenertová, Roland Meyer, Sandra Pappert, Petra Augurzky, Ina Mleinek, Nicole Richter, and Johannes Schließer (eds.), 152–179. Berlin /New York: Mouton de Gruyter. Baumann, Stefan and Martine Grice 2004 Accenting accessible information. Proceedings of the Second International Conference on Speech Prosody, 2004, Nara, Japan: 21–24. 2006 The intonation of accessibility. Journal of Pragmatics 38 (10): 1636 – 1657. Benincà, Paola, Lorenza Frison, and Giampaolo Salvi 2001 L’ordine degli elementi della frase e le costruzioni marcate. In Grande grammatica italiana di consultazione, Lorenzo Renzi, Giampaolo Salvi, and Anna Cardinaletti (eds.), 129 –239. Bologna: Il Mulino. Berruto, Gaetano 1986 Le dislocazioni a destra in italiano. In Tema-Rema in italiano/ThemeRheme in Italian /Thema-Rhema im Italienischen, Harro Stammerjohann (ed.), 55–69. Tübingen: Narr. Bertinetto, Pier Marco (ed.) 2001 AVIP: Archivio delle Varietà dell’Italiano Parlato. CD-ROM. Pisa: Ufficio Pubblicazioni della Classe di Lettere della Scuola Normale Superiore. Bertinetto, Pier Marco and Emanuela Magno Caldognetto 1993 Ritmo e intonazione. In Introduzione all’italiano contemporaneo: Le strutture, Alberto A. Sobrero (ed.), 141–192. Bari: Laterza. Bolinger, Dwight 1965 Forms of English. Cambridge, MA: Harvard University Press. Bonvino, Elisabetta 2004 Fenomeni sintattici e prosodici relativi alla distinzione fra due tipi di soggetto postverbale. Il parlato italiano: Atti del Convegno Nazionale, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, Renata Savy (eds.), CD-ROM: F02.pdf, Napoli: D’Auria Editore. Brown, Gillian 1983 Prosodic structure and the Given/New distinction. In Prosody: Models and measurements, Anne Cutler, D. Robert Ladd, and Gillian Brown (eds.): 67–77. Berlin: Springer. Brown, Gillian, Karen L. Currie, and Joanne Kenworthy 1980 Questions of Intonation. London: Croom Helm.

42 Claudia Crocco Büring, Daniel 1997 The Meaning of Topic and Focus: The 59th Street Bridge Accent. London: Routledge. Canepari, Luciano 1980 Italiano Standard e Pronunce Regionali. Padova: CLUEP. Caputo, Maria Rosaria 1993 L’intonazione delle domande sì/no in un campione di italiano parlato. Atti delle IV Giornate di Studio del Gruppo di Fonetica Sperimentale (GFS), 1993, Torino, Italy: 9–18. 1996 Presupposizione, focus, modalità e schemi melodici. Atti del XXIV Convegno Nazionale dell’Associazione Italiana di Acustica (AIA), 1996, Trento, Italy: 49–54. Carletta, Jean, Amy Isard, Stephen Isard, Jacqueline Kowtko, and Gwyneth DohertySneddon 1996 HCRC dialogue structure coding manual. Technical Report of the Human Communication Research Centre (HCRC/TR), 82, University of Edinburgh. Chafe, Wallace 1976 Giveness, contrastiveness, definiteness, subjects, topics and point of view. In Subject and topic, Charles N. Li (ed.), 25–55. New York: Academic Press. Chilin, Shih 1997 Declination in Mandarin. Proceedings of Intonation: Theory, Models, and Applications. Workshop of the European Speech Communication Association (ESCA), 1997, Athens, Greece: 293–296. Cresti, Emanuela 2000 Corpus di Italiano Parlato. Firenze: Accademia della Crusca. Cresti, Emanuela and Valentina Firenzuoli 2002 L’articolazione informativa topic-comment e comment-appendice: Correlati intonativi. Atti delle XII giornate di studio del Gruppo di Fonetica Sperimentale (GFS), 2001, Macerata, Italy: 153–166. Crocco, Claudia, Renata Savy, and Francesco Cutugno (eds.) 2003 API: Archivio dell’Italiano Parlato (www.parlaritaliano.it/parlare). Crocco, Claudia and Renata Savy 2007 Topic in dialogue: prosodic and syntactic features. Proceedings of Interspeech 2007, Antwerp, Belgium, CD-Rom. Cruttenden, Alan 1993 The de-accenting and re-accenting of repeated lexical items. Proceedings of the European Association of Speech Communication (ESCA), Conference on Prosody, 1993, Lund, Sweden: 16 –19. 2006 The de-accenting of given information: a cognitive universal? In Pragmatic organization of discourse in the languages of Europe, Giuliano Bernini, and Marcia L. Schwartz (eds.), 311–356. Berlin /New York: Mouton de Gruyter.

Topic accent and prosodic structure

43

Delais-Roussarie, Elisabeth, Annie Rialland, Jenny Doetjes and Jean-Marie Marandin 2002 The prosody of post-focus sequences in French. Proceedings of the First International Conference on Speech Prosody 2002, Aix-enProvence, France: 239–242. Di Cristo, Albert and Ludovic Jankowski 1999 Prosodic organisation and phrasing after focus in French. Proceedings of 14th the International Congress of Phonetic Sciences (ICPhS) 1999, San Francisco, USA: 1565–1568. D’Imperio, Mariapaola 1999 Tonal structure and pitch targets in Italian focus constituents. Proceedings of the 14th International Congress of Phonetic Sciences (ICPhS), San Francisco, USA: 1757–1760. 2001a Focus and tonal structure in Neapolitan Italian. Speech Communication 33 (4): 339–356. 2001b Tonal alignment, scaling and slope in Italian question and statement tunes. Proceedings of the 7th European Conference on Speech Communication and Technology (Eurospeech) 2001, Aalborg, Denmark: 99–101. 2002 Italian intonation: An overview and some questions. Probus 14 (1): 37–69. 2003 Tonal structure and pitch targets in Italian focus constituents. Catalan Journal of Linguistics 2: 55–65. D’Imperio, Mariapaola and Barbara Gili Fivela 2003 How many levels of phrasing? Evidence from two varieties of Italian. In Phonetic interpretation: Papers in laboratory phonology VI, John Local, Richard Ogden, and Rosalind Temple (eds.), 130 –144. Cambridge: Cambridge University Press. Endo, Reiko and Pier Marco Bertinetto 1997 Aspetti dell’intonazione in alcune varietà di italiano. Atti delle VII Giornate di Studio del Gruppo di Fonetica Sperimentale (GFS), Rome, Italy: 27–49. Ferrari, Giacomo, Claudia Soria, and Elena Milos 2001 AVIP: Annotazione morfosintattica e testuale-pragmatica. In AVIP: Archivio delle Varietà dell’Italiano Parlato. CD-ROM, Pier Marco Bertinetto (ed.), Pisa: Ufficio Pubblicazioni della Classe di Lettere della Scuola Normale Superiore. Féry, Caroline 1993 German Intonational Patterns. Tübingen: Niemeyer Frascarelli, Mara 1997 The phonology of focus and topic in Italian. The Linguistic Review 14: 221–248. 2000 The Syntax-Phonology Interface in Focus and Topic Constructions in Italian. Dordrecht /Boston / London: Kluwer.

44 Claudia Crocco Frascarelli, Mara and Roland Hinterhölzl 2007 Types of topics in German and Italian. In On information structure, meaning and form, Suzanne Winkler and Kerstin Schwabe (eds.), 87– 116. Amsterdam / Philadelphia: John Benjamins. Gili Fivela, Barbara 1999 The prosody of left-dislocated topicalized constituents in Italian read speech. Proceedings of the Sixth European Conference on Speech Communication and Technology (Eurospeech) 1999, Budapest, Hungary: 531–534. 2002 Tonal Alignment in two Pisa Italian peak accents. Proceedings of the First International Conference on Speech Prosody, 2002, Aix-enProvence, France: 339–342. 2003 The phonetics and phonology of intonation: the case of Pisa Italian. Ph.D. diss., Scuola Normale Superiore, Pisa. Gili Fivela, Barbara and Michelina Savino 2003 Segments, syllables and tonal alignment: A study on two varieties of Italian. Proceedings of the 15th International Congress of Phonetic Sciences (ICPhS), 2003, Barcelona, Spain: 2933–2936. Giordano, Rosa 2006 The intonation of polar questions in two central varieties of Italian. Proceedings of the Third International Conference on Speech Prosody, Dresden, Germany: CD-ROM: PS8-10-155.pdf. Giordano, Rosa and Claudia Crocco 2005 Sul rapporto tra intonazione e articolazione informativa. In Italiano parlato: Analisi di un dialogo, Federico Albano Leoni, and Rosa Giordano (eds.), 159 –188, Napoli: Liguori. Givón, Talmy 1983 Topic continuity in discourse: An introduction. In Topic continuity in discourse: A quantitative cross-language study, Talmy Givón (ed.), 1–41. Amsterdam: John Benjamins. Grice, Martine 1995 The Intonation of Interrogation in Palermo Italian: Implications for Intonation Theory. Tübingen: Niemeyer. Grice, Martine and Michelina Savino 1995a Intonation and communicative function in a regional variety of Italian, Phonus 1: 19–32. Universität des Saarlandes. 1995b Low tone versus ‘sag’ in Bari Italian intonation: A perceptual experiment. Proceedings of the 13th International Congress of Phonetic Sciences (ICPhS) 1995, Stockholm, Sweden: 658–661. 1997 Can pitch accent type convey information-status in yes-no questions? Proceedings of the Association for Computational Linguistics (ACL 97): Workshop on Concept-to-Speech Generation Systems, Madrid, Spain: 29–38.

Topic accent and prosodic structure

45

Grice, Martine, Michelina Savino, and Mario Refice 1997 The intonation of questions in Bari Italian: Do speakers replicate their spontaneous speech when reading? Phonus 3: 1–7, Universität des Saarlandes. Grice, Martine, Mariapaola D’Imperio, Michelina Savino, and Cinzia Avesani 2005 Towards a strategy for labelling varieties of Italian. In Prosodic typology and transcription: A unified approach, Sun-Ah Jun (ed.), 55– 83. Oxford: Oxford University Press. Gundel, Janette 1978 Stress, pronominalisation and the given-new distinction. Working Papers in Linguistics 10 (2): 1–13, University of Hawaii. 1988 Universals of topic-comment structure. In Studies in syntactic typology, Michael Hammond, Edith A. Moravscik, and Jessica R. Wirth (eds.) 209–239. Amsterdam: John Benjamins. 1998 On different kinds of focus. In Focus: linguistic, cognitive, and computational perspectives, Peter Bosch and Rob van der Sandt (eds.), 293–305. Cambridge: Cambridge University Press. Gundel, Jeanette, Nancy Hedberg, and Ron Zacharski 1997 Topic-Comment structure, syntactic structure and prosodic tune. Paper presented at the Workshop on Prosody and Grammar in Interaction, Helsinki, Finland, August 13–15. Gussenhoven, Carlos 1983 On The Grammar And Semantics Of Sentence Accent. Dordrecht: Foris. 2004 The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Hedberg, Nancy 2006 Topic-Focus controversies. In The architecture of focus, Valeria Molnar and Susanne Winkler (eds.), 373–399. Berlin /New York: Mouton de Gruyter. Hedberg, Nancy and Juan M. Sosa 2002 The prosody of questions in natural discourse. Proceedings of the First International Conference on Speech Prosody 2002, Aix-enProvence, France: 375–378. 2007 The prosody of topic and focus in spontaneous English dialogue. In Topic and focus: Cross-linguistic perspectives on meaning and intonation, Chungmin Lee, Mathew Gordon, and Daniel Büring, (eds.), 101–120. Dordrecht: Springer. Hirschberg, Julia and Gregory Ward 1985 Fall-Rise intonation and the place of intonational “meaning” in linguistic theory. Proceedings of the 11th Annual Meeting of the Berkeley Linguistics Society: 447–458.

46 Claudia Crocco Interlandi, Grazia 2004 Il continuum della variazione pragmatico-espressiva nell’intonazione dell’italiano parlato a Torino. In Il parlato italiano: Atti del Convegno Nazionale, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.), CD-ROM: F07.pdf, Napoli: D’Auria Editore. Jackendoff, Ray 1972 Semantic Interpretation in Generative Grammar. Cambridge, MA / London, England: MIT Press. Ladd, D. Robert 1980 The Structure of Intonational Meaning: Evidence from English. Bloomington /London: Indiana University Press. 1996 Intonational Phonology. Cambridge: Cambridge University Press. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press. Lambrecht, Knud and Laura Michaelis 1998 Sentence accent in information questions: Default and projection. Linguistics and Philosophy 21: 477–544. Lombardi Vallauri, Edoardo 2001 La teoria come separatrice di fatti di livello diverso. L’esempio della struttura informativa dell’enunciato. In Dati empirici e teorie linguistiche. Atti del XXXIII Congresso della Società di Linguistica Italiana (SLI), Federico Albano Leoni, Eleonora Stenta Krosbakken, Rosanna Sornicola, and Carolina Stromboli (eds.), 151–173. Roma: Bulzoni. Marotta, Giovanna 2002 L’intonation des énonceés interrogatifs ouverts dans l’Italien Toscan. Proceedings of the First International Conference on Speech Prosody 2002, Aix-en-Provence, France: 475–478. Marotta, Giovanna and Patrizia Sorianello 1999 Question intonation in Sienese Italian. Proceedings of 14th the International Congress of Phonetic Sciences (ICPhS) 1999, San Francisco, USA: 1161–1164. 2001 La teoria autosegmentale nell’analisi dell’intonazione interrogativa di due varietà di italiano toscano (Lucca e Siena). In Dati empirici e teorie linguistiche: Atti del XXXIII Congresso della Società di Linguistica Italiana (SLI), Federico Albano Leoni, Eleonora Stenta Krosbakken, Rosanna Sornicola, and Carolina Stromboli (eds.), 177– 204. Roma: Bulzoni. Maslova, Elena and Giuliano Bernini 2006 Sentence topics in the languages of Europe and beyond. In Pragmatic organization of discourse in the languages of Europe, Giuliano Bernini, and Marcia L. Schwartz (eds.), 67–120. Berlin/New York: Mouton de Gruyter.

Topic accent and prosodic structure

47

Mereu, Lunella 2004 La sintassi delle lingue del mondo. Roma / Bari: Editori Laterza. 2008 La sintassi dei circostanziali nel parlato. Atti del Convergno internazionale La comunicazione parlata/Spoken communication, Massimo Pettorino, Antonella Giannini, Marianna Vallone, and Renata Savy (eds.), 1747–1767. Napoli: Liguori Editore. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. Il Parlato Italiano: Atti del Convegno Nazionale, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.), CD-ROM: B10.pdf. Napoli: M. D’Auria Editore. Mereu, Lunella and Mara Frascarelli 2006 L’interfaccia sintassi-fonologia: interpretazione e implicazioni teoriche. In Atti del Secondo convegno nazionale dell’Associazione Italiana di Scienze della Voce (AISV), Renata Savy, and Claudia Crocco (eds.), Torriana: EDK (CD-ROM). Nakajima, Shin’ya and James F. Allen 1993 A study on prosody and discourse structure in cooperative dialogues. Phonetica 50: 197–210. Nespor, Marina 1993 Fonologia. Bologna: Il Mulino. Pierrehumbert, Janet 1980 The Phonology and Phonetics of English Intonation. Ph.D. thesis, MIT. Distributed by Indiana University Linguistics Club, Bloomington. Pierrehumbert Janet and Julia Hirschberg 1990 The meaning of intonational contours in the interpretation of discourse. In Intentions in communication, Jerry Morgan, Philip Cohen, and Martha Pollack (eds.), 271–312. Cambridge, MA: MIT Press. Prince, Ellen F. 1981 Toward a taxonomy of given-new information. In Radical pragmatics, Peter Cole (ed.): 223–256. New York: Academic Press. Savino, Michelina 2001 Non-finality and pre-finality in Bari Italian intonation: A preliminary account. Proceedings of the 7th European Conference on Speech Communication and Technology (Eurospeech) 2001, Aalborg, Denmark: 939–942. Selkirk, Elisabeth O. 1984 Phonology and Syntax: The Relation between Sound and Structure. Cambridge, MA: MIT Press. Silverman, Kim, Mary Beckman, John F. Pitrelli, Mori Ostendorf, Colin Wightman, Patti Price, Janet Pierrehumbert, and Julia Hirschberg 1992 TOBI: A standard for labeling English prosody. In Proceedings of the Second International Conference on Spoken Language Processing, (ICSLP) 1992, Banff, Canada: (2) 867–870.

48 Claudia Crocco Sluijter, Agaath and Jacques Terken 1993 Beyond sentence prosody: Paragraph intonation in Dutch. Phonetica 50: 180–188. Sornicola, Rosanna 1999 Labile orders and grammatical functions. The functional representation of one argument structures. In Boundaries of morphology and syntax, Lunella Mereu (ed.), 291–305. Amsterdam: John Benjamins. Steedman, Mark 1991 Structure and intonation. Language 67: 262–296. Swerts, Marc, Cinzia Avesani, and Emiel Krahmer 1999 Reaccentuation or deaccentuation: A comparative study of Italian and Dutch. Proceedings of the 14th International Congress of Phonetic Sciences (ICPhS), Berkeley, USA: 1541–1544. Swerts, Marc and Ronald Geluykens 1993 The prosody of information units in spontaneous monologue. Phonetica 50: 189–196. Swerts, Marc, Eva Strangert, and Mattias Heldner 1996 F0 declination in spontaneous and read-aloud speech. Proceedings of the 4th International Conference on Spoken Language Processing (ICSLP) 1996, Philadelphia, USA: 1501–1504. ’t Hart, Johan, René Collier, and Antoine Cohen 1990 A Perceptual Study of Intonation: An Experimental-Phonetic Approach. Cambridge: Cambridge University Press. Umeda, Noriko 1982 “F0 declination” is situation dependent. Journal of Phonetics 10: 279– 290. Vallduví, Enric 1992 The Informational Component. New York: Garland Publishing. Vallduví, Enric and Elisabet Engdhal 1996 Information packaging in HPSG. Working Papers in Cognitive Science 12, University of Edinburgh. Vallduví, Enric and Maria Vilkuna 1998 On rheme and kontrast. In The limits of syntax, Peter Culicover and Louise McNally (eds.) 79–108. New York: Academic Press. Vallduví, Enric and Ron Zacharski 1994 Accenting phenomena, association with focus, and the recursiveness of focus-ground. Proceedings of the Amsterdam Colloquium 9: 683– 702. Yule, George 1980 Speaker’s topics and major paratones. Lingua 52: 33–47. Xu, Yi 2005 Speech melody as articulatorily implemented communicative functions. Speech Communication 46: 220–251.

Topic accent and prosodic structure 2006a

49

Principles of tone research. Proceedings of International Symposium on Tonal Aspects of Languages, La Rochelle, France: 3–13. 2006b Speech prosody as articulated communicative functions. Proceedings of the Third International Conference on Speech Prosody, Dresden, Germany. Xu, Yi, Ching X. Xu and Xuejing Sun 2004 On the temporal domain of focus. Proceedings of the Second International Conference on Speech Prosody, Nara, Japan: 81–84. Xu, Yi and Ching X. Xu 2005 Phonetic realization of focus in English declarative intonation. Journal of Phonetics 33: 159–197.

The prosodic annotation of C-ORAL-ROM and the structure of information in spoken language Antonietta Scarano

1. Introduction The C-ORAL-ROM romance speech corpus (Cresti and Moneglia 2005) is a linguistic resource whose innovative aspect consists in enabling the analysis of utterances, which are assumed to be the units of reference of speech, and are prosodically identified. While C-ORAL-ROM fulfils the scientific community’s formal requirements on speech resources (EAGLES – Expert Advisory Group on Linguistic Engineering Standards; NERC – Network of European Reference Corpora; TEI – Text Encoding Initiative1), on the other hand it is mainly the result of specific choices concerning transcription formats and the access to the audio data (alignment by utterances) which are in line with the Informational Patterning Theory (Cresti 2000). The parsing of speech in prosodic units (rendered in the transcription) is determined by the perceptive relevance of terminal and non-terminal prosodic breaks, which enable the identification of both utterances and their informational patterning. The utterance, in such a perspective, is not defined syntactically but pragmatically, as the counterpart of a speech act. On the basis of the linguistic data made available by the corpus, the resource shows that, in speech, the informational structure of utterances is isomorphic to their intonational structure. The informational structure is directed to the expression of an illocution, and appears to operate at a higher level than the syntactic structure one. The first part of the article (§§ 2–3) presents the prosodic analysis system which underlies the tagging method in C-ORAL-ROM, and shows how this leads to the identification of a structural typology of utterance, which is unrelated to the syntactic structure. In the second part (§ 4), prosodic annotation is viewed in relation to both the structural and the “semantic” function of prosody. As a perspective which closely links the two functions, the theoretical frame underlying the choices made in the building up of C-ORALROM is then outlined. The last part of the paper (§ 5) shows revealing examples of the kind of complexity given by spoken language informational

52 Antonietta Scarano patterning. Moreover, such examples will show that this can be clearly distinguished by the syntactic complexity. 2.

C-ORAL-ROM

C-ORAL-ROM is a collection of spontaneous speech corpora of the four main romance languages (Italian, French, Portuguese and Spanish) consisting of approximately 300,000 words for each language. It is available on DVD with text-to-speech alignment. 2.1. Text lines section in transcription The CHAT standard (Codes Human Analysis of Transcripts; MacWhinney 1995) states that every transcription file should contain information on word level, utterance level, suprasegmental level and paralinguistic events. The indication of speaker turns, comments and context information is also required. This information is then split between different sections, which make up the transcription: a) metadata section regarding sessions and speakers; b) orthographically transcribed text lines; c) dependent lines with contextual information and possible annotation levels. The distinction between speech language research (conducted on the strictly phonic production of a language) and spoken language research (on widescale grammatical aspects) has long meant for phonetic transcription to be reserved to the former and orthographic transcription to the latter. Initiatives towards the encoding of texts in standard formats (Llisterri 1996; Barry and Fourcin 1992) have opted for orthographic transcription, which could be, fully or partially, linked to phonetic and/or phonemic transcriptions. C-ORAL-ROM features an orthographic transcription, with various annotations within the text lines. The CHAT format has been implemented (Moneglia and Cresti 1997) to express various functional properties of prosody, and the different levels at which they operate. Prosodic parsing identifies units greater than the word (utterances and prosodic units; see Cresti 2000, Cresti et al. 2002). The implemented version of the CHAT format features the parsing of each text turn by means of tags which mark terminal (//) and non-terminal (/) prosodic boundaries, as well as those due to fragmentation (retracting and interruptions). Terminal boundaries have been assumed to signal the end of an utterance. Word fragments (&) and overlaps (< >) have also been highlighted.

The prosodic annotation of C-Oral-Rom and the structure of information 53

3. The prosodic reference model in C-ORAL-ROM The prosodic parsing which appears in C-ORAL-ROM’s texts was introduced during transcription and not added later, at a certain level of transcription enrichment. The simultaneity between annotation and transcription was made possible by Cresti’s perspective of analysis, adopted in CORAL-ROM, which focuses on acoustic and perceptive aspects of intonation phenomena. The intonation reference model used is that of the Instituut voor Perceptie Onderzoek (IPO) team of Eindhoven, which, from the very first version of the theory (‘t Hart and Collier 1975), proposed a perceptive analysis as the link between the so-called abstract level of intonation (intonational models) and actual output. ‘t Hart and Collier (1990) assume that the perceptive level operates as a device which, given an input of pre-determined acoustic or physiological values, selects only the information needed for communication. In other words, the analysis of intonation conducted by the IPO team is based on the perceptive level of prosody and more specifically on the notion – proven by testing – that a speech listener is sensitive only to a very limited range of F0: namely, those values intentionally produced by the speaker. 2 Variations of the F0 parameter, which have perceptive pregnancy, are considered the entities of the first level of prosody structuring. The tonal unit corresponds to the one or more variations in F0 which form it, and is compatible with a class of profiles. Each class features certain compositional and distributional traits according to a given intonational pattern which is proper of the observed language. Cresti accepts the main guidelines of the IPO model, integrating them with concepts developed from her research experience on vast Italian speech corpora. It is assumed that certain pitch movement configurations exist, which are subject to distributional and recurrence rules: root units (nuclear units in themselves necessary and sufficient to build a pattern), prefix units (optional and melodically subordinated units, always placed before nuclear units), suffix units (optional units with a greater degree of melodic subordination compared to prefix units, placed after a nuclear or prefix unit) and dialogical auxiliary units (melodically subordinated and with variable distributional traits). Each pattern may contain a single root unit, whereas the prefix, suffix and other secondary units may appear more than once in the same pattern. A tonal pattern may even consist of just a single nuclear unit. The notation of prosody does not highlight trends, tonal or melodic accent values for each prosodic phenomenon; it does, however, signal the tonal

54 Antonietta Scarano pattern and the tonal units which the pattern consists of. Therefore, it is the utterance boundaries and the prosodic parsing within the utterance that are highlighted in the corpus, with no indication of melodic trend or quality of movement (ascending, descending, flat, ascending-descending or descending-ascending), nor of tone level (high, medium, low) or melodic accent, as is true for many prosodic notation methods3. The parsing adopted in C-ORAL-ROM is therefore consistent with a frame of analysis which identifies tonal patterns (prosodic reference entities), made up, in turn, of one or more tonal units. With regard to these properties of prosodic notation, it can be said that each utterance, corresponding to a tonal pattern, is made up of tonal units of two types: a) nonterminal with respect to the utterance, identified by a single slash; b) terminal with respect to the utterance, identified by various diacritics (double slash, question mark, suspension dots). As we’ll see in § 4.2.2, the tonal pattern is strictly related to the informational structure of spontaneous speech (informational patterning, Cresti 2000: 63–75) and the very definition of utterance at a pragmatic level (illocutionary criterion, Cresti 2000: 46–47). 3.1. Types of utterance by prosodic structure On the basis of the notation method used in C-ORAL-ROM, it has been possible to detail a typology of utterance structures in the four romance languages analyzed. Simple utterances consist of a single linguistic sequence (tonal root unit) which ends with a prosodic terminal break. They generally correspond to quite a brief and syntactically simple linguistic sequence, and not necessarily contain a VP: (1)

*ARA: eso es educación // ‘that means education //’

(efamdl154)

(2)

*RIT: o fruto probido // ‘the forbidden fruit //’

(pfamcv11)

Compound utterances consist of linguistic sequences – not necessarily featuring a verb phrase – that span the utterance from the beginning to a terminal prosodic break and contain at least one non-terminal prosodic break, so as to include a root unit plus at least one additional unit (prefix, suffix, auxiliary).

The prosodic annotation of C-Oral-Rom and the structure of information 55

(3)

*FAB: quella / è la cosa più bella / in assoluto // ‘that / is the most beautiful thing / of all //

(4)

*EST: avec du café toujours / bien sûr // ‘always with coffee / of course //’

(ifamcv12)

(ftelpv17)

Compound verbal utterances have been found to be the majority of romance speech utterances (Cresti 2005: 229 ).

4.

Prosodic annotation and functions of prosody

The classification of the various functions of prosody is one of the most important topics in research on prosody. Traditionally, a structural or distinctive function of prosody is separated from a so-called semantic (or modal) function and from an expressive function (Bolinger 1972, 1989; Crystal 1969; Crystal and Quirk 1964; Dane 1960; Ladd 1980; Rossi 1985, 1990). Prosody, therefore, carries out the following functions: a) segmentation of the phonic continuum into groups of words (structural function); b) expression of different modal contents (statements, questions, orders, etc.), regardless of their segmental content (“semantic” function); c) expression of emotions and moods (expressive function). 4.1. Prosodic parsing and structural function Traditionally, the need for prosodic annotation inside speech transcriptions is directly linked with prosody’s primary function, the so-called structural function. In fact, to include prosodic annotation in a transcription or not implies evaluating issues of semantic indeterminacy and syntactic structuring, that can emerge in the rendering of speech. The prosodic tagging used in C-ORAL-ROM highlights the effect of the main function of prosody – that is, the segmentation of a spoken chain – directly in the text. Prosodic annotation is not regarded, in this context, as a level of notation; rather, it is seen as a constructive part of the transcription, justified by the all-important role of prosody in oral production. The idea behind the use of tagging is that speech, being prosodically structured, is neither indefinite nor ambiguous, whereas a representation of speech which does not take into account its nature of “intoned linguistic production” may well result in uncertainty.5 As a matter of fact, the possibility of determin-

56 Antonietta Scarano ing the syntactic structure independently from the prosodic structure does not seem to reflect the true potential of spontaneous speech analysis. Let’s consider a few examples from the Table game conversation (ifamcv24), in which a group of kids plays a game on geography. One of them (GIO) can’t find the placeholder he had been using to mark his position on the map. The following is the bare transcription (sequence of items) of the verbal exchange, integrated by basic contextual information: (5)

*GIU: questo è il tuo ‘this is yours’ %act: GIU refers to the marker used by GIO to track his movements on the map *GIO: Giulia non urlare guarda il mio è qui ‘Giulia don’t shout look mine’s here’ %act: GIO has found his marker *AGO: uno due tre ‘one two three’ %act: counts the spaces along which to move

There are various possible groupings of the sequence in bold (“Giulia non urlare guarda il mio è qui”) in units pragmatically compatible with the circumstances: (6)

a. Giulia! non urlare. guarda. il mio? è qui. ‘Giulia! don’t shout. look. and mine? It’s here.’ b. Giulia! non urlare. guarda il mio. è qui. ‘Giulia! don’t shout. look at mine. it’s here.’ c. Giulia! non urlare, guarda. il mio è qui. ‘Giulia! don’t shout, look. mine is here.’ d. Giulia, non urlare. guarda il mio. è qui. ‘Giulia, don’t shout. look at mine. it’s here.’ e. Giulia, non urlare. guarda, il mio è qui. ‘Giulia, don’t shout. look, mine is here.’ f. Giulia, non urlare, guarda il mio. è qui. ‘Giulia, don’t shout, look at mine. it’s here.’ g. Giulia, non urlare, guarda. il mio è qui. ‘Giulia, don’t shout, look. mine is here.’

The prosodic annotation of C-Oral-Rom and the structure of information 57

h. Giulia, non urlare. guarda. il mio è qui. ‘Giulia, don’t shout. look. mine is here.’ The one actually performed is hypothesis e.: (7)

*GIU: questo è il tuo // ‘this one’s yours //’ *GIO: Giulia / non urlare // guarda / il mio è qui // ‘Giulia / don’t shout // look / mine is here //’ *AGO: uno / due / tre + ‘one / two / three +’

The unit is made up of two utterances (out of a possible five). The double slash, which highlights the terminal tonal unit ending each utterance (Karcevsky 1931), actually determines the boundaries of the syntactic relations domain, “interrupting” the structural relations between those elements of the sequence that don’t occur in the same domain. Every other possible grouping identifies different syntactic domains. The example in 7. shows that there is no syntactic information that can sufficiently identify a structure in units greater than the single word, as each sequence is also compatible with other structures. The information provided by the perceptive identification of terminal and non-terminal prosodic breaks makes it possible to select the appropriate units univocally. The syntactic criterion proves to be less objective and less usable than the prosodic criterion, as syntactic analysis can only be achieved on the basis of prosodic information. The fact that a non-prosodically transcribed text still remains interpretable does not mean that parsing is irrelevant: the interpreter of the transcription must still make guesses as to its parsing and therefore as to the grouping to be considered in the analysis. In conclusion, it is not a question of assessing whether the absence of prosodic notation lowers the degree of syntactic and/or semantic interpretability of a transcribed text, but of recognizing that prosody has a key role in defining a syntactic domain greater than the word. For this reason, prosody can but be regarded as a basic element of transcription. In other words, it is not a matter of establishing whether a transcription that lacks prosodic tagging determines or increases the indeterminacy of a sequence or makes its syntactic analysis uncertain, but of pointing out that not using prosodic parsing marker results in making something perfectly determined in speech underdetermined in transcription.

58 Antonietta Scarano 4.2.

From the structural function to the “semantic” function

4.2.1. Prosody and meanings Prosodic parsing has traditionally been considered an independent linguistic function, variously regarded in the different theoretical frames. The literature refers to entities highlighted by prosodic parsing, but these entities are defined by principles which stand at various levels of linguistic analysis (syntactic, semantic, textual or based on the merging of different criteria). The linguistic relevance of prosodic notation adopted in C-ORAL-ROM depends on the specific value attributed to the prosodic parsing within Cresti’s theory. Prosody is not regarded as an added level of execution with respect to the phonetic one; instead, it expresses a functional level which is proper of speech and acts as its primary structuring factor. The theoretical framework which the notation used in C-ORAL-ROM refers to, the Informational Patterning Theory, differs from the above in that it links the phonic continuum parsing function – regarded as primary in the literature – to a traditionally secondary function, that of conveying “modal” meanings. In this framework, this is achieved because of the key role attributed to illocution6, the bottom line being that the main output of prosody is, indeed, illocution.

4.2.2. Prosodic structure and informational structure: The Informational Patterning Theory Within the frame of analysis which constitutes the basis of C-ORAL-ROM speech representation, illocution is not regarded as one of the many meanings that can be expressed through variations of prosody; it is instead seen as “the content”, the expression of which determines the informational structure of utterances. Prosodic and informational structuring converge towards its expression. The definition of utterance proposed by Cresti is closely linked to the concept of speech act. Speaking is achieved through the performing of acts,7 which from a linguistics point of view correspond to utterances. A linguistic act is at the same time a locution (the product of a linguistic performance), an illocution (the attitude of the speaker towards the listener, their relationship) and a perlocution (the desired effect on the listener). Besides a method of recognition on the basis of pragmatic and semiological traits (Cresti 2000; Firenzuoli 2003), there is a formal criterion for

The prosodic annotation of C-Oral-Rom and the structure of information 59

identifying an expression which performs a linguistic act. According to Cresti – who bases her hypothesis from experiments on acoustic and linguistic data from various corpora – this is the evaluation of the prosodic traits of expressions. An evaluation that is at first perceptive, in keeping with IPO’s analysis guidelines, and thereafter an instrumental analysis of the acoustic parameters of sound, which is the formal criterion of recognition. Every linguistic system features a repertory of tonal patterns – and more specifically of root tonal units, the necessary nucleus of each pattern – conventionally encoded to express illocution. The criterion for the identification of utterances through recognition of patterns is known as illocutionary criterion: if an expression is intoned according to a root unit profile that fulfils certain requirements, it conveys a certain illocution. A given locutive content may produce different illocutions if it is adequately intoned. Different locutive contents, if intoned with the same root unit profile, can, obviously, produce the same illocution8. Within the reference theoretical frame, the prosodic pattern which identifies the utterance is generally isomorphic to an informational pattern which structures the utterance itself. The utterance, which corresponds to a linguistic act, a pragmatically interpretable entity, is therefore an informational pattern made up of informational units as read by prosody. It is built around a comment informational unit, which corresponds to a root informational unit, and which has, as its main function, the expression of the whole utterance’s illocution. Other groups of words may collect around the comment, beeing highlighted by various prosodic units of the pattern. This way, the tonal pattern expresses the informational articulation of the utterance. In this sense, then, the parsing function within the utterance is not independent from the illocutionary function, but is subordinated to it and built around it. Each informational unit features definite intonation, function and morphosyntactic characteristics. Only the comment unit is necessary and sufficient in building an utterance, as it expresses the illocution in itself. Informational units other than comment units carry out optional information functions, which are subordinated to the illocutionary function performed by comment units. The majority of utterances contains informational patterning between comment units and optional units. One of the main and most common types of informational patterning is the topic/ comment one. The identification and definition of the topic is based on the prosody (coincide with tonal unity prefix). It is clear that for this reason the perspective differs from both approaches syntactic/prosodic (Mereu 2004; Mereu and Trecci 2003; Frascarelli 2000) and classics functional approaches (Lambrecht 1994; Givòn 1983, 1995). The topic, in this framework, is the

60 Antonietta Scarano unit which opens the field for illocution, and sets the semantic premises for the locutive content of the comment. It is performed canonically, by means of a descending-ascending movement in the last syllables of the unit, and fast, with a suspended ascension, like an incomplete canonical movement (Cresti 2000, this volume; Signorini 2005; Signorini and Firenzuoli 2003): (8)

*LID: se lo prendo in collo / TOP non è che mi graffi // COM 9 (ifamdl02) ‘if I pick him up / he wouldn’t scratch me //’

(9)

*PER: à onze heure et demie du soir / TOP le dimanche soir / TOP forcément // COM (ffamcv09) ‘at half past eleven in the evening / on sunday evenings / no matter what //’

In light of what has been discussed above, the various types of utterance presented in paragraph 3.1 – simple verbal or verbless, compound verbal or verbless – can be linked, at this stage of our analysis, to their informational content. As we have seen, the recognition and the effectiveness of an utterance is formally ensured by a range of variously structured tonal pattern. This is not influenced by the presence of a verb phrase, and therefore not even by the syntactic sentence structuring. The prosodic pattern organises and performs informational contents, that guarantee the illocution expression. For example, the sequences shown in the column on the left in Table 1, extracted from the C-ORAL-ROM corpus, are all different in nature (sentences, clauses, adverbial, adjectival and prepositional phrases) but, from a prosodic point of view, they are all root units. Some of them, as shown in the column on the right, are the single and simple sequences corresponding to a pattern made up of just the root tonal unit ((10), (12), (14), (16), (19)); they are simple verbal ((14), (19)) and simple verbless ((10), (12), (16)) utterances. The others, besides having the same kind of grammatical structure, are, in their actual performance, part of complex sequences, articulated in a tonal pattern made up of two tonal units ((11), (13), (15), (17), (18)); they are part of compound verbal ((11), (17)) or verbless ((13), (15), (18)) utterances.

The prosodic annotation of C-Oral-Rom and the structure of information 61 Table 1. Root units and utterances root units

utterances

perfectamente // ‘perfectly //’

(10) *JUA: perfectamente // ‘perfectly //’

(emedts07)

está bien // ‘ok //’

(11) *PAT: superhorrible / está bien // ‘super horrible / ok //’

(efamcv01)

sympa // ‘funny //’

(12) *EST: sympa // ‘funny //’

(ffamcv11)

Londres // ‘London //’

(13) *GRA: e depois de Pequim / Londres // ‘and after Beijing / London //’

(pfamcv03)

vous avez pas de marques ? ‘you don’t keep labels?’

(14) *PAU: vous avez pas de marques ? ‘you don’t keep labels?’

(ffamdl13)

mh // ‘mh //’

(15) *NIC: perché io / mh // ‘because I / mh //’

(ifamdl03)

cinco ou seis pessoas lá // ‘five or six people there //’

(16) *RUI: cinco ou seis pessoas lá // ‘five or six people there //’

(pfamdl06)

tu t’ en lasses // ‘you go off them //’

(17) *CYR: si tu manges trop de pizzas / tu t’ en lasses // (ffamcv01) ‘if you eat too many pizzas / you go off them //’

più giovane // ‘younger //’

(18) *LIA: il figliolo / più giovane // ‘his son / younger //’

(ifamcv01)

porque sabe // ‘because he knows //’

(19) *JOS: porque sabe // ‘because he knows //’

(pfamcv03)

In examples (11), (13), (15), (17), (18), the first tonal unit has no nuclear profile, but is instead subordinate and temporally precedent (a prefix unit); the second is nuclear. In an experiment where the first unit is eliminated and the only unit made to listen to a competent speaker is the root, this could be interpreted as an independent linguistic act, in which something is approved, narrated, declared, concluded or stated about something else, the determination of which pertains to the context of enunciation instead of being linguistically expressed. On the contrary, if the omitted unit is the second, although the locution could still be understood, it could not be pragmatically interpreted on the basis of its intonation profile, and it wouldn’t be possible to associate any illocutionary force to it. This cannot depend on the semantic-syntactic incompleteness of the locution, as prefix units have a locutive content (clauses and phrases of various nature) similar to that of root units. Actually, what is prosodically complete corresponds to an informational unit whose function is to express the whole utterance’s illocution; what is prosodically incomplete has a secondary and subordinate informational function.

62 Antonietta Scarano Therefore, simple utterances consist of just a single comment informational unit, which is sufficient in conveying the illocutionary value of the whole utterance: (10’) *JUA: perfectamente //COM ‘perfectly //’

(emedts07)

(12’) *EST: sympa //COM ‘funny //’

(ffamcv11)

(14’) *PAU: vous avez pas de marques ? COM ‘you don’t keep labels?’

(ffamdl13)

(16’) *RUI: cinco ou seis pessoas lá //COM ‘five or six people there //’

(pfamdl06)

(19’) *JOS: porque sabe // COM ‘because he knows //’

(pfamcv 03)

Compound utterances are articulated from an informational point of view, and feature mark an informational relation between functionally distinct expressions: the comment unit is the core around which one or more secondary informational units are articulated. Examples (11’), (13’), (15’), (17’), (18’), are utterances which link the locutive content of the comment to locutions which represent the field of application of the illocutionary force and the semantic premise: these utterances are characterised by topic/comment informational articulation: (11’) *PAT: superhorrible /TOP está bien //COM ‘super horrible / ok //’

(efamcv 01)

(13’) *GRA: e depois de Pequim /TOP Londres //COM ‘and after Beijing / London //’

(pfamcv 03)

(15’) *NIC: perché io /TOP mh //COM ‘because I / mh //’

(ifamdl03)

(17’) *CYR: si tu manges trop de pizzas /TOP tu t’ en lasses //COM (ffamcv 01) ‘if you eat too many pizzas / you go off them //’ (18’) *LIA: il figliolo /TOP più giovane //COM ‘his son / younger //’

(ifamcv01)

The prosodic annotation of C-Oral-Rom and the structure of information 63

It is worth pointing out that the linearised content of compound utterances does not identify with the informationally articulated content. Example 18’. is especially clear in this respect. Its linear content (syntactically structured) would correspond to a noun phrase featured in comment (il figliolo più giovane) which wouldn’t be appropriate for the context in which the sequence is actually produced: (18’) a. *ELA: no &que ‘/’ queste persone / a quanto ho capito / però / erano abbastanza giovani // questi che son morti / // ‘no &th ‘/’ this people / as I know / instead / were rather young // this people that were died / that you talked about //’ ??*LIA: gli morì / ‘un so quant’ anni avrà avuto / una &cinqua + il figliolo più giovane //COM ‘well / her husband died / I don’t know how old was him / maybe he was &fifty + the youngest son //’ Its actual expression (informationally articulated), which perfectly fits the context, corresponds to an informational relation between a topic and a comment (the NP “il figliolo” is a prefix unit from an prosodic point of view, whereas from a semantic point of view it is the premise and the knowledge reference of the AP “più giovane” featured in the comment): (18’) b. *LIA: gli morì / ‘un so quant’ anni avrà avuto / una &cinqua + il figliolo /TOP più giovane //COM (ifamcv01) ‘well / her husband died / I don’t know how old was him / maybe he was &fifty + his son / younger //’ The topic and the comment may be integrated or corrected by additions, called in this framework appendices.10 From an informational point of view, appendices are the integration of a topic or comment unit, which they necessarily follow. They correspond to suffix tonal units. They have no intonational focus and they have a flat-descending movement (Cresti 2000, this volume; E. Tucci 2006): (20) *JOS: mas antigamente /TOP em que os alunos eram muito respeitoso / COM em relação ao professor ?APC (pfamdl 01) ‘in former times / schoolboys were more respectful / with regard to the teacher?’

64 Antonietta Scarano Dialogic auxiliary units can have incipit, phatic, allocutive or conative functions (Frosali 2005): (21) *VAL: donc /INP c’ est pas forcément en couple //COM ‘so / It’s not necessarily as a couple //’

(ffamcv01)

(22) *JES: nada /FAT es que tarda un poco más //COM ‘well / it is that he is a bit late //’

(efamcv09)

(23) *DAN: Daddo /ALL che fai ?COM ‘Daddo / what are you doing?’

(ifamcv15)

(24) *VAL: e dài /CON vèn via //COM ‘come on / don’t mind //’

(ifamcv27)

Other important informational units are speaker introducers and parentheses. The former are units that mark the fulfilment of a reported direct discourse, and also listings and other compound utterances forms (Cresti 2000; Giani 2005): (25) *CLA: poi arriva uno /TOP e mi fa /ILC birra //COM ‘and then a man appears / and says / beer //’ (26) *ARN: il dit /ILC eh bien /FAT tu n’ as plus rien ?COM ‘he says / well / don’t you have anything more?’

(ifammn03) (fnatpr03)

Parentheses are true metalinguistic auxiliaries, which are used to comment on the meaning of the very utterance in which they appear. They have an assertive-conclusive intonational profile, which is clearly distinguished in frequency from the rest of the utterance in terms of their F0 average value and, also, because of their variations in speed (Cresti 2000, this volume; Tucci 2002; Firenzuoli and Tucci 2003). (27) *GNA: cioè /INP la tua voce /TOP io dico /INX si sentiva //COM (ifamcv 02) ‘you see / your voice / I mean / we could hear it //’ (28) *MAR: mais ce qui m’ intéresse /TOP bien sûr /INX hein vous savez que je parle en me référant à Freud /INX &l ‘/’ la vieillesse c’ est quoi ?COM (fnatco 02) ‘but what I’m interested in / of course / well you know I’m basing this on Freud / what’s old age?’

The prosodic annotation of C-Oral-Rom and the structure of information 65

We believe that the idea of a structuring of prosody closely connected to informational patterning, which is what mainly and specifically characterises speech, and the recognition that prosodic structure is the formal identification index of structured informational contents, are the turning point beyond which a complete exploitation of oral resources can take place. 5. Informational complexity in spontaneous speech As we have observed, the reference unit used in C-ORAL-ROM is the utterance. The syntactic reading of utterances varies in accordance with the prosodic performance of the linguistic material. From a strictly syntactic point of view, the sequence Il mi’ bisnonno Pietro would be a noun phrase if considered outside its context. If analysed within the actual context: (29) *ELA: come si chiamava /COM il tu’ nonno?APC ‘what was the name / of your granddad?’ *LID: il / nonno mio /TOP si chiamava Gaetano // COM ‘my / granddad / was called Gaetano //’ *ELA: mh //COM ‘mh //’ *LID: il mi’ bisnonno /TOP Pietro //COM ‘my great-granddad / Pietro //’

(ifamdl02)

the prosodic reading and its pragmatic interpretation make it an utterance articulated in informational units, each of which consists of a noun phrase. Just one of these units (Pietro) can be interpreted by itself, and determines the illocutionary value of the whole utterance (it is a comment). A different prosodic performance would have determined a different syntactical and informational interpretation, either as a noun phrase (in a single utterance) or as two noun phrases (in two utterances): (Chi andò in guerra?) ‘who went to war?’ (30) Il mi’ bisnonno Pietro //COM ‘my great-granddad Pietro //’ SN

66 Antonietta Scarano (31) Il mi’ bisnonno //COM Pietro //COM ‘my great-granddad // Pietro //’ SN SN Those who learn the oral variety of a language must be able to verify that the complexity of speech is informational even more than syntactic; by listening, it is possible to identify nuclear blocks which can be utterances in themselves (as they can be pronounced alone and be pragmatically interpreted), like Pietro in our example, but can also be in variously related to other secondary informational blocks. Let’s consider the following utterance: (32) a. noi ci s’aveva già i biglietti //COM ‘we already had tickets //’ It is the informational unit (comment) of one of CORAL-ROM’s utterances. It is in itself a perfectly interpretable statement. Nevertheless, it is part of an actual utterance which is much more complicated from an information articulation point of view. The original example is extremely articulated around a nucleus, it can only be undertaken through the juxtaposition of melodically and informationally secondary units: (32) b. a fine marzo /TOP noi ci s’aveva già i biglietti //COM ‘at the end of March / we already had tickets //’ c. praticamente /TOP a fine marzo /TOP noi ci s’aveva già i biglietti //COM ‘actually / at the end of March / we already had tickets //’ d. niente /FAT praticamente /TOP a fine marzo /TOP noi ci s’aveva già i biglietti //COM ‘well / actually / at the end of March / we already had tickets //’ (32) *SAB: sicché /INP niente /FAT praticamente /TOP a fine marzo /TOP noi ci s’aveva già i biglietti //COM (ifamdl 09)11 ‘so / well / actually / at the end of March / we already had tickets //’ This example, though representative of a high degree of informational complexity, could in fact, because of the linguistic content of the comment (a sentence), make it seem as though the syntactic plot is always the primary agent, and make what we have interpreted as information articulation is as a sequence of adverbial phrases around an SVO-type core. Actually, there are many examples which account for syntactic “disconnectedness”. We

The prosodic annotation of C-Oral-Rom and the structure of information 67

must keep in mind that around 40% of utterances in all of the languages featured in C-ORAL-ROM contain no inflected verb forms and are therefore without an explicit SVO structure. The following example: (33) a. tutto pronto? COM ‘all ready?’ is a linguistic sequence which is perfectly interpretable as a question. It is, in fact, the informational nucleus of a more complex enunciative structure: an utterance whose content is syntactically de-structured (it is, in fact, what is normally considered an anacoluthon), although structured prosodically and informationally: (33) b. gli acidi / TOP tutto pronto? COM ‘the acids / all ready?’ (33) *CEC: di là / TOP gli acidi / TOP tutto pronto? COM ‘in there / the acids / all ready?’

(ifamdl17)

A critical view on the above concept would be to observe that, in truth, the comment content in 33. is still a sentence: nominal, verbless, but nonetheless structured in a Subject-Predicate form. This is not, however, the rule, as the production of a sentence – verbal or verbless – is not necessarily a consequence of the nature of a comment; some comments are very light in terms of their contents. Let’s consider two turns of the same dialogue from which the previous example was extracted: (34) a. *CEC: vuoi un biscottino?COM ‘want a biscuit?’ *NIC: sì //COM ‘yes //’ sì works perfectly as an answer in itself (it could be pronounced alone): it is the comment of the utterance which has the illocutionary value of an answer. In the original example (an entire clause) the linguistic content is much less structured compared to the informational unit which precedes it. (34) *NIC: se ce n’è ancora uno /TOP sì //COM (ifamdl17) ‘if there’s one left / yes //’

68 Antonietta Scarano The more structured material (se ce n’è ancora uno /TOP), appears to be secondary from an informational point of view; this is proven by the impossibility of interpreting it by itself, which is even more evident if the absolute autonomy of the comment (sì //COM) is taken into account. We could point out that it’s not a case of informational structuring as much as a question of weight of the context and, frequently, of the co-text. In the previous example, sì could be easily recognised as “the most significant” element of the utterance, as from a semantic point of view, it had the highest probability of being the answer to the question. In this perspective, then, identifying what is relevant for communication would be achieved on the basis of semantic coherence rather than on strictly informational premises. However, let’s look at another example: (35) a. *MAX: ma quant’ anni avevi / COM qui?APC che anno era?COM settanta ... COM ‘but how old were you / here? what was the year? seventy …’ *LIA: eh /INP &hr ‘/’ le nozze d’argento // COM ‘eh / our Silver wedding //’ as listening to the audio (ifamcv01) can easily confirm, the comment can be perfectly interpreted as an answer, even though this would require quite a strong conversational implication (“it was the year of my 25th wedding anniversary”). Nevertheless, in the original utterance, the verb – which from a strictly syntactical point of view would be responsible for the whole utterance’s structure – is an appendix informational unit, a secondary unit, a textual implementation which adds unnecessary information, and which, as we have seen, cannot be interpreted alone. The whole utterance is, in fact: (35) *LIA: eh /INP &hr ‘/’ le nozze d’argento /COM si fece //APC (ifamcv01) ‘eh / our Silver wedding / it was //’ Therefore it is not the case of syntactic complexity conveyed by a NP object moved before a VP; it is not necessary to postulate such movements in the syntactic structure, on the contrary, the utterance consists of a informational patterning made of a comment (which bares a NP) and of an appendix (which bares a VP).

The prosodic annotation of C-Oral-Rom and the structure of information 69

6. Conclusions Prosodic annotation in spoken language transcriptions is a means to convey speech as an “intoned production”. In the theoretical frame in which CORAL-ROM was conceived (isomorphism between prosodic structure and pragmatically based informational structure), prosodic parsing is not seen as just a performance, but as an index of formal recognition of an informational structure, making prosodic annotation – together with the access to audio – able to operate a pragmatic evaluation of speech. The degree of processing of oral data in C-ORAL-ROM, through the theoretical frame of reference, therefore enables a pragmatic analysis. The core of this pragmatic analysis are notions of illocution and informational patterning.

Notes 1. 2. 3. 4.

5. 6. 7. 8. 9.

10. 11.

See Llisterri (1996), Sinclair (1994), Payne (1992). On the concept of intentionality in this context, see Cresti (2000: 49). See, among others, Brazil (1995) and Pierrehumbert (1980). C-ORAL-ROM filenames provide details of three types: a) text language (f, for French; i, for Italian; e, for Spanish; p, for Portuguese); b) text type (e.g., fam indicates a text contained in the family/private part of the Informal section; dl identifies a dialogue, cv a conversation, mn a monologue; c) serial number, which labels each text within its section. For more details, see Moneglia (2005: 39–40). Moneglia (2004; 2005: 20 –24) proves that the syntactic structure of spontaneous speech is frequently underdetermined and that the index which determines the domain of syntactic structure is prosodic structure. Cresti (2000, 2003) and Tucci (2007) clearly set apart, redesigning their respective boundaries, modality from illocution. The direct theoretical reference is, of course, Austin (1962). For a summary and critical framework of the relationship between intonation, information structure and illocution in other perspectives, see Butler (2003), in particular the chapters I and II. Labels COM, TOP, APT, APC, INP, FAT, ALL, ILC, INX respectively refer to the comment, topic, topic appendix, comment appendix, incipit, phatic, allocutive and parenthesis informational units (see Cresti 2003). For a different interpretation of the appendix (as right topic), see Mereu (2004) and Lombardi Vallauri (2007 and this volume). Note that in this case the enrichment of the informational structure is contrary to the principle of linguistic progression of performance (the nuclear unit, necessary and sufficient in building an utterance, is at the end of the sequence of informational units).

70 Antonietta Scarano References Austin, John L. 1962 How to Do Things with Words. Oxford: Oxford University Press. Barry, William J. and Adrian J. Fourcin 1992 Levels of labelling. Computer Speech and Language 6: 1–14. Bolinger, Dwight. L. 1989 Intonation and its Uses: Melody in Grammar and Discourse. London: Edward Arnold. Bolinger, Dwight L. (ed.) 1972 Intonation: Selected Readings. Harmondsworth: Penguin. Brazil, David 1995 A Grammar of Speech. Oxford: Oxford University Press. Butler, Christopher S. 2003 Structure and Function: A Guide to Three Major Structural-Functional Theories, Part 2: From Clause to Discourse and Beyond. Amsterdam /Philadelphia: John Benjamins. Cresti, Emanuela 2000 Corpus di Italiano Parlato, Vol. 1. Firenze: Accademia della Crusca. 2003 Modalité et illocution dans le topic et le comment. In Macro-syntaxe et Pragmatique: L’analyse Linguistique de l’Oral, Antonietta Scarano (ed.), 133–182. Roma: Bulzoni. 2005 Notes on lexical strategy, structural strategies and surface clause indexes in the C-ORAL-ROM spoken corpora. In C-ORAL-ROM: Integrated Reference Corpora for Spoken Romance Languages, Emanuela Cresti and Massimo Moneglia (eds), 209–256. Amsterdam/Philadelphia: John Benjamins. this vol. The use of clitics in the information units: measurements and relations: Measurements and relations. Cresti, Emanuela, Massimo Moneglia, Fernanda Bacelar, Antonio M. Sandoval, Jean Veronis, Philippe Martin, Kalid Choucri, Valerie Mapelli, Daniele Falavigna, and Antonio J. Cid 2002 The C-ORAL-ROM Project: New methods for spoken language archives in a multilingual romance corpus. In Proceedings of LREC 2002, Vol. 1, Manuel Gonzáles Rodriguez and Carmen Suárez Araujo (eds.), 2–10. Paris: ELRA. Cresti, Emanuela and Massimo Moneglia 2005 C-ORAL-ROM. Reference Corpora for Spoken Romance Languages. Amsterdam/Philadelphia: John Benjamins. Crystal, David 1969 Prosodic Systems and Intonation in English. Cambridge: Cambridge University Press.

The prosodic annotation of C-Oral-Rom and the structure of information 71 Crystal, David and Randolph Quirk 1964 Systems of Prosodic and Paralinguistic Features in English. The Hague: Mouton. Dane, Frantisek 1960 Sentence intonation from a functional point of view. Word 16: 34–55. Firenzuoli, Valentina 2003 Le forme intonative di valore illocutivo dell’italiano parlato: Analisi sperimentale di un corpus di parlato spontaneo (LABLITA). Ph.D. diss., Dipartimento di Linguistica, Università degli Studi di Firenze. Firenzuoli, Valentina and Sabrina Signorini 2003 L’unità informativa di Topic: Correlati intonativi. In La Coarticolazione. Atti delle XIII Giornate di Studio del Gruppo di Fonetica Sperimentale (GFS 2002), Giovanna Marotta and Nadia Nocchi (eds.), 177–184. Pisa: Edizioni ETS. Firenzuoli, Valentina and Ida Tucci 2003 L’unità informativa di inciso: Correlati intonativi. In La Coarticolazione. Atti delle XIII Giornate di Studio del Gruppo di Fonetica Sperimentale (GFS 2002), Giovanna Marotta and Nadia Nocchi (eds.), 185–192. Pisa: Edizioni ETS. Frascarelli, Mara 2000 The syntax-phonology interface in Focus and Topic constructions in Italian. Boston: Kluwer. Frosali, Fabrizio 2005 Le unità di informazione di ausilio dialogico: Valori percentuali, caratteri intonativi, lessicali e morfosintattici in un corpus di italiano parlato (C-ORAL-ROM). Degree thesis, Dipartimento di Italianistica, Università degli Studi di Firenze. Giani, Daniela 2005 Il discorso riportato nell’italiano parlato e letterario: Confronto tra due corpora. Ph.D. diss., Dipartimento di Linguistica, Università degli Studi di Firenze. Givòn, Talmy 1983 Topic continuity in discourse: an introduction. In Topic Continuity in Discourse, Givòn, Talmy (ed), 3–41. Amsterdam /Philadelphia: John Benjamins. 1995 Functionalism and Grammar. Amsterdam /Philadelphia: John Benjamins. ‘t Hart, Johan and René Collier 1975 Integrating different levels of intonation analysis. Journal of Phonetics 3: 235–255. ‘t Hart, Johan, René Collier, and Antonie Cohen 1990 A Perceptual Study on Intonation: An Experimental Approach to Speech Melody. Cambridge: Cambridge University Press.

72 Antonietta Scarano Karcevsky, Serge 1931 Sur la phonologie de la phrase. Travaux du Cercle linguistique de Prague 4: 188–228. Ladd, D. Robert 1980 The Structure of the Intonational Meaning. London: Bloomington. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press. Llisterri, Joaquim 1996 EAGLES: Preliminary Recommendations on Spoken Texts. EAGTCWG-SPT/P (www.ilc.cnr.it/EAGLES96/spokentx/spokentx.html). Lombardi Vallari, Edoardo 2007 «Appendice» una categoria informazionale o semantica? In Lessico, grammatica, testualità, Anna-Maria De Cesare and Angela Ferrari (eds.), 75–95. Basilea Acta Romanica Basiliensa 18: Università di Basilea. this vol. “Appendix” or “postposed Topic”: Where does the difference lie? MacWhinney, Brian 1995 The CHILDES Project: Tools for Analyzing Talk. Hillsdale, New Jersey: Lawrence Erlbaum. Mereu, Lunella 2004 La sintassi delle lingue del mondo. Bari/Roma: Laterza. Mereu, Lunella and Alesssandra Trecci 2003 Focus sul topic. In Atti del Convegno Nazionale “Il Parlato Italiano”, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.), CD-ROM. Napoli: M. D’Auria. Moneglia, Massimo 2004 L’italiano come risorsa romanza nel corpus multilingue C-ORALROM. In Atti del Convegno Nazionale “Il Parlato Italiano”. CDROM, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.), 1–23. Napoli: M. D’Auria. 2005 The C-ORAL-ROM Resource. In C-ORAL-ROM: Integrated Reference Corpora for Spoken Romance Language, Emanuela Cresti and Massimo Moneglia (eds.), 1–70. Amsterdam /Philadelphia: John Benjamins. Moneglia, Massimo and Emanuela Cresti 1997 Intonazione e criteri di trascrizione del parlato. In Il Progetto CHILDES-ITALIA: Contributi di Ricerca sulla Lingua Italiana, Umberta Bortolini and Elen Pizzuto (eds.), 57–90. Pisa: Del Cerro. Payne, Jonathan 1992 Report on the Compatibility of J P French’s Spoken Corpus Transcription Conventions with the TEI Guidelines for Transcription of

The prosodic annotation of C-Oral-Rom and the structure of information 73 Spoken Texts. Working Paper, COBUILD Birmingham and IDS Mannheim. NERC-WP8/WP4-122 . Pierrehumbert, Janet B. 1980 The Phonology and Phonetics of English Intonation. Ph.D. diss., MIT. Rossi, Mario 1985 L’intonation et l’organisation de l’énoncé. Phonetica 42: 135–153. 1990 Ordre, organisation et intonation. In AA.VV., Scritti in Onore di Lucio Croatto: 211–226. Padova: Centro Studio per le Ricerche di Fonetica. Signorini, Sabrina 2005 Topic e soggetto in corpora di italiano parlato spontaneo. Ph.D. diss., Dipartimento di Linguistica, Università degli Studi di Firenze. Sinclair, John M. 1994 Spoken language ‘3B’, Phonetic/phonemic and prosodic annotation ‘5.2’. NERC-(1994) NERC-1 Network of European Reference Corpora, Final Report. Pisa: Istituto di Linguistica Computazionale – CNR. Sperberg-McQueen, Michael C., and Lou Burnard (eds.) 1994 TEI P3. Guidelines for Electronic Text Encoding and Interchange. Chicago /Oxford: Association for Literary and Linguistic Computing. Tucci, Elena 2006 L’unità informativa di appendice in un corpus di italiano parlato (CORAL-ROM): Caratteristiche intonative, semantiche e morfosintattiche. Degree thesis, Dipartimento di Italianistica, Università degli Studi di Firenze. Tucci, Ida 2002 Caratteristiche sintattiche e frequenza degli incisi in un corpus di parlato spontaneo. Degree thesis, Dipartimento di Italianistica, Università degli Studi di Firenze. 2007 L’espressione lessicale della modalità nel parlato spontaneo. Analisi del corpus C-ORAL-ROM italiano. Ph.D. diss., Dipartimento di Linguistica, Università degli Studi di Firenze.

Universals of information structure Lunella Mereu

1. Introduction Although a well-defined concept since the 1930s of the past century, information structure is a topic to which much discussion has been devoted recently in syntactic studies. The reason for the interest is due to the attention given to syntactic variation in typological studies on the languages of the world. In these works it has often emerged that the sources of variation are mainly to be looked for in the different strategies of information structure that the languages adopt. In other words the different grammatical behaviour shown cross-linguistically is very often due to the ways in which the theme – rheme, and/or topic – focus, articulation within the sentence is realized. The aim of this paper is therefore to discuss universal principles of information structure under the hypothesis that these can only be found in the interaction between the morpho-syntactic organization of languages and the discourse grammar properties according to which utterances are built in a language. Though this issue is an old one in typology, and a number of proposals are already available, we think that we have not yet found an accurate cross-linguistic characterization of information structure. To reach this aim, the interplay between structural and pragmatic information needs to be characterized through a serious cross-linguistic survey of typologically different languages and taking into account data from spoken language. We will first discuss language-specific ways of building sentence structure. This will lead us to take into account the range of syntactic and morphological variation, that is the different morpho-syntactic coding languages allow. We will then proceed to single out the kinds of pragmatic principles associated with the organization of utterances in the different languages. From this it will emerge that basically languages are either syntactically or pragmatically organized, in other words they either belong to those in which syntactic order shows a correspondence between the grammatical relations and the given – new information structure sequence, or to those in which there is no correspondence in that utterances are simply sequences of pragmatically [+ salient] – [– salient ] information.1

76 Lunella Mereu 2. Structural diversity among the languages of the world Let us start from the typological assumption according to which structural diversity is an important property of languages; this means that we must assume that all languages do not have the same structure, even at deep level, as in generative grammar. Leaving apart the phonological organization, occurrences of diversity in the grammar of languages are at least the following: (1)

syntactic order; morphological markers/particles/adpositions; grammatically relevant suprasegmental features; zero anaphora or pro-drop; pronominal markers; discontinuous sequences2; constituent structure; function changing phenomena3; dislocation, fronting and extraposition phenomena; binding phenomena

The list in (1) already shows how the grammatical organization of a language has to be viewed as an interface problem, that is as due to the interplay of at least three levels, syntax, morphology and prosody. Is this all? To answer this question we need to share the typologists’ characterization of languages as entities ‘tailored’ to favour communicative efficiency among speakers. This means that a language is built to facilitate communication among its speakers or, more specifically, in Van Valin’s and LaPolla’s words: “In order to decrease the chance of misunderstanding, the speaker, in creating the sentence, tailors the form of the sentence to allow the hearer to create the proper context for interpretation with minimal processing effort. For his part, the hearer assumes that the sentence will be tailored in just this way…” (Van Valin and La Polla, 1997: 198). This leads us to consider primarily the way in which the information in the sentence is delivered, what has been called by the linguists of the Prague school “the functional perspective of the language” and since the 1960s its “information structure” (Halliday 1967). A number of slightly different pairs of notions, proposed by different linguists and linguistic schools, are relevant to characterize sentences in these terms:

Universals of information structure

(2)

Topic Theme Given Topic Presupposition Background Presupposition

77

Comment Rheme New Focus Focus Foreground Assertion

The pairs in (2) show the semantico-pragmatic distinctions according to which the constituents of a sentence, sometimes sub-portions of it, are linearly organized to allow communication. They do not overlap, rather they emphasize different and complementary aspects of the pragmatic organization of a sentence4. We will mainly deal with the topic – focus pair as the literature distinguishes between topic- and focus-prominent languages as generally they grammaticalize either topic as opposed to comment information or focus as opposed to topic5 (Li and Thompson 1976; Abraham and De Mey 1986; Kiss 1995a). To give an example of how a language can be pragmatically organized, let us consider two types of sentences in Italian, the first a sentence with a predicate requiring two arguments (a SVO structure) , the second one with a mono-argumental predicate (a VS structure): (3) Syntactic order: Subj – verb – (Obj) Semantic structure: Agent – predicate – (Patient or other sem. roles) Information structure: (Given) – New (Topic) – Focus (4) Syntactic order: Verb –Subj Semantic structure: Pred. – Theme or other semantic roles Information structure: (Topic) – Focus (3) and (4) show the correspondences between the syntactic, semantic and informational structures of syntactically and pragmatically neutral sentences such as (5) and (6) respectively: (5) Gianni ha colpito il ragazzo G. has hit the boy ‘Gianni hit the boy’

78 Lunella Mereu (6)

E’ partito Gianni Is left G. ‘Gianni left’

(5) is neutral in that it is characterized either as carrying all new information, in other words as being broad focus, or as carrying given information on the subject, generally an agent, and new information on the predicate. In monoargumental or intransitive sentences such as (6) in Italian, the interplay among the three levels of structure is generally given by the reverse sequence6. Given the behaviour above, we can say that Italian sentences are built according to the pragmatic principle of progression from given to new information: (7)

Principle of progression from given to new information: “More indefinite/discontinuous/new information follows more definite/ continuous/old information” (Givón 1984: 207)

(7) is a well known principle in the literature, as quite a number of languages adopt it. But can we say that all languages share this principle? We can split this question into the following two: a) do all languages obey the same principles ruling the distribution of pragmatic functions such as topic and focus in an utterance? b) in what relationship are these principles with the different morpho-syntactic structures realized by languages? Certainly we have already received some significant answers to question (a) and (b) but these vary according to the theoretical frameworks adopted. Let us start by examining syntactic order in languages other than Italian, and see how this correlates with pragmatic principles of sentence construction.

3. Syntactic order in the literature Basically there are two main ways of viewing syntactic order variation in linguistics, that is I) the typological, and II) the generative. Typology distinguishes between OV and VO types of languages (Greenberg 1963; Comrie 1981), and it also admits the presence of a free or flexible, word order type which, as Givón (1984) points out, may not be

Universals of information structure

79

governed by the principle in (7). Notice that, if we recognize the existence of free word order languages, we must characterize order phenomena not only at the level of constituent structure but also at word level and hypothesize that some languages do not have constituent structure, that is a hierarchical organization of the sentence. We will come back to the typological approach to syntactic order later on in the text. Let us now consider how generative grammar deals with this phenomenon. First of all we need to consider two approaches. The first completely denies the relevance of syntactic order variation in the languages of the world, as it is assumed that all languages share the same configuration at deep structure, that is they are all SVO (Kayne 1994). In the minimalist programme (Chomsky 1995) this is called the ‘cartographic’ approach 7 (Belletti 2004; Cinque 2002; Rizzi 1997, 2004). This means that the syntactic representation allows us to identify the grammatical relations underlying a sentence and that we can interpret the sentence in the right way, even if something is not in the right position at surface level. In case a subject or an object is not in the canonical position it should have according to the SVO configuration, its surface realization is derived through movement. There is a second approach which was introduced by Hale in the 80s (Hale 1981, 1983), but which is no longer practiced nowadays. This was based on a macro-distinction between two types of languages, a configurational type, and a non-configurational one. The first type can be dealt with according to the cartographic approach sketched above, the second according to a non-cartographic approach or in terms of syntactic trees which represent flat structures with no hierarchical positions for subject, object and verb. We think that this distinction is interesting even outside the generative approach, and that to detect universals of information structure it is useful to analyze the relationship between the syntax and pragmatics of this kind of languages. The best example of a non-configurational language is Warlpiri, an Australian language which has been broadly analysed by a number of generative linguists (Hale 1981, Jelinek 1984, among the others). As the following sentence shows: (8)

a. Ngarrka-ngku ka wawirri panti-rni man-ERG AUX kangaroo hit-NONPST ‘The man is hitting a kangaroo’ b. Wawirri ka panti-rni ngarrka-ngku c. Panti-rni ka ngarrka-ngku wawirri d. Panti-rni ka wawirri ngarrka-ngku

(SOV) (OVS) (VSO) (VOS)

80 Lunella Mereu e. Wawirri ka ngarrka-ngku panti-rni f. Ngarrka-ngku ka panti-rni wawirri

(OSV) (SVO)

this is a free word order language, as all six possibilities of order are allowed, with the only restriction being the occurrence of the auxiliary in second position. Zero anaphora, discontinuous constituents and presence of a number of morphological markers both on the verb and on the nominals, all properties which Hale associates with non-configurational languages, are present in Warlpiri. Therefore Hale (1983), Jelinek (1984, 1989) as well as Baker (1996) adopt the traditional hypothesis, as old as in Meillet (1903) for classical languages, according to which the freedom of words in a language such as Warlpiri is related to its morphological richness. If this hypothesis is correct then, the relevant distinction between the languages of the world is between syntactically oriented languages, the fixed word order type, and morphologically oriented ones, the flexible order type (Baker 1996). Now the important question is again: (9)

Must we assume a Given – New structure for non configurational languages?

Let us take into account the typologists’ view of free word order languages, in particular Mithun’s (1987) proposal, according to which the distinction between the languages of the world is between ‘syntactically based’ languages, which is the fixed word order type, and ‘pragmatically based’ languages, the flexible order type8. The latter is only guided by a principle of ‘newsworthiness’, or, again in Givón’s (1984) terms, by the pragmatic principle in (10): (10) New/more unpredictable/more unexpected information precedes given/ more predictable information If the typologists’ hypotheses are correct, morphology does not need to be the universal condition to freedom of word order in the languages of the world, at least not the only one, in that it should cooperate with the pragmatic principle in (10). It is useful now to take into account the linguistic resources languages adopt to realize information structure.

Universals of information structure

81

4. How languages mark information structure As far as topicalization strategies are concerned, we can say that these include at least the following: (11) a. topic markers (Japanese); b. specific positions (Chinese); c. dislocations and/or resumptive pronouns (Italian) As for focalization strategies, these are well described in the literature and can be summarized drawing from Dik et al. (1981): (12) a. phonological 

– intonational prominence (primary accent or pitch on focus) b. morpho-syntactic  – specific positions (Hungarian); – focus markers which separate focus from the other components of the sentence (Somali); – specific costructions such as cleft sentences (Italian)

Let us give a few examples of the applications of information structure strategies in some languages. In Hungarian, for example: (13) Évát János várta a mozi el tt Eve-ACC John waited the cinema in-front-of ‘Eve was waited for in front of the movie by JOHN’ (Kiss 1995b, (2a): 208) the focused constituent is marked by the immediately pre-verbal position, while the topic is signalled by the position before the focused element. The situation is similar in Somali (Puglielli 1981): (14) a. Cali moos buu (baa+uu) cunay Cali banana FOCM+SUBJM ate ‘(As for) Cali, he ate a BANANA’ b. moos buu cunay Cali ‘(As for) Cali, he ate a BANANA’ In this language the focus is always marked by the pre-verbal position but also by the focus marker baa right-adjacent to the focused element; to signal

82 Lunella Mereu the topic, a subject marker is attached to the focus marker as a resumptive pronoun whenever the topic is in the position before the focused element (14a) or in the post-verbal position (14b). More interesting are the data from Mithun (1987) who analyses three typologically different languages. One of these is Cayuga, belonging to the Iroquian group, which according to the question-answer test to detect focus information behaves as in (15) and (16): (15) S: sne: ‘? who you.two.go.FUT ‘Who are you going with?’ Sám ya: khne: ‘ Sám we.two.go.FUT ‘I am going with SAM’

(Mithun 1987, (66))

(16) T ’h  ’te’ a:y : ‘ ihse: a: shni:no’? what seem you.think you.buy.COND ‘What do you think you would like to buy?’ O:, akya’tawi’thrá’ ki’ a: y : ‘ kihsa:s a:khní:n oh, dress just seem I.try I.buy.COND ‘Well, I am looking for a DRESS’ (Mithun 1987, (67)) To answer both questions in Cayuga, the initial position is used for the focused constituent confirming the newsworthiness principle Mithun hypothesizes for pragmatically based languages, a principle which we know is similar to the one in (10). Mithun associates the property of being morphologically rich with her groups of pragmatically based languages, as they all share this property, being all polysynthetic. Therefore both for Mithun and for generativists such as Hale and Baker morphological richness is the property allowing freedom of word order or surface non-configurationality in the languages of the world. However morphological richness is not shared universally. In particular, an important language which can be considered pragmatically based and which has no inflectional morphology at all is Chinese. We will say more about Chinese below, but before concentrating on this language, let us summarize schematically the different kinds of morpho-syntactic coding that the languages considered up to now show, maintaining the macrodistinction between a configurational and a non-configurational type for convenience (Mereu 2004):

Universals of information structure

(17)

83

hierarchical relations configurational precedence relations

morphosyntactic coding non-configurational

case marking agreement marking case+agreement marking  marking 9

The scheme in (17) tells us that morphological richness is not a universal property of non configurational languages, as there are languages of this type which are morphologically -marked. Therefore we need to have a deeper look at the behaviour of nonconfigurational languages to see what they really have in common. As has emerged in some recent work on information structure (Mereu 2004; Mereu and Frascarelli 2006), non-configurationality can be associated with the following groups of languages and/or phenomena: (18) a. ‘topic-prominent’ languages (Li and Thompson 1976, 1981); b. pragmatically oriented languages (Mithun 1987)10; c. syntactically-marked phenomena in configurational languages (dislocations, constituent fronting) Let us characterize now the two kinds of languages in (18a,b) and the phenomena in (18c). From their common properties a new view of non-configurationality will emerge and finally two universal principles of information structure will be proposed, the first matching configurational languages and the second non-configurational languages or phenomena. 5.

For a new look at non-configurationality

5.1. Topic-prominent languages In their interesting work in the 1980s Li and Thompson (1976, 1981) distinguish between subject-prominent and topic-prominent languages on the basis of the syntactic properties of Mandarin Chinese. While subject-prominent languages build the sentence on the basis of the subject – predicate sequence, topic-prominent languages build it on the basis of the topic – comment sequence. Chinese is topic-prominent because it obligatorily realizes the topic in initial position, as the following example shows:

84 Lunella Mereu (19) Nèi-chang hu xìngku xi ofang-duì lái de kuài that-CLASS fire fortunate fire-brigade come CSC quick ‘That fire, fortunately the fire-brigade came quickly’ (Li and Thompson 1981, (34): 96) In (19) the topic is neither the subject, nor any other argument in the sentence; it is only semantically related to the subject, realizing what in many languages is called ‘hanging topic’; of course subject and topic may coincide as in (20). (20) Zh ngs n d w le Z. hit I LE ‘Zhngsn hit me’

(Li and Thompson 1981, (6): 87)

But all sorts of elements can fill the first position in the sentence, for example an oblique, as in (21), or even a complement clause, as in (22): (21) zài Tàibi kéyi ch de hn ho at Taipei can eat DE very good ‘(In) Taipei one can eat really well’ (Li and Thompson 1981, (30): 95) (22) Zh ngs n mìngti n qù Miguó w juéde hn qìguài Z. tomorrow go America I feel very strange ‘Zhangsan’s going to the United States tomorrow, I feel is very strange’ (Li and Thompson 1981, (49): 99) As Li and Thompson (1976) point out, any sentence in Chinese is interpretable as the following syntactically-marked sentence in English: (23) As for education, John prefers Bertrand Russell’s ideas TOPIC

COMMENT

This was a very important insight in the early 1980s as it was the first hint at a distinction between syntactically oriented and pragmatically oriented languages. However it did not tell the whole story about pragmatically oriented languages. Actually, if we think of the latter in terms of Givón’s pragmatic principle in (10) or of Mithun’s (1987) newsworthiness principle, we should think of pragmatically oriented languages as focus-prominent, rather than topic-prominent. Our idea instead is that this kind of languages can be both topic- and focus-prominent.11

Universals of information structure

85

Let us consider a few more data from Chinese. In Li and Thompson’s book, we find the following question-answer pair: (24) A: t gn shéi niàn-sh? 3SG with who study-book ‘Who does s/he study with?’ B: Gn Lsì niàn-sh With Lisi study-book ‘S/he studies with Lisi’

(Li and Thompson 1981, (156): 558)

(24) shows that even the focused constituent can occur in initial position, although this is not its canonical position. As a matter of fact, focus can occur both in pre-verbal or post-verbal position, though with a difference in meaning: (25) a. t gngkè yjing zuò-wán le 3SG homework already do-finish LE ‘S/he’s already done his/her homework’ (Li and Thompson 1981, (235): 161) b. t yjing zuò-wán gngkè le ‘S/he’s already finished his/her homework’ (Li and Thompson 1981, (236): 161) In Li and Thompson’s words (25a) is used when the event is described as being “contrary to the expectation”, while (25b) realizes “neutral meaning”.12 Now, if we consider the following example: (26) yfu xn de ho; péngyóu jiù de ho Clothes new NOM(inaliser) good friend old NOM good ‘Clothes, new ones are good; friends, old ones are good’ (Li and Thompson 1981, (55): 101) in which we have a contrastive topic in first position and go back to (24), in which we have a narrow focus, we can conclude that Chinese is a pragmatically oriented language not exactly in the sense of the principle in (10), but rather in the following sense: (10’) Principle of prominence or pragmatic salience:13 Pragmatically salient information precedes pragmatically non salient information

86 Lunella Mereu Therefore we can classify Chinese as being a non-configurational language which, being Ø-marked, is not subject to any morphological constraint. This does not mean that the language has no syntactic order. Actually very often there is a correspondence between the SVO sequence and the topic – comment structure, but this is not an obligatory requirement. What is more important about Chinese is that it behaves according to pragmatic constraints and that it undergoes the principle which we have proposed in (10’).

5.2. Pragmatically oriented languages We hypothesize that all non-configurational languages obey the principle in (10’). One interesting datum is provided by Warlpiri again, although there are no specific studies on the information structure of Warlpiri utterances. The datum concerns the initial position, that is the one before the auxiliary where either a single word or a whole constituent other than the VP can occur. In one of his papers Hale (1992) pointed out that this position is the locus of the focused element; although much work needs to be done to characterize how topic and focus are marked in Warlpiri, this confirms our hypothesis about the relationship between non-configurationality and the pragmatic orientation of languages with free word order. In this perspective we could go on to analyze Hungarian or Somali too, which we know mark both topic and focus14, but rather than dealing with these languages, we prefer to concentrate on analyzing syntactically-marked phenomena in configurational languages such as Italian.

5.3. Syntactically-marked phenomena in configurational languages With the term ‘syntactically-marked phenomena’, which we take from Benincà, Salvi and Frison (1988), we refer to all those sentences in which the constituents do not occur in their canonical positions to express specific pragmatic meanings (Benincà, Salvi and Frison 1988: 115–116). We are referring to cases which in the syntactic literature are called left or right dislocations, constituent fronting or wh-questions. We consider these phenomena in Italian, a language in which they present interesting characteristics. This will allow us to compare Chinese and Italian and to single out that both Chinese and the syntactically-marked phenomena in Italian obey the principle in (10’). The comparison is inter-

Universals of information structure

87

esting because the two languages are both typologically and genetically different and one, Chinese, is pragmatically oriented while the other, Italian, is syntactically oriented. There is no doubt that we can group syntactically-marked phenomena in Italian in the following ways: (27) left dislocations right dislocations hanging topic fronting cleft sentences wh- questions

topic realizations

focus realizations

Now let us consider data from spoken language, focusing both on the morpho-syntactic and prosodic properties of topic and focus realizations in Italian. This interface perspective is very important as it allows us to discover very relevant properties of these phenomena which have been unnoticed up to very recently. We therefore discuss very briefly the results of our research work on spoken Italian in the interplay between syntax, phonology and pragmatics, based on a corpus of semi-spontaneous informal conversations recorded in the Roman area. (Mereu 2004, 2005, in press; Bonvino 2004, 2005; Frascarelli 2004). The specific phenomena we describe now are cases of left and right dislocations and of fronting of constituents. We know that we have an occurrence of a topic on the left in Italian when a constituent other than the subject occurs in initial position. An example is given by the following utterance: 15 (28) Francese l’ho fatto alle medie per tre anni16 French (I) studied it at school for three years The object in (28) is in initial position and it is connected to the resumptive pronoun, the clitic form lo left-adjacent to the verb; this is obligatory when the object is left-dislocated. The occurrence of the resumptive pronoun is further evidence that the object is given information, therefore a topic. As for an instance of right dislocation or of a right topic, we can consider an utterance in which a subject is right-dislocated or occurring in post-verbal position:

88 Lunella Mereu (29) non è male il palazzo ti dirò (It) is not bad the building, I will tell you ‘The building is not bad, I tell you’ In Italian the subject can be post-verbal in two cases, first when it is monoargumental, as in (6) above, and in that case it is focused, and second when it comes after the whole predicate or VP, or other material too, and in that case it is a topic as in (29).17 As for focus manifestations we have fronting whenever a constituent other than the subject occurs in initial position as in the following example: (30) LI’ ho preso degli appunti, ma sugli altri tre no. There (I) have taken notes, but on the other three (I did) not ‘I took notes THERE, but on the other three I didn’t’ in which the adverb, which is a circumstantial normally occurring after the verb, is in initial position. The other property which is connected with focus is that the fronted constituent cannot be doubled by a resumptive pronoun. In other words we could never have the locative clitic vi (30’) in Italian: (30’) *LI’ vi ho preso degli appunti, ma sugli altri tre no In (30) we have an instance of a contrastive focus, that is information which is new in that it corresponds to an element selected out of a set. It is just because of its contrastive interpretation that the focus is fronted in Italian, otherwise it would occur ‘in situ’. Now, if we go back to the cross-linguistic resources languages adopt to realize information structure summarized in (11) and (12), we would expect that, as far as prosodic marking of topic and focus is concerned, focus should be marked by intonational prominence, that is by a primary accent or pitch, while there should be no pitch accent on the topic.18 Interestingly, however, prosodic marking on our Italian data does not seem to distinguish between topic and focus but between initial and postverbal position of the pragmatic functions. Let us see now the intonational properties of some of the utterances from our corpus and see where the pitch accent falls in each utterance. 5.3.i. Prosodic analysis of topic and focus in Italian We now discuss the intonational contours of the utterances manifesting topic and focus occurrences. The feature we concentrate on to detect prosodic marking of these pragmatic functions is the height of the f0.

Universals of information structure

89

If we go back to the utterance in (28), in which an object is left-dislocated to mark the topic, and consider its intonational contour: (31) Francese l’ho fatto alle medie per tre anni French I studied it at school for three years

we notice the rising of the f0 on the tonic syllable of the topic in (31), while what comes after, the focus, is not marked by any rising, rather it has a flat contour. Let us consider now a case of a right-dislocated topic, that is the utterance in (29) whose prosodic representation is the following: (32) non è male il palazzo ti dirò (It) is not bad the building, I will tell you ‘The building is not bad, I tell you’

In this case we have a flat contour on the post-verbal topic.

90 Lunella Mereu Now let us consider the following utterance: (33) la strada che porta all’aeroporto so riuscita a beccarla la Roma Fiumicino (AUR04) the street that takes to the airport, (I) have managed to take it, the Roma Fiumicino in which the same topic is both right- and left-dislocated, though through different but pseudo-synonymous expressions. Again we notice that only the left topic is marked by prosodic prominence:19 (34)

Interestingly, as our data show, there are also cases of contrastive topics20 as in (35): (35) in inglese ho avuto sempre problemi con i professori In English I have always had problems with professors

and again we have a peak on the tonic syllable of the topic.

Universals of information structure

91

To interpret these data we need to consider the semantic difference between left and right topics. We think that the prosodic behaviour of the two types of topics confirms the hypotheses by Bossong (1981), Berruto (1986) and Simone (1997) on their different functions. The topic in initial position, rather than expressing an afterthought (Givón 1976), denotes pragmatic salience of the theme of the utterance, while the topic in final or post-verbal position is a mechanism to make salient what occurs on its left. Now let us consider the prosodic behaviour of focus in Italian. Drawing from Frascarelli (2004), we notice that focus is prosodically prominent only when it is in initial position, as (36) shows: (36) LI’ ho preso degli appunti, ma sugli altri tre no. There (I) have taken notes, but on the other three (I did’n)t 400

300

200

100 db 40 20

in which case it is an instance of a contrastive focus, while, when it is in situ, there is no rising of the f0: (37) “eh comunque lo vedi sulla delibera dell’AIPA.” Oh however (you) (can) see it on the decision of the AIPA

92 Lunella Mereu 300

200

100 db 40 20

The only case in which the post-verbal focus is prosodically marked is when it is an instance of narrow focus, therefore a contrastive kind of focus again: (38) “io cercavo di applicare le VECCHIE regole alla nuova lingua” I was trying to apply the old rules to the new language 500

400 300 200

100 db 40 20

Universals of information structure

93

In (38) there is a rising of the f0 on ‘vecchie’, though the peak is lower than for topic and focus in initial position, in accordance with the f0 declination. To sum up, our data show that, rather than marking differently topic and focus realizations, Italian marks both topic and focus by means of a pitch accent when these occur in pre-verbal position, while it reserves a flat contour for topic and focus in post-verbal position (see (31), (33), (34), (35) and (36) as opposed to (32), (34) and (37) above), unless of course we have an instance of a narrow contrastive focus on the right.

6. Information structure revisited In her book on information structure Choi (1999) interestingly distinguishes between topic and focus in the following way: (39) focus: (1) new information(=completive focus); (2) alternatives in a set (=contrastive focus) topic: (1) topic (= + prominent) (2) tail21 (= – prominent) We are already familiar with the distinction between the two kinds of focus, instead the way in which she characterizes the right-dislocated topic, for which she adopts the term ‘tail’ as in Vallduvì (1992), is interesting. To explain it let us show the feature analysis she presents in an early paper (Choi 1997)22: (40)

– N(ew)

topic

+N

tail + prominent

contr. focus

compl. focus

– prominent

As (40) show, she considers right-dislocated topics, the tail in her terms, and completive focus as sharing the [– prominent] feature, while the [+ prominent] feature is associated to both the left-dislocated topic and the contrastive focus. Evidence for her feature analysis is given by data from Korean in which the topic and contrastive focus are marked in the same way:

94 Lunella Mereu (41) a. Swuni-ka Inho-lul manassta Swuni-NOM Inho-ACC meet-PST-DCL ‘Swuni met Inho’ b. Swuni-nun Inho-lul manassta Swuni-MTOP Inho-ACC meet-PST-DCL ‘As for Swuni, she met Ihno’ (42) Swuni-ka Inho-nun manassta Swuni-NOM Inho-MTOP meet-PST-DCL ‘Swuni met Inho (but may be not others)’

(neutral)

(contrastive focus)

Now the feature analysis in (40) is just what distinguishes topic and focus occurring in initial position from those occurring in final or post-verbal position at the prosodic level in Italian. Therefore it seems that when we have syntactically-marked phenomena in Italian, we can say that, rather than having a syntactic order according to which given information precedes new information, we have the [+ prominent] – [– prominent] sequence just as in pragmatically oriented languages. This means that there are two universal principles of information structure: 1) the principle of progression from given to new information and 2) the principle of pragmatic salience. The first holds for configurational languages, as far as neutral, that is syntactically and pragmatically unmarked, utterances are concerned, the second holds for non-configurational languages, which we have characterized as being pragmatically oriented, and for syntactically and pragmatically marked phenomena in configurational languages. Of course our hypothesis for configurational languages still needs to be confirmed by considering prosodic data in a broad sample of languages, but some indications in this direction are already available23. 7. Conclusions In this paper we have shown that we can find universal principles of information structure if we study this area according to an interface approach taking into account the interplay of prosodic, morpho-syntactic and discourse grammar properties of languages and if we compare spoken data through a serious cross-linguistic survey of typologically different languages. This has led us to state that the distinction between configurational and non-configurational languages is still valid to characterize morpho-syntactic variation, but the difference between the two macro-types lies in the pragmatic principles they obey.

Universals of information structure

95

There are two principles of information structure: (1) the principle of progression from given to new information, and (2) the principle of pragmatic salience. Basically, configurational languages are syntactically oriented and obey the first principle, while non-configurational languages are pragmatically oriented and obey the second principle. This means that non-configurational languages do not necessarily have a basic syntactic order in that they realize the following pragmatic sequence: A.

[+ salient information] – [– salient information]

while configurational languages have the following two possibilities: B. i. [given information] – [new information] (in unmarked contexts) B. ii. [+salient information] – [–salient information] (in syntactically-marked contexts). Acknowledgements I wish to thank Giorgio Banti and Edoardo Lombardi Vallauri for their comments on this paper; special thanks to Nigel Vincent for his useful suggestions. I am, of course, fully responsible for any mistakes or inaccuracies. Notes 1.

2.

3.

By pragmatically salient information we mean information which the speaker considers the relevant part of his/her utterance, it can be either a topic or a focus, that is +/– new. The concept is close to Givón’s (1988) notion of ‘task urgency’ and to his principle: “Attend first to the more urgent task”. (252). See sections 3–5 for a deeper characterization of our concept. See also note 5 below for a definition of topic and focus. By discontinuous sequences we mean constituents whose internal material does not occur in adjacent positions, as other elements occur in between, for example a modifier occurring far away from its head. See data from Warlpiri in Hale (1983). Function changing phenomena is the term that Relational Grammar (Perlmutter 1983) or Lexical-Functional Grammar (Bresnan 1982, 2001) use to refer to cases such as passives, that is cases in which the function of a base constituent changes in surface structure.

96 Lunella Mereu 4. 5.

6.

7.

8.

9. 10. 11. 12.

13. 14.

15. 16. 17. 18.

19.

For an overview about the differences and similarities concerning the different pairs distinguished in the text see Lombardi Vallauri (2001, 2002). With the term ‘topic’ we mean given or presupposed information, but “not an arbitrarily selected part of the given information ...what the rest of the sentence is about” (Büring 1997; Mereu and Trecci 2004); ‘focus’ corresponds to new, but also contrastive or emphatic information. Mono-argumental sentences are complex structures in Italian; for some linguists only inaccusative structures can have the reverse order (Burzio 1986). The complexity of the phenomenon is well explained in Sornicola (1995, 1999, 2006) even in cross-linguistic terms. The cartographic approach more or less corresponds to the configurational approach of the ‘80s in that subject and object and the left periphery with the topic and focus projections (Rizzi 1997) have all fixed hierarchical positions in the syntactic tree representing the sentence in all languages. Notice that a similar distinction in generative terms is made by Kiss (1995a) according to whom there are “discourse configurational languages”, that is “languages in which topic and focus form key constituents of sentence structure, i.e., languages in which primary sentence articulation serves to express discourse-semantic functions” (Kiss 1995a: 5). By  marking coding we mean the absence of inflectional morphology in a language. As the scheme in (17) above shows, and as it will be pointed out in Section 5.2, topic-prominent languages and pragmatically oriented languages are all instances of non-configurationality. See note 5 for a definition of topic and focus and Section 2 for the distinction between topic- and focus-prominent languages. It is not our aim to characterize the whole set of properties and realizations of focus in Chinese. We are just interested to show that pragmatically prominent information in Chinese tends to occur pre-verbally. See note 1 for a characterization of our concept of pragmatic salience. See Frascarelli and Puglielli (this volume) for an analysis of topics in Somali; see Svolacchia, Mereu and Puglielli (1995), and Lecarme (1991) for a characterization of how focus and topic are organized in Somali. All utterances are taken from the Bonvino corpus (2005). Italics indicate topics, while capital letters focus. See Bonvino (2004, 2005) for a characterization of post-verbal subjects in Italian and for the prosodic behaviour of post-verbal subjects as topics and focuses. In the literature on syntax prosodic prominence has not been associated to topics up to very recently: See Crocco (this volume) for a survey of works on the prosodic marking of topics and Cresti (2000) for a non-syntactic model of information structure which analyzes the prosody of topics and comments. The three segments which are visible on the final part of the topic are there because of background noises, so they just indicate that the acoustic signal is disturbed.

Universals of information structure

97

20. Büring (1997), Mereu (2004), Mereu and Trecci (2004), Crocco (this volume) show that even topics can have contrastive meaning. Contrastivity is associated to topics also in Steube (2004), Molnár and Winkler (2006), and Erteschik-Shir (2007). 21. Tail corresponds to a right-dislocated topic, or to an appendix in the terminology used by Cresti (2000, this volume) and by Lombardi Vallauri (this volume). 22. Interestingly Choi builds her feature analysis considering data not only on Korean but also on right dislocations in Catalan, which show a similar behaviour to Italian. 23. See Crocco (this volume) and the references in her paper.

References Abraham, Werner and Sjaak De Meij (eds.) 1986 Topic, Focus and Configurationality. Papers from the 6th Groningen Grammar Talks. Groningen 1984. Amsterdam/ Philadelphia: John Benjamins. Albano Leoni, Federico, Franco Cutugno, Massimo Pettorino, and Renata Savy (eds.) 2004 Il Parlato Italiano. Atti del Convegno Nazionale (13–15 febbraio 2003), Napoli: M. D’Auria Editore – CIRASS (CD-ROM). Baker, Mark 1996 The Polysynthesis Parameter. Oxford: Oxford Univerity Press. Belletti, Adriana (ed.) 2004 The cartography of syntactic structures. Structures and beyond, Vol. 3. Oxford: Oxford University Press. Benincà, Paola, Giampaolo Salvi, and Lorenza Frison 1988 L’ordine degli elementi della frase e le costruzioni marcate. In Grande Grammatica italiana di Consultazione, Lorenzo Renzi (ed.), 115–226. Bologna: Il Mulino. Berruto, Gaetano 1986 Le dislocazioni a destra in italiano. In Tema/Rema in italiano, Harro Stammerjohann (ed.), 55–70. Tübingen: Narr. Bonvino, Elisabetta 2004 Fenomeni prosodici in relazione al soggetto postverbale. In Albano Leoni, Cutugno, Pettorino and Savy (eds.), (CD-ROM). 2005 Le sujet postverbal en italien parlé: syntaxe, zones et intonation. Paris: Ophrys Bossong, Georg 1981 Séquence et visée. L’expression positionelle du theme et rhème en français parlé. Folia Linguistica 15: 237–252. Bresnan, Joan (ed.) 1982 The Mental Representation of Grammatical Relations. Cambridge, MA: MIT Press.

98 Lunella Mereu Bresnan, Joan (ed.) 2001 Lexical-Functional Syntax. Oxford: Blackwell. Büring, Daniel 1997 The Meaning of Topic and Focus. The 59th Street Bridge Accent. London: Routledge. Burzio, Luigi 1986 Italian Syntax. A Government-Binding Approach. Dordrecht: Reidel. Choi, Hye-Won 1997 Information Structure, Phrase Structure, and their Interface. In Proceedings of the LFG97 Conference (on line), Miriam Butt and Tracy Halloway King (eds.). Stanford: CSLI Publications. 1999 Optimizing Structure in Context: Scrambling and Information Structure. Stanford: CSLI Publications. Chomsky, Noam 1995 The Minimalist Program. Cambridge, MA: MIT Press. Cinque, Guglielmo (ed.) 2002 The Cartography of Syntactic Structure. Functional structure in DP and IP, Vol. 1. Oxford: Oxford University Press. Comrie, Bernard 1981 Language Universals and Linguistic Typology. Syntax and Morphology. Oxford: Blackwell. Cresti, Emanuela 2000 Corpus di italiano parlato, Vol. 1 /2, CD-ROM. Firenze: Accademia della Crusca. Dik, Simon, Maria E. Hoffman, Jan R. de Jong, Sie Ing Djiang, Harry Stroomer, and Lourens de Vries 1981 On the typology of Focus Phenomena. In Perspectives on Functional Grammar, T. Hoekstra, H. van der Hulst and M. Mortgat (eds.), 41– 74. Dordrecht: Foris. Erteschik-Shir, Nomi 2007 Information Structure. The Syntax-Discourse Interface. Oxford: Oxford University Press. Frascarelli, Mara 2004 L’interpretazione del focus e la portata degli operatori sintattici. In Albano Leoni, Cutugno, Pettorino and Savy (eds.), (CD-ROM). Givón, Talmy 1976 Topic, Pronoun and Grammatical Agreement. In Li (ed.), 149–188. 1984 Syntax: a Functional-typological Introduction, Vol. 1, Amsterdam / Philadelphia: John Benjamins. 1988 The Pragmatics of Word-order: Predictability, Importance and Attention. In Studies of Syntactic Typology, M. Hammond, E. Moravcsik and J. Wirth (eds.), 243–284. Amsterdam /Philadelphia: John Benjamins.

Universals of information structure

99

Greenberg, Joseph H. 1963 Some universals of Grammar with Particular Reference to the Order of Meaningful Elements. In Universals of Grammar, J. H. Greenberg (ed.), 73–113. Cambridge, MA: MIT Press. Hale, Kenneth 1981 On the Position of Warlpiri in a Typology of the Base. Blomington: Indiana University Linguistics Club. 1983 Warlpiri and the Grammar of Non-configurational languages. Natural Language and Linguistic Theory 1 (1): 5–47. 1992 Basic word order in two «Free Word Order» Languages. In Pragmatics of Word Order Flexibility, Doris Payne (ed.), 63–82. Amsterdam / Philadelphia: John Benjamins. Halliday, Michael A. K. 1967 Notes on Transitivity and Theme in English. Journal of Linguistics 3 (part I): 37–81, (part 2): 199–244. Jelinek, Elinor 1984 Empty Categories, Case and Configurationality. Natural Language and Linguistic Theory 2: 39–76. 1989 The Case Split and Pronominal Arguments in Choctaw. In Configurationality. The Typology of Asymmetries, L. Marácz and P. Muysken (eds.), 221–256. Dordrecht: Foris. Kayne, Richard 1994 The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Kiss, Katalin É. (ed.) 1995a Discourse Configurational Languages. Oxford: Oxford University Press. 1995b NP Movement, Operator Movement, and Scrambling in Hungarian. In Kiss 1995a (ed.). Lecarme, Jacqueline 1991 Focus en Somali: syntaxe et interprétation. Linguistique Africaine 7: 33–65. Li, Charles N. and Sandra Thompson 1976 Subject and topic: A new typology of language. In Subject and Topic, Charles N. Li (ed.), 457–490. New York: Academic Press. 1981 Mandarine Chinese: a Functional Reference Grammar. Berkeley: University of California. Lombardi Vallauri, Edoardo 2001 La teoria come separatrice di fatti di livello diverso. L’esempio della struttura informativa dell’enunciato. In Atti del XXXIII Congresso Internazionale della Società di Linguistica Italiana. Dati empirici e teorie linguistiche, Napoli, 28–30 ottobre 1999, F. Albano Leoni, E. S. Krosbakken, R. Sornicola, and C. Stromboli (eds.), 151–173. Roma: Bulzoni. 2002 La struttura informativa dell’enunciato. Milano: La Nuova Italia.

100 Lunella Mereu Meillet, Antoine 1903 Introduction à l’étude comparative des langues indoeuropéennes, Paris (Reprint, Alabama 1964: University of Alabama Press). Mereu, Lunella 2004 La sintassi delle lingue del mondo. Roma: Laterza. 2005 Tipologia sintattica nell’interfaccia con la pragmatica. In Tipologia linguistica e società. Considerazioni inter- e intralinguistiche (Due giornate italo-danesi di studi linguistici. Roma, 27–28 novembre 2003), I. Korzen and P. D’Achille (eds.), 209–228. Firenze: Franco Cesati Editore. 2008 La sintassi dei circostanziali nel parlato. In La comunicazione parlata, Atti del Congresso Internazionale su Comunicazione parlata, Napoli, 23–25 febbraio 2006, Tomo III, M. Pettorino, A. Giannini, M. Vallone and R. Savy (eds.), 462–482. Napoli: Liguori Editore. Mereu, Lunella and Mara Frascarelli 2006 L’interfaccia sintassi-fonologia. Interpretazione e implicazioni teoriche. In “Analisi prosodica”. Teorie, modelli e sistemi di annotazione, Atti del 2° Convegno Nazionale AIS (Associazione italiana Scienze della Voce) Università di Salerno, 30 Nov. – 2 Dec. 2005, Renata Savy and Claudia Crocco (eds.). Università di Salerno, CD-ROM. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. In Albano Leoni, Cutugno, Pettorino and Savy (eds.), (CD-ROM). Mithun, Marianne 1987 Is basic word order universal? In Coherence and Grounding in Discourse, R. S. Tomlin (ed.), 281–328. Amsterdam /Philadelphia: John Benjamins. Molnár, Valéria and Susanne Winkler (eds.) 2006 The Architecture of Focus. Berlin /New York: Mouton de Gruyter. Perlmutter, David M. (ed.) 1983 Studies in Relational Grammar, Vol. 1. Chicago, University of Chicago Press. Puglielli, Annarita 1981 Frase dichiarativa semplice. In Sintassi della lingua somala 2, Annarita Puglielli (ed.), 1–44, Roma: Ministero degli Affari Esteri. Dipartimento per la Cooperazione allo Sviluppo. Rizzi, Luigi 1997 The fine structure of the Left Periphery. In Elements of Grammar. Handbook of Generative Syntax, Liliane Haegemann (ed.), 281–337. Dordrecht: Kluwer. Rizzi, Luigi (ed.) 2004 The Cartography of Syntactic Structure. The Structure of CP and IP. Vol. 2. Oxford: Oxford University Press.

Universals of information structure

101

Simone, Raffaele 1997 Une interprétation diachronique de la ‘dislocation à droite’ dans les langues romanes. Langue Française 115: 48–61. Sornicola, Rosanna 1995 Theticity, V S order and the interplay of syntax, semantics and pragmatics. Sprachtypologie und Universalienforschung 48 (1/2): 72–83. 1999 Labile Orders and Grammatical Functions: the Functional Representation of One-Argument Structures. In Boundaries of Morphology and Syntax, Lunella Mereu (ed.), 291–305. Amsterdam /Philadelphia: John Benjamins. 2006 Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe. In Pragmatic Organization of Discourse in the Languages of Europe, Giuliano Bernini and Marcia L. Schwartz (eds.), 357–544. Berlin /New York: Mouton de Gruyter. Steube, Anita (ed.) 2004 Information Structure. Theoretical and Empirical aspects. Berlin / New York: Mouton de Gruyter. Vallduvì, Engdahl 1992 The Informational Component. New York: Garland. Van Valin, Robert D. and Randy J. LaPolla (eds.) 1996 Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press.

Section 2 Grammar and information structure in Italian and some of its dialects

Constructions with preposed infinitive: Typological and pragmatic notes Giuliano Bernini

1. Introduction Constructions with preposed infinitive such as the ones illustrated in examples (1) and (2) for Italian and Yiddish are found in quite a number of languages. Their interpretation presents various problems of a semantic, pragmatic and typological nature, whose investigation may shed light on the structural potential of individual language systems in the expression of discourse functions. (1)

Italian (field notes) [Speaker A had previously indicated a tobacconist’s to speaker B] Speaker A: l’ha trovato? ‘Have you found it?’ Speaker B: per esserci, c’era ma era chiuso for be-there it-was-there but it-was closed ‘The tobacco shop was there, but it was closed’

(2)

Yiddish (Sholem Aleykhem, Motl Peyse dem khazns [Motl son of Peyse the Cantor], text reported in Lockwood 1995: 219) [He is called Yoyne, Yoyne the baker]

dos heyst, er aleyn bakt nisht. bakn bakn andere. this means he alone bakes not bake-INF they-bake others er koyft nor mel un he buys only flour and

farkoyft broit. sells bread

‘that is he doesn’t bake alone. Others bake. He only buys flour and sells bread’ Constructions of this type involve the articulation of the clause in two parts. The first part contains a predicate in a non-finite form, which in Italian may

106 Giuliano Bernini be accompanied by the preposition per ‘for’ as in (1). In the second part the same predicate is repeated in the finite form required by the context, such as the imperfect in (1) and the present in (2); the predicate in the finite form may be accompanied by arguments and adjuncts, such as the subject NP andere in (2) and the indirect object mi ‘to me’ in (3). (3)

Italian (Fucini, Il matto delle giuncaie) [A father to the man who has admitted his interest in his daughter] “Dunque,” dice lui “state a sentire. Dispiacere non mi so says he listen displease-INF not me dispiacete, perché de’fatti vostri nessuno m’ ha detto nulla you-displease because of affairs yours nobody me has told nothing di male, ma a mezzi come si sta? of bad but for resources how it is ‘ “So,” he said “listen to me. You don’t displease me, because nobody has told me anything bad about you, but as regards your private means, how are you fixed?”’

These constructions have been investigated by Gideon Goldenberg in a non-recent paper (1971/1998), where they are referred to as “tautological infinitive” and are illustrated for Indo-European, Turkic and Afro-Asiatic languages with a particular regard for the Semitic languages. On the basis of his abundant and detailed documentation, Goldenberg is able to distinguish two major types of “tautological infinitive”, to which a third minor type is added. The first type, called type A by Goldenberg, is the construction under examination here, reported to be common in several European languages. In this type the infinitive is “in extraposition in front of the sentence” (Goldenberg 1971 / 1998: 36). In the second type, called type B by Goldenberg, the “tautological infinitive” is found in the main clause of a cleft sentence, whereas its finite counterpart is found in the subordinate pseudorelative clause of the cleft construction (Goldenberg 1971/1998: 51), as shown in the tentative gloss it is go that he goes. This type seems to be attested primarily in Semitic languages (Goldenberg 1971/1998: 60) and is illustrated in (4).

Constructions with preposed infinitive 107

(4)

Babylonian-Talmudic Aramaic (Goldenberg 1971 /1998: 52)

       milq hu d-l lq aval qurban mti flogg-INF it that-not flogged but offering brings ‘It is flogging that it is not flogged, but an offering he brings’ Goldenberg (1971/1998: 71–72) mentions also a third type of “tautological infinitive” which he calls type C. This type resembles type A and is distinguished from it at a prosodic level. Goldenberg (1971/1998: 72) reports this type for Russian, where it is characterized by the fact that the infinitive and the inflected form of the predicate belong to the same prosodic unit with the main stress on the inflected form, as in (5). In Russian type A, on the contrary, the infinitive is separated from the rest of the sentence by a break in intonation and may be followed by the particle to, as in (6). (5)

Russian (Goldenberg 1977/1998: 72) znat’ ne znaju know-INF NEG I-know ‘I absolutely do not know’

(6)

Russian (Anton exov, Djadja Vanja) Ja stal udakom, njan’ka… Poglupet’-to I became whimsical: INST nanny PFV.grow.stupid: INF-PTCL ja e ë ne poglupel, Bog milostiv, mozgi na svoëm meste I yet not grew.stupid god merciful brains on their place no uvstva kak-to pritupilis’ but feelings somehow dulled-PL-REFL ‘I have become whimsical, nanny. As for growing stupid, I have not yet done so, by the grace of God, my brain is in its place, but my feelings are dulled’

A distinction which recalls Goldenberg’s types A and C has been proposed more recently by Benincà, Salvi and Frison (1988: 193) for Italian at a syntactic and a prosodic level. In Italian a construction with preposed infinitive and level intonation (“intonazione piana”) as in (7) is claimed to oppose a similar construction where the infinitive is separated from the rest of the sentence by a break and is pronounced with a slightly rising intonation (“l’intonazione leggermente ascendente”); furthermore the infinitive of this

108 Giuliano Bernini construction may be accompanied by per ‘for’ (cf. example 1) or by quanto a ‘as for’ as in (8). (7)

Italian (Benincà, Salvi and Frison 1988: 191, their example 192) Mangiare, mangio poco I-eat a.little eat-INF ‘I eat not very much’

(8)

Italian (Benincà, Salvi and Frison 1988: 194, their example 206d) Quanto a venire, verrà senz’altro. as for come-INF he-will-come certainly ‘He will certainly come’

Constructions with preposed infinitive, i.e. Goldenberg’s types A and C with no further differentiation, have been pointed out in Maslova and Bernini (2006: 81–83) as exemplifying non-nominal topics with no further specification than the assumption “that infinitives in topic positions can be used to encode states of affairs as topics” (p. 83).1 With respect to Maslova and Bernini (2006), this contribution presents a closer examination of the typology and of the pragmatic functions coded by these constructions. In this perspective, only the constructions with preposed infinitive followed by an inflected form of the same verb as illustrated by examples (1) and (2) – i.e. Goldenberg’s types A and C –, will be considered here. Instead, Goldenberg’s type B, i.e. tautological infinitives in cleft sentences, is not examined here further because of its different structure and of its function probably linked to focalization.2 In this contribution three questions will be tackled. The first pertains to the structure of such constructions and the semantic status of the infinitive and of the inflected verb forms. This question will be discussed in the framework of Wolfgang Klein’s (1998, forthcoming) approach to the notion of finiteness. The second question relates to the articulation of pragmatic functions within the constructions with preposed infinitive and in particular to the types of topic constructions they may be attributed to according to Maslova and Bernini (2006). The third question relates to the functions these constructions are used for with respect to the discourse contexts they seem to mostly occur in, which are adversative contexts. Whereas the discussion of the first question is based on the comparison of similar constructions found in a number of mainly European languages, the second and third questions will be cautiously approached on the basis of

Constructions with preposed infinitive 109

Italian data, in order to arrive at a result which might be further verified in other languages. The reason for this methodological choice lies in the fact that the assessment of such constructions in pragmatic terms needs a careful examination of the context they appear in. Since these constructions are not really frequent, we must restrict their examination to languages where sufficient, reliable data are readily available, such as in Italian. Finally, no distinction will be attempted between Goldenberg’s types A and C, and the constructions with preposed infinitive will be dealt with as a single macro-type. In fact the prosodic clues claimed to differentiate the two types in Russian by Goldenberg (1971/1998: 72) and in Italian by Benincà, Salvi and Frison (1988: 193) can hardly be detected in actual speech. Furthermore, consideration of a single macro-type also allows the investigation of constructions with preposed infinitive drawn from written sources.

2. The status of the verb forms in constructions with preposed infinitives As to the first question dealt with in this paper, i.e. the structural analysis of the construction considered here and the status of the infinite and of the finite verb forms which appear in it, Goldenberg (1971 / 1998: 37) had already recognized that in this and in the related constructions mentioned above “the lexical substance of the verb is, in principle, disconnected from its accidence”. In other words, in this construction the expression of the lexical and grammatical components of the semantic content of the verb are split with respect to their synthetic expression found in regular clauses. This is illustrated in table 1 for the first example mentioned above. In the upper section of table 1 the synthetic expression of the two components in the Italian finite verb form c’era (‘there was’) is contrasted with their split expression reported in the lower sections. The split expression of the two components of the verb semantics may be argued for on the basis of the behaviour of Italian stative predicates formed by the copula and an adjective, as in essere bello (‘to be nice’). In these predicates the grammatical and lexical components are already split: the latter component is expressed by the adjective, the former by the copula.

110 Giuliano Bernini Table 1. Semantics of constructions with preposed infinitive LEXICAL+GRAMMATICAL COMPONENTS

c’ er-a there be: PAST.IMPERFECTIVE-3SG LEXICAL COMPONENT

GRAMMATICAL COMPONENT

esserci

c’era Mood: INDICATIVE Tense & Aspect: PAST IMPERFECTIVE SUBJECT: 3SG (shop, implicit)

‘existence’

When these predicates appear in the construction under discussion here, only the adjective is fronted, whereas the whole compound predicate with finite copula is required in the second part of the sentence, as illustrated in (9).3 (9)

Italian (fragment of conversation between carabinieri taken from a popular TV series; speaker C is the commanding officer of the barracks mentioned) Speaker A: questa è la Sua caserma ‘this is your [= of speaker C] barraks’ Speaker B: bella, vero? ‘it is nice, isn’t it?’ Speaker C: bella, è bella nice is nice ‘It is very nice indeed’

A second piece of evidence in favor of the split expression of the lexical and semantic components in the construction dealt with here is found in those languages which possess a proform for verbs, such as do in English and tun in German. In these languages the fronted predicate in the infinitive is not repeated in the second part of the sentence, but represented by the verbal proform, inflected for the relevant categories required by the context. Examples (10) and (11) show the function of the verbal proforms in English and German. (10) English (P. D. James, A Certain Justice, London, Penguin, 1997: 369)4 She had been opening letters. The paper-knife was there on the desk. Carpenter seized it and drove it in. She may not have meant to kill, but kill she did.

Constructions with preposed infinitive 111

(11) German (Die Zeit 10 / 02.03.2006: 37; interview) Stern: […] Eine sinnvolle Didaktik für das frühe Sprachenlernen fehlt bislang weitgehend. ‘A reasonable teaching methodology for language learning is still needed’ ZEIT: Aber schaden tut die Englischstunde nicht. but damage-INF does the English-lesson not ‘But the English lesson causes no damage’ Stern: Wahrscheinlich nicht. Aber es ist ja auch eine finanzielle Frage und damit ein soziales Problem. ‘Probably not. But it is also a question of financing and a social problem as well’ Dropping of the copula in compound stative predicates appearing in the first part of the construction examined, and representation of the lexical verb by a verbal proform in the second part support the interpretation of this structure as the split expression of the lexical and grammatical components of the verb semantics. However, in most cases and in many languages the split is not apparent, both components being necessarily coded by the (same) verbal lexeme. These three types of split expression are reported in table 2. The role of the grammatical component of verbal forms such as the ones found in Indo-European languages has been highlighted by Wolfgang Klein (1998, forthcoming) within a functional theoretical framework where finiteness is not considered as “an epiphenomenon of verb inflection” (Klein, forthcoming: § 10), but a category pertaining to the utterance and located at an abstract level. According to this theory (cf. in particular Klein 1998: 238– 243), finiteness, notated as FIN*, is an operator with scope over a non-finite component notated as INF*, containing the lexical representation of the situation being talked about in an utterance in terms of predicate, arguments and optional elements. Table 2. Types of split expression LEXICAL+GRAMMATICAL COMPONENTS

c’era / è bella / killed LEXICAL COMPONENT

copula dropping

esserci bella kill

GRAMMATICAL COMPONENT

c’era è bella she did

repetition proform

112 Giuliano Bernini The linking of these two components at an abstract level results in an actual utterance. The linking involves the joining of a lexical element of the INF* component to the FIN* component, allowing the expression of the validity of the utterance in terms of a speech act (e.g. assertion) and of topic time, i.e. the time span for which the speaker considers his utterance valid. In Indo-European languages the lexical element which is part of this process is the lexical verb, which is inflected at least for mood and tense in the actual utterance.5 The linking process is also affected by the topic-focus structure, which depends on the speaker’s intention and on contextual factors. Topic elements usually precede FIN*, since they are outside its scope (Klein 1998: 241). Adopting Wolfgang Klein’s theoretical framework which has been briefly outlined here, the structure of the constructions with preposed infinitive can be analyzed as resulting from two major operations: (i) assignment of topic function to the lexical content of the predicate contained in the INF* component, which is then moved in front of FIN*;6 (ii) choice of the lexical carrier of the FIN* variables pertaining to speech act and topic time. The lexical predicate in the non-finite component is assigned topic status, moved out of the non-finite component and placed in an initial position in the actual utterance. After this first operation, two major choices are available for the linking operation, whereby a lexical element is assigned the function of carrying the expression of the speech act type and of the topic time, and is then inflected at least for mood and tense: (a) the same verbal lexeme contained in the non-finite component and already assigned topic status is used and therefore repeated in the actual utterance as in the first type in table 2, resulting in the constructions reported for Italian, Yiddish and Russian (cf. examples 1, 2, 3, 5, 6); (b) a substitute for the verb is used and the lexical verb is not repeated in the actual utterance as in the proform type of table 2, resulting in the constructions reported for English and German (cf. examples 10, 11). The case illustrated by compound stative predicates of Italian (cf. example 9) represents a third, intermediate type, as shown in table 2. The operation of topic assignment affects only the lexical elements, i.e. the adjectival component of the predicate, and the linking operation may resort to the copula as carrier of the speech act and topic time variables; however, the

Constructions with preposed infinitive 113

copula does not behave as a real verb substitute and the lexical part of the compound must be repeated.7 After having accounted for the presence of two verb forms and of the potential repetition of the same verbal lexeme in this type of constructions, deriving from Wolfgang Klein’s theoretical framework, we may now look more closely at the morphological form of the predicate assigned topic status in the utterance. The topic predicate is exempt from the task of carrying the finite variables of the speech act and topic time and need not be inflected for mood and tense. It may therefore be coded by a bare lexical form, if available, as in the English kill in example (10).8 In languages where bare lexical forms are not allowed, the coding of the predicate in topic function might resort to a wide range of inflected forms. However a major tendency is evidenced in the data available so far. Many languages tend to resort to inflected forms with the least amount of specification with respect to the major variables of speech act form and topic time, such as the infinitive forms of Italian, Yiddish, Russian and German, shown in the examples above. In Indo-European and in other languages, forms of this kind are removed from the prototype of the verb category and overlap with nouns in many aspects of their behaviour, as argued in Ramat (2002). The overlap may also be shown by the supine form of the verb used in Rumanian, a language which has almost lost the inherited infinitive form found in other Romance languages (cf. 12a), and furthermore by the verbal noun in –ma/-me used in Turkish in constructions comparable to the ones examined here and reported in Goldenberg (1971/1998: 60f.). As illustrated in (12b), the Turkish verbal noun in front position is also inflected for nominal categories, such as third person singular possessive and dative cases. (12) a. Rumanian (Jinga 2006: 148) De v zut v d, dar nu pot s cred of see: SUPINE I-see but not I-can that I-believe ‘As for seeing, I see, but I can’t believe it’ b. Turkish (Goldenberg 1971/1998: 60) Ol-ma-sın-a ol-du, amma nasıl? how be-INF-3SG-DAT be-PAST(3SG) but ‘Yes, it’s done, but how’ (Original translation in Goldenberg) In an apparent minority of cases, languages may inflect the topic predicate for the same mood and tense as the main verb. In fact in this case the same morphology is found in the predicate topic and in the main verb, as illustrated

114 Giuliano Bernini in (13) for Swedish. In this example the finite component of the utterance is expressed by means of the proverb göra – cf. the proform type of table 2 –, and the lexical verb springa ‘run’ is also inflected for tense and mood. (13) Swedish (Holmes and Hinchliffe 1994: 562) Springer gör jag sällan rarely run-IND.PRES do-IND.PRES I ‘I do rarely run’ The major tendency of resorting to an infinite/nominal form for the coding of the topic predicate, and the minor tendency of inflecting the topic predicate for the same categories as the main verb may actually reflect the difficulty in inflecting the verb for a function – the topic function – which does not correspond to its prime function of carrying the expression of finiteness in the utterance. The potential conflict between the two tendencies is actually attested, although in marginal cases, by irregular infinitives found in constructions of the type under discussion here and formed by affixing the infinitive morpheme to the lexical allomorph of the main verb inflected for the required mood and tense. Irregular infinitives of this sort are found in Hungarian and Yiddish.9 For Hungarian Goldenberg (1971/1998: 72–73) mentions a formation like volni in the sequence volni volt ‘for being there, it was there’. Volni is composed of the lexical allomorph vol- of lenni ‘to be’, found in the inflected past form vol-t, literally: ‘be.PAST-PAST(3SG)’ of the quoted example.10 Yiddish irregular infinitives found in the constructions with preposed infinitives are exemplified in the second column of table 3; the inflected form they are built upon are shown in the right-hand column of table 3. Table 3. Yiddish irregular infinitives (cf. Jacobs et al. 1994: 407, 414; Lockwood 1995: 117) Regular infinitive zeyn ‘to be’ visn ton

‘to know’ ‘to do’

Irregular infinitive binen izn veysn tun

Inflected form bin iz veyst tu

(IND PRES 1 SG) (IND PRES 3 SG) (IND PRES 2 PL) (IND PRES 1 SG)

In Yiddish the forms of preposed infinitives in the case of allomorphy in verb inflection show alternations such as the ones illustrated in (14a) for the regular infinitive and in (14b) for the irregular one.

Constructions with preposed infinitive 115

(14) Yiddish (Goldenberg 1971/1998: 72) a. gebn hob ikh shoyn gegebn already given give-INF have I ‘I have already given’ b. gibn gib ik give-INF give I ‘I do give’ Besides being a clue to the potential difficulty in choosing an inflected form suitable for the non-usual topic function of the predicate, the presence of analogical infinitives built upon the main verb form found in these constructions might shed some light upon the linking operation in Klein’s (1998) finiteness framework. In fact analogical infinitives such as the ones discussed here may be evidence for the assignment of the topic position to the verb after this has moved from the infinitive component into the finite component and has become the carrier of the speech act and topic time variables. This line of research must be left to future investigations based on a more complete database allowing a better insight of the behaviour recorded for Hungarian, Yiddish (and Hebrew) and illustrated here, but not recorded in other languages such as Italian or German. After having examined the structure of the constructions with preposed infinitive and observed some of their typological regularities, we may now consider the pragmatic functions they perform. 3. The pragmatics of the constructions with preposed infinitives In the investigation of the pragmatics of the constructions dealt with here, we shall first look at the topic-comment articulation of the sentence they occur in, in order to define the kind of topic constructions they exemplify and what is focussed on in the comment section of the sentence. The discourse functions of this type of constructions will be discussed on this basis.

3.1. Assumption of the state of affairs expressed in the semantic content of a lexical verb as a topic distinguishes the constructions with preposed infinitive from ordinary nominal topic constructions where the topic referent is an entity, as also pointed out in Maslova and Bernini (2006: 83). As with

116 Giuliano Bernini nominal topics, topicality results from the activation of the states of affairs expressed by a predicate in the context of previous discourse. Activation may be triggered by the explicit mention of the state of affairs, as shown by (essere) bella ‘be nice’ in (9) and kill in (10), but also by the agent noun beker ‘baker’ of example (2) with respect to bakn ‘bake’. Besides being explicitly mentioned, a state of affairs may also be evoked in the universe of discourse either by tighter or looser semantic associations with previous discourse matters, or else by shared knowledge or assumptions. Tighter semantic associations are shown in example (1), where existence (cf. esserci ‘to be there’) is implied for the object of trovare ‘to find’; looser semantic associations may be found in examples (6) (between stat’ udakom ‘to become whimsical’ and poglupet’ ‘to grow stupid’) and (11) (a lack of language teaching methodology and potential damage caused by unsuitable language teaching methods). Finally, shared assumptions may be illustrated by example (3), referring to a social context where a maid’s suitor is bound to win her father’s approval (non dispiacere ‘to not displease’). Nominal topics are claimed by Maslova and Bernini (2006) to be coded by three major types of constructions across languages. Topics can be encoded as subjects of a sentence by means of subject promoting constructions such as voice (cf. 15a), they can be fronted (cf. 15b), and finally they can be construed as reference points (also known as “hanging topics”, cf. 15c).11 (15) a. John was kissed by Mary. b. Everything else we’ll look at first. c. As for version 1.1., the package includes three utilities. These three construction types reflect the three major types of semantic relationships between the topic referent and the states of affairs described in the comment, although not exclusively (Maslova and Bernini 2006: 83), i.e.: a. the relationship between an event and its primary participant (John with respect to Mary in 15a); b. the relationship between an event and its spatial/temporal frame (as the anteposed direct object in 15b);12 c. the relationship between an event and an entity closely related to one of its participants (the program version contained in the package mentioned in 15c). With respect to nominal topics, the kind of infinitive topics under discussion here bear a different relationship to their comments. As argued in section 2,

Constructions with preposed infinitive 117

the relationship involves a state of affairs represented by the lexical content of the predicate (the topic) and the realization of this state of affairs in terms of utterance type, as assertion (together with polarity and epistemic modality, i.e. the speaker’s degree of confidence with respect to the validity of the propositional content of his utterance13), topic time, participants involved. On the basis of the data available so far, such topics, despite their particular relationship with their comments, appear to be construed either as reference points or as spatio-temporal frames, their coding as primary participants being excluded for obvious reasons. However, the assignment to one or the other type of construction is not always straightforward, as shown e.g. in Italian, where only infinitives preceded by the preposition per (as in example 1: per esserci, c’era) can be said to be construed as reference points, while bare anteposed infinitives (as in example 3: dispiacere non mi dispiacete) are ambiguous between the constructions used for encoding nominal topics as reference points and as spatio-temporal frames.14 The situation discussed for Italian is apparently shared by other Romance languages, as shown in (16) for French and in (17) for Spanish.15 The French example shows a reference point construction where the preposed infinitive is accompanied by the preposition pour, corresponding to the Italian per; the bare preposed infinitive in the Spanish example is once again ambiguous with respect to topic-encoding constructions. (16) French (Malet, Léo, Les enquêtes de Nestor Burma et les nouveaux mystères de Paris, vol. II, Paris, Laffont, 1985: 195 f.) [from the nearest bistrot I called the jeweller. Ranelagh 89-10] Ça sonna. Pour sonn-er, ça sonna. […] it rang for ring-INF it rang Ça sonna. Mais personne ne décrocha. it rang but nobody not picked.up ‘It rang. As for ringing, it did. But nobody picked up the receiver’ (17) Spanish (El habla culta de ciudad de Buenos Aires. Materiales para su estudio, Tomo 1, Universidad Nacional de Buenos Aires, Facultad de Filosofia y Letras. Buenos Aires, 1987: 19). Bueno, pero es una decisión que no la toma qualquiera, ¿eh? […] ‘Fine, but it is a decision that nobody would take, eh?’ Tom-ar-la, se puede tomar. take take-INF-it IMPERS can ‘Take a decision, one can do so’

118 Giuliano Bernini At this point it might be useful to explore the potentialities of the coding of predicates as topics in Italian in other types of sentences than those examined here as defined in section 1 in order to better limit the cases of ambiguous coding as pointed out above. The evaluation of examples to follow starts with reference point constructions and then goes on to anteposition. As to the reference point, a clear case of an infinitive as a topic bearing a loose relationship with one of the participants in the situation described in the comment is reported in (18) (cf. “eating” and “fat people”).16 (18) Italian (Benincà, Salvi and Frison 1988: 194, their example 204) Mangiare, risulta che molte persone sono grasse eat-INF it.is.found that many persons are fat anche se digiunano. even if they-stay.off.food ‘As for eating, it was found that many people are fat even if they stay away from food’ Outside Italian, an analogous case for English may be illustrated by example (19) (cf. “being afraid/frightened” and the speaker in the dialogue). Note that in this case English resorts to an “as for” construction for the coding of the reference point topic. (19) English (Nabb, Magdalen, 2005, The Monster of Florence, London, Arrow Books, p. 436) ‘Are you afraid of him?’ ‘I don’t know […] I haven’t spoken to him since I was twelve […] As to being frightened, I don’t really know.’ On the other hand, a case more akin to the constructions at issue here may be illustrated in (20). In this example the compound stative predicate in the infinitive form is explicitly construed as a reference point with the preposition per; however, in the comment, the lexical component is treated as a kind of complement (actually as the nominal complement of the predicate) and represented by the clitic object pronoun lo, instead of being repeated as, e.g. in (9). (20) Italian (from the daily newspaper la Repubblica, 26.09.06, p. 7) (Title) Così gli spioni costruiscono i dossier ‘This is the way spies manufacture files’

Constructions with preposed infinitive 119

(Beginning of the article) Per essere articolati e ricchi di dettagli, for be-INF articulated and rich of details questo è certo, lo erano this is sure it: OBJ they-were ‘As for being articulated and rich in detail, they certainly were’ As regards fronting or anteposition, i.e. the major constructions used to encode nominal topics as spatio-temporal frames with respect to their comments, only compound stative predicates are present in the examples available to date.17 In these examples the lexical component of the stative predicate is anteposed (as e.g. in 9), and represented in the comment by the clitic pronoun lo as in typical cases of left dislocation (unlike e.g. in 9), as illustrated in (21). In other words, as already seen in (20), the lexical component is treated as a complement of the stative predicate. (21) Italian (Beninca, Salvi and Frison 1988: 186, their example 165) Stupido, non lo è mai stato stupid not it: OBJ is never been ‘He has never been stupid’ Besides non-repetition of the lexical component within the comment, cases like the one exemplified in (21) are also distinguished by the possibility of right-dislocating the topic, a possibility not permissible in the constructions with preposed infinitives being considered here (e.g. *è bella, bella, cf. 9, or *non mi dispiacete, dispiacere, cf. 3). Right dislocation of the lexical component of a stative predicate is shown in (22). (22) Italian (radio news broadcast) l’ascensore si è rotto l’altro ieri ‘the lift broke down the day before yesterday’ e ieri sera lo era ancora, rotto and yesterday evening it: OBJ was still broken ‘and last night it was still out of order’ The restriction on right position as found in Italian may be a matter of typological variation. In Turkish, e.g., the non-finite verb form of the verb in the type of constructions in consideration here may also follow the comment, as shown in (23), which may be contrasted with (12b).18

120 Giuliano Bernini (23) Turkish (from Aziz Nezin, quoted in Goldenberg 1971/ 1998: 61) – […] iki lira yet-er … de-di say-PAST(3SG) two lira be.enough-AOR(3SG) ‘two pounds are enough… he said’ – Yet-er yet-me-sin-e ama, be.enough-AOR(3SG) be.enough-INF -3SG-DAT but bura-dan nasıl çık-aca -ım? here-ABL how go.out-FUT-1SG ‘well they are enough, so far as being enough is the problem (or: as to being enough they are enough), but how shall I get out of here?’ (Goldenberg’s translation) Turning now our attention to the comment side of the constructions with preposed infinitives, the split expression of the lexical component of the predicate as the topic and of the finiteness component of the predicate within the comment, makes it possible to focalize on one of the grammatical components of the predicate. As a matter of fact, in most of the examples presented here what seems to be focused on in the comment is assertion of the state of affairs, together with either positive or negative polarity, as shown e.g. in (1) and respectively in (11).19 These constructions represent another way of expressing what has been called “Verum-Fokus” by Höhle (1988, 1992), i.e. the focus on the truth of the predicate in an utterance, which may be also marked in Germanic and other languages by the accentual prominence of the finite form of the verb.20 In other words, example (1) might be paraphrased as: “As for the (potential) attribution of existence to the shop (as part of both speakers’ shared universe of discourse), this is asserted as true at the established topic time”. Assertion may be characterized by different degrees of the speaker’s confidence in the validity of the propositional content of his utterance. The degree of confidence may be high, as in example (1), or low. In this case focalization involves also epistemic modality, as shown in (24) for Italian, where it is lexicalized by means of the finite verb deve ‘you must’. (24) Italian (Fucini, Il merlo di Vestro) [With reference to a bird which has flown away from its cage] E lei, signor Piovano, se mai domani quella bestia capitasse ‘And you, Vicar, if that beast ever happened to come’ alla su’ uccelliera... mi raccomando a lei signoria… ‘to your aviary… I beg you, Sir’

Constructions with preposed infinitive 121

riconoscere lo deve riconoscere di certo anche lei recognize-INF it must recognize-INF certainly also you dall’ ugnòlo che gli manca, se ne ricorda?” from.the claw that he misses you-remember it ‘Vicar, if that beast should ever fly to your aviary, you are bound to recognize it by its missing claw. Do you remember?’ Finally, the data available allow us to claim that also the realization of a certain argument of the predicate involved in the construction may be focalized on, as in example (2), which might be paraphrased as: “As for the action of baking (at the centre of attention in current discourse), it is asserted at the established topic time that it holds for other people”.21 On the basis of these brief considerations it may be concluded that the major function of the constructions dealt with here, where the lexicalsemantic component of a predicate is split from its grammatical component, allows the focalization of grammatical meanings belonging to the finite component of the utterance and expressed by inflectional means, at least in some languages such as the Indo-European languages. We can now look at the broader context in which these constructions appear to be used, in order to understand their discourse functions. 3.2. A feature common to all the examples discussed so far and drawn from different languages is their occurrence in adversative discourse contexts.22 At least for the data available to date, two major adversative contexts may be envisaged in which these constructions appear, namely concessive contexts and denial of expectations, either legitimized by preceding discourse contexts or otherwise implicit in the discourse. In the first type of adversative contexts the constructions under examination here appear as the first item of a cluster parallel to the pre-concessive one of the type “it is true that p, but q”. According to Berretta (1998: 80), these clusters fully realize concessivity on a semantic level, although they represent a pre-syntactic expression of concessivity. The concessive type of contexts is represented by those examples where the clause with preposed infinitive is followed by a clause introduced by an adversative conjunction, as in (1) and (3) for Italian, (6) for Russian and (12b) for Turkish. In the second type of adversative contexts the clause with preposed infinitive denies what can be legitimately expected on the basis of the preceding discourse contexts, as in the Yiddish example (2), and may be introduced

122 Giuliano Bernini by an adversative conjunction as in (10) from English and (11) from German. In example (2) the denied expectation arises in the preceding context from the fact that Yoyne, the protagonist of the novel, is called “the baker”; in (10) from the affirmation that the protagonist didn’t mean to kill; in (11) from the fact that the lack of a reasonable language teaching methodology may make any language teaching in primary school harmful. In this respect it is worth considering also the following example, where the German translation in (25b) of the original Turkish version of (25a) shows a preposed infinitive as denial of the reasonable expectation that children understanding Ladino can also speak it. The Turkish original simply joins the two clauses by means of an adversative conjunction. (25) Turkish, German (Özcan, C., Ho geldin. Die Türkei in kleinen Geschichten, München, DTV, 1994: 130–131).23 a. Benim çocuk-lar-ım Ladino anl-ıyor-lar, 1SG:GEN child-PL-1SG Ladino understand-PRES-PL ama, konu -amı-yor-lar. but speak-POT:NEG-PRES-PL b. Meine Kinder verstehen zwar my children understand PTCL aber sprechen können sie esi but speak-INF can they it

Ladinoi, Ladino nicht. not

‘My children understand Ladino, but they can’t speak it’ Finally, the third type shares with the second type the denial of legitimate expectations, which in this case are implicit in general discourse. A relevant example from Italian is reported in (26); this example is drawn from TV fiction and its function can only be assessed by referring to the scene it belongs to. (26) Italian [From a TV series: in a hospital, one policeman to another policeman about a suspect to be interrogated and who is undergoing medical treatment] tu rimani qua perché per parlare parla you remain here because for speak-INF speaks ‘Stay here, because he is going to have to talk’ [contrary to the expectation that he won’t be able to after medical treatment] A similar denial is probably represented by example (9), although it is difficult to reconstruct the expectations implicit in that sphere of discourse. The

Constructions with preposed infinitive 123

final example (27), from French, may better illustrate the third type of discourse function of the constructions with preposed infinitive. In this example the construction denies the potential negative expectation involved in the hope that somebody will read the article being talked about.24 (27) French (Malet, Léo, Les enquêtes de Nestor Burma et les nouveaux mystères de Paris, vol. II, Paris, Laffont, 1985, pp. 289) Maintenant, il ne restait plus qu’à souhaiter que cela serve, que quelqu’un lise cet article et agisse... Oui, mais qui? Et agisse comment? ‘Now it could only be wished that this could be useful, that somebody should read this article and take action… Yes, but who? And what should the action be?’ Oh! Pour être lu, ça serait lu. oh for be-INF read-PAST.PART this it-should-be read-PAST.PART ‘As to being read, it will be read’

4. Concluding remarks The discussion in the preceding section 3.2. has pointed out that the constructions with anteposed infinitive serve a rather limited discourse purpose in adversative contexts. However, these constructions allow the finer articulation of discourse at local level, representing an option whose choice may have a greater interactive or rhetorical effectiveness, as evidenced in examples (1) and (2) respectively, among the others discussed in this contribution. The discourse effectiveness of these constructions is brought about by the possibility of focalizing semantic functions expressed by inflection, such as assertion and epistemic modality, as in the many examples presented in the preceding sections. As shown in sections 3.1. and 2., the focalization of those semantic functions is made possible by the splitting of the lexical semantic component and of the grammatical component of the predicate and by their encoding in a topic-comment structure. The consideration of these particular constructions has a certain impact also at a theoretical, typological and methodological level. As discussed in section 2, these constructions, although rare, allow a better insight into the organization of the utterance and in particular into the notion of finiteness, supporting Klein’s (1998) claim that it should not be an epiphenomenon of verb inflection, but an abstract operator involved in the actualization of utterances.

124 Giuliano Bernini At a typological level, these constructions point to two further questions whose further inquiry can only be carried out by future research. The first question pertains to the diffusion of this type of constructions across languages. On the basis of the available data this type should actually be restricted to those languages possessing a nominal form of the verb, such as the infinitive of many Indo-European and Turkic languages. Can languages without an infinitive or without other infinitive forms such as the supine in Rumanian (cf. ex. 12a), split the lexical and the grammatical component of a predicate? If not, do these languages have any means of expression allowing them to modulate the discourse in the finer way described in section 3.2.? The second question is more general and relates to the position of these constructions within the general typological frameworks of languages (and of types of languages). In Romance languages these constructions seem to be consistent with the general iambic prosodic structure found in utterances, whereby the focalized, stressed constituent is in second position, as in the case of verb-subject order where the subject is focussed on (cf. Italian Chi è arrivato? – È arrivato GIOVANNI, vs. English Who has come? – JOHN has come). However, although Germanic languages such as English and German can stress the finite form of the verb in an utterance, they do also possess the type of split constructions under discussion here. Finally, at a methodological level the researcher is confronted with constructions which appear to be rare and to have a limited discourse function and furthermore cannot be elicited with standard techniques, since their occurrence is dependent upon particular discourse conditions. However, as this contribution tries to show, these constructions appear to be particularly valuable pieces of evidence of the potential of expression in language and its general organization.

Abbreviations 123 ABL AOR DAT FUT GEN IMPERS IND INF INST

first, second, third person ablative aorist dative future genitive impersonal indicative infinitive instrumental

NEG OBJ PART PFV PL PRES PTCL REFL SG

negation object participle perfective plural present particle reflexive singular

Constructions with preposed infinitive 125

Acknowledgements The research presented here was carried out within the project “Analisi e sintesi nella costituzione di lessemi verbali” (grant 60BER06 of the Università degli Studi di Bergamo, Dipartimento di Scienze dei linguaggi, della comunicazione e degli studi culturali). I am grateful to Sandro Caruana, Richard Dury, Alessandro Mengozzi for their assistance in the evaluation of various examples. I am also indebted to Paolo Ramat, Anna Giacalone Ramat and Sonia Cristofaro, Giovanna Marotta and Romano Lazzeroni for their helful comments submitted on the occasion of the presentation of this piece of work at the Universities of Pavia and Pisa. I am also grateful to Jennifer Pearson (University of Bergamo) for her valuable help in the revision of the English text. It goes without saying that I am the only person responsible for all the errors and shortcomings found in this paper.

Notes 1.

2.

3.

4.

Actually some of the examples discussed here have already been quoted in Maslova and Bernini (2006): example (1) was quoted as example (31) on page 81 and example (5) as example (32) on page 82. A predicate cleft construction with predicate copy analogous to this type of construction is found in creole language forms, where it has a focalizing function. Cf. Haitian Se mache Jan mache al lekol (literally: FOC walk John walk go school) ‘John WALKED to school’ (i.e. did not run) (Veenstra and den Besten 1994: 308). As for Italian, an analogous construction may be found in exclamative sentences, as shown in: Che scappare che fa questa piccola!, literally ‘what run.away-INF that does this little-F (cat)’, seen in a casual comment following the running away of a scared kitten. The infinitive scappare is fronted and clefted and referred to in the following clause by means of the pro-verb ‘do’. Copula deletion in speaker B’s statement (bella for è bella) is not related to the question discussed here and is dependent on conditions governing colloquial speech. With respect to anteposition, note that the fronting of the lexical component of statives as illustrated here is possible only without the preposition per. Presence of this preposition triggers also the fronting of the infinitive of the copula, cf. per essere bella, è bella vs. *per bella, è bella. The lexical component may be represented by an object clitic too, as lo in bella, lo è; however this is a different type of construction, as discussed in 3.1. with respect to examples (20) and (21). According to a native speaker, the prosody of this sentences is noted as follows: KILL she DID, with main stress on the verbal proform.

126 Giuliano Bernini 5.

6. 7.

8. 9. 10.

11. 12. 13. 14.

15. 16.

17.

18.

19.

Other languages may resort to different choices in the linking of the INF* and of the FIN* components, as in the case of Chinese aspectual particles (Klein, Li and Hendriks 2000). In this respect Klein’s approach to the notion of finiteness can result in a powerful tool for language typology. Discussion of the assignment of topic status to the lexical content of the predicate is dealt with later in section 3. This type is identified here on the basis of Italian; however the behaviour of stative predicates might be subject to typological variation and is open to further investigation. Cf. also Goldenberg (1971/1998: 75). Goldenberg (1971/1998: 72–73) also mentions analogical formations in the area of the niph’al conjugation in Hebrew. Vol- as the allomorph for past tense is opposed to van ‘he is’ and lesz ‘he will be’. No mention is made of these formations in the reference grammar of Hungarian by Kenesei, Vago and Fenyvesi (1997). Examples (15b) and (15c) are drawn from Maslova and Bernini (2006), pp. 105 (their example 72b) and 89 (their example 41b) respectively. Dislocations, in particular left dislocations, code this same type of relationship. The inclusion of epistemic modality may be claimed on the basis of examples such as (24), to be discussed later. The two types of constructions may, however, be identified by their prosodic behaviour, as claimed by Benincà, Salvi and Frison (1988: 193). Cf. the remarks made on this issue in section 1. These examples have already been mentioned in Maslova and Bernini (2006: 83 and 82 respectively). According to Benincà, Salvi and Frison (1988: 194), the anteposed infinitive is pronounced here with a slightly rising intonation and is separated from the rest of the sentence by a slightly marked pause. Anteposition of infinitives with modal verbs such as Correre, non dovrei (literally ‘run-INF not I-should’) ‘I shouldn’t run’, are actually cases of complement clauses construed as topics, cf. Benincà, Salvi and Frison (1988: 189– 190). Therefore they are not relevant in the perspective adopted here. The presence of a verbal proform and its role in anaphoric chains allows the addition of the bare form of the verb as an afterthought in English in order to render less ambiguous the reference of the proform, as in following example: “My cousin knows one of the cops who found her, and he said whoever did it must really have hated her. At least that’s what my cousin said he said.” He looked at his audience. “But I guess she did, huh? Hate her, I mean.” (Leon, Donna, Doctored evidence, London, Arrow Books, 2006, pp. 26–27). Such cases can hardly be considered as instances of the construction dealt with here with postposition of the infinitive. In the Italian example (1) the focus can only be on the finite verb c’era, the only constituent of the comment; in the German example (11) the position of

Constructions with preposed infinitive 127

20. 21.

22. 23. 24.

the negative particle shows that the subject constituent die Englischstunde is not focussed on. See also Klein (1988: 228) and Lombardi Vallauri (1998). It is questionable whether this is different from argument-focus as discussed in Lambrecht (1994: 228–233), besides the fact that the structures involved are different. This is true for authentic examples available to date. Examples drawn from grammars and provided with no context cannot be considered in this respect. Also reported in Maslova and Bernini (2006: 82). In this example epistemic modality appears to be focalized as in (24). However, unlike in (24), modality is here expressed by inflectional means, i.e. the present conditional.

References Benincà, Paola, Giampaolo Salvi, and Lorenza Frison 1988 L’ordine degli elementi della frase e le costruzioni marcate. In Grande grammatica italiana di consultazione. Vol. 1: La frase. i sintagmi nominale e preposizionale, Lorenzo Renzi (ed.), 115–225. Bologna: il Mulino. Berretta, Monica 1998 Il continuum fra coordinazione e subordinazione: il caso delle preconcessive. In Ars linguistica. Studi offerti da colleghi ed allievi a Paolo Ramat in occasione del suo 60° compleanno, Giuliano Bernini, Pierluigi Cuzzolin, and Piera Molinelli (eds.), 79–93. Roma: Bulzoni. Goldenberg, Gideon 1971 Tautological infinitive. In Israel oriental Studies 1: 36–85. Reprinted 1998 in Studies in Semitic Linguistics. Selected Writings, 66 –115. Jerusalem: Magnes Press, The Hebrew University. Höhle, Tilman N. 1988 Verum-Fokus. In Sprache & Pragmatik 5: 1–7. 1992 Über Verum-Fokus im Deutschen. In Informationsstruktur und Grammatik, J. Jacobs (ed.). Linguistische Berichte Sonderheft 4/1991– 1992: 112–141. Holmes, Philipp and Ian Hinchliffe 1994 Swedish. A comprehensive grammar. London /New York: Routledge. Jacobs, Neil G., Ellen F. Prince, and Johan van der Auwera 1994 Yiddish. In The Germanic languages, Ekkehard König and Johan van der Auwera (eds.), 388–419. London: Routledge. Jinga, Lauren ia 2006 Grammatica romena per italiani. Alessandria: Edizioni dell’Orso.

128 Giuliano Bernini Klein, Wolfgang 1998 Assertion and finiteness. In Issues in the Theory of Language Acquisition: Essays in honor of Jürgen Weissenborn, Norbert Dittmar and Zvi Penner (eds.), 225–245. Bern: Lang. forthc. On Finiteness. In Semantics meets acquisition, V. Van Geenhoven (ed.). Dordrecht: Kluwer. Klein, Wolfgang, Ping Li, and Henriëtte Hendriks 2000 Aspect and assertion in Mandarin Chinese. In Natural language and linguistic theory 18 (4): 723–770. Kenesei, István, Robert M.Vago, and Anna Fenyvesi 1997 Hungarian. London /New York: Routledge. Lambrecht, Knud 1994 Infomation structure and sentence form. Topic, focus and the mental representation of discourse referents. Cambridge: Cambridge University Press. Lockwood, William B. 1995 Lehrbuch der modernen jiddischen Sprache. Mit ausgewählten Lesestücken. Hamburg: Buske. Lombardi Vallauri, Edoardo 1998 Focus esteso, ristretto e contrastivo. In Lingua e stile 33 (2): 197–216. Maslova, Elena and Giuliano Bernini 2006 Sentence topics in the languages of Europe and beyond. In Pragmatic organization of discourse in the languages of Europe, Giuliano Bernini and Marcia L. Schwartz (eds.), 67–120. Berlin /New York: Mouton de Gruyter. Ramat, Paolo 2002 La natura dell’infinito. In L’infinito e oltre. Omaggio a Gunver Skytte, Hanne Jansen, Paola Polito, Lene Schøsler, and Erling Strudsholm (eds.), 409–417. Odense: Odense University Press. Veenstra, Tonjes and Hans den Besten 1994 Fronting. In Pidgins and creoles. An introduction, Jacques Arends, Pieter Muysken and Norval Smith (eds.), 303–315. Amsterdam: John Benjamins.

Negation and «Focus Clash» in Sardinian Franck Floricic

1.

Introduction

Sardinian is a Romance language that resorts to various strategies to express total and partial questions. Apart from intonation that may mark the interrogative character of a utterance, Sardinian makes a particular use of the marker a (< lat. aut ‘or’) and of fronting. The syntactic properties of interrogative utterances presented show the incompatibility of fronting and marking through the interrogative marker a. We shall discuss the reasons for this incompatibility, and we shall see how negation interferes with the interrogatives. The hypothesis presented is that negation and interrogative marking are incompatible because they are associated to different focus domains that come to clash. Given that Sardinian has three main different varieties which can be quite distinct with respect to their phonology in particular, reference is made here to the Logudorese variety, other varieties nonetheless exhibiting similar phenomena. The discussion will include a consideration of so-called interrogative conjugations in Northern varieties of Italian dialects, which demonstrate particularly interesting properties concerning the relations between negation and interrogation: when total questions and negation are not completely incompatible, their joint use can lead to profound syntactic reorganisations which recall the ones found in Sardinian.

2.

Questions

2.1. Questions and their properties It is obviously beyond the scope of this paper to present anything particularly new on interrogative utterances. The various proposals allow us to identify recurring elements that characterise interrogative structures. On the one hand, they constitute a type of utterance that makes the interlocutor the main pole of the information structure; on the other, the interlocutor is asked to identify a value that the speaker cannot or doesn’t want to satisfy.

130 Franck Floricic Finally, the expected identification can bear on one of the terms of the predication (arguments, adjuncts) or on the predicative relation as a whole. French for instance is a language where the identification procedure is marked by morphosyntactic features that have characteristic functions. In the cleft construction C’est X que Y, the function of the anaphoric item c’ is to point to a class of objects which are to be identified; this class thus becomes the informational foreground of the utterance, the rest of the predication being relegated to the background of the informational structure. It is precisely the role of the complementizer que to signal the pre-constructed status of the predication: an utterance such as C’est quoi que t’as acheté? ‘It is what that you has bought’ “What is it that you bought?” supposes a scheme You bought X, and it is the identification of the entity functioning as second argument that constitutes the informational foreground of the interrogative. Thus, interrogative pronouns represent a class of objects from which the interlocutor is expected to identify and extract an occurrence or a set of occurrences. In the case of total questions, the interrogative also invites an identification; the interlocutor is expected to identify the right value among two polar values p and p’. Thus the interlocutor is put in a position to (in)validate a propositional content by the identification he is asked to provide. The complexities of the morphosyntactic mechanisms marking interrogatives no doubt reflect that of the cognitive mechanisms at stake. Negative interrogatives are even more complex in their morphosyntactic realisations and their processing costs. This is exemplified by positive questions such as Where did you go this summer? or When did he come to see you? which suppose on the part of the interlocutor the identification of determined sites locating the events referred to by the predication. The negative versions Where didn’t you go this summer? and When didn’t he come to see you? are more complex as they bring together the acts of negation and interrogation. They imply the identification of a site locating a non-event. The localisation of what is other than what is concerns a potentially infinite set of locating sites. In other words, given that an individual necessarily occupies a determined spatio-temporal position (Locke 1694), potential localisation sites are infinite with respect to actual ones – as proposed by Givón (1979: 135), “It is […] statistically much easier to fix a point in time by the occurrence of an event than by the infinite number of points in time at which it did not occur”. This explains the further complexity of Where didn’t you go this summer? Such a complexity explains the tendency of such utterances to convey an exclamative value (as in French Qu’est-ce qu’il a pas dit!). Of course, when this kind of negative interrogatives is not concerned with the localisation of an event but with the assignation of a

Negation and «Focus Clash» in Sardinian 131

subjective value, the processing of the sentence is less costly. A question such as What didn’t you like in Venice? is a priori more plausible than When didn’t he come to see you? (Andrée Borillo, p. c.), would it be only because the verb like involves a scenario which is independent from any particular spatio-temporal setting: the function of the interrogation is the identification of an element from the class of y in the schema ((x) ((neg) like (y)). The fact that a question such as Where didn’t you look for it? to someone who has lost something is well-formed depends on the partition of potential sites, some in which a given object could be found, and some others where it is not expected to be. Interrogatives are complex construction involving among other things word order, cliticization, and subordination. Their bleaching leads them to be constantly renewed, with correlative increase in complexity, as shown by the structural complexity of French Quand tu pars? as opposed to C’est quand est-ce que tu pars? “When are you leaving?” (see Foulet 1921).

2.2. Total questions in Sardinian Sardinian resorts to two different strategies to express total questions: the use of marker a and fronting. The marker a derives from latin aut ‘or’ (Meyer-Lübke 1903; Wagner 1984: 34–35 § 17 and 332 § 370; Pittau 1972: 37–38, 41, 143–145, 157; Contini 1985, 1986; Jones 1988, 1993: 24ff.) – and figures in sentence-initial position to signal total questions. This is illustrated in examples (1)–(4): (1)

A mi podes agiuare? A CLIO1SG can2SG help ‘Could you help me?’

(2)

a. Anne’, a nos cùmbidas? Anne’, a CLDO1PL invite ‘Annedda, do you invite us?’ b. *A Anne’, nos cùmbidas?

(3)

a. Un’ àteru cafè, a mi lu batis? An other coffee a CLIO1SG CLDO3MSG bring ‘An other coffee, do you bring it to me?’ b. *A un’àteru cafè, mi lu batis?

132 Franck Floricic (4)

a. Frutta, a nde cheres? Fruit a CL.part want2SG ‘Fruits, do you want any?’ b. *A frutta, nde cheres?

Example (1a) shows that the particle a precedes the proposition that is queried.1 Example (2a) illustrates the scope of the marker: Vocative Anna being outside the propositional core thus cannot figure within the scope of a. The same goes for (3a) and (4a): the particle cannot precede the thematic constituent, whether it be the DP un ateru cafè or the bare noun frutta, and must be in the initial position of the proposition at stake. Notice that the particle a is the strong element of the clitic cluster it forms with partitive nde in (4a). Involvement in such clusters might lead one to conclude that a is itself a clitic and that its clitichood dictates the observed distributional restrictions2. However, dialogues such as (1’) can be observed where a appears in final position, and where no repair strategy is used despite final stress, which is otherwise a marked configuration in Sardinian: a salient feature of Sardinian phonology is to impose a trochaic binary and bi-syllabic structure on the right boundary of the word: (1’)

– A mi podes agiuare? A CLIO1SG can2SG help ‘Could you help me?’ – E a? No lu podes fagher tue? you And a? NEG CLDO3MSG can2SG do ‘And why? Can’t you do it yourself?’

While the prosodic constraints of this marker remain to be examined more closely, the possibility to have an example such as (1’) constitutes a strong argument against a clitic analysis of a. The other strategy resorted to by Sardinian to express questions is that of fronting. Examples involving the fronting strategy are given in (5): (5)

a. Torradu est? Came.MSG be3SG ‘Did he come back?’ b. Famidu ses? Hungry.MSG be2SG ‘Are you hungry?’

Negation and «Focus Clash» in Sardinian 133

c. Cumpresu m’ as? Understood.MSG CLIO1SG have2SG ‘Did you understand me?’ d. Cumprendende ses, Francu? Understanding be2SG, Francu ‘Do you understand, Franck?’ e. La Heineken, bona est? The Heineken, good.FSG be3SG ‘Is the Heineken good?’ f. Drommire cheres? Sleep want2SG ‘Do you want to sleep?’ g. Babbu, nudda cheret? Dad nothing want3SG ‘Dad i, doesn’t hei want anything?’ h. Innoghe istas? Here live2SG ‘Do you live here?’ i. In domo fit? In house was3SG ‘(S)he was home?’ In these examples, the fronted element is of different nature. It can be the auxiliated in a split verb-phrase structure (5a–f); it can be a direct object negative indefinite (5g); it can even be an adjunct (5h–i). The periphrastic future (5j) (aere ‘have’ (present indicative) + a + infinitive) and conditional (5k) (devere ‘must’ (imperfect indicative) + infinitive) can however not be fronted (Jones 1988: 185 and 1993: 147; Molinu 1999): (5)

j. *A lu faghere, as? A CLDOMSG do have2SG ‘Shall you do it?’ k. *Lu faghere, tias? CLDOMSG do should2SG ‘Should you do it?’

The main point is that fronting is thus incompatible with particle a; the marker a cannot be used along with the fronted material in the examples under (5):

134 Franck Floricic (6)

a. *A torradu est? ‘Did he come back?’ b. *A cumprendende ses, Francu? ‘Do you understand, Franck?’ c. *A cumpresu m’as? ‘Did you understand me?’ d. *A famidu ses? ‘Are you hungry?’ e. *La Heineken, a bona est? ‘Is the Heineken good?’ f. *Babbu, a nudda cheret? ‘Dad i, doesn’t hei want anything?’ g. *A innoghe istas? ‘Do you live here?’

The same goes for negation: the negative marker no is incompatible with fronting: (7)

a. *Famidu no ses? ‘Aren’t you hungry?’ b. *Innoghe no istas? ‘Don’t you live here?’

(see No istas innoghe?)

c. *E / de nudda no as bisondzu? ‘Don’t you need anything?’ (see No as bisondzu (d)e nudda?) d. *Frutta, a no nde cheres? ‘Fruits, don’t you want any?’ e. *Sa Heineken, bona no est? ‘Isn’t the Heineken good?’ f. *Compresu no m’as? ‘Didn’t you understand me?’ g. *Drommire no cheres? ‘Don’t you want to sleep?’

(see Frutta, no nde cheres?) (see Sa Heineken, no est bona?) (see No m’as compresu?) (see No cheres drommire?)

Examples (6a–g) and (7a–g) show that Sardinian has a strong constraint that excludes the joint use of negation and the interrogative particle on the one hand, and of this particle with fronting in total questions. What then of partial questions?

Negation and «Focus Clash» in Sardinian 135

2.3. Partial questions Constituents involving partial questions are generally used in preverbal position, which is a privileged informational position. (8)

a. Ite cheres faghere? What want2SG do ‘What do you want to do?’ b. Chie at appidu su café? the coffee Who have3SG had ‘Who had the coffee?’ c. Inue istas? Where live2SG ‘Where do you live?’ d. Proite b’ andas? Why CL.loc go2SG ‘Why do you go there?’ e. Comente as drommidu? How have2SG slept ‘How did you sleep?’ f. E frade tou, cando (est chi) benit? And brother your when (is that) come3SG ‘And your brother, when does he come?’ g. Sa domo, cantu l’ as pagada? The house how much CLDO3MSG have2SG paid.FS ‘Your house, how much did you pay it?’

Unlike what is shown by (5a-g), (9) illustrates that Sardinian negation is in principle compatible with partial question markers: (9)

a. Ite / ite est chi no as cumpresu? What be3SG that NEG have2SG understood.MSG ‘What didn’t you understand?’ b. Chie / chie est chi no at appidu su café? Who be3SG that NEG have3SG had the coffee ‘Who didn’t have the coffee?’ c. *Inue no istas? *‘Where don’t you live?’

136 Franck Floricic (cf. Inue no ti tiat piaghere de istare? ‘Where wouldn’t you like to live?’) c’. Inue no ses ancora istadu? been.MSG Where NEG be2SG yet ‘Where didn’t you go yet?’ d. Proite no b’ andas? Why NEG CL.LOC go2SG ‘Why don’t you go there?’ e. *Comente no as drommidu? How NEG have2SG slept.MSG *’How didn’t you sleep?’ f. ??E frade tou, cando no benit? And brother your when NEG come3SG ‘And your brother, when doesn’t he come?’ f’. E frade tou, cando est chi no podet bennere? And brother your when be3SG that NEG can3SG come ‘And your brother, when cannot he come?’ g. ?? Sa domo, cantu no l’ as pagada? The house how much NEG CLDOFSG have.2SG paid.FSG ‘The house, how much didn’t you pay it?’ It can already be seen that partial questions are constrained at different degrees: while (9a) and (9b) are fine, (9c) is unacceptable: The question Inue istas? asks the interlocutor to identify an event located in the here and now that is already validated; location is already provided by the semantics of the stative verb istare. Unacceptability comes from the contradiction of locating an event that did not take place. The same goes for (9e): comente asks the interlocutor to provide a modality for the existence of the process: in order to assign a modality for the existence of the process, the latter must however be actual at some constructional level: if the event of ‘sleeping’ didn’t take place, no modality of existence can be attributed to such a “nonevent”. The asymmetry between cases under (7) and those under (9) raises the question of why negation is compatible with (some) partial questions but not with total ones. This question can be answered through the informational and semantic properties of negation and those of yes-no questions.

Negation and «Focus Clash» in Sardinian 137

3.1. Negation as a type of judgement A number of indications confirm the hypothesis that negative markers constitute a particular type of judgement – a negative judgement. The informational nucleus of a negative sentence thus lies in this particular type of judgement expressed by negation. Nouns in Aghem for instance – a bantu language of Cameroon – have two different forms; an in-focused one and an out-of-focus one. Direct objects normally take in-focus form, as they are associated to salient information. Under negation however, the direct object nominal takes out-of-focus form, as if informational saliency had been usurped by negation. It follows that the negative sentence expresses a negative state of affairs whose exponent is the negative marker (cf. Reinach 1911). That is what is illustrated by examples (10a–c) from Hyman (1984): (10) a. èm mˆO zèi kéi-béE néE I PAST ate fufu:FOC today ‘I ate fufu today’ (object in focus) béE-’ kéO b. èm mˆO zèi néE I PAST ate today fufu:OUT-OF-FOC ‘I ate fufu today’ (object out of focus) néE c. èù kéa zèi béE-’ kéO I NEG ate fufu:OUT-OF-FOC today ‘I didn’t eat fufu today’ (object out of focus) In (10a), the designation of the entity with object function is the informational core of the utterance, which is why the noun béE is marked as infocus. In (10b), it is the localisation of the event which is the core of the message; that is the reason why the adjunct néE in immediate postverbal position brings about the out-of-focus marking of the direct object. (10c) shows that the negative marker kéa induces a shift of the informational core from the direct object, which is out-of-focus, to the negation itself. Hyman thus concludes that negation is endowed with an intrinsic focus that is responsible for the suspension of focus marking on other elements of the utterance. Likewise in Grébo and Wobé, some particles cliticising unto the verb signal the event as the core of the message; which is why they are known as assertive focus markers. Examples (11a) and (12a) show vowel E and syllable ne related to the verb to be bring the focus unto the verb (Marchese 1983: 122):

138 Franck Floricic (11) wobé a. O di-E ko He ate-AF rice ‘He ate rice’ b. O se ko di He NEG rice eat ‘He didn’t eat rice’ (12) grébo

nE ne a. ne du-da I pound-PAST it AF ‘I pounded it’ b. née yi nE du I NEG it pound ‘I didn’t pound it’ / ‘I have not pounded it’ These examples thus clearly establish that assertive focus markers are excluded with negation; being itself the exponent of a focus domain, negation suspends or blocks the marking of any other element. That is also demonstrated by Yukaghir, a language spoken in several small communities in the Yakut (Saha) Republic in the North-East of Russia. Yukaghir has a complex focus marking system where four types can be distinguished: predicate, subject, direct object and neutral foci. In the case of predicate focus, the verb is marked by prefix me(r)- and is conjugated in person and number, as shown by example (13a) (Comrie 1992: 56).3 (13) a. met mer--jeù I FOC-go-1SG ‘I went’ In the case of subject focus, the verb keeps the number agreement marks, but looses those of person (Maslova, p.c.)); with transitive verbs, the verbal form corresponds to the root as in (13b); with intransitives however, the non-finite verb form is suffixed with an -l and the subject takes either suffix -le(ù ) (see (13c)), or -(e)k (see (13d)): (13) b. (ileù ) met ai reindeer I shoot-NONFIN ‘I shot the reindeer’

Negation and «Focus Clash» in Sardinian 139

-l c. ile-leù reindeer-FOC go-NONFIN ‘The reindeer went’ d. met-ek -l I-FOC go-NONFIN ‘I went’ As for direct object focus, the DP has the same form as the focal subject of an intransitive verb, which agrees in person and number with the subject: (13) e. met ile-leù ai-meù I reindeer-FOC shoot-1SG (obj-FOC) ‘I shot the reindeer’ The fourth type of focus is identical to the predicate focus, except that the verb is not prefixed with prefix me(r)-, a configuration defined by Comrie as neutral focus. With no explicit marking, this focus is realised when a constituent other than the subject, object or predicate is highlighted (13f) (Comrie 1992: 58–59): (13) f. i´dilwej-mirije la´til-ùi´n kewe-´t PN-woman firewood-DIR go-3SG ‘I´dilwej went for firewood / firewood’ The behaviour of negation seems very similar in Yukaghir and in the bantu languages just mentioned. The same structure as predicate focus sentences is manifested by negative sentences, with the caveat that the negative marker is in complementary distribution so to speak with the predicate focus marker. Examples (13g–h) show the absence of prefix mer- in the negative utterance which contains the negative marker el instead: (13) g. met mer--jeù I FOC-go-1SG ‘I went’ h. met el--jeù I NEG-go-1SG ‘I didn’t (did not) go’ Comrie thus concludes (p.64) that «the negative prefix is the focus, thus usurping the position of the predicate focus marker me(r) and also preventing focus from being marked on any other constituent of the clause».4

140 Franck Floricic A parallel case is found in Hungarian where verbal prefixes occupy the preverbal focus position that negative marker nem fills in negative sentences:5 (14) a. el-ment-él az iskolá-ba Away-go:PAST-2SG the school-DIR ‘You went away to school’ b. nem ment-él el az iskolá-ba Not go:PAST-2SG away the school-DIR ‘You didn’t go (away) to school’ The negative version of (14a) shows nem in preverbal focal position, with the verbal prefix relegated to another position (Dezs 1982: 161)6. Negation marking therefore patently clashes with focus marking, and a possible analysis of this clash is to assign negation an ‘on focused’ value. More support to this analysis is provided by the language of Gascony. Gascon is an Occitan language that is endowed with enunciative particles, notably que, be and e. Examples (15a) and (15b) show that particle que occupies preverbal position (Rohlfs 1970: 205–206): (15) a. Que las ey entenudos That CLDO3FPL have1SG heard.FPL ‘I have heard them’ b. Ta pay qu’ ey arribàt Your father that be3SG come ‘Your father has come’ The origins and nature of enunciative que is controversial; one of the most convincing hypotheses is that a former focus process is at play (Push 1998). The enunciative que in (15b) would be the trace of the focalisation of the preceding constituent, the focus moving from the nominal to the verbal. This focalisation marker, Push concludes on the basis of a corpus analysis, co-occurs very rarely with negatives7. Negation and ‘enunciative’ marker thus do look like they are in complementary distribution. Interestingly, Gascon imperatives also tend to exclude que, which might be explained if we assume that the imperative also has intrinsic focus which is itself incompatible with the focus trace of the preverbal marker que8. The data mentioned above show that negation may interfere with other types of morpho-syntactic markers in such a way as to reconfigurate the whole syntactic shape of the sentence.

Negation and «Focus Clash» in Sardinian 141

3.2. The notion of Focus Clash It is now possible to go back to the Sardinian data and to try to understand why negation is incompatible with the total negation marker on the one hand and with fronting on the other. In the light of typological data, this incompatibility, illustrated in (6) and (7), can be hypothesised to result from a focus clash. In other words, negation being the exponent of a specific judgement, it can enter in conflict with markers and configurations that mark the saliency of a constituent. It is however at the level of the predication globally and at that of the judgement of which it is a sign that the conflict arises: though in examples such as (5) the last stressed syllable of the fronted element alone bears main stress (cf. (5a) Torradu est?, (5b) Famidu ses?, etc.), focus extends to the whole predication. That is also the case of the Sardinian interrogative particle a, which essentially operates at the level of predication as a whole, not at the level of any of its constituents. Given that the negative particle no shares the same property, their incompatibility follows. This incompatibility can be represented in the following fashion: (16)

*[[ X ] [ Y ]] | | +F +F

The hypothesis can be made however that constituent question markers are compatible with negation because their focal value is circumscribed to a domain that does not concern the predication as a whole. Utterances such as (8) Ite cheres faghere? («What do you want to do?») involve a predicative scheme ((tu) fagher (y)) where what is queried only is the identification of the entity filling the second argument slot. Constituent questions are thus based on a presupposed or preconstructed predicative scheme that remains outside the scope of the question marker. Likewise in (8b) Chie / chie est chi no at appidu su café? («Who didn’t have the coffee?»), the question bears on the identification and extraction of the entity filling the first argument slot; the question is therefore built on the pre-existing scheme ((x) neg ((abere) (café))). It is precisely because of this preconstructed scheme that constituent questions are compatible with negation. It is in this respect interesting to look at the Paduan interrogative conjugation.

142 Franck Floricic 3.3. The interrogative conjugation in Paduan Paduan is a Romance language which, like other Northern varieties, has an interrogative conjugation. This can be agreed to emerge from a morphologisation of the enclitic in interrogatives (see Fava 2001)9. (17) The interrogative conjugation in Paduan (Zamboni 1974: 42) kántoÁi ‘do I sing?’ kántito ‘do you sing?’ kánteo (kántelo) ‘does he sing?’ kanté moÁi ‘do we sing?’ kantçéeo ‘do you sing?’ kánteÁi (kánteli) ‘do they sing?’

pçéerdoi ‘do I loose?’ pçéerdito ‘do you loose?’ pçéerde(l)o ‘does he loose?’ perdé moÁi ‘do we loose?’ perdío ‘do you loose?’ pçéerde(l)i ‘do they loose?’

As shown in (17), interrogative conjugation is the result of person markers’ affixation to the verb. The tableau given in (17’) further illustrates the extension of such affixation in the interrogative forms of the verb: it covers the whole paradigms, while Paduan assertive conjugation only shows person markers’ affixation in the second (singular) and third person forms. (17’) Assertive conjugation 1SG 2SG 3SG 1PL 2PL 3PL

 te el (M) / la (F)   i (M) / le (F)

Interrogative conjugation -i -to -lo (M) / -la (F) -i -o -li (M) / -le (F)

The interrogative form of the verb, let us note, is compulsory with partial and total questions. That is shown in examples (18), where the use of assertive forms produces ill-formed sequences (Portner & Zanuttini 1996: 257– 258). (18) a. El vien S.CL come ‘he comes’ b. Vien-lo? (* El vien?) come-S .CL ‘does he come?’

Negation and «Focus Clash» in Sardinian 143

c. Quando vien-lo? (*Quando el vien?) when come-S .CL ‘when does he come?’ d. La ga magnà tuto S.CL has eaten all ‘she ate everything’ e. Cossa ga-la magnà? (*Cossa la ga magnà?) what has-S.CL eaten ‘what did she eat?’ The fundamental point is that the interrogative form of the verb is incompatible with negation (Parry 1997: 182; Zanuttini 1997: 42): the assertive form of the verb is therefore required in negative interrogatives, the person marker cliticising unto the negative (Portner & Zanuttini 1996: 258–259; Zanuttini 1997: 46–47): (19) a. *No vien-lo? NEG come-S .CL ‘Doesn’t he come?’ b. *Cossa no ga-la magnà? What NEG has-S.CL eaten ‘What didn’t she eat?’ c. Nol

vien? come ‘Doesn’t he come?’

NEG -S.CL

d. No la ga magnà? NEG S.CL has eaten ‘Hasn’t she eaten?’ e. Cossa ze che no la ga magnà? What is that NEG S.CL has eaten ‘What it is that she didn’t eat?’ Paduan data show that it is perfectly possible to formulate a question in the negative; this possibility however imposes restructuring the morpho-syntactic make-up of the sentence. The restructuring involved is quite similar to that considered for Sardinian, and it is therefore reasonable to relate it to similar causes. Completeness brings us to mention that in Paduan negation co-occurs with the interrogative verb form in some contexts. Wh-exclamatives are one such context (Zanuttini & Portner 1996: 261–262; Zanuttini 1997: 53):

144 Franck Floricic (20) a. Cossa no ghe dise-lo! (* Cossa no (e)l ghe dise!) What NEG him say-S.CL ‘What things he is telling him!’ As the translation of (20a) indicates, the interrogative pronoun evokes as in many other languages a high degree value, whether this bears on the quantity or the quality of the concerned entities. The intensive reading results from a value that closes unto itself, starting and closing with the same subjective reference point. In this kind of sentence, the question-like morphosyntactic frame only superficially requires the hearer’s implication. If no implication is expected on the part of the hearer, it is because the dialogical structure is short-circuited; in this case, the subjective origin of the sentence turns out to be at the same time the terminus a quo and the terminus ad quem of the propositional content, and no fixed value can be brought about by the question marker. In other words, when no referential class can be associated with the interrogative marker (either because the verb valency is incompatible with such a marker or because the verb argument slots are already filled), the only available issue is to refer to the internal quantitative / qualitative make-up of the process.10 The negative exclamative in (20a) thus points to the exclusion of a content whose qualitative properties preclude that it should have been expressed: topologically, negation marks crossing not from the positive to the negative (the non occurrence of an expected event or object), but from the positive to the negative (the occurrence of a non expected event or object). On the other hand, where the negative clause is made dependent, it precludes the interrogative form of the verb (20b). (20) b. *Cossa che no ghe dise-lo! What that NEG him say-S.CL ‘What things he is telling him!’ The interrogative verb is also required in partial questions introduced by parcossa (‘why’) (20c), as long as the proposition is not embedded (Zanuttini 1997: 54)11; embedding imposes the assertive form of the verb, as shown in (20d): (20) c. Parcossa no ve-to anca ti? Why NEG go-S.CL also you ‘Why aren’t you going as well?’

Negation and «Focus Clash» in Sardinian 145

d. Parcossa ze che no te ve anca ti? Why be3SG that NEG S.CL go also you ‘Why aren’t you going as well?’ According to Zanuttini & Portner (1996: 262), the interrogative in (20c) would signal that the speaker knows that the interlocutor didn’t come, the question presupposing that the refusal to come is unjustified. (20d) would on the contrary be a simple request of information concerning the ‘non-going’. We can now turn back to the interrogative conjugation and ask whether there is any common denominator among the examples where the Paduan negative marker no is compatible with the interrogative verb form. We may suggest that negation is compatible with the interrogative verb form whenever the question is built upon a presupposed value or a pre-constructed propositional scheme (cf. Culioli 1999). In other words, negation is compatible with the interrogative verb form whenever interrogation takes as a structural basis a content which remains outside the scope of interrogation, or whenever it implies a polarization on a given value. Given that negation fundamentally operates at sentence level, its possible co-occurrence with interrogative focus on a given value of predication is thus accounted for. We also find now an explanation of compatibility of negation with interrogative markers whenever the content of predication is relegated to a dependent status. In examples such as (20c) – Parcossa ze che no te ve anca ti!? – we are dealing with a nuclear question whose focussed term is marked as such by the identifying copula ze; complementizer che signals that the whole predication is informationnally backgrounded. This backgrounding also is responsible for the use of assertive conjugation in this type of context – suffice it to mention French interrogatives such as C’est quand que tu viens? vs. * C’est quand que viens-tu? We may account in the same way for the presence of negation in Sardinian sentences like (8a) Ite / ite est chi no as cumpresu? or (8b) Chie est chi no at appidu su café?: identificational focus has as its domain the entity filling the direct object and the subject slots, and not the predication as a whole; in this case as well, backgrounding of the predicative core may account for the wellformedness of the sentence. It should be pointed out that interrogative sentences are not the only ones which require re-ordering of the sentence elements in presence of negation. Exclamatives show the same kind of restrictions, as illustrated in (21) (Jones 1993: 346): (21) a. Bellu est! Nice be3SG ‘It’s nice!’

146 Franck Floricic b.  Kale keres? Which want2SG ‘Which one do you want?’  Cudda keldzo! That one want1SG ‘That one I want!’ As Jones (1993: 338) puts it, «(…) the fronted item is interpreted as the focus of the sentence with an effect similar to that of cleft or pseudo-cleft constructions in English». Interestingly, none of the examples above allows as such the insertion of the negative marker no(n) (cf. *Bellu no est! and *Cudda no keldzo!). Putting to the fore an actual property or a given object thus excludes its concomitant and correlative denial or rejection.12 The only option once more is to reorganize the sentence elements (cf. No est bellu! and No keldzo cudda!): such a reorganization entails defocusing of the fronted elements in the examples in (21) and maintaining in preverbal position the only negative marker. Placed in post-verbal position, the adjective bellu and the deictic cudda are of course not deprived of informational saliency, but they are rather informationally integrated into the wider VP in the scope of negation. The same naturally holds for the interro-negatives in (7): the presence of negation implies the informational demotion of the fronted constituent, a demotion realized via syntactic reordering, leaving thus the sole negation in preverbal position.13

4. Conclusion It has been shown that Sardinian interrogatives resort to various strategies whose effect is to assign informational prominence on a given content; this is particularly clear with fronting strategy, in which the fronted constituent is a head and bears primary stress (Jones 1993: 332). This strategy has been argued to be incompatible with negation, as well as with the use of particle a, whose function is to assign focal value to the content in its scope. Naturally, interrogation and negation are not incompatible per se (see the very existence of negative interrogatives!), but their co-occurrence tends to induce syntactic restructuring; this in turn brings evidence that the exponents of these two major modalities can clash: this clash has been argued to depend on the ‘in focus’ value of both interrogatives and negation.

Negation and «Focus Clash» in Sardinian 147

Acknowledgements I wish to thank Pierre Larrivée and Elisabeth Connellan for the translation of this paper.

Notes 1.

2.

3. 4.

5. 6.

The interrogative particle a must not be confused with the homophonous preposition a, which can both occur in the same utterance. This is shown by examples such as A mamma tua, a l’as bida custu mandzanu? [a ‘mam: a ‘·ua a l al ‘vi·a ‘kustu man’dzanu] ‘Your mother, you saw her this morning?’ “Did you see your mother this morning?”, where the first a introduces the topical direct object, whereas the second marks a total question. Contini (1986: 531) classifies this particle amongst the unstressable monosyllabic forms inherited from latin forms with final consonant: e (< lat. et ‘and’), a (< lat. ad ‘to’), ne (< lat. nec ‘neither’). This final consonant would motivate the consonant lengthening of the following word (see examples such as 'a kkan'tamusu? ‘do we sing?’). According to Maslova (1997: 458), predicate focus sentences represent the neutral version of nominal focus constructions. Comrie mentions the exception of an object having an attributive complement; example (13i) shows that the object is marked for focus, with a still transitive verb – the intransitive conjugation being normally used with negated transitive verbs: (13) i. met amat e d-ile-k el -bu-meù I good-reindeer-FOC NEG-kill-1SG (Obj: FOC) ‘I did not kill the good reindeer’ Elena Maslova (p.c.) points out that in negative interrogatives such as kin-ek el-kel-ul (‘who didn’t come?’), the nuclear interrogative is marked with the focus mark -ek, despite the adjacency of negation. The complexity of the marking system would deserve a fuller assessment that cannot be done here. It is worth pointing out, however, that co-occurrence of focus and negative marking is allowed when focus is restricted to an interrogative constituent (vs. the predication as a whole). Hungarian verbal “prefixes” occur in preverbal position as a “default” rule and they constitute a prosodic unit with the verb (cf. Rainò 1984: 127). Actually, the preverb may co-occur with the negative, as demonstrated by (14c) from Dezs (1982: 161) (see also Rainò 1984: 128): (14) c. A fiú nem elolvassa az újságot ‘the boy does not peruse the newspaper (but he puts it on the fire)’ Dezs (op. cit.) points out that in this kind of context “(…) the verb is markedly emphatic and needs to be complemented”.

148 Franck Floricic 7. 8.

9.

10.

11.

12.

13.

Rohlfs (1970: 208) notes that the particle que is only used in assertives, while noting an extension to negative sentences in certain varieties. Comrie (1992: 63) also points out that Yukaghir predicate focus particle me(r) disappears in the imperative, where the verb only has person morphology (see i. and ii.): i. wie-k ii. kelu-k Make-IMPER: 2SG Come- IMPER: 2SG ‘Do [it]!’ ‘Come!’ It is particularly interesting to note that in the Kru languages signaled by Marchese (1983: 124), the imperative verb is incompatible with the assertive focus markers, as shown by the examples below in Guéré and Tepo respectively: iii. mu O8O iv. di dE Go market Eat thing ‘Go to the market!’ ‘Eat!’ *mu-e O8O *di dE nI Go-AF market Eat thing AF This is still a controversial point: some see a syntactic case of subject clitic inversion, others a morphological process that would justify the recognition of an interrogative conjugation distinct from the assertive one (see De la Grasserie 1899): this would be supported by the fact that the paradigms for prefixed series and for suffixed series are distinct (see Fava 1993 and 2001 for a discussion). See French sentences such as Qu’est-ce qu’il pleure?!, where the neutral interrogative marker qu’est-ce que doesn’t / cannot refer to any argument of the sentence; as default, it refers to the internal quantification of the process, or to its causal value. In sentences like Qu’est-ce qu’il a (pas) mangé comme cochonneries! on the other hand, both qualitative and quantitative polarization are available. It can be pointed out that « why » questions behave in a very different way from other question types, because such questions do not bear on the identification of a given entity or on the circumstances of the event, but rather on the causal relationship between such and such event. All things being equal, the same reasons exclude negation in French sentences such as Ya le train qui arrive (cf. *Ya pas le train qui arrive): it would be somewhat nonsensical on the one hand to bring about the existence of a given entity or situation, and at the same time to deny or reject it. This phenomenon, it will be mentioned, is reminiscent of a well known kind of prosodic restructuring that occurs when adjacent stressed constituents are to be found: configurations such as Italian metà tempo [me'ta 'tempo] ‘part time’ are thus “repaired” via stress retraction, hence ['meta 'tempo] (cf. Nespor & Vogel 1989). It is then tempting to ascribe to the same general principle – some kind of OCP constraint – the raison d’être of these apparently very different syntactic and phonological phenomena.

Negation and «Focus Clash» in Sardinian 149

References Andersen, Henning (ed.) 1986 Sandhi Phenomena in the Languages of Europe. (Trends in Linguistics. Studies and Monographs 33). Berlin /New York: Mouton de Gruyter. Aronson, Howard I. (ed.) 1992 The non-Slavic languages of the USSR. Chicago: Chicago Linguistic Society. Benincà, Paola 2001 La struttura della frase esclamativa alla luce del dialetto padovano. Quaderni di Lavoro dell’ASIS 3: 13–60. Frasi esclamative e strutture correlate. Benincà, Paola, Gugliemo Cinque, Tullio De Mauro, and Nigel Vincent (eds.) 1997 Italiano e dialetti nel tempo. Saggi di grammatica per Giulio C. Lepschy. Roma: Bulzoni. Cinque, Guglielmo 1991 ‘Mica’: note di sintassi e pragmatica. In Teoria linguistica e sintassi italiana. Bologna: Il Mulino. Comrie, Bernard 1992 Focus in Yukaghir (Tundra dialect). In H. I. Aronson (ed.), 55–69. Contini, Michel 1985 L’intonation des phrases affirmatives et interrogatives avec inversion en sarde. Analyse et synthèse. Actes du XVIIème Congrès International de Linguistique et de Philologie Romanes (Aix-en-Provence, 29 août – 3 septembre). Linguistique descriptive: phonétique, morphologie et lexique, Vol. 3, 15–29. Aix-en-Provence: Université de Provence. 1986 Les phénomènes de sandhi dans le domaine sarde. In H. Andersen (ed.), 519–550. Culioli, Antoine 1999 Pour une linguistique de l’énonciation. Domaine notionnel. Tome 3. Paris, Ophrys. Dezs, Lazslo 1982 Word order types, topic and comment in Hungarian and Russian. In Studies in Syntactic Typology and Contrastive Grammar, 103–210. The Hague/ Paris/ New York: Mouton Publishers. Erteschik-Shir, Naomi 1992 Focus structure and predication. The case of negative Wh-Questions. Belgian Journal of Linguistics 7: 35–51. Fava, Elisabetta 1993 Sulla pertinenza della pragmatica nell’analisi grammaticale: un esempio della cosiddetta coniugazione interrogativa nel dialetto alto-vicentino. In M. A. Cortelazzo, E. Leso, P. V. Mengaldo et al. (eds.), 2495– 2520.

150 Franck Floricic Fava, Elisabetta 2001 On word, clitic and affix distinctions in some North Eastern Italian Dialects. In C. Muller et al. (eds.), 365–379. Feuillet, Jack 1994 Typologie de l’interrogation globale. In Interrogation. Des marques aux actes, Paul Boucher and Jean-Michel Fournier (eds.), 7–33. Rennes: Presses Universitaires de Rennes. Foulet, Lucien 1921 Comment ont évolué les formes de l’interrogation. Romania 47: 243– 348. Givón, Talmy 1979 On understanding grammar. New York /San Francisco /London: Academic Press. De la Grasserie, Raoul 1899 De la conjugaison négative ainsi que de l’interrogative et de la dubitative. Louvain: J. B. Istas (Extrait du Muséon). Haiman, John 1991 From V/2 to subject clitics: evidence from Northern Italian. In E. C. Traugott and B. Heine (eds.), 135–157. Hyman, Larry M. 1984 Auxiliary Focus. Studies in African Linguistics 15 (3): 233–273. Jones, Michael A. 1988 Auxiliary verbs in Sardinian. Transactions of the Philological Society 86 (2): 173–203. 1993 Sardinian Syntax. London /New York: Routledge. Maiden, Martin and Mair Parry (eds.) 1997 The Dialects of Italy. Cambridge: Cambridge University Press. Marchese, Lynell 1983 On assertive focus and the inherent focus nature of negatives and imperatives: evidence from Kru. Journal of African Languages and Linguistics 5: 115–129. Maslova, Elena 1997 Yukagir Focus in a typological perspective. Journal of Pragmatics 27: 457–475. 2005 Information structure in Tundra Yukaghir and typology of focus structures. In M. M. Jocelyne Fernandez-Vest (ed.), 599–610. Meyer-Lübke, Wilhelm 1890 [1906] Grammaire des langues romanes. II. Morphologie. III. Syntaxe. (transl. Auguste et Georges Doutrepont). Genève: Slatkine Reprints / Marseille: Laffitte Reprints. 1903 Zur kenntnis des Altlogudoresischen. Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften zu Wien. Phil. Hist. Kl. 145 (5), 1–76.

Negation and «Focus Clash» in Sardinian 151 Molinu, Lucia 1999 Morfologia logudorese. In La lingua sarda. L’identità socioculturale della Sardegna nel prossimo millennio, Atti del Convegno di Quartu Sant’Elena, 9–10 maggio 1997, Roberto Bolognesi and Karin Helsloot (eds.), 127–136. Cagliari: Condaghes. Muller, Claude 1994 La négation comme jugement: une application aux interronégatives. LINX (special issue “La Négation”): 205–221. Nespor, Marina and Irene Vogel 1989 On clashes and lapses. Phonology 6: 69–116. Parry, Mair M. 1997a La negazione italo-romanza: variazione tipologica e variazione strutturale. In P. Benincà, G. Cinque, T. De Mauro, and N. Vincent (eds.), 225–257. 1997b Negation. In M. Maiden and M. Parry (eds.), 179–185. Pittau, Massimo 1972 Grammatica del sardo-nuorese, il più conservativo dei parlari neolatini. Bologna, Riccardo Pàtron Poletto, Cecilia 1996 Three kinds of subject clitics in Basso Polesano and the theory of pro. In A. Belletti and L. Rizzi (eds.), 269–300. 1997 Pronominal syntax. In M. Maiden and M. Parry (eds.), 137–144. Portner, Paul and Raffaella Zanuttini 1996 The syntax and semantics of scalar negation: evidence from Paduan. In K. Kusumoto (ed.), 257–271. Push, Claus D. 1999 La phrase principale affirmative en gascon: un cas de focalisation figée? In C. Guimier (ed.), 107–119. Rainò, Nicola 1984 Los prefijos verbales en húngaro. A propósito de algunos estudios recientes. Estudios de Lingüística 2 (Universidad de Alicante), 121– 135. Reinach, Adolf 1911 Zur Theorie des negativen Urteils. In Münchener Philosophische Abhandlungen. Festschrift für Theodor Lipps, A. Pfänder (ed.), 196–254. Leipzig: J. A. Barth, (also in Gesammelte Schriften, A. Reinach, 56 – 102. Halle, 1921: Max Niemeyer. (English translation by Don Ferrari: Toward the Theory of the Negative Judgement. Aletheia 2, 1981: 15– 64; English translation by Barry Smith: On the Theory of the Negative Judgment. In Parts and Moments, B. Smith (ed.), 315–337. München, 1982: Philosophia; French translation by Marc B. de Launay: Théorie du jugement négatif. In Revue de Métaphysique et de Morale 96, 1996: 384–436.

152 Franck Floricic Rohlfs, Gerhard 1968 Grammatica storica della lingua italiana e dei suoi dialetti. Morfologia. Torino: Giulio Einaudi 1969 Grammatica storica della lingua italiana e dei suoi dialetti. Sintassi e formazione delle parole. Torino: Giulio Einaudi 1970 Le gascon. Études de philologie Pyrénéenne. Tübingen: Niemeyer. Tesnière, Lucien 1988 Eléments de Syntaxe Structurale. Paris: Klincksieck. Virdis, Maurizio 1987 Appunti per una sintassi del sardo. Biblioteca Francescana Sarda 1: 409–440. Wagner, Max Leopold 1938 Flessione nominale e verbale nel sardo antico e moderno. L’Italia Dialettale 14: 93–170. 1941 Historische Lautlehre des Sardischen. Halle, Max Niemeyer. (trad. it. de Giulio Paulis avec introduction et appendice, Fonetica storica del sardo. Cagliari, Gianni Trois Editore 1984) Zamboni, Alberto 1974 Veneto. Pisa: Pacini Editore. (Profilo dei Dialetti Italiani 5.) Zanuttini, Raffaella 1997 Negation and Clausal Structure: A Comparative Study of Romance Languages. New York /Oxford: Oxford University Press.

Do-support in a Sicilian variety, an Italian pseudo-cleft, and the packaging of information Ignazio Mauro Mirto

A parallel is drawn between a Sicilian structure with do-support strategy, and a type of pseudo-cleft in Italian in which a do-verb occurs. These structures are examined from three different perspectives. First, morpho-syntax: the two clause-types share a number of formal aspects, which include the do-verb, a lexical verb in the infinitive, and the sharing of the same subject by both verbs. The second has to do with their packaging of information: in both cases, higher communicative prominence is given to an event (VP) rather than a participant (NP or PP). The expression of the event is broken down into two separable components, its existence (an event is expressed without being specified, e.g. All she does is complain), and denomination (i.e. the naming of the event, e.g. All she does is complain). These components can be considered as features encapsulated in the do-verb and the lexical verb in the infinitive respectively. Such encoding is closely related to the third perspective called for in the analysis: pragmatics. Both clausetypes presuppose the existence of a certain event, which works as a background against which only the denomination of the event is foregrounded.

1. Introduction This paper deals with the information structure of utterances such as (1), which exemplifies a Sicilian clause-type constructed with the verb fari ‘do’, and spoken in one area of Sicily:1 (1)

Babbiari fa. to.joke does ‘He is (only) joking.’

This periphrastic clause-type, previously discussed in Mirto 2004/to appear, is practically unknown either in the literature on the subject or in scholarship on the Sicilian language, and coexists with the non-periphrastic type illustrated in (2):

154 Ignazio Mauro Mirto (2) Babbia. joke.3RD.SG ‘He is joking.’ For reasons that will shortly become evident, the analysis of this Sicilian structure will be conducted in parallel with that of Italian sentences such as (3): (3)

Ciò che fa è scherzare. this that does is to.joke ‘What he does is (to) joke.’

Also this type of construction, a pseudo-cleft in which the information unit with higher salience, i.e. scherzare ‘to joke’, is an infinitive (see Salvi 1991: 177–189), regularly has a non-cleft counterpart, as in (4), structurally parallel to (2): 2 (4)

Scherza. joke.3RD.SG ‘He is joking.’

The sentences (1) and (3) must be distinguished from (2) and (4) respectively, because only the former structures convey the presupposition of the existence of an event. It is argued (a) that (1) and (3) share the presupposition of an unspecified event; (b) that this presupposition is induced in distinct ways; (c) that in both clause-types its presence calls for some sort of do-support strategy, and (d) that the clause-types exemplified with (1) and (3) fulfil the same goal, namely a division into information units in which that with higher salience only includes the denomination of a certain event, which in (1) and (3) is formally expressed with a lexical verb in the infinitive. On account of discourse factors, the mere existence of an unspecified event can be presented as presupposed (‘mere existence’ meaning ‘without lexically specifying the nature of the event’).3 If this is the case, the speaker’s choice regarding the appropriate structure will be constrained, considering that non-periphrastic structures (e.g. (2) or (4)) are not suitable to achieve this purpose, whilst clause-types with a do-verb such as those of (1) and (3) are. A few words of caution with regard to terminology are called for. It is well-known that work with the “structure of utterances according to com-

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 155

municative criteria” (Bussmann 1996: 483) is made more difficult by the terminological hotch-potch found in the literature (see Chafe 1992: 215; Gundel 1988: 209; Kotschi 2006: 677; Sornicola 2006: 766). A number of the dichotomies commonly found, all conceptually different from each other (see Lombardi Vallauri 2002), are: ‘given/new information’, ‘topic/comment (or topic/focus)’, and ‘theme/rheme’. The last pair will be used in the following pages, though – given the data examined – ‘given/new’ and ‘topic/ comment’ could be considered equally suitable choices. For space reasons, and to partially avoid the complication of this terminological variation, in the following analysis of the internal organization of information the clause-type illustrated with (1) will be considered with no regard to marked stress patterns, and that exemplified in (3) as characterized by the typical intonation and word order patterns of basic (i.e. non reversed) pseudo-clefts (see Di Tullio 2006: 489). Section 2 offers a pre-theoretical description of the syntactic properties of both clause-types that turn out to be pertinent to their information structure. Section 3 shows that both clause-types presuppose the existence of an unspecified event, and the different modes in which their common presupposition is induced. Section 4 briefly illustrates word order in Sicilian, whilst Section 5 presents an analysis of their information structure. The last section is dedicated to a few concluding remarks.

2.

Syntactic properties related to information structure

2.1. Italian Salvi (1991: 180-189) inserts Italian pseudo-clefts such as (3) in a section devoted to copular sentences. In (3) the copula is preceded by what appears to be a nominal relative clause and followed by an infinitive: Ciò che fa ‘what he does’, and scherzare ‘to joke’ respectively. According to Gundel (2006: 518), “Most dialects of English allow the clefted (sic) constituent to be an adjective or VP only in wh-cleft/pseudocleft”. McArthur’s definition (1992: 222) is more restrictive: “Unlike the cleft sentence, the pseudo-cleft can have a verb (and other elements that follow it) as the focused item: What we did was replace all the carpets.”.4 In McArthur’s parlance, the focused information unit of (3) is the uninflected verb scherzare ‘to joke’, a non-argumental phrase in the corresponding non-cleft (4). The referent of this discourse entity is not an object/individual (as, for instance, would be the case with Mary in the it-cleft

156 Ignazio Mauro Mirto It is Mary I asked). Nevertheless, some analysts maintain that the referential status of NPs and VPs can be “treated analogously” (Kotschi 2006: 678). As observed above, in Italian, but also in other languages, this type of pseudo-cleft requires a do-verb in the relative clause. The verb fare in pseudo-clefts such as (3) is termed “vicario” (an Italian word translatable as ‘proxy’) by Salvi (1988: 82–85).5 The verb fare, in this context, appears to be cataphorically linked to the infinitive scherzare ‘to joke’.6 In (3), fare carries the person and number of the subject, as in the English pseudo-cleft (5). Sentences (3) and (5) show that the copula and the do-verb share the same tense and mode. (5)

What he does is (to) complain.

In the analysis put forward in this study, a crucial syntactic property of the pseudo-cleft in (3) has to do with the subject of the infinitive scherzare ‘to joke’. Unlike the infinitive occurring in Italian sentences such as Scherzare è sempre piacevole ‘Joking is always fun’, or that of impersonal sentences such as Non bisogna scherzare su certe cose ‘One must not joke about certain matters’, in (3) scherzare ‘to joke’ turns out to have a subject that can only be interpreted as co-referential with that of fare (see Salvi 1991: 185).

2.2. Sicilian The Sicilian causative fari ‘make/let’, as in Nni fici ririri ‘He/she made us laugh’, works just like its Italian counterpart fare. Whilst causative fare introduces a new subject in the clause (see La Fauci and Mirto 2003), the fari in (1) is zero-valent: the two verbs combine and the outcome is a monoclausal sentence (see Rosen 1997). Mirto (to appear) analyzes this fari as a function-word (an auxiliary). The Sicilian construction is found in the affirmative form, as in (2) and (6), but also in the negative,7 and in the interrogative, as in (7) and (8)8 respectively:9 (6)

U Palermu sempri pareggiari fa. the Palermo always to.tie (a match) does ‘The Palermo team always draws (in its matches).’

(7)

Cu stu vastuni mancu grapiri po fari. with this stick not.even open can.you do ‘With this stick, you can’t even open (it) [it = the curtain].’

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 157

(8)

L’armali sempri suppurtari avi a fari? the animal always put.up.with has to do ‘Does the pet always have to put up with (this)?’

Some of the syntactic features of this construction related to its information structure are: (a) in all the utterances collected, the subject is either omitted, as in (2) and (7), or definite as in (6) and (8); (b) as with the pseudo-cleft in (3), the two verbs always share the same subject. Thus the ‘doer’ is the team whose match ends in a draw in (6), the person who cannot open the curtain in (7), and the pet that always has to put up with something in (8); (c) the absolute (i.e. intransitive, see Bussmann 1996: 2) use of the verb, as in (7) and (8), in which no lexical object occurs along with the lexical verbs grapiri ‘open’ and suppurtari ‘to put up with’, both allowing for transitive structures. Occasionally, a pronominal direct object occurs, as in (9) and (10), where the clitic is in bold print: (9)

U piattu sempri lavari l’am(a) a fari. the dish always wash it.have.we to do ‘In any case, this dish has to be washed (by us).’

(10) (Sti cristiani) mancu licinziari i po fari. RD these people even sack 3 .M.PL can.you do ‘That lot cannot even be sacked.’ (d) most of the utterances contain a sentence adverbial, for instance sempri ‘always’, as in (6), (8), and (9) (the sentence is then in the affirmative) or mancu ‘not even’, as in (7) and (10) (in this case the sentence takes on a negative meaning). These adverbs always precede the lexical verb.10

3.

Presuppositions

3.1. Italian If Max were the subject of fare in (3), as in the pseudo-cleft below, only Max could be the joker:

158 Ignazio Mauro Mirto (11) Ciò che Max fa è scherzare. this that Max does is to.joke ‘What Max does is (to) joke.’ It can be shown that (11) conveys the presupposition in (12): (12) Max fa qualcosa. Max does something ‘Max does something.’ The semantic relationship between (11) and (12), a truth relation (see Saeed 1997: 94, 98), can be demonstrated with a test on negation11 (see Stalnaker 1973 [1970]: 520, and the comment by Levinson (1983: 182) on Karttunen’s list of presupposition-triggers):12 (12) turns out to be entailed not only by (11) but also by (13), the sentence resulting from negating the main verb in (11). What follows is that (11) presupposes (12) (p-entailment, see Seuren 2006). (13) Ciò che Max fa non è scherzare. this that Max does not is to.joke ‘What Max does is not joking.’

3.2. Sicilian Besides the marker of negation mancu ‘not even’, the Sicilian clause-type can also be negated with nun ‘not’, as in the reply in (14b) below. Unlike mancu, the negative adverbial nun immediately precedes the inflected verb.13 (14) a. Speaker 1: Question Ma tua figlia si prepara da mangiare da sola? but your daughter REFL prepares to eat by alone ‘But does your daughter prepare her own food?’ b. Speaker 2: Answer Sì, sì. Studiari nun fa. yes, yes study not does ‘Yes, that is so. What she does not do is study.’ Significantly, the answer to the question above cannot be Nun studia ‘She does not study’, the non-periphrastic counterpart of (14b). In (14), such a

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 159

reply would simply be inappropriate. This inappropriateness depends on discourse factors (see Van Valin 2001: 328). The first part of the answer, “sì, sì”, provides confirmation that the sentence ‘the daughter does something’ is ‘common ground’ for both speakers (see Horn 1992). The VP of the non-periphrastic Nun studia functions as an assertion with no background assumption. On the other hand, the VP of the periphrastic clausetype conveys a background assumption that contrasts with what the sentence asserts (see Levinson 1983: 173). A good translation of (14b) thus appears to be ‘True, she (= my daughter) does prepare her own food; but what she does not do is study”. It is worth observing that the same result would be achieved with a presupposition induced by an adverb of contrast such as però ‘but’: Sì, sì. Nun studia però ‘Yes, that is so. But she does not study’. Another appropriate translation could then be: ‘Yes, she does do something, but that something is not studying’. What then appears to make (14b) suitable for the question in (14a) is its conveying the presupposition ‘She (= my daughter) does something’.14 As previously noted, most of the utterances contain a sentence adverbial. According to Saeed (1997: 98): “Many presuppositions are produced by the presence of certain words”. These presupposition-triggers can be factive predicates, viz. verbs, (Saeed 1997: 98f.) and adjectives (Chierchia 1997: 136–138), or time adverbial clauses (Levinson 1983: 182). The presupposing force of adverbials such as ‘(not) even’ is well-known. For instance, the presupposition in (7) and (10) arises from the presence of mancu ‘not even’ (for its Italian counterparts manco, nemmeno, see Manzotti and Rigamonti 1991: 309–317). There are valid reasons to analyze the adverb found in most of the utterances as a lexical trigger of presupposition. In the discourse below, this appears to be the case with ancora ‘still, more’ (see Lombardi Vallauri’s (2002: 12) brief remarks on this and other adverbials in Italian): (15) a. Speaker 1 Ora mi manciu a nsalata. now for.me eat.1st.SG the salad ‘I am now going to eat the salad.’ b. Speaker 2 Ancora manciari a fari? now to.eat have.to.2 ND.SG do ‘You are not going to continue eating, are you?’

160 Ignazio Mauro Mirto The presupposition in (15b) is that the referent of the 2nd person subject had been eating. The role played by the adverbial in triggering the presupposition is here apparent. This is also the case in (16), where the sentence adverbial is puru ‘also, even’. (16) Puru lamintarisi fa. also complain does ‘On top of it all, he/she is complaining.’ In (16), the assertion is “He/she complains”, but this works against the background triggered by puru ‘also’, which can be expressed as follows “He/she has done (a number of) things, and in addition to this…”. As with other adverbials, this background includes the existence of (at least) an event in which the subject is a ‘doer’. In only a few of the utterances collected, no adverb occurs, as in (17), addressed to a person who was about to get into a car and ready to leave. In cases such as (17), the presupposition concerning the existence of a given event is brought about by the extra-linguistic context. (17) Iri m’Palermu fai? to.go in Palermo do.you ‘Is it Palermo you are heading for?’ The Sicilian clause-type thus appears to convey the same presupposition conveyed by the Italian pseudo-cleft. In Sicilian, however, the presupposition is either lexically induced by means of adverbials or due to inferences arising from the context in which the sentence is uttered, whereas in Italian pseudo-clefts such as (3) this is not the case. 4. Word order in Sicilian

A number of constructions in Sicilian are verb-final (see Rohlfs 1969: 323f.). When the inflected verb is an auxiliary, for instance, it generally occupies the last position, as in the sentences below: (18) Friddu fa. cold does ‘It is cold.’

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 161

(19) Stancu si. tired are.2 ND.SG ‘You are tired.’ Such sentences are generally entirely rhematic, but contexts in which only the lexical item is highlighted (e.g. a contrastive context) cannot be excluded. If this is the case, in (18) and (19) the verbs fari ‘do’ and essiri ‘be’ carry thematic information, whilst the content-words friddu ‘cold’ and stancu ‘tired’ are construed as the sentence rheme.

5. Existence and Denomination In addition to the presupposition they have in common, (1) and the pseudocleft (3) share other important aspects. First, they both contain a single lexical word, i.e. babbiari in (1), and scherzare in (3) (recall that the verbs fari/fare are analyzed as auxiliaries). Second, in both this content-word is a verb in the infinitive. Third, both clause-types contain a do-verb. Fourth, the subject of the do-verb is also that of the lexical verb in the infinitive, and fifth, given the word-order differences between Italian and Sicilian, in the case of unmarked linear order the content-word of both (1) and (3) is rhematic information, as the tables below illustrate: Table 1. Information packaging in (1) Babbiari

fa.

RHEME

THEME

Table 2. Information packaging in (3) Ciò che fa è

scherzare.

THEME

RHEME

The difference between a Sicilian sentence such as (1) and the corresponding non-periphrastic (2), i.e. Babbia ‘He is joking’, can be conceived of as follows: in both clause-types, the VP carries (among others) two pieces of information: (a) it says that an event exists, i.e. ‘someone is doing something’, and (b) it specifies the nature of the event. In (2) this is done synthetically and simultaneously, that is in a single verbal form, whereas in (1)

162 Ignazio Mauro Mirto the same amount of information is encoded analytically: the existence of a given event (‘someone’s doing something’) is encapsulated in the verb fari, whereas the infinitive, a content-word, specifies the nature of the event. Mirto (2004) advances the idea that these distinct pieces of information can be regarded as features: the first, [E], concerns the existence of an unspecified event (it provides “the spatiotemporal anchoring of the event being expressed”, Kotschi 2006: 678); the other, [D], is the denomination of a certain event by means of a word that carries descriptive (lexical) content. In the non-periphrastic VP in either (2) or (4), [E] and [D] are distributed in a single word (though in two morphemes carrying lexical and grammatical meaning respectively, e.g. the root scherz- and the suffix -a), whilst in the periphrastic (1) they are distributed in distinct words. The latter distribution requires two types of words (or morphemes): [E] requires grammatical-words (or morphemes), whereas [D] needs content-morphemes. In the Sicilian clause-type in (1) and in the Italian pseudo-cleft in (3), [E] is encapsulated in the theme, and [D] in the rheme. Thus, in case of unmarked word-order, the packaging of the information in such clause-types can be summed up as the encapsulation of Denomination in the rheme only. Another Italian construction, termed esclusivo-restrittivo by Manzotti and Rigamonti (1991: 316f.), works in the same way (see Mirto 2004 /to appear): (20) Non ha fatto che scherzare tutto il pomeriggio. nothe.has done that to.joke all the afternoon ‘He has done nothing but joke all afternoon long.’ Some English expressions with do-support work in a similar fashion, as happens in the translation in (20). Consider also (21) and (22), a couple of possible answers to the wh-question What did he do? (which presupposes ‘he did something’, see Levinson 1983: 184): (21) Some shopping. (22) He shopped. In such a case, the ellipsis in (21) comes from He did some shopping (a support verb construction), which differs from (22), He shopped, in the ways [E] and [D] are distributed:

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 163 Table 3. Information packaging in English with do-support He did

some shopping.

[E]

[D]

THEME

RHEME

Table 4. Information packaging in English without do-support He

shop-(p)ed. [D] [E]

THEME

RHEME

In table 3, i.e. with do-support strategy, the rheme contains [D] only, whereas in table 4, without do-support, the rheme contains both [D] and [E]. In the pseudo-cleft in (3), [D] and [E] are split as in table 3: table 5 illustrates that [E] is encapsulated in the do-verb and [D] in the infinitive. Table 6 shows the same distribution for the Sicilian clause-type. Table 5. Distribution of [E] and [D] in (3) What Ciò che

he does fa

is è

[E]

joke scherzare [D]

THEME

RHEME

Table 6. Distribution of [E] and [D] in (1) joke Babbiari

he does fa

[D]

[E]

RHEME

THEME

The feature [E] in table 3 and that in table 5 have different origins. The first one comes from the question What did he do? and thus originates outside the sentence, whilst the second arises from the subordinate clause Ciò che fa ‘What he does’ contained in the pseudo-cleft. Nonetheless, in both cases the presupposition of existence seems to be induced by pronouns such as what.

164 Ignazio Mauro Mirto 6. Concluding remarks In spoken discourse, Sicilian periphrastic sentences such as (1) and their nonperiphrastic counterparts are generally not interchangeable. Not surprisingly, the two clause-types turn out to be semantically different. The presupposition in the Sicilian periphrastic clause-type and its packaging of information were shown to be parallel to that of pseudo-clefts such as (3), which accounts for how Babbiari fa and Babbia differ semantically and pragmatically. The sentences examined show that there are constructions in which the information contained in lexical verbs is broken down into two distinct components: the existence of an unspecified event and its denomination. The linguistic material necessary to obtain this separation includes a doverb and an uninflected verbal form such as an infinitive. With unmarked word order, the information is then packaged to give communicative prominence to only one of these components, i.e. denomination. The other component, the existence of an event, is presupposed and packaged as part of the sentence theme. No light on the two clause-types and on their common features can be shed by approaching the comparison from a single viewpoint. A thorough understanding of their similarities calls for an analysis in which morphosyntax, information structure, and pragmatics converge.

Acknowledgements I owe a special thank-you to Nunzio La Fauci for his valuable comments on a previous draft of this work. I would also like to thank two anonymous reviewers for their comments and suggestions. Any errors or oversights remain my responsibility.

Notes 1.

2.

This area includes Cefalù and Pollina, two small towns on the northern coast of Sicily, midway between Palermo and Messina. All instances of the Sicilian clause-type are authentic utterances. The clause-type in (1) is also found in the regional variety of Italian (Italiano regionale di Sicilia). If compared to the non-periphrastic types in (2) and (4), which have a wider distribution in Sicilian and Italian respectively, the clause-types in (1) and (3) appear to be marked.

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 165 3.

4.

5.

6.

As an example, consider a sentence such as Someone did something, which asserts the existence of an event without specifying what it was, in contrast, for instance, with Someone bought something, which also asserts the existence of an event but simultaneously specifies its nature. Di Tullio (2006: 485) notices that clefts are unable to foreground clauses: “Clauses, in contrast, are nearly exclusively foregrounded by means of basic pseudocleft sentences. In these, free relatives may introduce lexical units that are not present in the simpler sentence: the verb in (5c) [What Jeremy did was to leave the paper on the desk yesterday] signals that the focused nonfinite clause is headed by an action verb”. For other proxy uses of do-verbs, see what Tsujioka writes about the Japanese verb suru (2002: 154 –157), and Mirto 2003 for additional such uses of Italian fare. A comparison can help shed light on the cataphoric value of the do-verb in (3). Consider (i), a right-dislocated sentence in which the pronoun lo ‘him’ is cataphorically connected to the referential NP l’idraulico ‘the plumber’. It is common knowledge that a pronoun carries only grammatical, rather than lexical, information. In (i) the pronoun guarantees the existence of a referent while not specifying what this referent is.

(i) Luca lo chiamò, l’idraulico. Luca him called the plumber ‘Luca called the plumber.’ In a parallel way, in a pseudo-cleft such as (3), the do-verb fare carries grammatical information only, but it also guarantees the existence of an event, though it does not specify its nature. The parallel between (i) and (3) can be schematically represented as in the table below: DENOMINATION EXISTENCE

7.

8.

9.

lo – +

l’idraulico + +

fa – +

scherzare + –

A different type of negative form, with nun ‘not’ rather than mancu ‘not even’, is analyzed in section 3. In the collected utterances, nun occurs less frequently than mancu ‘not even’. (7) was uttered during a conversation in which a stick had been used while trying to open a curtain. (8) was used after a reaction by the family cat due to its undergoing repeated annoyance. The Sicilian clause-type is dissimilar to a construction with do-support in a dialect spoken in Lombardy (Benincà and Poletto 2004), which occurs only in the interrogative form. A Piedmontese construction, analyzed in Ricca 2001, presents similarities with the clause-type in (1), but differs in other ways, for example in allowing negation of the lexical verb: Faccio che non produrlo (I.do that not to.produce.it) ‘I don’t produce it.’ (from Ricca 2007).

166 Ignazio Mauro Mirto 10. Another adverbial is sulu ‘just, only’. In the few utterances collected in which sulu occurs, the adverb still regularly precedes the infinitive, but the two verbs appear in the reversed word order, as (i) illustrates: (i) I turisti fannu sulu dormiri. the tourists do just sleep ‘The only thing (those) tourists do is sleep.’ A remarkable aspect of the utterances with sulu is their word order, identical to that of causatives (in which, however, a pronominal or lexical object often occurs). 11. In discussing pairs of utterances in English such as Nielsen won and It was Nielsen who won, Gundel (2006: 517) notes that only in the latter, an it-cleft, “it is taken for granted that someone won […] made structurally explicit by being expressed as a separate constituent in the embedded cleft clause” and that “this would be equally true in the negated version, It wasn’t Nielsen who won, which has the same presupposition”. For presuppositions in Italian itclefts, see Chierchia 1997: 133 f. 12. Stalnaker (1973 [1970]: 520) distinguishes between semantic presupposition and pragmatic presupposition. Only the former, which Stalnaker ascribes to Van Frassen, is involved in the examples considered. 13. In (14b), the position of the marker of negation nun ‘not’, distinct from that of mancu ‘not even’, might be due to its being a clitic. The sentence has no causative interpretation (see fn. 10). 14. Particles such as but give rise to ‘Gricean conventional implicatures’, which according to Levinson (1983: 127) “are non-truth-conditional inferences that are not derived from superordinate pragmatic principles like the maxims, but are simply attached by convention to particular lexical items or expressions”. In Horn’s opinion (1992: 262), “the addressee is entitled to draw certain inferences about the speaker’s beliefs and intentions, based on non-truth-functional components of the utterance”.

References Benincà, Paola and Cecilia Poletto 2004 A case of do-support in Romance. Natural Language and Linguistic Theory 22: 51–94. Bussmann, Hadumod 1996 Routledge Dictionary of Language and Linguistics [translated and edited by Gregory Trauth and Kerstin Kazzazi]. New York / London: Routledge. Chafe, Wallace 1992 Information Flow. In International Encyclopedia of Linguistics, Vol. 2, William Bright (editor in chief), 215–218. New York /Oxford: OUP.

Do-support in a Sicilian variety, Italian pseudo-cleft, packaging of information 167 Chierchia, Gennaro 1997 Semantica. Bologna: il Mulino. Di Tullio, Ángela 2006 Clefting in spoken discourse. In Encyclopedia of Language and Linguistics, Vol. 2, Second Edition, Keith Brown (editor-in-chief), 483– 491. Oxford: Elsevier. Gundel, Jeanette K. 1988 Universals of topic-comment structure. In Studies in Syntactic Typology, Michael Hammond, Edith A. Moravcsik, and Jessica R. Wirth (eds.), 209–242. (Typological Studies in Language 17) Amsterdam / Philadelphia: John Benjamins. 2006 Clefts in English and Norwegian: Implications for the grammar-pragmatics interface. In The Architecture of Focus, Valéria Molnár, and Susanne Winkler (eds.), 517–548. (Studies in Generative Grammar 82) Berlin /New York: Mouton de Gruyter. Horn, Laurence R. 1992 Pragmatics, Implicature, and Presuppositions. In International Encyclopedia of Linguistics, Vol. 3, William Bright (editor in chief), 260– 266. New York /Oxford: Oxford University Press. Kotschi, Thomas 2006 Information structure in spoken discourse. In Encyclopedia of Language and Linguistics, Vol. 5, 2nd Ed., Keith Brown (ed.), 677–683. Oxford: Elsevier. La Fauci, Nunzio and Ignazio Mauro Mirto 2003 Fare. Elementi di sintassi. Pisa: ETS. Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press. Lombardi Vallauri, Edoardo 2002 La struttura informativa dell’enunciato. Milano: La Nuova Italia. Manzotti, Emilio and Alessandra Rigamonti 1991 La negazione. In Grande grammatica italiana di consultazione, Vol. II, Lorenzo Renzi and Giampaolo Salvi (eds.), 245–320. Bologna: il Mulino. McArthur, Tom 1992 The Oxford Companion to the English Language. Oxford: Oxford University Press. Mirto, Ignazio Mauro 2003 Che fare? Analisi di costrutti di un verbo critico in italiano. In Il verbo italiano. Studi diacronici, sincronici, contrastivi, didattici, Mathiée Giacomo-Marcellesi and Alvaro Rocchetti (eds.), 277–291. Roma: Bulzoni. 2004 Uno strano fari siciliano. Quaderni di semantica 25 (2): 289–302. to appear Predicazione scissa in una varietà siciliana: fari come ausiliare. In Tradizione e innovazione. Il parlato: teoria – corpora – linguistica dei

168 Ignazio Mauro Mirto corpora, Atti del VI Convegno SILFI, Duisburg, June 28 –July 2, 2000, Elisabeth Burr (ed.). Firenze: Franco Cesati Editore. Ricca, Davide 2001 Facciamo che andare: sulla semantica di una tipica perifrasi dell’italiano regionale piemontese. In La parola al testo. Scritti per Bice Mortara Garavelli, Gian Luigi Beccaria and Carla Marello (eds.). Alessandria: Edizioni dell’Orso. 2007 La perifrasi fare che + infinito in Piemonte. Handout delivered on January 17th, Romanisches Seminar, Universität Zürich. Rohlfs, Gerhard 1969 Grammatica storica della lingua italiana e dei suoi dialetti. Sintassi e formazione delle parole. Torino: Einaudi. Rosen, Carol 1997 Auxiliation and serialization: On discerning the difference. In Complex Predicates, A. Alsina, J. Bresnan and P. Sells (eds.), 175–202. Stanford: CSLI. Saeed, John I. 1997 Semantics. Oxford: Blackwell. Salvi, Giampaolo 1988 La frase semplice. In Grande grammatica italiana di consultazione, Vol. I, Lorenzo Renzi (ed.), 29–114. Bologna: il Mulino. 1991 Le frasi copulative. In Grande grammatica italiana di consultazione, Vol. 2, Lorenzo Renzi and Giampaolo Salvi (eds.), 163–190. Bologna: il Mulino. Seuren, Pieter A. M. 2006 Presupposition. In Encyclopedia of Language and Linguistics, Vol. 10, Second Edition, Keith Brown (editor-in-chief), 80–87. Oxford: Elsevier. Sornicola, Rosanna 2006 Topic and Comment. In Encyclopedia of Language and Linguistics, Vol. 12, Second Edition, Keith Brown (editor-in-chief), 766–773. Oxford: Elsevier. Stalnaker, Robert C. 1973 Pragmatica. In La struttura logica del linguaggio, Andrea Bonomi (ed.), 511–530. [translated by Gabriele Usberti, original edition: Pragmatics in Synthese 22, 1970: 272–289]. Sonzogno: Bompiani. Tsujioka, Takae 2002 The Syntax of Possession in Japanese. New York/London: Routledge. Van Valin, Robert D. Jr 2001 Functional Linguistics. In The Handbook of Linguistics, Mark Aronoff and Janie Rees-Miller (eds.), 319–336. Oxford: Blackwell.

Clitics and anaphoric relations in informational patterning: A corpus-driven research on spontaneous spoken Italian (C-ORAL-ROM) Emanuela Cresti

1. Introduction Clitic morphemes are one of the most relevant features that set Romance languages apart from Latin. Their importance has always been underlined in the linguistic literature.1 The description of Clitics in early Italian language grammars dates back to the Renaissance (Castelvetro 1563; Buommattei 1643; Cinonio 1644). More recently, from the 1970s onwards, cliticization and anaphoric relations have been important issues in theoretical linguistics (Lo Cascio 1970; Seuren 1974; Evans et alii 1978; Berretta 1985; Conte 1990; Kleiber 1994; Muller 2001; Korzen 2007), and very special subjects in generative linguistics (Kayne 1975; Rizzi 1982; Reinhart 1983). Few research works, however, focus on the actual use of clitics and fewer still, even in corpus linguistics, consider their quantitative weight in language use. This paper presents the findings of a corpus-driven research run on the spoken Italian corpus of the C-ORAL-ROM collection (Cresti & Moneglia 2005) and is devoted to the quantitative and qualitative description of cliticization and anaphoric relations in the everyday usage of Italian.2 Automated searches run on C-ORAL-ROM Italia showed that this corpus records around 10,000 clitic tokens, including Tuscan unstressed subject pronouns.3 Therefore, clitics account for around 3% of the total tokens recorded in the corpus (310.969). The amount of clitics can be more perspicuously examined when compared to the total number of utterances and, more specifically, to the number of verbal utterances (23,873)4 which are the necessary context in which clitic morphemes occur.5 Considering that utterances with more than one verb can occur, we can roughly consider that in CORAL-ROM Italia, one out of three verbal utterances is involved in cliticization phenomena. Therefore, it is safe to say that cliticization, as a whole, is a highly frequent language structure, and a fundamental challenge in spoken language processing.

170 Emanuela Cresti The present research work has been led on a sub-corpus of C-ORALROM Italia, consisting in a balanced sample of 5,498 utterances (Tucci Sub-corpus).6 In this set, only verbal utterances are relevant as far as the occurrence of clitics (4,179) is concerned. For reasons that will be clarified later, a second set of 201 utterances taken from a different C-ORAL-ROM sub-corpus has been added to the sample. Therefore, the overall corpus consists of 4380 verbal utterances. This corpus was already parsed into utterances and tagged with regard to the functions performed by the information units (henceforth, IUs) which make up each utterance, such as Comment, Topic, Parenthesis, Locutive Introducer, Appendix and Dialogic units (See paragraph 2.). The annotation of IU types has been carried out strictly in line with the Teoria della Lingua in Atto [Language in Act Theory] and the Informational Patterning Hypothesis (Cresti 1994, 2000; Cresti & Moneglia forthc.) deriving from it. This annotation allows to infer the distribution of clitics and their referential linguistic heads7 with respect to the utterance boundaries, and across the various types of IUs. This research is part of a larger program led by LABLITA8, intended to verify the accuracy of the pragmatic definition given to the IUs in our theoretical framework. From all previous works, it clearly emerges that every stretch of speech used with a specific informational function presents prosodic, morpho-syntactic and also semantic correlations. The paper is specifically dedicated to the study of the subset of clitics which convey phoric relations, and excludes a detailed investigation of the other functions performed by clitics in spoken language [See paragraph 4.3]. It only deals with the description of anaphoric clitics deriving from corpus analysis, with their distribution, and finally with the constraints they show. The novelty of the present approach rests in the correlations found between the distribution of clitics and the informational structure of spoken language. Indeed, a set of regularities of anaphoric relations emerges in accordance with a given informational structure. We will argue that these are a direct consequence of the pragmatic value which defines each IU. 9

2. The pragmatic definition of IUs According to the Speech Act theory (Austin 1962), a given utterance accomplishes a speech act. When we apply this principle to the analysis of spoken corpora, we find that a large part of utterances (58% in C-ORAL-ROM) features an internal

Clitics and anaphoric relations in informational patterning

171

organization with functionally characterized chunks. The Teoria della lingua in atto and the Informational Patterning Hypothesis maintain that each chunk is an IU, which can perform a restricted set of informational functions. Consequently, an utterance corresponds to an informational pattern. Moreover, in this theoretical framework, it is assumed that an informational pattern necessarily corresponds to a prosodic pattern, formed by one or more prosodic units (henceforth, PUs). In this perspective, spontaneous spoken language is governed by a one-to-one correspondence between PU types and IU types. Each of a pattern’s PUs is characterized by a perceptively relevant prosodic form (‘t Hart et al. 1990) and is set apart from the subsequent one by a Non- terminal prosodic break. A pattern is invariably concluded by a Terminal prosodic break.10 Monologues and formal speech in spoken corpora frequently also contain other textual entities apart from utterances, which we identify with the term “Stanza”. A Stanza is concluded by a Terminal prosodic break, as an utterance is, but unlike utterances, it doesn’t fully accomplish an illocutionary act, having instead a generic assertive value. Moreover, it can be made up of various IUs, each marked by a Non-terminal prosodic break. Its informational patterning is only partially consistent with the utterance’s, but it still shows an internal informational organization. Stanzas seem to reflect the transition from a pragmatically based spontaneous exchange to textual construction in speech.11 The correspondence between prosodic patterns and informational patterns is one of the most useful devices in analysing large spoken corpora, since it allows to clarify both how the speech-flow is parsed into utterances, and how these are internally organized from an informational point of view. In this paper, the correspondence between IUs and PUs will be taken for granted and we will not deal with the prosodic correlations of the IUs.12 IUs are divided into Textual and Dialogical units. Textual units participate in the construction of the utterance’s text and to its semantic cohesion, while Dialogical units are those parts of the utterance which regulate the exchange with the interlocutor. As for Textual units, a list of the IU types with their main characters follows: Comment (COM).13 It is the IU dedicated to the accomplishment of the illocutionary force of an utterance. A repertory of roughly 80 illocutionary types has been detected in spontaneous spoken Italian.14 A Comment is necessary and sufficient to produce an utterance. 42% of C-ORAL-ROM’s utterances are simple, i.e. they consist of one only Comment IU. Comments always have a modal value and bear an informational focus.15 The distribu-

172 Emanuela Cresti tion of IU types in an informational pattern and their values are functions of the Comment. Topic (TOP). It represents the field of application of a Comment’s illocutionary force; it specifies the domain in which the speech act conveyed by the Comment is relevant, thus allowing the distantiation of the Comment from the context.16 The Topic necessarily precedes the Comment, has its own modality (which can only be Alethic or Epistemic)17 and an informational focus. Whether the Topic is an existing or a new one is not relevant from a semantic point of view, but because of its distantiation function, it must allow the Comment to create a textually independent utterance, i.e. an utterance that does not depend on the deictic space or on the previous utterances. The Topic necessarily precedes the Comment due to informational requirements.18 As it comes first, it isolates the utterance from previous utterances and prevents the interlocutor from referring the illocutionary force of the Comment to the context.19 As a matter of fact, the Topic-Comment order seems to follow an informational principle shared by all languages and reveals a valid path for ontogenesis development.20 Parenthesis (PAR). It acts as a meta-linguistic insertion within the utterance (in modal evaluations, terminological explanations, and judgements on the activity of the interlocutor).21 It specifically refers to an IU, most likely a Comment or a Topic, but it may also be linked to another Parenthesis, and very rarely to a Locutive Introducer. It bears a proper modal value (mostly epistemic) which can confirm or weaken the modal value of the IU it is related to.22 It has no informational focus. As for its distribution, a Parenthesis immediately follows, or even interrupts, the IU it is related to. Locutive introducer (INT). It is the IU that typically introduces reported speech. More generally, it makes the interlocutor aware of the occurrence of a secondary utterance whose illocutionary force does not perform any act (neutralization). Its only distribution constraint requires for it to precede the neutralized utterance. The Locutive introducer has a proper modal value, which is necessary to determine the value of the reported utterance. It behaves as an evidentiality medium, i.e. it specifies whether the secondary utterance can be trusted as reported speech, or as advice, an instruction, a list, etc. It does not bear an informational focus.23 Locutive introducers are characterized from a lexical point of view, i.e. they can be verba dicendi, general nouns, or presentational formulas. The linguistic arguments of the Locutive introducer (for instance the subject and the object of the verbum dicendi) offer the deictic reference for the secondary utterance (See paragraph 5.2).

Clitics and anaphoric relations in informational patterning

173

Appendix. (APC) It is defined as a “textual integration” of the Comment,24 whose role is to conclude the utterance in agreement with the interlocutor. In other words, the speaker uses various textual integration strategies (60% repetitions of previously mentioned expressions or leitmotifs; 25% formulaic expressions; 15% new information) to ensure mutual consent on the statement. The Appendix necessarily follows the Comment it integrates, and strictly refers to it. In no case can it be the field of application of the illocutionary force, i.e it cannot distantiate the Comment, as the Topic does. Appendices don’t bear a proper modal value, because they are included in the scope of the modal value of the Comment. They have no informational focus.25 An Appendix is uttered “once the utterance is already done”. When the Appendix adds new information, this is not crucial and could be erased without loss of functionality.26 For these reasons, despite the fact that the Appendix follows a Comment, it does not contain the arguments of a verb in Comment; frequently it corresponds to a specification, or to a circumstantial, or to a non-restrictive relative clause. Dialogical IUs can be roughly classified according to five types: 27 Incipit (INP). It signals the will of the speaker to start their dialogic turn or to continue it after the conclusion of an utterance. It necessarily opens the utterance. Phatic (PHA). It is dedicated to the control of the communication channel, keeping it open and stimulating the perception of the interlocutor. It can also help the speaker take time to organize speech, or it can be used to manifest agreement on receipt of a message. Its position is undetermined. Allocutive (ALL). It controls the communication exchange, refreshing the interlocutor’s attention. It mentions the interlocutor’s name or addresses him or her with a deictic or with a kind or insulting expression. Its position is free, but it frequently appears at the opening. Conative (CON). It is directed at pushing the interlocutor to take part in the exchange or to restrain from behaving in an unsuitable way according to the verbal relation. Its position is free, but frequently final. Expressive (EXP). It can accompany the utterance with expressions, often vivid or even vulgar, intended to stimulate a common point of view on the utterance. Its position is free, but it frequently appears mid-utterance. Dialogical units don’t bear either a modal value or an informational focus. Contrary to Textual units, all Dialogical IUs are bound to a closed lexical repertory.

174 Emanuela Cresti With regard to Stanzas, which occur frequently in our reference corpus, their informational patterning is only partially consistent with the informational patterning of the utterance, because they are frequently formed by many weak Bound Comments (^COM). They can be accompanied by other IUs, such as Topics or Parentheses or Dialogical units. From a syntactic point of view, these IUs deal with sentences, or clauses, or phrases, which stay juxtaposed and do not form a whole syntactic configuration.28

3.

The repertory of clitics in spoken Italian

3.1. Quantitative data The corpus taken as reference (4,380 verbal utterances) contains 3,174 tokens of clitics, but only 2,418 of these verbal utterances (52.2%) are involved in cliticization. 3,107 clitic tokens were allocated to a morpho-syntactic and functional category, while the remaining 67 tokens (2.1% of the total) are unclassified for various reasons, i.e. being isolated forms, retracting elements or elements whose function could not be appreciated in the text. Grammars usually present a paradigmatic list of 17 different clitic forms according to the person, gender, singular number and object function, which is considered as neutral (1°sing. mi/me, 2°sing. ti/te, 3°sing. lo/la), to plural number (1° plur. ci/ce, 2° plur. vi/ve, 3° plur. li/le), to other functions (dative gli, /le; reflexive si/se; partitive ne) and 5 clitic clusters (glielo, gliela, glieli, gliele, gliene). But where true spoken usage is concerned, we must consider the distributional properties of clitics; that is, whether a clitic is employed: a) in isolation or in a cluster; b) in a proclitic or enclitic position; c) as a subject (Tuscan variety) or with other argumental functions; d) in conjunction with the verb essere [to be] or in other lexical locutions. Each of these cases presents specific distributional properties: this is the reason why each instance of the previous variation was evaluated as a type. The set of classified clitics found in the corpus corresponds to the following range: a) Single clitics in pro-clitic postition (si, mi, lo, ci, la, gli, ne, etc..); b) Clitic clusters (two or more clitics) in pro-clitic position (ci si, me lo, te lo, glielo, tu te la, ci se la, etc…); c) Single clitics in enclitic position (-lo, -si, etc…);29 d) Clitic clusters in enclitic position (-selo, -sene, etc…);

Clitics and anaphoric relations in informational patterning

175

e) Clitics found in conjunction with the verb essere (ci essere[there is], ce ne essere [there some is]); f) Clitics as unstressed subjects in the Tuscan variety, often featuring an apocope (tu, l’, gl’, e’, etc…). A list of 96 types of clitics has been identified. Naturally, the identified forms do not cover the whole range of possible combinations and clusters in Italian: they only represent the actual usage of clitics in our corpus.30 We will not consider the distribution of the repertory in this paper; however, the quantitative prominence of clitic types in spoken Italian can easily be derived from the statistics. For instance, roughly 26.5% of occurrences is covered by the proclitic plus the enclitic form si, performing mostly morpho-syntactic functions (reflexive, impersonal, passive, Tuscan 1st plural person). The set of enclitics (15.3%) is of course less frequent compared to proclitic types, but certain types, respectively -si,- lo, -mi, occur with a very high frequency. Clitic clusters are frequent too (ci si, te lo, me lo). More specifically, considering clitic clusters as types allows to perceive that certain types are surprisingly absent (-meli, tu te ne), despite the fact that one could imagine expressions in which they should be commonly employed (dammeli, [give to-me them], tu te ne sei fregato [you for-you about-it didn’t care]).

4.

Suggested classification

The study of anaphoric relations in spoken language calls for a preliminary identification of the relations that actually occur in the corpus. This activity is particularly relevant for spoken language analysis. As the literature on this matter has shown, clitics which play functions other than anaphorae are the majority in spoken Italian. Indeed, only a small percentage of clitics is fully anaphoric (17.5%, according to Berretta 1985). Our data show, however, that the percentage of anaphoric clitics in the corpus is nearly 24% and that their anaphoric functionality is full: 31 it can therefore be the object of detailed research. 4.1. Annotation Criteria The annotation of clitics and anaphoric relations within our corpus has been carried out in accordance with the following criteria:

176 Emanuela Cresti 1) the distribution of clitics, i.e. the IU in which each clitic occurs; 2) the different functions performed by clitics in Comment IUs and in the other IUs; 3) whether the anaphoric relation: – crosses the utterance boundaries; – crosses IU boundaries; – is established within a given IU (linearized). 4.2. The distribution of clitics The occurrence of clitics within the textual IUs of the utterance shows both preferential positions and restrictions that seem to be a consequence of informational constraints. The IUs in which a clitic may appear are listed below, followed by their percentage of occurrence over the total clitics. We have added in parentheses the number of verbal utterances which are the seat of the specific IU in our reference corpus, and their percentage on the total, as this is relevant in the correct evaluation of the clitics’ occurrence: Comment: 82 % (of 4,179 verbal utterances, all of course containing a Comment) Topic: 7.5 % (of 2,492 verbal utterances containing a Topic, that is, 59.6 % of the 4,179 verbal utterances) Parenthesis: 7.9% (of 1,754 verbal utterances containing a Parenthesis, corresponding to 42 % of the 4,179 verbal utterances) Locutive introducer: 1.8 % (of 201 verbal utterances containing a Locutive introducer, corresponding to 4.8 % of the 4,179 verbal utterances) Appendix: 0.5 % (of 222 + 201 verbal utterances containing an Appendix; corresponding to 9.6 % of a total 4,380 verbal utterances) On the basis of the above figures, it is obvious that the probability of occurrence of a clitic in a given IU varies consistently. Although this probability also correlates with the distribution of verbs in the IUs, we can highlight some interesting points. First of all, a high likelihood of occurrence of clitics within a Comment unit remains evident. Secondarily, even though the frequency of Topics and Parentheses seems quite similar, the presence of a clitic in a Parenthesis is actually more relevant, due to its relative frequency in the corpus. The pres-

Clitics and anaphoric relations in informational patterning

177

ence of clitics in Locutive introducers is quite significant, given the low percentage of occurrence of this IU. Nevertheless, the most remarkable restriction concerns the Appendix, since we found hardly any such clitics in our first corpus, and only 12 cases in the enlarged corpus.32 Moreover, all these clitics were part of an idiom (non ci potevo credere [I couldn’t believe (in) it], farci i cazzi nostri [to mind our own business]) or participated in the utterance as ecolalic repetitions (l’avevo / COM l’avevo // APC [(I) had it, (I) had it]).33 Therefore, no anaphoric relations were in place, unlike in all the other Textual IUs. Such a datum deserves an explanation, especially once the absence of such restriction in the Topic unit was not observed. Finally, in our corpus, only the above five Textual IUs are involved in cliticization phenomena and neither clitics, nor heads of clitics, have been found in Dialogical IUs.34 The above distribution regards all cliticization phenomena and not specifically the anaphoric use. Later, the various clitic functions and their frequency within the Comment IU will be discussed in detail, and so will their frequency in the other IUs, even if briefly. 4.3. The various functions of clitics The variety of functions that clitics play in spontaneous speech emerges clearly when examining them in Comments, which are their preferential position. Indeed, only few clitics establish anaphoric/cataphoric relations (21% of cases), while the other occurrences are to be found in lexical, morphological and grammatical functions. The percentages and typical instances of the functions played by clitics found in Comments are listed below. The following is the overall picture: a) Anaphoric (1) and cataphoric functions (2) (roughly 21%) (1) i42035*ANT: secondo me / TOP tu se dovessi fare il regista / TOP attori / TOP non ne troveresti mai //COM [according to me / if you were to become a director / actors / you would never find any (of them)] Anaphora: head in TOP and clitic in COM2) i1299*LOR: no /CON un anno glieli feci / COM mi pare /PAR i bicchieri //APC [no / one year I did buy them / I think / glasses]Cataphora: clitic in COM and head in APC

178 Emanuela Cresti b) Lexical and semantic functions: fulfilling of verbal regency through 1st and 2nd person, singular and plural, deictic clitics (3); stereotypical reference in idioms (4) (roughly 28%). (3) i271*PAO: m’era venuto in mente / TOP ti volevo chiedere // COM [it had occurred (to) me / I wanted to ask you] (4) f654*CUS: non ce la faccio // COM [I can’t do it] c) Morphological functions: Reflexive diathesis of the verb (5); impersonal locutions (6); passive (7) and “benefactive” diathesis (8), which have been recently developing in spontaneous spoken Italian (roughly 23%). (5) i269*PAO: pensavo tu dicessi / mi sono sposata apposta // COM [I thought you were saying / I (myself) got married on purpose] (6) f200*GER: si tratta di una situazione di emergenza // COM [it’s a case of emergency] (7) f1007*VER: […] il problema di fondo / è che &n / difficilmente / si ottengano / miglioramenti / sul piano della civiltà // COM [the root of the problem / is that / it’s hard / (we) to achieve / improvements / from a civility point of view] (8) f248*CIN: ma intanto / INP ci sono quattro barche / ICO 36 per me / PAR che possono cercare di arrivare nei quattro / e farsi il match race // COM [but meanwhile/ there are four boats/ according to me / that can try to get in the last four / and run (themselves) the match race] d) Unstressed filler of the subject position (9). This morphological function is typical of the Tuscan variety, and thus recorded aside from the other morphologic functions (roughly 10 %) (9) f753*CAP: però / INP se gli si dice di chiudere magari il sabato / TOP o forse un altro giorno / INX forse non sarebbero così d’accordo // COM [but / if (we to them) tell to close perhaps on a Saturday / or maybe another day / they may not agree //] As for clitics placed in other IUs, their functions show less variation than in Comments, and their preferential functions also vary in accordance with the IU type. Topics record anaphoric clitics in 35% of cases (with antecedent heads in preceding utterances), 36% morphologic functions, nearly 16% verbal deictic regencies and a very little usage of formulaic expressions. Therefore, clitics

Clitics and anaphoric relations in informational patterning

179

used in Topics in anaphoric function appear more frequently than those in Comments. Among Parentheses, nearly 29% of clitics are devoted to anaphoric/cataphoric relations within the same utterance (with the head in Topic or in Comment). Nearly 40% of clitics are employed as idioms and only the remaining 30% varies in accordance to morphologic functions and deictic regency. Consequently, in Parentheses, clitics used in anaphoric mode also reach a high percentage, but such IUs are characterized by a massive use of formulaic clitics, while the remaining variation seems to be less significant. On the contrary, in Locutive Introducers, more than 70% of clitics are the deictic regency of a verbum dicendi, and 30% are anaphors (with heads in preceding utterances, rarely in a preceding Topic within the same utterance).

5. Levels of anaphoric and cataphoric relations Anaphoric/cataphoric relations exist only if: a) one or more clitics perform the argument role of verbal regency; b) a semantic head of reference (which may be phrasal, sentential, discursive or contextual) is in place for each clitic. Moreover, according to the Informational Patterning Hypothesis, the above relation can operate at different levels: a) crossing the utterance boundaries; b) crossing the IU boundaries; c) within the IU (linearized).37 a) Cross-utterance anaphoric relations between a clitic and its head in a preceding utterance (10) i125*MAR: il caffè com’ era ? COM ma lo possiamo paragonare a quello del bar / COM come qualità ?APC [the coffee what was it like? but can we compare it to the bar’s / in quality?] (11) i171*MAR: quanti altri colori ci stanno ?COM […] *MAR: […] quindi / INP gliene devi mettere altri due // COM [how many other colours are there? […] […] so / you have to put another two in (of these to them)//] (12) i230*ALE: […] lei è una biondina /^COM lui con gli occhi azzurri // COM […] *FRA: se hanno la mia età / TOP è possibile che li conosco //COM [she’s a blonde / he’s got blue eyes // […] if they’re my age / chances are I know them //]

180 Emanuela Cresti (13) f926*PRI: ma dire di un uomo /INT questo /TOP è l’agnello di Dio /^COM è dire qualcosa di così grande / COM che svela qualcosa della sua natura //APC […] chi mi aveva detto di andare a battezzare /TOP mi aveva detto /INT lo riconoscerai // COM’’ [but to say of a man / this / is God’s lamb / is to say something so great / that reveals something of his nature // who told me to go and baptise / told me / you will recognise him //] In the previous cases the head is placed in an utterance which precedes the one containing the clitic, but the relation can involve: a) two contiguous utterances, as in (10); b) utterances in a same speaker’s different dialogical turns, as in (11); c) utterances in different speakers’ dialogical turns, as in (12); d) non-contiguous utterances in a formal monologue, as in (13). Moreover, the head may occur also in a Topic as in (14) and (15) (See paragraph 8.1). (14) f85*RUF: la verità / TOP la cerco prima di tutto / dentro di me // COM cercando / conferma / forse negli altri // COM ma / se non [/] se non la trovo dentro di me / TOP non posso trovarla negli altri // COM [truth / I search for it first of all / looking for / confirmation / perhaps in others // but / if I [/]if I don’t find it in myself / I can’t find it in others //] (15) f132*DON: se tu non hai i soldi / TOP rimani malato e muori // COM quelli che hanno soldi / TOP siccome le medicine sono troppo costose / INX se hanno i soldi / TOP le medicine se le possono comprare / e si curano // COM chi non ce l’ ha / TOP non può fare altro che farsi la malattia e poi…COM [if you haven’t got the money / you stay ill and die // those who have money / because medicines are too expensive / if they have the money / they can buy medicines / and cure themselves // those who don’t have it / can only let the illness take its course and then… ] The percentage of cross-utterance anaphoric relations is quite consistent, corresponding to almost 53% of the whole phoric usage. As a matter of fact, we have never found examples in which the head is hosted in an utterance following the one in which the clitic is placed. In other words, so far, no cross-utterance cataphoric relations have been found.38

Clitics and anaphoric relations in informational patterning

181

b) Anaphoric and cataphoric relations crossing IU boundaries within the utterance (16) i253*LAU: quindi / INP anche la parte [/] parte museale / TOP se è in vetro / TOP cioè /PHA puoi tu insonorizzarla come vuoi // COM [so / also the part [/] museum part / if it’s glass / actually / can you soundproof it as you want //] Anaphora: phrasal head in TOP and clitic in COM (17) f899*ISP: io / TOP sono convinta / INT lo dico da tutte le parti / INX tutti mi guardano così /INX però io sono profondamente convinta // COM [I / am convinced / I say it everywhere / everybody looks at me like that / but I am deeply convinced //] Anaphora: Sentential head in INT and clitic in PAR (18) i189* GCM: ma bisognava seccarlo / COM però / il fieno //APC [but we should have dried it / though / the hay //] Cataphora: clitic in COM and phrasal head in APC (19) i389*POL: ‘un l’avrei mai pensato/ COM ci fosse uno studio così // APC [I’d never ‘ave thought (it) / there was a study like this //] Cataphora: clitic in COM and sentential head in APC Therefore, in the present framework, anaphoric/cataphoric cliticization within the utterance may include the relation between IUs; as we will prove in detail, the distribution of clitics and heads undergoes restrictions that are strictly connected to the informational patterning. It corresponds to approximately 35.8 % of the phoric usage. c) Anaphoric and cataphoric relations between a clitic and its head, linearized within the same IU (20) f871*PRF: in realtà / TOP quando deve mettere le mani su un monumento e deve provare a ripararlo / TOP se non ha fatto una diagnosi corretta / TOP fa come si è fatto moltissimo negli anni /COM int. direi /INX nei decenni precedenti // COM [actually / when he has to put his hands to a monument and try to fix it / if he hasn’t come up with a correct diagnosis / he does as it has been done very often over the years / I should say / over decades //] Anaphora: antecedent phrasal head in TOP and clitic in the same TOP

182 Emanuela Cresti (21) f812*COA: io / TOP questo lo porterei a regolamento // COM [I / this one (it) I would reckon up //] Anaphora: antecedent pronominal head in COM and a contiguous clitic in the same COM 39 (22) i258*LUI: eh / FAT poi devi vedere cosa ci metti in mezzo tra un blocco e un altro // COM [hm / then you have to think what to put in (there) between one block and the other //] Cataphora: clitic in COM and subsequent phrasal head in the same COM In the above cases, clitics and their heads appear linearized within the same IU, both in anaphoric and cataphoric relation (although the cataphoric relation seems less frequent). Crucially, linearized anaphoric and cataphoric relations have been found in the most important IUs as far as textual construction is concerned (Comment, Topic, Parenthesis). Linearized relations correspond to nearly 11.2 %. This paragraph has outlined the general features of anaphoric relations and their levels according to the Informational Patterning Hypothesis; in the following we will present findings and restrictions regarding both heads and clitics at the various levels. 6. Informational constraints on the position of the head Interesting constraints at the informational level have been found in the distribution of heads of cataphoric and anaphoric relations. The relation with the head can of course be cross-utterance, intra-utterance, or linearized. As for cross-utterance relations, both anaphoric and cataphoric relations could in principle cross the utterance boundaries, but cross-utterance cataphoric relations, as we have noted, were never found in our corpus. Moreover, despite the quantitative limits of this research, we must highlight that in cross-utterance anaphoric relations, the head was found either in Comments or Topics. In other words, the head of a cross-utterance anaphoric relation was never found in Parentheses, Appendixes, Locutive introducers or Dialogical units. This clear tendency calls for an explanation at informational level, and probably has a high predictive value with regard to the restrictions on anaphoric relations in spoken texts. It appears that if the anaphoric relation refers to a head outside the utterance boundaries, this occurs within an IU

Clitics and anaphoric relations in informational patterning

183

characterized by an informational focus. Therefore, we can assume that the heads of cross-utterance anaphoric cliticizations never involve non-focal IUs, and that an informational focus is needed. This constraint seems to operate in order to ensure an easy recalling of cross-utterance heads, and can be considered a principle of textual cohesion. In intra-utterance relations, the Comment can typically accommodate the head of an anaphoric relation with a clitic placed in a subsequent Parenthesis, but it can also contain the head of a cataphoric relation with a clitic placed in a previous Parenthesis. Surprisingly, Comments do not contain the head of a cataphoric relation with a clitic placed in a Topic. We have never encountered what the literature refers to as the backward anaphora, which is by all means a grammatical entity and therefore within our area of competence. In other words, the following theoretical possibility, where a clitic in Topic finds its head in a subsequent Comment, never occurs in our corpus: (23) “Quando li’ho guardato / TOP Marioi ha voltato la testa // COM” [When (I) looked at him/ TOP Mario turned his head // COM]. Cataphora: clitic in TOP and phrasal head in COM Although we didn’t investigate phoric relations between heads and 0 Subjects in detail (see Footnote 2), we can recall that examples showing a 0 subject in Topic and its head in a subsequent Comment were not found in the specific studies carried out at LABLITA on this matter either (Signorini 2005): (24) “Quando 0i è partito / TOP Marioi ha ringraziato tutti // COM” [When 0i left / TOP Marioi thanked everyone // COM] Of course, the absence of such relation does not mean that it is not possible to produce them in spoken Italian, but it does indicate that it is highly unlikely. The explanation of their absence is quite straightforward from the utterance’s informational organization point of view. While the informational function of the Topic unit is “specifying the domain of relevance of the speech act”, the tendency to fill the Topic with individuated entities can find a strong informational motivation, which should be counteracted by a clitic (or a 0 subject) with a cataphoric reference. However, anaphoric cross-utterance clitics in Topics may well occur, being individuated by their

184 Emanuela Cresti heads in the preceding utterances. In summary, this negative finding on spoken corpora contradicts all expectations created by classical studies on backward anaphora, and is not confirmed by competence-based evidence. In accordance with its function, which requires an identified reference, the Topic very often bears the head of an anaphoric relation, with clitics placed in subsequent IUs capable of bearing a clitic in our corpus: Comment, Locutive Introducer and Parenthesis. As far as intra-utterance anaphoric relations are concerned, a Parenthesis may contain the head of an anaphoric clitic placed in a subsequent Comment. Only rarely is it the seat of the head of cataphoric clitics placed in a preceding Comment. Locutive introducers never contain the head of an intra-utterance relation, neither cataphoric nor anaphoric. This is also a consequence of their informational and semantic function. The Locutive introducer mainly introduces reported speech, whose deictic expressions, by definition, refer to a different space with respect to the original utterances. For this reason it cannot establish any anaphoric relations. Therefore the restriction in a Locutive introducer is strictly due to its role. Appendixes only host heads of cataphoric relations, with clitics placed in Comments. This severe limit is parallel to the absence of anaphoric clitics in this IU. The Appendix can be considered “inert” with regard to anaphoric relations. More specifically, Appendixes never feature heads of intra-utterance anaphoric relations. Such a restriction cannot depend only on distributional reasons, (it is frequently in final position, but it can be followed by a Parenthesis), but is rather due to the fact that the Appendix is non-active and non-cohesive with respect to other IUs of the utterance.40 This restriction is also consistent with the informational role of the Appendix, which is specifically one of textual integration of the Comment unit (not of the whole utterance), so the scope of heads in an Appendix is consequently restricted to clitics in Comments. 7. Cataphoric relations crossing the IU boundaries We observed above that at intra-utterance level, the Topic cannot host a clitic whose head is placed in a subsequent IU. The following is the positive distribution of intra-utterance cataphoric relations crossing the IU boundaries.

Clitics and anaphoric relations in informational patterning

185

7.1. Cataphoric relations with Appendix units Intra-utterance cataphoric relations are quite rare, but, as we just said, cataphoric relations for clitics placed in Comments are the only possible phoric relation phenomenon involving Appendixes (36 contexts in the corpus, 2.8% of the total anaphoric relation phenomena, almost all involving phrasal heads). A few such examples follow: (25) f222 *BIA: però / INP non lo so / COM quale vuoi // APC [but / I don’t know (it) / which (one) you want] Cataphora: clitic in COM and clausal head in APC (26) f533 *CMA: perché non me lo potete dimostrare / COM il contrario //APC [why can’t you prove it / the contrary] Cataphora: clitic in COM and phrasal head in APC 7.2. Cataphoric relations with a Parenthesis The following was the only example to be found of a cataphoric relation between a clitic in a Comment and a head in a subsequent Parenthesis: (27) f262*POZ: su queste credo ci potessimo lavorare / INT con ci intendo noi / PAR […] o sempre o mai // COM [on these I think we could have worked / by we I mean us / […] always or never] Cataphora: clitic in INT and phrasal head in PAR 7. 3. Cataphoric relations with a Comment Only six utterances showing a cataphoric relation with a head in a Comment have been found. In all of these cases, the clitic was placed in a previous Parenthesis. (28) f646*PMA: io/ TOP direi / INT. ma credo che chiunque lo potrebbe dire/ PAR che si tratti / di un processo unico / nella storia giudiziaria di questo paese // COM [I / would say / but I think anyone could say (it) / that it is / a unique trial / in this country’s judicial history] Cataphora: clitic in PAR and clausal head in COM

186 Emanuela Cresti 7.4. Cataphoric relations with Bound Comments Three very special examples of cataphoric relations concern Bound comments in the Stanza, involving a clitic in a Bound Comment and a sentential head in the subsequent Bound comment. (29) i1239*PAO: anche se dobbiamo dirlo molto chiaramente /^COM Gianluca / ALL che / non è che i rapporti / &ee / i problemi del centro sinistra / i rapporti / con il partito maggiore / i Diesse / non ci siano // COM [although we must make it very clear / Gianluca / that / it’s not as if the relations / &ee / the mid-left wing’s problems / the relations / with the head party / the DS / don’t exist] Cataphora: clitic in a Bound Comment and sentential head in the subsequent Comment In accordance with the above remarks, we must conclude that cataphoric relations in spontaneous speech are quite rare. Apart from the typical and most frequent case, which relate to Appendixes, it may be interesting to note that cataphoric relations mostly concern the clitic form lo and a sentential head, involving Comments or Bound comments within the Stanza. 8.

Anaphoric relations

Unlike cross-utterance cataphoric relations, cross-utterance anaphoric relations are very frequent. The following paragraphs deal with the positive distribution of cross-utterance and intra-utterance anaphoric relations. In paragraph 9. it will emerge that no constraints to anaphoric relations within the IU were detected. 8.1. Cross-utterance anaphoric relations A clitic can be placed respectively in a Topic, in a Comment and in a Locutive introducer and find its head in the Comment; see examples (10), (11), (12), (13) or in the Topic (14) of a preceding utterance, especially in formal spoken texts.41 Cross-utterance anaphoric relations were never found in Parentheses within our reference corpus. This negative result seems to be connected with the nature of the Parenthesis unit, which is essentially a meta-linguistic insertion within the utterance.42 Therefore, the restriction requiring clitics in Parentheses to refer to a head within the utterance itself, is a consequence of this feature, and is in line with the principles ensuring textual coherence.

Clitics and anaphoric relations in informational patterning

187

8.2. Intra-utterance anaphoric relations When the clitic and its head are placed in two IUs of the same utterance, the position of both the clitic and the head undergoes severe limits, which are due to informational constraints. a) As we have explained in paragraph 6, the Topic frequently hosts the head of a clitic which can be placed in any other IU (Comment, Locutive introducer and Parenthesis). (30) f408*SCA: &de / devi fare farine /TOP ma non più per mandarle all’alimentazione / ma per andare alla distruzione // COM [&de / you have to produce (animal) flour / but no longer to send it as food / but to be destroyed] Anaphora: head in TOP and clitic in COM (31) f308*MAR: se lei dovesse dare un voto / TOP e qui pubblicamente lo sta dando /PAR di fronte a milioni di italiani // COM [if you were to give marks / and here in public you are giving them / in front of millions of Italians] Anaphora: head in TOP and clitic in PAR (32) i1116*PAO: allora / INP voglio dire / PAR se lei è giustamente &he / animalista / TOP se lei è &de/ buddista / TOP io sono &he / cattolica / TOP io gli dico / INT guarda / CON io sono della tre effe / ^COM però insomma / FAT dovresti convertirti // COM [so / I mean / if she is obviously &he/ an animal activist / if she is &de / a buddhist / I am & he / catholic / I’ll tell her / look / I belong to Tre Effe / but really / you should convert] Anaphora: head in TOP and clitic in INT Of course a Topic cannot establish intra-utterance anaphoric relations with any preceding head, being necessarily the first Textual IU of the utterance. b) A Comment can host the head of an anaphoric relation only in conjunction with a subsequent Parenthesis. (33) F634*PMA: ecco / INP il pubblico ministero non si commisura / COM non può farlo / PAR a questa / variabile / o a questo insieme di variabili / che si condensano nella figura dell’imputato // COM

188 Emanuela Cresti [well / the prosecutor doesn’t measure himself / he can’t do (it) / against / this / variable / or this set of variables / which condense into the figure of the suspect] Anaphora: sentential head in interrupted COM and clitic in PAR This is a restriction parallel to those just observed in Topics and Appendixes: 1) Topics may have intra-utterance heads of anaphoric relations, but no clitics with cataphoric intra-utterance relation; 2) Appendixes only feature heads of intra-utterance cataphoric relations with Comments, and hardly any clitics. This limitation of the possibility for a head of an anaphoric relation to occur in a Comment is therefore a consequence of the positive constraints on clitics in Topics and Appendixes. c) A Parenthesis may host the head of a subsequent Comment, but this does not happen frequently (34) f1001*VER: è un confronto che / COM ammesso che propone e pone gli stessi obiettivi / PAR deve determinare una linea politica chiara per raggiungerli // COM [it’s a debate which / given that it proposes and poses the same goals / must determine a clear political path to reach them //] Anaphora: Phrasal head in PAR and clitic in COM A clitic placed in a Parenthesis finds its head in a preceding Comment or Topic, and, as discussed, never outside the utterance, because of its informational function which is relative to one IU within the utterance.43 It seems interesting to highlight, however, the absence in the corpus of any cataphoric or anaphoric relation between a clitic in a Parenthesis and a head in an Appendix. Such constraint is not caused by the functional role of the Parenthesis, but rather derives from the functional definition of the Appendix, whose role is specifically one of textual integration of the Comment, its scope being thus restricted to the clitics in that IU. d) A head in a Locutive Introducer cannot establish anaphoric relations with expressions featured in the Comment it introduces. This is due to the necessary discrepancy in deictic space between the basic utterance and the secondary one. Clitics placed in Locutive introducers with a head in a Topic have rarely been found, see (32).

Clitics and anaphoric relations in informational patterning

189

9. Anaphoric and cataphoric relations between linearized elements of an IU No restrictions seem to operate on cataphoric and anaphoric relations between a clitic and its head when both are placed within the same IU. Such relations have been found in all textual IUs which can bear a clitic (Comment, Topic and Parenthesis)44 and are relatively frequent. More specifically, Comment recorded 75 cases of head-clitic anaphoric relation, and 53 cases of clitic-head cataphoric relations. (35) f599*BOR: e / eh / quindi / INP le forme [/] le funzioni linguistiche le ritroviamo poi nelle prime frasi / COM le prime parole // APC [and / um / so / the forms [/]the linguistic functions we find them again in the first sentences / the first words//] Anaphora: contiguous phrasal head and clitic in the same COM (36) f539*DIR: no / INP scusate / CON voglio dire / INX la commissione l’ha svolto un lavoro // COM [no / sorry / I mean / the commission did carry (it) out a job] Cataphora: non-contiguous clitic and phrasal head in the same COM (37) i368*LRT: io / TOP non lo so se ci credo // COM [I / don’t know (it) if I believe that //] Cataphora: non contiguous clitic and clausal head in the same COM In Topics, ten examples of head-clitic anaphoric relation have been found, but only one clitic-head cataphoric relation. (38) i354*BAB: se quello / che spacca la legna / TOP come ho visto io / PAR che non la deve spaccare / TOP e poi dice / INT […] [if he / who cuts timber / like I have seen / that shouldn’t cut it / and then says] Anaphora: non contiguous clitic and phrasal head in two parts of the same TOP In Parenthesis we found four instances of head-clitic anaphoric relations and 3 clitic-head cataphoric relations. 39) f500*DOM: allora magari / TOP ecco / PHA lo so che anche loro / cercano di seguire al meglio il cliente / PAR però il più delle volte / TOP sono portati a [/] a pensare al loro tornaconto // COM

190 Emanuela Cresti [so actually / well / I know (it) that they too / try to satisfy the customer as best they can / bust most of the time / they tend to [/] to think about their own interest //] Anaphora: non-contiguous clitic and clausal head in the same PAR In conclusion, no restrictions on linearized cataphoric and anaphoric relations were found; both relations occur in the most important IUs as far as textual construction is concerned (Comment, Topic, Parenthesis), of course with the exception of Appendix. However, preferential behavioural patterns can be observed. Both relations occur by preference in Comments and Topics. Anaphoric relations are more frequent, but the number of cataphoric relations is high, when compared to the small number of cataphoric relations which cross IU boundaries (mainly 36 cases involving Appendixes).

10. Conclusions Reference relations achieved by means of clitics are among the most relevant forms of textual cohesion (Halliday & Hasan 1976; Dressler 1974; Conte 1999). The “Teoria della lingua in Atto” and the Informational Patterning Hypothesis can consistently explain the distribution of anaphoric clitics in spoken language. We must stress that the specific behaviour of each IU type with regard to the presence of clitics and their quantity, the possibility to host a head of reference and the complex constraints on the distribution of anaphoric relations are also dealt with in this framework, in conjunction with a pragmatic identification of each IU type. A summary of the main characters of their distribution follows. In general, it is evident that the distribution of anaphoric and cataphoric relations varies strongly according to their occurring between a clitic and a head that are placed: a) in the same IU; b) in two different IUs of the same utterance; c) in two different utterances. Therefore, it is safe to say that the rules governing relations between clitics and their heads are sensitive to illocutionary and informational features. It is possible to claim that the semantic processing involved in cataphoras can be somewhat “heavy” for spoken language, while it doesn’t seem to be a problem in writing, where it is recorded with a certain frequency.45 Moreover, our data show that cataphoric clitic-head relations in spoken corpora never cross the utterance boundaries, appear rarely and with strong constraints when the IU boundaries are crossed, whereas they can freely occur within the same IU. We could suggest that cataphoric relations, in

Clitics and anaphoric relations in informational patterning

191

order to be processed, require contiguity between the clitic and its subsequent head. Such a constraint can be clearly appreciated in our framework, since we assume that the IU is the only syntactic and semantic domain where regency and semantic compositionality exist (Cresti 2000). This assumption directly explains why cataphoras are not limited to the IU’s boundaries, when clitics and heads participate in the same syntactic configuration and in the same modality.46 This constraint also explains why a cataphora never involves two distinct utterances, whose syntactic and semantic domains are fully independent. The interpretation of cross-utterance cataphoras would require the addressee to wait for the accomplishment of a new speech act, which may or not be uttered by the speaker, in order to find the reference head of the clitic – already performed in the main utterance – in such new utterance. In writing, the subsequent head need only answer to a criterion of visibility expressed by proximity. Nonetheless, cataphoras appear to be specifically constrained at the informational patterning level, in the rare cases in which they occur across IU boundaries. Such cases only occur provided a cataphora: a) never involves a clitic in Topic; b) typically occurs between a clitic in Comment and its head in Appendix; c) concerns Parentheses or Bound comments in a Stanza. The above distribution can be seen as a function of the strong divergent informational role of both Topic and Appendix. The Topic, whose role is to specify the domain of application of the Comment, requires identified entities, and therefore prevents cataphora or so-called backward anaphora. On the contrary, the cataphoric interpretation of a clitic in Comment is coherent with the definition of an Appendix as a “textual integrator of the Comment”. Since these integrations don’t change either the syntactic configuration or the modal value of the Comment, they only constitute a lexical explication of the clitic in Comment. As for intra-utterance cataphora, Parentheses behave according to their distribution against a Comment, but actually allow for all possibilities: a cataphora between a clitic in Parenthesis and its head in a subsequent Comment, or more rarely a cataphora between a clitic in a Comment and its head in a subsequent Parenthesis. A Parenthesis can be more generally defined as an “open” IU for the internal anaphoric/cataphoric relations with other IUs. The case of Bound comments in a Stanza proves to be quite interesting, as they enable some specific cataphoric references to be established. In principle this appears to contradict our assumptions. But, as we have found, no single Comment in a Stanza bears the illocutionary force of the whole, and when an addressee listens to a Stanza, prosodic cues alert them that

192 Emanuela Cresti another Comment will surely be found with the reference head. A negative finding concerning cataphoras in Locutive introducers must be noted, given that a clitic in this IU cannot have its head in the reported Comment of the secondary utterance (which by definition refers to another deictic space). The Informational Patterning Theory also provides an explanation for the distribution of anaphoric relations, which are more frequent and less constrained than cataphoras in spoken language. Head-clitic anaphoras are not heavy to process, as the reference remains within the linguistic memory, although there is a distance limit, as the literature shows. However, when crossing the utterance boundaries, in order for a head to be recovered, it should by preference be in a Comment or a Topic. This is because the latter two are the only IUs bearing an informational focus; in other words, utterance boundaries still limit possible anaphoric relations, and crossing such boundaries strictly follows an informational requirement. Moreover, crossing utterance limits is also restrained by the internal informational role of the IU where the clitic is placed, such as in the case of Parentheses which never show references out of their host utterance. In conclusion, only clitics placed in Comments, Topics and, more rarely, in Locutive introducers can activate an anaphoric relation with a head in a preceding utterance, and more specifically in an IU with a focal point. On the contrary, anaphoric relations which cross the IU boundaries within an utterance appear to be almost free, and can regard all IU types, except Appendixes, which never bear an anaphoric clitic in our reference corpus. In any case, despite the fact that all textual IUs are able to host a clitic, the possibility of finding their head in a preceding IU is strongly limited by both positive and negative informational requirements: a) clitics in Parentheses are the only ones that can freely have their head in a preceding IU of the utterance (any IU other than an Appendix); b) clitics occurring in Topics can never have an anaphoric head within the utterance; c) clitics in Locutive introducers rarely find their head in a preceding Topic; d) clitics in Comments may have their head in a preceding Topic or Parenthesis. Finally, no constraints operate on linearized anaphora and cataphora. These are, however, only present in Comments, Topics and Parentheses; that is, in those IUs which usually take part in the construction of the utterance text in the most significant way.

Clitics and anaphoric relations in informational patterning

193

Notes 1.

2. 3. 4. 5.

6.

7.

8.

9.

10. 11. 12.

13.

14.

See Simone (1994), A. Lepschy & G. Lepschy (1981), Serianni (1988), Renzi et alii (1985–1994) for a general description of the Italian system. See Cresti (forthc. b) for a corpus driven description. Anaphoric null subjects are out of the scope of this research, which is restricted to positive evidence that can be derived from corpus analysis. C-ORAL-ROM Italia mainly represents the Tuscan variety. See below for the definition of utterance. A referential linguistic head is the expression (word, phrase, sentence or even discursive argument) which the clitic refers to, both when the head precedes the clitic and when follows it, and when it occurs in a different utterance, in a different IU and in the same IU. The corpus corresponds to the full set of utterances bearing a lexical modal index in C-ORAL-ROM Italia (see Tucci 2007 and in this volume). The corpus is balanced in its diaphasic variation with respect to the C-ORAL-ROM sampling strategy, but records a high percentage of verbal utterances; this is a desirable feature for a research on cliticization. A referential linguistic head is the expression (word, phrase, sentence or even discursive argument) which the clitic refers to, both when the head precedes the clitic and when follows it, and when it occurs in a different utterance, in a different IU and in the same IU. LABLITA (Laboratorio linguistico del Dipartimento di Italianistica) currently runs a research programme on its spoken corpora archive, which is dedicated to the morphosyntactic and prosodic correlations of the informational structure in spoken Italian. See http://lablita.dit.unifi.it for the description of the data base and the full bibliographical references. Bibliographical references for each IU are given below. In this perspective, we don’t deal with relevant topics which have been debated in the literature (in other words, which are the best semantic features to be the reference head of a clitic or of a clitic chain, from proper names to personal pronouns; how long a clitic chain can be; or why a head of reference must be new, or accessible or near). See for instance Korzen (2007). See Hirst & Di Cristo (1998). See Scarano in this volume. See Cresti (1994, 2000) for the general framework and Scarano (in this volume) for a detailed presentation. See also Cresti & Firenzuoli (1999, 2002), Cresti, Moneglia & Martin (2003) for experimental researches on the core prosodic feature of the IUs. Here and below, in parentheses, are the tags used to identify the IU types in this paper. See Cresti & Moneglia (forthc.) for a detailed definition of the IUs and the criteria for their annotation in spontaneous speech-corpora. See Cresti & Firenzuoli (1999), Firenzuoli (2003) and also Cresti (2006) for a recent overall picture.

194 Emanuela Cresti 15. A focus is an informational prominence in an IU necessarily correlated to a perceptual prosodic nucleus (Cresti 2002). 16. The general term “context” refers here to both the pragmatic context and the preceding textual information. See Tamburini (1994), Cresti & Firenzuoli (1999) and Signorini (2004) and (2005) for studies focusing on this IU in the present framework. 17. The Topic has a severe restriction on the deontic modal value. See Tucci (2007, 2008) and Tucci (in this volume) 18. The theoretical definition of Topic and the question of its distribution, and consequently those of Appendixes, are hardly debated (Li 1976). See for different assumptions Givón (1983), Lambrecht (1994), and especially Mereu (2004, in this volume) and Lombardi Vallauri (2007, in this volume). 19. This could be evaluated as anti-intuitive. Actually the performance of a Topic IU implies for the speaker to have already programmed at least the TopicComment patterning of the utterance. This condition is not strictly necessary for the other IUs, which can be added by the speaker during the performance of the utterance. 20. The informational Topic-Comment order has been identified in all languages studied as regards their informational structure, and more specifically in creole languages (Li 1976). For specific assumptions, see also Mereu in this volume. According to Moneglia (1994); Moneglia and Cresti (1993); Moneglia and Cresti (2000), the ontogenesis of a Topic sees an early stage where the Comment-Topical sequence is established, and a later development into the adultlike Topic-Comment pattern after a long process. This adult-like structure seems quite complex, as it usually follows the onset of compositional structures of head dependency (VP or NP). 21. See Tucci (2002) and (2004) for a detailed study regarding Parenthesis in this framework. 22. See E. Tucci (2007). 23. See Giani (2003 and 2005) for a detailed study regarding Locutive introducers in this framework. 24. This paper does not deal with Appendix IUs used to integrate a Topic, which are mostly found in formal speech (Cresti 2000 and Signorini 2005). 25. See E. Tucci (2007) for a detailed study regarding Appendix in this framework. 26. Appendixes can be erased from the signal without losing the full pragmatic consistence of the utterance. This property has been verified many times in experimental studies at LABLITA (See Tamburini 1998, 1995, Signorini 2005) and also regards those Appendixes which contain the head of a clitic placed in a Comment. 27. See Frosali (2008) for a detailed study regarding Dialogical units in this framework. 28. See Cresti (forthcoming a) for the distinction between coordination and juxtaposition.

Clitics and anaphoric relations in informational patterning

195

29. Here and below, enclitic positions are marked by a hyphen before the clitic form. 30. Those types that are missing in this corpus may of course occur in larger corpora, but remain nonetheless low-frequency morphemes. 31. According to Sobrero (1994) the anaphoric function of clitics in Spoken italian should be “weak”. 32. This finding emerged in the early phase of this research and has been closely verified because of its relevance. For this reason, a special subcorpus taken from a research focussing on Appendix has been added to the original set of utterances tagged in the reference corpus. This subcorpus records 201 utterances and is derived from another C-ORAL-ROM sampling (around 5000 utterances). The addition is formed by the full set of utterances containing an Appendix (less than 5 % of total complex utterances). 33. In all examples, here and below, strokes mark Terminal (//) and Non- terminal (/) prosodic breaks. The information function of the IU is marked with the corresponding apex tag in the prosodic break. 34. The five occurrences of the formulaic expression scusami [excuse me] which have been found in Conatives, do not change the overall picture. 35. Here and below the examples taken from the reference corpus are marked with their reference numbers in the Data Base. 36. An IU can be suspended by the insertion of a Parenthesis after which the IU continues until its conclusion. 37. In our terminology, linguistic relations occurring within the same IU are referred to as “linearized”. 38. Of course this finding must be confirmed by a more extensive analysis, but our sample is sufficiently large to at least highlight a trend. 39. In our approach contiguous expressions that can be analyzed from a syntactic point of view as a dislocation, belong to the same IU, if they are not separated by a prosodic break. Consequently they develop together the same informational function. Every IU must be identified trough a dedicated PU. 40. However, this could not explain why an Appendix never hosts the head of cross-utterance anaphoric relations, as a Topic does. We believe that this is a consequence of its specific semantic content and of the fact that it doesn’t bear an informational focus, as we explained in paragraph 6. 41. Not necessarily an immediately preceding utterance. The literature has frequently dealt with this topic looking for constraints (See Mortara-Garavelli 1994; Kleiber 2007; Korzen 2007). 42. In this sense a Parenthesis cannot be considered a reduced utterance nor a syntactic reduced clause. For a different evaluation see Schneider (2006). 43. More in detail, Parentheses commonly develop anaphoric relations with a head in a preceding Comment (17 instances) and with a Topic (7 instances). 44. Only one case in Locutive introducer. 45. For instance, Mortara-Garavelli (1993) records that in recent years cataphoric relations in Italian newspapers and advertising have increased. In this context,

196 Emanuela Cresti cataphora is interpreted as a rhetoric device creating suspense. The author also reports some spoken examples of cataphora, which of course are not tagged with regard to utterance boundaries and IU patterning. For this reason, a comparison with our data is not possible. 46. Evidence of this emerges from the scope of Modal values. While two IUs belonging to the same utterance can bear different modal values, only one modal value is allowed in one IU. See Tucci in this volume.

References Austin, John L. 1962 How to Do Things with Words. Oxford: Oxford Univerity Press. Berretta, Monica 1985 I pronomi clitici nell’italiano parlato. In Gesprochenes Italienisch in Geschichte und Gegenwart, G. Holtus and E. Radtke (eds.), 185–224, Tübingen: Narr. 1990 Catene anaforiche in prospettiva funzionale: antecedenti difficili. In Anaphoric relations in sentences and texts, Conte M. E. (ed.), Rivista di linguistica 2 (1): 91–120. 1994 Il parlato italiano contemporaneo. In Scritto e parlato, L. Serianni and P. Trifone (eds.), 239–270. Einaudi. Buommattei, Benedetto 1623–43 Della lingua toscana. Florence: Zanobbi-Pignone. Castelvetro, Lodovico 1563 Giunta fatta al ragionamento degli articoli et de verbi di Messer Pietro Bembo. Modena: Eredi di Cornelio Gadalino. Conte, Maria-Elisabeth (ed.) 1990 Anaphoric relations in sentences and texts. Rivista di Linguistica 2 (1). 1999 Le condizioni di coerenza. Ricerche di linguistica testuale. Alessandria: Edizioni dell’Orso. C-ORAL-ROM See http://lablita.dit.unifi.it/coralrom. Corblin, Francis 2007 Pronouns and mentions. In Comparing Anaphors, between sentences, texts and languages, I. Korzen and R. Lundquist (eds.). Copenhagen Studies in Language 34: 27–43. Cornish, Francis 2007 Text-type and anaphor-type distribution in terms of discourse-functional units: a comparison. In Comparing Anaphors, between sentences, texts and languages, I. Korzen and R. Lundquist (eds.). Copenhagen Studies in Langage 34: 45–66.

Clitics and anaphoric relations in informational patterning

197

Cresti, Emanuela 1994 Information and intonational patterning. In Accent, intonation and modèles phonologiques, P. Martin, B. G. Ferguson and H. Gezundhajt (eds.), 99–140. Toronto: Edition Mélodie. 2000 Corpus di italiano parlato, Vol. 1/2, CD-ROM. Florence: Accademia della Crusca. 2002 Alcune riflessioni sulla marcatezza e sul concetto di focus. In L’infinito and oltre. Omaggio a Gunver Skytte, H. Jansen, P. Polito, L. Schlosser and E. Strudsholm (eds.), 107–129. Odense University Press. 2003 Illocution et modalité dans le Comment et le Topic. In Macrosyntaxe et pragmatique: l’analyse linguistique de l’oral, A. Scarano (ed.), 133–182. Roma: Bulzoni. 2006 Some comparisons between UBLI and C-ORAL-ROM. In Spoken Language Corpus and Linguistics Informatics, Y. Kawaguchi, S. Zaima and T. Takagaki (eds.), 125–152. Amsterdam: John Benjamins. forthc. a La parataxe: articulation informative dans le parlé spontané vs juxtaposition syntaxique dans l’écrit litteraire? In Actes du I ° Colloque international de macro-syntaxe “La parataxe”, Neuchâtel 12–15 février 2007, M. J. Beguelin and M. Avanzi (eds.). forthc. b Frequenza ed uso dei clitici in un corpus di parlato spontaneo (CORAL-ROM Italia). In Actes du XXV CPRL, P. Danler (eds.), Niemeyer. Cresti, Emanuela and Valentina Firenzuoli 1999 Illocution et profils intonatifs de l’italien. Revue française de linguistique appliquèe 4 (2): 77–98. 2002 L’articolazione informativa topic-comment e comment-appendice: correlati intonativi. In La fonetica acustica come strumento di analisi della variazione linguistica in Italia. Atti delle XII° GFS, Agostino Regnicoli (ed.), 153–161. Roma: Editrice “Il Calamo”. Cresti, Emanuela and Massimo Moneglia 1993 Formazione dell’atto linguistico complesso e intonazione: ontogenesi delle relazioni informative in italiano. In Ricerche sull’acquisizione dell’italiano, E. Cresti and M. Moneglia (eds.), 63–114. Roma: Bulzoni. 2000 The value of prosody in the transition to complex utterances. Data and theoretical implications from the acquisition of Italian. In Proceedings VIIIth International Congress IASCL, 12–16 luglio 1999, S. Sebastian, 851–873. Cascadilla Press. 2006 C-ORAL-ROM. Prosodic boundaries for spontaneous speech analysis, In Spoken Language Corpus and Linguistics Informatics, Y. Kawaguchi, S. Zaima and T. Takagaki (eds.), 89–114. Amsterdam: John Benjamins. forthc. Specifications for the annotation of the informational patterning in spontaneous spoken Italian. Florence: Florence University Press.

198 Emanuela Cresti Cresti, Emanuela and Massimo Moneglia (eds.) 2005 C-ORAL-ROM. Integrated reference corpora for spoken romance languages, DVD + Vol. Amsterdam: John Benjamins. Cresti, Emanuela, Massimo Moneglia, and Philippe Martin 2003 L’intonation des illocutions naturelles répresentatives: analyse et validation perceptive. In Macrosyntaxe et pragmatique: l’analyse linguistique del’oral, A. Scarano (ed.), 243–264. Rome: Bulzoni. Dressler, Wolfgang U. 1974 Introduzione alla linguistica del testo. Rome: Officina Edizioni. Evans K., Giulio C. Lepschy, S. Morris et al. 1978 Italian clitic clusters. Studi italiani di linguistica teorica e applicata 7: 153–168. Firenzuoli, Valentina 2003 Le forme intonative di valore illocutivo dell’italiano parlato. Analisi sperimentale di un corpus di parlato spontaneo (LABLITA), Ph.D. diss., University of Florence. Frosali, Fabrizio 2008 Le unità d’informazione di Ausilio dialogico: valori percentuali, caratteri intonativi, lessicali and morfo-sintattici in un corpus di italiano parlato (C-ORAL-ROM). In Prospettive nello studio del lessico italiano, E. Cresti (ed.), 417–424. Florence: Florence University Press. Giani, Daniela 2003 Le discours direct rapporté dans l’italien parlé et écrit. In Macrosyntaxe et pragmatique: L’analyse linguistique de l’oral, A. Scarano (ed.), 203–213. Rome: Bulzoni. 2005 Il discorso riportato nell’italiano letterario e parlato: confronto tra due corpora, Ph.D. diss., University of Florence. Givón, Talmy 1983 Topic continuity in discourse: an introduction. In Topic continuity in discourse, T. Givón (ed.), 3–41. Amsterdam: John Benjamins. Halliday, Michael A. K. and Ruqaiya Hasan 1976 Cohesion in English. London: Longman. Hirst, Daniel and Albert Di Cristo 1998 Intonation systems: A survey of twenty languages. Cambridge: Cambridge University Press. Kayne, Richard S. 1975 French Syntax. Cambridge, MA: MIT Press. Kleiber, Georges 1994 Anaphores et pronoms. Paris: Duclot 2007 Anaphore(s), Théorie de l’accessibilité et pronoms. In Comparing Anaphors, between sentences, texts and languages, I. Korzen and R. Lundquist (eds.). Copenaghen Studies in Langage 34: 75–91.

Clitics and anaphoric relations in informational patterning

199

Korzen, Iørn 2007 Linguistic typology, text structure and anaphors. In Comparing Anaphors, between sentences,texts and languages, I. Korzen and R. Lundquist (eds.). Copenaghen Studies in Langage 34: 93–109. Jensen, Anne 2006 The absence of appositions in Danish speech. In Appositions in selected European languages, P. Colliander (ed.). Copenhagen Studies in Language 33: 9–28. LABLITA See http://lablita.dit.unifi.it. Lambrecht, Knud 1994 Information structure and sentence form. Cambridge: Cambridge University Press. Lepschy, Anna Laura and Giulio C. Lepschy 1981 La lingua italiana. Milan: Bompiani. Li, Charles N. (ed.) 1976 Subject and Topic. New York: Academic Press. Lo Cascio, Vincenzo 1970 Strutture pronominali e verbali italiane. Bologna: Zanichelli. Lombardi Vallauri, Edoardo 2007 «Appendice» una categoria informazionale o semantica? In Lessico, grammatica, testualità, Acta Romanica Basiliensa 18, A-M. de Cesare and A. Ferrari (eds.), 75–95. Università di Basilea. this vol. “Appendix” or “postposed Topic”: Where does the difference lie? Mambelli, Marcantonio (Cinonio) 1644 Osservazioni della lingua italiana. Raccolte dal Cinonio Accademico Filergita, le quali contengono il trattato delle particelle. Ferrara: Giuseppe Gironi. Mereu, Lunella 2004 La sintassi delle lingue del mondo. Bari: Laterza. this vol. Universals of information structure. Moneglia, Massimo 1994 The ontogenetic foundation of informational patterning. In Intonation, stress and phonological models, P. Fergusson, H. Gezundhajt and P. Martin (eds), 155–186. Toronto: Editions Mélodie. Moneglia, Massimo and Ida Tucci forthc. Strategies of modalization in spoken language. A corpus-based analysis. In Proceedings of the 3rd Workshop “Corpora and Pragmatics”, C. D. Pusch (ed.), Freiburg, 14–17 September 2006. Mortara-Garavelli, Bice 1993 Strutture testuali e retoriche. In Introduzione all’italiano contemporaneo: le strutture, A. Sobrero (ed.), 371–402. Bari: Laterza. Muller, Claude (ed.) 2001 Clitiques et cliticisation. Paris: Champion.

200 Emanuela Cresti Renzi, Lorenzo, Giampaolo Salvi, and Anna Cardinaletti (eds.) 1985–94 Grande grammatica italiana di consultazione. Bologna: Il Mulino. Reinhart, Tanya 1983 Coreference and bound anaphora: a restatement of the anaphora questions. Linguistics and Philosophy 6 (1): 47–88. Rizzi, Luigi 1982 Issues on Italian Syntax. Dordrecht: Foris. Scarano, Antonietta this vol. The prosodic annotation of C-ORAL-ROM and the structure of information in spoken language. Scarano, Antonietta (ed.) 2003 Macrosyntaxe et pragmatique: L’analyse de l’oral. Rome: Bulzoni. Schneider, Stefan 2007 Reduced parenthetical clauses as mitigators. A corpus study of spoken French, Italian and Spanish. Amsterdam: John Benjamins. Serianni, Luca 1989 Grammatica italiana. Lingua comune and lingua letteraria. Torino: UTET. Seuren, Pieter A. M. 1974 Pronomi clitici in italiano. In Fenomeni morfologici e sintattici nell’italiano contemporaneo, M. Medici and A. Sangregorio (eds.), 309–327. Rome: Bulzoni. Signorini, Sabrina 2004 Il topic: criteri di identificazione e correlati morfosintattici e intonativi in un corpus di italiano parlato. In Atti del convegno nazionale “Il parlato italiano”, F. Albano Leoni, F. Cutugno, M. Pettorino and R. Savy (eds.), Napoli, 13–15 February 2003. 2005 Topic and Soggetto in corpora di parlato italiano, Ph.D. diss., University of Florence. Simone, Raffaele 1990 Fondamenti di linguistica, Bari: Editori Laterza. 1993 Stabilità e instabilità nei caratteri originali dell’italiano. In Introduzione all’italiano contemporaneo: le strutture, A. A. Sobrero (ed.), 41–100. Bari: Laterza. Sobrero, Alberto A. 1993 Pragmatica. In Introduzione all’italiano contemporaneo: le strutture, A. A. Sobrero (ed.), 403–450. Bari: Laterza. Tamburini, Guido 1998 L’ordine dei costituenti e l’articolazione dell’informazione. Studi di Grammatica Italiana XVII: 399–433 Tucci, Elena 2007 L’unità informativa di appendice in un corpus di italiano parlato (CORAL-ROM): caratteristiche intonative, semantiche e morfosintattiche. Masters thesis, University of Florence.

Clitics and anaphoric relations in informational patterning Tucci, Ida 2002 2004

2007 2008

this vol.

201

Caratteristiche sintattiche e frequenza dell’inciso in un corpus di italiano parlato. Degree diss., University of Florence. L’inciso: caratteristiche morfosintattiche e intonative in un corpus di riferimento. In Atti del Convegno “Il parlato italiano”, F. Albano Leoni, F. Cutugno, M. Pettorino, R. Savy (eds.), 1–14. Napoli: D’Auria. L’espressione lessicale della modalità nel parlato spontaneo. Dati dal corpus C-ORAL-ROM Italiano. Ph.D. diss., University of Florence. La modalizzazione nel parlato spontaneo. Dati dal corpus C-ORALROM Italiano. In Atti del IX Congresso SILFI “Prospettive nello studio del lessico italiano”, E. Cresti (ed.), 377–386. Florence: Florence University Press. The informational structure and the scope of lexical modality in spoken Italian (C-ORAL-ROM).

The informational structure and the scope of lexical modality in spoken Italian (C-ORAL-ROM) Ida Tucci

1.

Modality

1.1. A definition of modality In the history of philosophy of language and linguistics, many definitions of “modality” have been proposed.1 The most classic characterization of this notion dates back to Bally (Bally 1932 and 1942), who defines modality as “la forme linguistique [Modus on Dictum] d’un jugement intellectual, d’un jugement affectif ou d’une volonté qu’un sujet pensant énonce à propos d’une perception ou d’une représentation de son esprit”, i.e. the speaker’s cognitive, emotive, or volitive attitude towards a state of affairs (Kiefer 1994: 2516). Coherently with this original approach, we assume that modality represents all kind of subjective/evaluative meanings. This basic idea has been reported, with many other characters, also in Lyons (1977: 452): “[modality represents] the speaker’s opinion or attitude towards the proposition that the sentence expresses or the situation that the proposition describes […]”. The identification of subjective/evaluative meanings in the actual language performance is, however, puzzling. Palmer (1986 and 2001), presents a general survey of modality as a typological category. He draws attention to the subjective nature of modality, but he also defines it as “the grammaticalization of speakers’ (subjective) attitudes and opinions” (Palmer 1986: 16). The linguistic analysis of real data takes advantage of the latter contribution, as we can tie modal values to specific linguistic indexes. In Bybee and Fleischman’s view we again find a reference to modal operations as “additions” to “neutral” verbalizations: “When the proposition of an utterance in the most neutral semantic status, i.e. factual or declarative, is subject to further addition or overlay of meaning, this extension represents modality” (Bybee & Fleischman 1995: 2). The concept of further addition or overlay of meaning in a “neutral” utterance hints towards a lexicalization of modal meanings in languages that modifies non-modalized entities.

204 Ida Tucci In order to prevent possible misunderstanding caused by the huge literature on the topic, it is also necessary to distinguish modality from sentence typology (Graffi 1994; Schneider 1999) and from illocution. In our theoretical framework, illocution concerns the attitude of the speaker towards the listener (Cresti 2002), while, according to a traditional definition, modality consists in the evaluation of the speaker towards his own verbalization. Therefore, modality belongs to semantics and not to pragmatics, as it is a cognitive process, not an act. (Tucci & Moneglia, to appear).

1.2. Our research This view has been adopted in this paper for the study of modality in a large spoken Italian corpus (C-ORAL-ROM Italian)2, taking into consideration those lexical indexes that modify the “neutral status” of the spoken Dictum. More specifically, the C-ORAL-ROM-Italian spontaneous speech corpus has been the object of a detailed analysis of all contexts in which the main lexical modal indexes occur, the properties of lexical modalization have thus been derived from corpus analysis. This procedure forced us to consider modality in connection with the specific linguistic units that characterize the speech performance, such as utterances and informational units (see below §3), and to investigate into the relation between modal values and the informational structure of spoken language. As a matter of fact, in spontaneous speech a few actual propositions occur and the Dictum is rather organized into utterances showing an informational structure (Cresti 2000, 2005). As far as spontaneous speech is concerned, each linguistic filling of an informational unit behaves as a sole Dictum. Therefore, the scope of a modal index cannot be set on hypothetical “propositions”, as the written idealization of language leads us to do, but has to be considered within the lexical, morphological and semantic domain of the informational structure of the utterance. A closer look at the distribution of modal indexes in our reference corpus pointed out that: a) on 38,741 total utterances, 5,498 are lexically modalized, (14.2%); b) on 5,498 modalized utterances 25% are simple utterances (a single information unit) and 75% are compound (patterned in more information units); c) 30.4% of compound utterances feature more modal indexes distributed each in a different information unit of the same utterance;

The informational structure and the scope of lexical modality in spoken Italian 205

d) 92.4% of the informational units feature a sole modal index ; e) only 7.6% of informational units feature more than one modal index. Various kinds of results can be derived from the data; anyway it seems to be relevant that compound utterances preferentially record more than one modal index, each distributed in a different information unit, and modal indexes occur, in the great majority of cases, in isolation within an information unit. All these aspects lead us to formulate a question regarding the scope of modal indexes: if modality is basically a property of propositions in the written domain, what is the linguistic entity which undergoes modal modification in spontaneous speech? The paper will offer positive evidence towards considering the information unit as the pertinent “unit”, the Dictum, on which a modal index operates, as a Modus. 1.3. A summary In section 2 the definition of modal values adopted for corpus annotation (Alethic, Epistemic, Deontic) is presented. The informational functions which pattern speech are then defined in section 3 in accordance with the Informational Patterning Theory (Cresti 2000). In 4, the set of lexical indexes taken into account in the analysis is selected and the operational setting for corpus annotation is proposed. Sections 4, 5, 6 and 7 present quantitative data regarding the percentage of modalized utterances, the distribution of modal indexes and modal values in the corpus, and the relation between modal values and informational values which emerges from corpus analysis. Specifically, we will show that only a sub-set of informational functions may bear a modal index; a striking find is the fact that Appendix units are never modalized. Moreover, within this set, modal values have a preferential distribution: Topic units bear a restriction regarding Deontic values and Parentheticals are, by preference, Epistemic, while Comment units are on the contrary unrestricted. These aspects are positive evidence to support that the information unit is, in speech, the linguistic domain where a modal value counts. The paper finally presents -in 8.- the impossibility of a compositional result of modal indexes, each distributed in different information units within a same utterance, in comparison with the necessary composition of modal indexes within a same information unit. All these reasons lead to consider information units, rather than the utterance itself, the unit that is properly concerned by modality.

206 Ida Tucci 2. Selected modal values Among the various types of modal values recognized in the literature, those that have more importance from a linguistic point of view are the alethic, epistemic and the deontic modality. Let’s briefly describe these modal typologies:  Alethic modality: the point of view of the speaker refers to the necessity or possibility of the truth of the propositions – that is, to propositions that can be verified in the actual world or in possible worlds, by means of logical, factual or perceptual judgments (Lyons 1977; Perkins 1983; Kiefer 1994). To these fundamental features, we can add what has been called dynamic modality (Palmer 1986; Huddleston & Pullum 2002), which expresses the ability/disposition of a subject to do something in a possible world. Ex. a. A leopard must be spotted. b. A swan can be black. c. An athlete can run faster than you! Let’s see some examples from our reference corpus: (1)

[itelpv13] *CRI: tirandosi dietro la porta / può aver rimbalzato //3 [when he dragged the door / it could have bounced //]

(2)

*PMA: il Pubblico Ministero / deve illustrare i fatti // [a Public Prosecutor / must 4 illustrate the facts //]

[inatla01]

Alethic modalities range from very general physical or natural necessities (what is possible or necessary in view of physical/natural laws) and the possibility of very specific restrictions encompassing the full range of dynamic modalities – be they agent-oriented (what is necessarily possible in view of certain internal features of an agent) or circumstantial (what is necessarily possible in view of certain facts of the external world).  Epistemic modality: the speaker’s point of view refers to the possibility or necessity of a proposition to be verified in possible worlds specifically related to the speaker’s belief, opinion or attitude (evaluative modality) (Lyons 1977; Palmer 1986; Venier 1991; Hoye 1997; Nuyts & Wietske 1999; Nuyts 2001; Papafragou 2001; Pietrandrea 2005). Ex. d. Mario may leave tomorrow (he told me something like that) e. Mario must be gone (I don’t see him any more) f. Mario is depressed, I believe.

The informational structure and the scope of lexical modality in spoken Italian 207

(3)

*PAL: dovrebbero essere in sei / a mangiare // [they should be six / eating //]

[ifamcv04]

(4)

*GIA: cioè / è / immagino / un lavoro allucinante // [well / it is / I suppose / a horrible job //]

[ifamdl16]

(5)

*GUI: è andato via / fortunatamente // [he went away / fortunately //]

[ifammn22]

 Deontic modality: the speaker’s point of view, intended as morally responsible agent, refers to the duty, obligation, permission, wish expressed into a proposition (see von Wright 1951; A. G. Conte 1977; M. E. Conte 1995; Pottier 2000). Ex. g. We had to finish this work in three days. h. I want to talk with him. i. You can’t lie to me! Deontic values also extend to “duties” that are social or moral “obligations” in relation to an axiological5 manifestation of attitudes (see Hare 1961; Galvan 1991). Let’s see some examples from our reference corpus: (6)

*DAN: ci sono militanti di partito / che possono averlo e hanno il diritto di averlo // [ipubcv01] [there are party militants/ who can have it and who have the right to have it //]

(7)

*ALE: noi vogliamo che l’imposizione si riduca // [we want the levy to be reduced //]

(8)

*ROS: bisogna essere ironici / perché non bisogna mai prendere troppo sul serio / quello che ci succede // [imedin01] [we must be ironical / because we must never take too seriously / what happens to us //]

[ifammn22]

3. Prosody and informational structure of spoken language The reference unit suitable for linguistic analysis in spontaneous speech is the “utterance”. In the C-ORAL-ROM framework, the utterance is “the minimal linguistic entity such that it can be pragmatically interpreted; i.e. the linguistic entity that is ‘concluded’ and ‘autonomous’ from a pragmatic point of view” (Cresti & Moneglia 2006: 91).

208 Ida Tucci This definition, which follows the direction of Austin’s works (Austin 1962), does not imply any necessary correlation between “utterance” and “proposition”, but rather highlights the relations between prosody and the accomplishment of speech acts. In C-ORAL-ROM the operative criterion for the segmentation of the speech flow into utterances is prosodic: prosody marks boundaries as terminal breaks, and the former are assumed to define the utterance limits. However, within the Informational patterning theory, that is used here as a general frame for the study of information structure in speech, the correlation between prosodic envelopes and linguistic units goes beyond this. According to this theory, each utterance corresponds to an informational pattern, which is isomorphic to a prosodic pattern (see Cresti 1987, 1994, 2000). More specifically, each group of words within a prosodic envelope is assumed to receive a specific functional role, which is also marked by a specific prosodic form, which ends with a non terminal break. 6 The set of functional values identified within the Informational patterning theory is defined in a closed list (see Tab 1):

Dialogical informational units

Textual Informational units

Table 1. Informational tagset 7 (Cresti 2000) Type

Informational function

Tag

Comment

specifies the illocutionary force of the utterance COM and is the only necessary and sufficient information unit within the utterance

Topic

the application field of the Comment that specifies the object, state or event the speech act is about

TOP

Parenthetical

an evaluative metalinguistic insertion in the text of the utterance

INX

Appendix

the textual integration of a Comment

APC

Incipit

dedicated to turn-taking

INP

Phatic

dedicated to regulation of the communication channel

FAT

Allocutive

for a direct summoning of the interlocutor

ALL

Conative

To push the interlocutor to do something or to avoid a certain behaviour

CON

Expressive

performed to emphasize the performance of the speech act

EXP

Speaker introducer

the unit that marks reported speech and introduces exemplifications, listings, etc.

ILC

The informational structure and the scope of lexical modality in spoken Italian 209

An utterance observed in a spoken corpus can be simple, i.e. made up by just one information unit (necessarily a Comment), or compound, i.e. a pattern of different information units (a Comment unit plus other information units with different functions). Therefore C-ORAL-ROM prosodic tagging, by marking terminal and non-terminal prosodic boundaries, also highlights whether or not spoken language utterances have an informational structure. Indeed it turns out that roughly 57% of spoken utterances in the corpus correspond to an informational pattern, while 43% are simple utterances. However, the C-ORALROM data-base does not specify which informational function is conveyed by each prosodic unit. In order to investigate into the relation between modality and informational structure of spoken Italian, this annotation has been added to the database, for what concerns all utterances bearing at least one of the modal indexes identified in the section below.

4. Modal markers A number of scholars maintain that modality covers a broad range of semantic contents that is not easy to define (e.g., Coates 1983: 4; Perkins 1983: 1; Palmer 1986: 2; Bybee & Fleischman 1995: 2). They also believe that it can be expressed by means of different formal devices in different languages. Therefore, the semantic definition of modality may be subject to modification if a better match with the corresponding named formal category is to be achieved. In Italian, for instance, a defined set of lexical entries (i.e. Modals and modal adverbs) shares common formal characteristics that serve as a guideline for researchers to decide where to set the limit for their semantic range. In any case it is not always easy to decide what conveys modal significance and what doesn’t. For this research we analyzed all the utterances in C-ORAL-ROM Italian (38,741 utt. – 274,356 words)8 which contain an “explicit” lexical and/or morphological index of modality (see Tucci 2007). More specifically, we considered verbal and adverbial structures such as the following items: – Modals (potere, dovere, volere) – Belief verbs (credere, pensare, supporre, etc.) – Periphrastic and analytical forms (essere da + infinito, andare + participio passato, avere da + infinito, etc.). – Copular verbs (sembrare, parere) – Desire and necessity verbs (desiderare, bisognare, occorrere, etc.)

210 Ida Tucci – Evaluative adjectives in nominal predicates (è giusto, è doveroso, è improbabile, etc.) – Modal adverbs9 (probabilmente, evidentemente, giustamente, per forza, etc.) – Verbal moods (Future indicative, Conditional) We wish to stress that in our approach, no specific modal value is necessarily selected by a lexical index of modality. For this reason, each time we assigned the modal value of the index according to: a) the lexical meaning of the index; b) the informational function of the hosting informational unit; c) the communicational context (conversational background) of the whole utterance.

4.1. Operational tools of analysis The modal value expressed by an index appears not to be lexically determined but context dependent and subject to subtle shifts in conversation that are likely to go undetected. Therefore, on the basis of the previously reported definition of modality (§1.1), we assigned a modal value to each information unit which bears one or more modal lexical indexes. In order to interpret the speaker’s attitude, which emerges in any information unit, as we have already discussed, we have examined the lexical meaning of the index, the informational function of the unit where it occurs and, more generally, the utterance context in a holistic way. Let’s look at the examples below: (9)

*MAX: necessariamente deve essere un nome solo //COM [it must be necessarily one name //]

[ipubdl 04]

This utterance in (9) is a request of confirmation from the interlocutor (a financial agent). The two modal entries (necessariamente and deve) are embedded in the same Comment unit. The informational unit has a unique deontic value despite the fact that two lexical indexes occur in it. In this case, we assigned a deontic value to the whole utterance that is patterned only by a comment unit (simple utterance). A different case is (10): (10) *CAT: beh / INP voglio dire / INX la cosa è certamente diversa // COM [well / I mean / it is of course a different matter //] [imedts 06]

The informational structure and the scope of lexical modality in spoken Italian 211

In the utterance (10) we find a modal verbal locution (voglio dire) embedded in a Parenthetical information unit and a judgment adverb (certamente) in a Comment unit. The context of the conversation is a television debate about genuine food. We assigned an epistemic value to the Parenthetical unit mostly depending on its informational function (a Parenthetical evaluates the content of the utterance according to a subjective point of view). More specifically, we considered the value of the index and the function of the information unit in a holistic way. In the same way, we assigned an epistemic value to the Comment unit that asserts the speaker genuine opinion about the matter of discussion. We can say that in (10) an epistemic modal value expressed by the Comment is reinforced by the modal value still epistemic, born by the Parenthetical unit. A more complex case is in (11): (11) *IDA: in realtà / TOP Basilicata / TOP dovrebbe significare la terra dei boschi // COM [ifamdl18] [as a matter of fact / Basilicata / should mean land of woods //] The example in (11) is an assertion regarding the etymological definition of a name of a region (Basilicata) uttered during an informal conversation. We find two modal indexes, one within a Topic unit (in realtà) and the other in a Comment unit (dovrebbe significare). We assigned an alethic value to the index in Topic (as a matter of fact) because the speaker is providing a “descriptive” alethic application field of the following statement, and to that in Comment an epistemic value (it should mean), because in the following part of the utterance the speaker mitigates the “strength” of the statement proposing it as a supposition. In this case our choice was to describe the semantic structure of the utterance as a pattern of coexistent modal values within the utterance itself (see § 7.1, 7.2). 5. Frequency of morpho-lexical indexes of modality The proportion of modal indexes in the reference corpus shows that the verbal strategy is more represented than the adverbial one: respectively 73.4% vs. 26.6%. In detail, within the verbal strategy, Modals account for more than half of the forms (53%) and belief verbs for almost a fifth (18.5%). Based on the frequency lists of all the lemmas in the reference corpus, we stress that

212 Ida Tucci modals and belief verbs are high rank verbs: they altogether represent approximately 8% of the total verbal tokens. If we consider their percentage on the total of verbal utterances, we find that their weight reaches almost 20%, which means that nearly one in five utterances has a verb that belongs to this category. Periphrastic and analytic forms are represented in lower percentages (respectively 4.6% and 4.9%). Desire and necessity verbs equal 6.2%. Concerning modalization by moods, conditional and indicative future as modal morphemes of a verbal form are less frequent than would be expected (altogether 12.8%) (Fig. 1).

Figure 1.

The more frequent verbal modalization strategies let us appreciate how the domain of a modal value is related to the syntactic configuration, i.e. the syntactic regency of the verbal entry, and how this copes with an information unit.

6. Frequency of modalization in the reference corpus Lexical modalization is very frequent in spoken language. In our reference corpus, we found 5,498 lexically modalized utterances, corresponding to 14.5% of the total utterances. Considering that in the same corpus subordination (che) concerns 20% of utterances, and coordination (e, ma) 17% together with verbal negation (non) 11%, it is clear that lexical modalization is a structural strategy of spoken grammar.

The informational structure and the scope of lexical modality in spoken Italian 213

6.1. Modality and information units Let’s now discuss some relations between the modal lexicon and the information units within the utterance. An important result of our corpus-based research is that there is a very different behaviour of informational unit types according to lexical modalization: some unit types are lexically modalized and they are frequently modalized, other types do not record any kind of modalization. In our reference corpus we found that only Comment information units, Parenthetical, Topic and Speaker introducer units contribute to the expression of modality into the utterance. On the contrary, information units such as Incipit, Allocutive, Phatic, Expressive, Appendix are never modalized. The figure below shows the occurrence of all information units in the reference corpus and the occurrence of each type in its lexically modalized form:

Figure 2.

The graph below shows the absolute percentage of modalization of the select modalized information unit type (figure 3):

214 Ida Tucci

Figure 3.

Considering their absolute occurrence, Topic information units are lexically modalized in 26% of the occurrences; Speaker introducers are modalized in 16% of their occurrences. On the contrary, the most frequent modalized units are Comment units (65.8%) and Parenthetical units (53.1%), which also represent the typical units that concur in the modalization of a compound utterance (33%). The fact that only specific types of information units host modal expressions and that also the frequency of modalization differs in relation to the types of information unit does not seem accidental. In our view, it shows that the expression of subjective/evaluative meanings can be accomplished in speech only within certain expressions (Topic, Comment, Parenthetical and Speaker Introducer) that put into practice some specific functional values and not others. The dialogical information units (Phatic, Conative, Expressive, Allocutive) provide the regulation of the communication channel by making contact with the interlocutor while not contributing to the expression of a subjective evaluation of the speaker in his verbalization. In addition, the Appendix, that is, mere repetition of forms or textual integrations, doesn’t involve the expression of a modal judgment. Finally, this unambiguous distribution of modalization within the informational patterning of an utterance shows that lexical modality involves information units depending on their type and their function. Therefore, the information unit seems a pertinent unit for assigning a modal value to any modal entries within the utterance.

The informational structure and the scope of lexical modality in spoken Italian 215

7.

Frequency of alethic, epistemic and deontic modal values

The proportions of modal values (alethic, epistemic and deontic) were considered in relation to all the lexical occurrences and with regard to their characterization within the different information units (e.g. Topic, Comment, Parenthetical and Speaker Introducer). 7.1. Modal values and lexical occurrences On 6,459 total lexical and morphologic indexes of modality (3,202 in Informal, 3,257 in Formal C-ORAL-ROM Italian), alethic and deontic values are represented in similar proportions, respectively in 27.9% and 26.5% of occurrences, while epistemics are widely predominant (45.6%), with a substantial similarity between the two sociolinguistic partitions of the corpus (Informal and Formal). From this point of view, it seems that in spoken language there is a constant attitude of the speakers to express a certain type of modal meaning, independently from the sociolinguistic and diamesic variations.

Figure 4.

7.2. Modal values and information units On the contrary, a clear difference in the distribution of modal values is pointed out by their proportions among information unit types. Indeed, on 5,668 total modalized units (2,845 in Informal – 2,823 in Formal) each one of them has a special modal characterization (figure 5).

216 Ida Tucci

Figure 4.

More specifically, Topic information units mainly perform alethic functions (in roughly 48% of cases) or epistemic functions (52%).10 Parenthetical units are mostly epistemic (75.7%), but also alethic (18.4%) or deontic (5.8%). Only Comment and Speaker introducer units are quite “free” to perform all modal functions (even though not in comparable percentages: both concur to the utterance modalization respectively in 41% and 10% of the cases). More in detail, Comment units assume alethic values in 31.2% of cases, deontic values in 32.2% and epistemic values in 36.6%. The distribution of modal values among the information unit types allows us to formulate hypotheses on the nature of the relationship that, within the utterance, is established between the informational function of a specific unit and the modal values mainly represented in it. Firstly, it is not unexpected for us to find that Comment units have the highest possibility of variation in expressing all modal values. In fact, in our theoretical framework, the Comment represents the necessary and sufficient information unit for the pragmatic interpretation of an utterance, also when the latter is composed by more than one information unit. In our view, the free distribution of alethic, epistemic and deontic values in such unit type seems to be related to the variety of illocutionary typologies realized in Comment,11 and also to the kind of their lexical fillings, which is preferentially verbal (61.9%). For the other information units (in special way Topic and Parenthetical units) the relation is more complex, since the restrictions on the performing of modal values are related to their particular functions within the utterance. It is worth making some considerations on these restrictions.

The informational structure and the scope of lexical modality in spoken Italian 217

7.3. Remarks on modal values in Topic and Parenthetical units By way of example, I propose some instances of modalization in Topic and Parenthetical information units, in order to highlight how their function is related to the type of modal values. As we have seen before, in Topic units it is alethic and epistemic values which are mainly achieved. From our point of view, such restriction is due to the informational function that the Topic carries out in the utterance: always preceding the Comment, Topic units correspond to the “application field of the illocutionary force of the Comment”, and constitute the “semantic premises” which enable the utterance to move away from the linguistic context. Alethic modal judgments (12)– (13) and epistemic ones (14)– (15) can precede the fact or the event to which they are applied. (12) *VER: in realtà / TOP si darà ai genitori di Simone // COM [in reality / we will give it to Simon’ parents //]

[ifamdl14]

(13) *GPA: perché in effetti / TOP avere il mixer / TOP vuol dire inizia’ a registrare anche a noi // COM [ifamcv02] [because actually / to have the mixer / means we can start recording //] (14) *MIS: fosse così /TOP 12 si aprirebbe uno spiraglio//COM [imedrp04] [if that was the case / we could have a slight chance //] (15) *MAX: secondo me /TOP ne dimostrava di più //COM [in my opinion / she looked older //]

[ifamcv01]

With regard to Parenthetical modal values, we notice an absolute predominance of epistemic productions (75.7%). The reasons for such a restriction of modal productions can be found in the nature of the informational function of Parentheticals which, by definition, is “to add a subjective evaluation on the text of the utterance” (see Tucci 2004). (16) *BER: visto che / il corpo di pace /TOP che in Europa dovrebbe essere costituito /INX ancora non è stato costituito //COM [imedts03] [since / the peace-keeping force / that should have been constituted in Europe / has not been constituted yet //] Given that Parenthetical units always include a different point of view within the utterance, if the Comment unit itself manifests an explicit modal-

218 Ida Tucci ization, the Parenthetical can intensify it (17), waken it (18) or contradict it (19): (17) *ROD: ci sarà qualche condizione / COM immagino // INX [there will be some conditions / I suppose //]

[ifamcv07]

(18) *CLA: poteva esse’ interpretato così / COM probabilmente // INX [ifammn03] [it could have been interpreted in that way / probably //] (19) *MIC: al limite si può / COM INT anche se non ci credo / INX si potrebbe riscoprire Troisi / &he / attore // COM [ifamdl01] [at least we can / even if I don’t believe it / we could rediscover an interest in Troisi / as an actor //] Beside the function of expressing the genuine point of view of the speaker within an utterance, Parenthetical information units also offer the possibility to conjugate the modal function to a metalinguistic utility (come posso dire / how can I say), which works directly on the lexical choice or on the enunciation context (20). (20) *VER: è / COM INT come posso dire / INX l'humus della democrazia // COM [inatps03] [it is / how can I say / the humus of democracy //] More specifically, Parenthetical units show an “expressive effectiveness” within a rhetoric or argumentative speech, when the speaker employs a term, such as a useful device to keep up the interlocutor’s interest (Tucci 2002). In conclusion, the expression of a specific modal value seems to be sensitive to, if not strictly determined by, the function which is developed by the information unit type hosting the modal index. In this regard too, the information unit seems to be the proper domain of modal semantic influence.

8. Compositionality vs. non compositionality of modal values The scope of lexical modalization in spoken language is a relevant theoretical point. During our corpus-based research we have been wondering whether or not a modal value affects the whole text of an utterance. As we already anticipated, spoken language data shows that the scope of lexical modality is strongly conditioned by the informational patterning of the utterance.

The informational structure and the scope of lexical modality in spoken Italian 219

8.1. Some data on the composition of different modal values in the utterance Lexical modalization can occur in an utterance in various manners in accordance with the patterning strategy adopted and also in connection with the number of modal indexes. In regard to the informational composition of modal utterances (5,498 total), we found 25% simple utterances vs. 75% compound. 1) if a single modal index occurs in the utterance this can be placed: (1a) in a simple utterance; or (1b) in a complex utterance13; 2) if more than one modal index occurs in an utterance, they can be placed in: (2a) a simple utterance; (2b) in one information unit of a complex utterance; (2c) in more than one information unit of the utterance. In the first case (1a) the modalization concerns both the utterance and the Comment unit, and no scope problems arise (21). (21) *SIM: è da pulire quel camino // COM [that chimney has to be cleaned //]

[ifamcv07]

In this case we considered that the modal status (deontic) of the information unit characterizes the modal status of the entire utterance (simple utterance). However, this is not obvious when one modal index occurs in a complex utterance (1b). If an utterance is patterned into information units, a scope problem arises with regard to the modal interpretation: (15) *MAX: secondo me /TOP ne dimostrava di più //COM [in my opinion / she looked older //] (5)

*GUI: è andato via / fortunatamente // [he went away / fortunately //]

[ifamcv01] [ifammn22]

In (15) the Topic unit bears an epistemic index and also in (5) the Parenthetical unit bears an epistemic index. Indeed we do not have obvious reasons to claim that the Comment unit in (15) and (5) is not concerned by the epistemic modality too, although there is reasonable evidence of this assumption.14 If a simple utterance bears two modal indexes (2a), the scope problem does not arise, and it is also evident that, crucially, only one modal value emerges.

220 Ida Tucci (9)

*MAX: necessariamente deve essere un nome solo //COM [it must be necessarily one name //]

[ipubdl 04]

In other words, considering the values conveyed by two or more modal indexes in the same simple utterance, or in the same informational unit, we adopt a compositional criterion 15 (a semantic and syntactic dominance of one index within the same information unit). Under these conditions, in (9) only one modalization appears around a deontic value which is dominant in the above instance. In this case both indexes record the same modal value (deontic)16. Modal dominance enters in force and determines the interaction between two modal indexes even when they are not harmonious and/or placed within one information unit of a complex utterance (2b) (22): (22) *SCA: oggi ho proposto al consiglio dei ministri / di avviare l’organizzazione di un piano per il miglioramento degli allevamenti / COM perché dobbiamo sicuramente fare qualcosa //INX [imedts06] [today I proposed to the government’s council / to initiate an improvement for farm planning / because surely we must do something about it] In a complex utterance such as (22), for example, two modal entries, not similar, are embedded in a Parenthetical unit that assumes a resulting (dominant) deontic value. (23) *CAP: visto che stasera dobbiamo parlare in fiorentino il più possibile /TOP lo dirò //COM [inatpd02] [since we must speak in Florentine as much as possible tonight / I’m going to say it] In a complex utterance such as (23) we come across two modal entries; this time the first dobbiamo assumes an alethic value such as a representation of a “duty” given by the context and the second il più possibile stresses it through an intensification. So the Topic information unit assumes a unique alethic modal value. (24) *VAL: l’attuare questa strategia /TOP credo che possa solo far bene / al migliore sviluppo dell' azienda //COM [imedin02] [to put this strategy into action / can only be very useful / for the growth of the business]

The informational structure and the scope of lexical modality in spoken Italian 221

In a complex utterance such as (24) we find two modal entries, the first epistemic (credo) and the second alethic (possa), within a Comment unit. The locutive filling of the unit shows a syntactic and semantic dominance of the epistemic judgement (credo che possa). As a compositional result, the information unit assumes a unique epistemic modal value. The scope problem emerges when the occurrence of two modal indexes regards two distinct information units (17, 11): (18) *CLA: poteva esse’ interpretato così / COM probabilmente // INX [ifammn03] [it could have been interpreted in that way / probably //] (11) *IDA: in realtà / TOP Basilicata / TOP dovrebbe significare “la terra dei boschi” // COM [ifamdl18] [as a matter of fact / Basilicata / should mean land of woods //] In order to assign a proper interpretation to the Modus in its Dictum, in this case we are forced to consider each modal value by itself, and it doesn’t seem possible to adopt a compositional criterion of the two different modal values. In our framework, no semantic or syntactic compositional relationship is necessarily established between information units in a complex utterance (Cresti 2000: 176; Scarano 2003; Scarano, this volume). As a matter of fact, in (18) the epistemic modalization occurring in the Parenthetical does not affect the alethic modalization of the Comment unit; the utterance contains both the expression of an alethic point of view having the Dictum of the Comment as its scope, and an epistemic point of view having the Dictum of the Parenthetical as its scope. The same non-compositional interpretation is required in (11), which contains both an alethic point of view in the Topic unit and an epistemic one in the Comment unit. Therefore, when we find more modalized information units within the same utterance, the compositionality of modal indexes in a single modal value is not allowed. As we claimed, in these cases our choice is to describe the modalized utterance as a pattern of coexistent modal values. That is the case of all complex utterances in the corpus expressing modalization in more than one information unit (see examples 10, 11, 14, 17–19, 23, 24). In other words, the impossibility of a compositional solution of modal values in different modalized information units of a patterned utterance, considered against the compositionality of modal indexes within a same information unit, is in our opinion the most relevant proof that the locutive entity (Dictum) for a speech modal operation (Modus) is the information unit.

222 Ida Tucci Summarizing, there is a whole set of positive evidences supporting, in an harmonious way, that the scope of a modal value in speech can be considered the information unit: quantitative data of distribution (see § 1.2); the fact that only specific types of information units can bear modal lexical indexes (see § 6.1), the preference for specific modal values shown by each type of information unit,(see § 7.2) and finally the impossibility of a compositional solution of modal values in different modalized information units. All these aspects can hardly been explained if a semantic entity such as the proposition is taken as the reference unit for modality in speech.

Notes 1. 2.

3.

4. 5.

6.

7.

See Lyons (1977) and Palmer (1986) for an overview of the different proposals. C-ORAL-ROM (Cresti & Moneglia 2005) provides a comparable set of corpora of spontaneous spoken language of the main romance languages (French, Italian, Portuguese and Spanish). It consists of 772 spoken texts and 123:27:35 hours of speech (roughly 300,000 words for each romance language). CORAL-ROM Italian records 285,010 words in 38,741 utterances and 16h 32’ recording sessions. The corpus annotation has been accomplished within my PhD dissertation (Tucci 2007). In accordance with the C-ORAL-ROM dialogue annotation system, speakers are indicated by three capital letters preceded by an asterisk (*ABC:). Double (//) and simple (/) slashes respectively refer to terminal or non-terminal prosodic breaks. See Cresti & Moneglia (2005: ch. VI). must has an alethic value because it is the “natural” “peculiar” function “to illustrate facts” of a Public Prosecutor in a legal process. Moral and ethical judgments are not only introduced by the speaker such as “personal reports of the psychological states, attitudes, of the speaker” (epistemic reading), but they also contain an intention to advise the listener to assume an analogue “attitude” to what is expressed by the judgment, i.e. the persuasion that such attitude is the “best” one, the most “appropriate”. In this sense we considered such utterances as characterized by a deontic-axiological connotation (see Lay 2003). The Informational patterning theory has been proposed in many publications. See Cresti (1995, 2000) and Cresti & Moneglia (2006) for the general framework. See also the following specific studies concerning the various information units: Signorini & Firenzuoli (2003), for the Topic unit, Frosali (2008), for Dialogical units, Tucci (2003) for Parentheticals, Firenzuoli (2003) for the typologies of the Comment units. For more details see Scarano (this volume).

The informational structure and the scope of lexical modality in spoken Italian 223 8. 9.

10.

11. 12.

13.

14.

15.

16.

1,661 utterances and 10.644 words of C-ORAL-ROM Telephone Man-Machine interactions were excluded from our analysis. We took into account a set of 65 modal adverbs recording 1,814 total occurrences. Lemmas have been chosen in accordance with the classical grammatical view (such as probabilmente, giustamente, necessariamente, realmente, etc.) but this set was partially integrated with other lemmas emerged from the corpus-driven research (ignorantemente, tendenzialmente, senz’altro, praticamente, per forza, etc.). All the occurrences have been systematically analyzed. For the complete list of modal adverbs employed, see Tucci (2007). Rarely we find deontic lemmas in Topic (5.2%), but they are “fixed” locutions like when I have to / if I have to / having to, etc. They contextually shift to an alethic value, such as “descriptions” of the state of affairs because of the informational function of Topic information unit. So the percentage of alethic values in Topic would reach roughly 48%. The discussion of this point is presented it in Tucci & Moneglia (forthcoming). We follow Papafragou (2001) in considering these epistemic uses a particular case of metarepresentation where the locution p is not used “as a truth conditional representation of a state of affairs in the external world but as a representation of an abstract hypothesis, which is considered to be […] entailed by the speaker's set of beliefs” (Papafragou 2001: 70). Compound utterances can be compound by a Comment and one or more non-modalized informational units (dialogical unit) (20.6 %), otherwise a they can be patterned by a Comment and an informational unit (textual units) that is subject to lexical modalization (54.4 %). In this way we can recalculate the percentage of the utterances that are modalized only within Comment units adding the 25 % of simple utterances to 20.6 % of compound by a Comment and a dialogical units (45.6 %). Actually, even if the Comment units in (15) and (5) are lacking traditional modal indexes, we could consider other features, such as the type of the verbal entry and the illocutionary force of the unit (assertive/evaluative) that should make us assign to both Comments an alethic modal value on which the epistemic Topic and epistemic Parenthetical are applied. Our theoretical framework claims that there is no kind of necessary syntactic relation in terms of “regency” among linguistic expressions of different information units, but there is a clear syntactic domain inside every information unit (see Cresti 2000: 176). We assume that also from a semantic point of view, there is not a kind of “regency” among the meaning of different information units but a semantic pattern of contents, which allows every information unit to bear a proper modal value. On the contrary, within the same information unit, all the expressions must share the same modality and in the case of many modal indexes these must be solved in a compositional process. As we claimed (§1), only 7.6% of modalized information units bear more than one modal index. 4.1% of plural modal entries in the same information unit show a harmonious modal value, 3.5% an inharmonious value.

224 Ida Tucci References Austin, John L. 1962 How to do things with words. Oxford: Oxford University Press. Bally, Charles 1932 Linguistique générale et linguistique française. Bern: Francke. 1942 Syntaxe de la modalité explicite. Cahiers Ferdinand de Saussure 2: 3–13. Bybee, Joan and Suzanne Fleischmann (eds.) 1995 Modality and grammar in discourse. Amsterdam /Philadelphia: John Benjamins. Chierchia, Gennaro 1997 La semantica. Bologna: Il Mulino. Conte, Amedeo G. 1977 Aspetti della semantica del linguaggio deontico. In Logica deontica e semantica, G. Di Bernardo (ed.), 147–165. Bologna: Il Mulino. Cresti, Emanuela 1987 L’articolazione dell’informazione nell’italiano parlato. In Gli italiani parlati. Sondaggi sopra la lingua di oggi. Incontri del Centro di studi di grammatica italiana, Florence March 29 – May 31, 1985, Accademia della Crusca (ed.), 27–90. Florence: Accademia della Crusca. 2000 Corpus di italiano parlato. Florence: Accademia della Crusca. 2002 Illocuzione e modalità. In Scritti in onore di Bice Mortara-Garavelli, P. Beccaria and C. Marello (eds.), 133–145. Torino: Dell’Orso. 2005 Enunciato e frase: teoria e verifiche empiriche. In Italia Linguistica: discorsi di scritto e di parlato. Scritti in onore di Giovanni Nencioni, M. Biffi, O. Calabrese and L. Salibra (eds.), 249–260. Siena: Prolagon. Cresti, Emanuela and Massimo Moneglia 2006 C-ORAL-ROM. Prosodic Boundaries for Spontaneous Speech Analysis. In Spoken Language Corpus and Linguistic Informatics, Y. Kawaguchi et al. (eds.), 89–113. Amsterdam: John Benjamins. Cresti, Emanuela and Massimo Moneglia (eds.) 2005 C-ORAL-ROM. Integrated reference corpora for spoken romance languages. Amsterdam / Philadelphia: John Benjamins. Firenzuoli, Valentina 2003 Repertorio delle forme intonative di valore illocutivo dell’italiano. Analisi sperimentale di un corpus di parlato spontaneo (corpus LABLITA). Ph.D. diss., Dipartimento di Linguistica, University of Florence. Galvan, Sergio 1991 Logiche Intensionali. Sistemi proposizionali di logica modale, deontica, epistemica. Milan: Franco Angeli. Graffi, Giorgio 1994 Sintassi. Bologna: Il Mulino.

The informational structure and the scope of lexical modality in spoken Italian 225 Hare, Richard M. 1961 Il linguaggio della morale. Rome: Ubaldini editore. Hoye, Leo 1997 Adverbs and Modality in English. London: Longman. Huddleston Rodney and Geoffrey K. Pullum 2002 The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Kiefer, Ferenc 1994 Modality. In The Encyclopedia of language and linguistics, R. E. Asher (ed.), 2515–2520. Oxford: Pergamon Press. Lay, G. 2003 Dal credere al conoscere la conseguenza non è buona. Tecniche conversazionali 29: 55–64. Lyons, John 1977 Semantics. Vol. 2. Cambridge: Cambridge University Press. Nuyts, Jan 2001 Epistemic modality, language and conceptualization: a cognitive pragmatic perspective. Amsterdam: John Benjamins. Nuyts, Jan and Vonk Wietske 1999 Epistemic modality and focus in Dutch. Linguistics 37: 699–737. Palmer, Frank R. 1986 Modality and the English Modals, 2nd ed. London: Longman. 2001 Mood and Modality. Cambridge: Cambridge University Press. [1st ed. 1986.] Papafragou, Anna 2001 Modality: Issues in the Semantics-Pragmatics interface. Amsterdam: Elsevier. Perkins, Michael R. 1983 Modal expressions in English. London: Frances Pinter. Pietrandrea, Paola 2005 Epistemic modality: functional properties and the italian system. Amsterdam: John Benjamins. Pottier, Bernard 2000 Représentations mentales et catégorisations linguistiques. Louvain/Paris: Peeters. Scarano, Antonietta this vol. The prosodic annotation of C-ORAL-ROM and the structure of information in spoken language. Schneider, Stefan 1999 Il congiuntivo tra modalità e subordinazione. Rome: Carocci. Sewell, Julie 2004 Diagnostic assessment with the Skills for Life strategy. Presented at the 30th International Association for Educational Assessment Conference, Philadelphia, Pennsylvania, USA, 13–18 June.

226 Ida Tucci Signorini, Sabrina and Valentina Firenzuoli 2003 L’unità informativa di Topic: correlati intonativi. In Atti delle XIII Giornate di Studio del Gruppo di Fonetica Sperimentale (A.I.A.), G. Marotta and N. Nocchi (eds.), 177–184. Pisa: ETS. Simone, Raffaele and René Amacker 1977 Verbi modali in Italiano. Per una teoria generale della modalità nelle lingue naturali. Italian Linguistics 3: 7–12. Tucci, Ida 2002 Caratteristiche sintattiche e frequenza dell’inciso in un corpus di italiano parlato. Degree diss., University of Florence. 2004 L’inciso: caratteristiche morfosintattiche e intonative in un corpus di riferimento. In Atti del Convegno “Il parlato italiano”, F. Albano Leoni, F. Cutugno, M. Pettorino, R. Savy (eds.), 1–14. Napoli: D’Auria. 2005 L’espressione della modalità nel parlato: i verbi modali nei corpora italiano e spagnolo C-ORAL-ROM”. In Atti del VIII convegno internazionale della SILFI “Lingua, cultura e intercultura”, I. Korzen (ed.), 295–308. Copenhagen: Samfundslitteratur Press. 2007 L’espressione lessicale della modalità nel parlato spontaneo. Dati dal corpus C-ORAL-ROM Italiano. PhD diss., University of Florence, Dipartimento di Linguistica. 2008 La modalizzazione nel parlato spontaneo. Dati dal corpus C-ORALROM Italiano. In Atti del IX Congresso SILFI “Prospettive nello studio del lessico italiano”, Florence, June 14–17, 2006, E. Cresti (ed.), 377–386. Florence: Florence University Press. Tucci, Ida and Massimo Moneglia forthc. Strategies of modalization in spoken language. A corpus-based analysis. In C. D. Pusch (ed.), Proceedings of the 3rd Workshop “Corpora and Pragmatics”, Freiburg 14–17 September 2006. Van der Auwera, Johan and Vladimir Plungian 1998 Modality’s semantic map. Linguistic Typology 2: 79–124. Venier, Federica 1991 La modalizzazione assertiva, avverbi modali e verbi parentetici. Milan: Franco Angeli. Von Wright, Georg H. 1951 Essays on Modal Logic. Amsterdam: North-Holland.

Section 3 Diachronic aspects in Italian

On two functions of verb-subject order in Old Italian Ludwig Fesenmeier

1. Introduction In Old Italian texts,1 it is easy to find sentences that are quite similar with regard to the states of affairs described, but differ as far as the position of the subject with respect to the finite verb is concerned. As becomes apparent from the pairs of examples given in (1a)/ (1b) and (2a) / (2b), the subject (S) may precede or follow the finite verb (V):2 (1) a. Tre sono le parti de la penitenzia, cioè contrizione, confessione e satisfazione. La satisfazione, cioè questa terza parte, hae più parti, … (to be continued in (1b)) ‘Repentance consists of three parts, namely contrition, confession and penance. Penance, that is this third part, consists of several parts, …’ b. … sì come ti dicessi: il corpo hae molte membra, l’uno de’ quali è il braccio, ma il braccio ha più parti, cioè la mano e le dita, e le dita hanno più altre parti; così de la penitenzia e de la sadisfazione. Hae dunque questa terza parte, cioè la satisfazione, altre tre parti, cioè digiuno, limosina e orazione; […]. (Giordano da Pisa, Quaresimale Fiorentino [1305–1306]) ‘… as if I said: a body has many parts, one of which is the arm, but the arm has several parts, that is the hand and the fingers, and the fingers still have several parts: the same holds for repentance and penance. This third part then, that is penance, has three further parts, that is fasting, giving alms and praying’ (2) a. Tutti i santi profeti ànno predette tutte queste cose: tutti gli elementi nella natività di Cristo mostrarono letizia, e nella sua passione tristizia, e tutti i corpi celestiali; il sole, la luna iscurò: […]. (Anonymous, La Legge di Maometto [Magl. Strozz. XXXV 169], 1355) ‘All the holy prophets have predicted all these things: all the forces of nature showed gladness at the nativity of Christ and sadness at his suffering, and so did all orbs: the sun, the moon eclipsed’

230 Ludwig Fesenmeier b. […] la divina potenzia degnò d’incarnare nella graziosa vergine Maria, ed esere Idio e uomo nato di vergine, e sofferire passione e morte, e ne la passione scurò tutto il sole nel mezzodì […]. (Giovanni Villani, Nuova Cronica [ed. Porta], 1348) ‘the divine power condescended to become flesh through the graceful Virgin Mary, and to be God and man born by a virgin, and to suffer passion and death, and during the passion, the sun eclipsed completely in the midday’ The common denominator of the foregoing examples consists in the fact that their subject referents are “definite” in the sense of Chafe (1976: 39): It is […] of some interest in the communicative situation whether I think you already know and can identify the particular referent I have in mind. If I think you can, I will give this item the status of definite. The assumption in this case is not just ‘I assume you already know this referent’, but also ‘I assume you can pick out, from all the referents that might be categorized in this way, the one I have in mind’. [all emphasis ours] 3

In (1a) la satisfazione has just been introduced in the preceding sentence, questa terza parte in (1b) is an anaphoric expression pointing back to la satisfazione (which, in turn, is again repeated immediately after the subject); as far as the examples (2a) and (2b) are concerned, that there is a sun (and only one) can be reasonably assumed to be part of a medieval reader’s everyday knowledge.4 The aim of this paper is to show that in Old Italian, the contextual criterion of definiteness has to be clearly distinguished from the pragmatic function of ‘topic’, and that definite subjects, when following the finite verb, need not, necessarily, fulfil the topic function: it is argued, rather, that in Old Italian, VS ordering is multifunctional, the different functions being linked to quite distinct contexts and serving different purposes on the level of textual structure.5

2. Methodological background We will distinguish between three levels of description when analyzing the examples: the syntactic level, to which belong syntactic functions such as ‘subject’ and ‘predicate’ (finite verb), the semantic-propositional level, i.e. the propositional content (‘state of affairs’) described by the sentence, and

On two functions of verb-subject order in Old Italian

231

the pragmatic level, to which belong pragmatic functions such as ‘topic’ and ‘comment’.6 As far as the concept of ‘topic’ is concerned, following one of the standard interpretations, the basic criterion is aboutness, which is “the most pervasive, if not central, characteristic” (Schlobinski and Schütze-Coburn 1992: 99). In this perspective, the “topic is what the sentence is about. That part of the sentence that says something about the topic is the predication [= comment]” (Primus 1993: 880). However, as Reinhart (1981: 58) has pointed out, many linguists who define “topic” in such a way take the notion of ‘aboutness’ “as a primitive, without attempting to explain what it means for a sentence to be about (the referent of) one of its expressions”. Consider, e.g., the following dialogue, proposed in Vanelli (1986: 252, ex. 6, 7 and 5b): (3) Q1: Cosa è successo? ‘What has happened?’ Q2: Ma come mai Carlo non c’è? ‘Why is Carlo not here?’ A: Carlo è stato fermato dai carabinieri ‘Carlo has been arrested by the police’ Suppose that Carlo is known to the interlocutors in (3). If utterance A is an answer to question Q1, it delivers completely new information, and Vanelli (1986: 253) interprets Carlo here indeed as “il […] tema di una frase tutta nuova [the topic of an all-new utterance]”. Such an interpretation is perfectly in line with the author’s definition of ‘topic’ as “il punto di partenza per lo sviluppo successivo del discorso, ciò su cui verte la predicazione [the starting point for the further development of the discourse, the element the predication culminates in]” (1986: 270, note 2). However, A fits also perfectly well as an answer to Q2, in which Carlo is explicitly mentioned. Thus, the definition of ‘topic’ proposed by Vanelli is not able to catch the difference in Carlo’s pragmatic status, which depends on the context represented by the two questions, and her concept of ‘topic’ must be assigned, in the end, to the semantic–propositional level, and not to the pragmatic one. In line with Reinhart (1981: 58–60, 78–87), we shall therefore adopt the more precise criterion of ‘pragmatic aboutness’, which is based on what Strawson (1964: 115) has termed the Principle of Relevance: “We do not, except in social desperation, direct isolated and unconnected pieces of information at each other, but on the contrary intend in general to give or add information about what is a matter of standing or current interest or concern [emphasis ours]”.7 With this specification in mind, the difference stated

232 Ludwig Fesenmeier above regarding (3) can be accounted for as follows: it is only as an answer to Q2 that A is ‘pragmatically about’ Carlo. We would like to introduce other aspects which appear to be of some relevance for the discussion to follow. The first one concerns the fact already noticed in Morgan (1975: 434), that “it is not NP’s that are topics, but the things they refer to”.8 The second aspect regards another “platitude”, equally mentioned in Morgan (1975: 434): “it is not sentences that have topics, but speakers”. Especially in written discourse, in particular in its typical instances characterized by what Koch and Oesterreicher (1985) have termed “communicative distance” (kommunikative Distanz),9 one must take into consideration that, as Arcangeli (2004: 58) puts it, “l’ordine dei costituenti di un enunciato, nella circostanza, possa essere appunto regolato dalla sintassi dell’intorno testuale di riferimento, possa essere cioè il frutto del condizionamento esercitato dalle aspettative del textus [the constituent order in an utterance may possibly be ruled by the syntax of the relevant textual environment, that is, may be the result of the conditioning exerted by the expectations which, in turn, have arisen from the textus]”. Arcangeli illustrates his ideas with the help of two passages taken from Novellino tales:10 (4) a. Andar li ambasciadori ‘The ambassadors left’ b. Adomandò lo signore ‘The lord asked’ As far as these examples are concerned, the author states the following: Gli ambasciadori, nel primo esempio, e il ‘signore che porta […] corona di re’, nel secondo, sono infatti già noti al lettore quando si giunge all’altezza degli enunciati in grassetto [cf. (4a) and (4b)]: il narratore ha semplicemente deciso [emphasis ours] di riservare la posizione di incipit alle azioni espresse da quegli enunciati […] per dare loro, appunto, un qualche rilievo. (2004: 58–59) [The ambasciadori in the first example and the ‘lord who wore the crown of a king’ in the second are indeed already known to the reader when one arrives at the utterances in bold: the author has simply decided to reserve the sentence-initial position to the events denoted by these utterances in order to assign them some emphasis.]

What deserves particular attention here is Arcangeli’s suggestion that the factual VS ordering (Andar li ambasciadori, Adomandò lo signore) is the

On two functions of verb-subject order in Old Italian

233

result of a choice made by the author, who could just as well have chosen SV ordering (Li ambasciadori andar, Lo signore adomandò): what needs to be explained, thus, is the reason why he presumably had decided in favor of VS (and, thereby, against SV). Lastly, we would like to comment on the relation between the contextual feature of ‘definiteness’ and the pragmatic function of ‘topic’ of the referent of a given constituent. As have confirmed several studies, definiteness is a necessary, but by no means sufficient condition for a constituent to function as the topic of an utterance.11 Equating definiteness and topic function leads to a restricted perspective, as can be seen from the statement in Renzi (1988: 130, note 7) with regard to the following quotation taken from the Novellino tales:12 (5) Dopo ’l pranzo parlò Socrate [alli ambasciadori] ‘After dinner Socrates spoke to the ambassadors’ In italiano moderno una frase come quella del Novellino [cf. (5)] […] va interpretata con Dopo il pranzo come un elemento di cornice, fuori dalla frase (che è anche una delle possibili interpretazioni per l’italiano antico), mentre l’intero segmento parlò Socrate, o il solo Socrate sono Nuovi. In italiano antico Socrate può essere Dato, ed in effetti lo è, se consideriamo il contesto della novella [emphasis ours]. [In Modern Italian, a sentence such as the following taken from the Novellino must be interpreted with Dopo il pranzo as a framework element, outside the sentence (which is also one of its possible interpretations in Old Italian), while the whole segment parlò Socrate, or even the only Socrate, are New. In Old Italian, Socrate can be Given, and indeed it is, if we take into consideration the context of the novella.]

Lombardi Vallauri (2004: 295) describes Renzi’s account in the following way: Renzi segnala che […] in it. mod. il S sarebbe obbligatoriamente nuovo, mentre qui può benissimo essere, ed è, dato. In effetti, se si guarda il contesto, la frase è equivalente a un it. mod. dopo il pranzo Socrate parlò agli ambasciatori, dove il R[ema] è costituito dal verbo, e il S è tematico: […]. [Renzi points out that in Modern Italian, the subject would be obligatorily new, while here it can be, and is, in fact, given. If one considers the context, the sentence is equivalent to Modern Italian dopo il pranzo Socrate parlò agli ambasciatori, where the comment consists of the verb and the subject is the topic.]

234 Ludwig Fesenmeier Socrates has in fact been mentioned in the preceding text (not quoted by Renzi), but, as we shall argue below, it is far from clear that he can be also considered the topic of the utterance being examined here, its alleged “modern” equivalent being of no evidential value here. A further argument in favor of a differentiation between the categories ‘given/new’ and ‘topic/comment’ can be derived from examples such as the following ones, taken from Primus (1993: 881, ex. (1a) and (1b)): (6) a. What do you know about pet animals? DOGS BARK. HUNDE BELLEN. b. What’s the noise? DOGS are barking. HUNDE bellen. Primus (1993: 881) comments on the examples as follows: (1) [cf. (6)] presents the intonational differences between all-new sentences with a topic (cf. (1a) [cf. (6a)]) and all-new sentences without a topic (cf. (1b) [cf. (6b)]), in English and German [emphasis ours]. Capital letters mark a word carrying a primary accent. […] In all-new structures with a topic, both the topic and the predicate receive a primary accent. In all-new thetic utterances on the other hand, the non-topical nominative argument behaves accentually like a verb phrase internal argument in forming a close unit with the verb.

Against the background of the foregoing considerations, the following definitions proposed in Lambrecht (1994: 131) seem particularly suitable for the purpose of our analysis: topic: A referent is interpreted as the topic of a proposition if in a given situation the proposition is construed as being about this referent, i.e. as expressing information which is relevant to and which increases the addressee’s knowledge of this referent. topic expression: A constituent is a topic expression if the proposition expressed by the clause with which it is associated is pragmatically construed as being about the referent of this constituent. Following the definition of ‘topic’ proposed by Lambrecht, the comment turns out to be “information which is relevant to and which increases the

On two functions of verb-subject order in Old Italian

235

addressee’s knowledge” of the topic, i.e. of what is “matter of standing interest or concern” (Lambrecht 1994: 119). Therefore, it seems logical to consider ‘topic’ and ‘comment’ as being two correlative concepts, i.e. we assume that there are no so-called “comment-only” utterances.13 Utterances which contain a topic (and, as a consequence, a comment) will be called “categorical”, others will be termed “thetic”.14 Finally, it should be pointed out that following Stark (1997: 40–42), we consider those elements which set “a spatial, temporal, or individual framework within which the main predications holds” (Chafe 1976: 50), typically adverbs of time, manner, etc., not as topics, but precisely as mere “framework-providing elements”.15 In spite of all tentatives, it is difficult to find one reliable test which makes it possible to identify, in any given utterance, the topic element (if there is any, cf. examples (3) and (6b) above). Among the various strategies proposed,16 the most suitable ones still seem to be paraphrases such as ‘as far as X is concerned’, where the referent of “X” turns out to be the topic, and the procedure for which Hatcher (1956a: 239) has coined the felicitous expression “underlying question”: “In attempting to establish the point of view of a predication I proceed on the assumption that any statement may be considered as the answer to an underlying question, which means that I must attempt to construct a set of questions which might account, in their way, for all predications”.17 The author had chosen “the basic criterium of such questions […] in the distinction between the relatively known and relatively unknown elements of a predication” (1956a: 239), but she reveals herself well aware of the problems that can arise from such a criterion: do the baby began to cry, the baby was put to bed represent Z[ero is given]: ‘what happened, what was done?’, or does the first announce the baby’s reaction (S[ubject is given]), the second, the treatment the baby received (O[bject is given])? The context should contain the answer, which will hardly be so simple as to depend on whether ‘the baby’ had been recently mentioned. (1956a: 241)

As becomes apparent from these considerations, there is no clear-cut relation between a given utterance and the “appropriate” underlying question, when the former is considered in contextual isolation; furthermore, a referent’s feature ‘having been recently mentioned’ is of no direct relevance for choosing it as the topic; finally, it is the hearer/reader who appears to play a central role in determining what is the appropriate underlying question – not being excluded thereby that the speaker/writer may linguistically provoke certain expectations as a consequence of the foregoing context.

236 Ludwig Fesenmeier 3. Two pragmatic functions of Old Italian VS order As early as 1980, Blumenthal has discussed a large number of examples taken from Modern and Old Italian texts, trying to demonstrate daß […] in der Literatursprache bisher […] nicht diskutierte Inversionen mit textlinguistischer Funktion vorkommen, die sich […] insbesondere nicht aus der […] Theorie der Thema-Rhema-Gliederung des italienischen Satzes ableiten lassen. (1980: 123)18 [that in literary language, one finds cases of text-structurally motivated inversion which have not been discussed so far, and which cannot be explained by the theory of the topic-comment structure of the Italian sentence.]

Two of the examples quoted by the author are reproduced in (7) and (8), but are provided with a somewhat larger context:19 (7) E secondo il pensier fatto mandò a essecuzione: per ciò che, mandato avanti ogni uomo, esso con poca compagnia e di gentili uomini entrò in cammino; e, avvicinandosi alle terre del marchese, un dì davanti mandò a dire alla donna che la seguente mattina l’attendesse a desinare. La donna, savia e avveduta, lietamente rispose che questa l’era somma grazia sopra ogn’altra e che egli fosse il ben venuto. E appresso entrò in pensiero che questo volesse dire, che uno così fatto re, non essendovi il marito di lei, la venisse a visitare: né la ’ngannò in questo l’aviso, cioè che la fama della sua bellezza il vi traesse. Nondimeno, come valorosa donna dispostasi a onorarlo, fattisi chiamar di que’ buoni uomini che rimasi v’erano, a ogni cosa oportuna con lor consiglio fece ordine dare, ma il convito e le vivande ella sola volle ordinare. E fatte senza indugio quante galline nella contrada erano ragunare, di quelle sole varie vivande divisò a’ suoi cuochi per lo convito reale. Venne adunque il re il giorno detto e con gran festa e onore dalla donna fu ricevuto. Il quale, […], riguardandola, gli parve bella e valorosa e costumata, […]. (Giovanni Boccaccio, Decameron, 1370) ‘He proceeded to carry out his scheme: after sending his troops ahead, he set out with a small escort and some noblemen, and when he neared the lands of the Marquis, he sent one of them ahead to the lady to tell her she should expect him to dine with her the next day. The lady, who was prudent and wise, cheerfully replied that she considered this a great honor and that he would be most welcome. But then she began to wonder what all this meant – the fact that a king

On two functions of verb-subject order in Old Italian

237

would come to visit her when her husband was away; nor was she deceived when she concluded that he was drawn there by the renown of her beauty. Nevertheless, since she was a worthy lady she prepared to honor him, and she gathered together all the eminent gentlemen still remaining at her court in order to make all the appropriate arrangements with their advice. She alone, however, would see to the banquet and the food. Without delay she collected as many chickens as there were in the countryside and ordered her cooks to use only these chickens in the various dishes for the royal banquet. The King then arrived on the appointed day and was received by the lady with great festivity and honor. When he saw her, she seemed beautiful, virtuos, and courtly’ (Musa and Bondanella 1983: 43) (8) E quantunque pallida e assai male in ordine della persona per la fatica del mare allora fosse la donna, pur pareano le sue fattezze bellissime a Pericone: per la qual cosa subitamente seco diliberò, se ella marito non avesse, di volerla per moglie, e se per moglie aver non la potesse, di volere avere la sua amistà. Era Pericone uomo di fiera vista e robusto molto; e avendo per alcun dì la donna ottimamente fatta servire e per questo essendo ella riconfortata tutta, veggendola esso oltre a ogni estimazione bellissima dolente senza modo che lei intender non poteva né ella lui e così non poter saper chi si fosse, acceso nondimeno della sua bellezza smisuratamente, con atti piacevoli e amorosi s’ingegnò d’inducerla a fare senza contenzione i suoi piaceri. (Boccaccio, Decameron) ‘Although the lady was pale and dishevel[l]ed as a result of her harrowing experience at sea, she nevertheless seemed most beautiful to Pericone; and because of this he immediately decided to take her for his wife, if she had no husband, or to have her as his mistress, if he could not have her as his wife. Pericone was a very robust, bold-looking man. He saw to it that the lady was served in the best of fashions, and after several days, when she was completely recovered, he saw that she was even more beautiful than he had imagined, and he was most unhappy that they could not understand each other’s language, for he was unable to learn who she was. But he remained moved beyond measure by her beauty, and with gracious and amorous deeds he kept trying to induce her to fulfil his desires without resistance.’ (Musa and Bondanella 1983: 111–112) According to the author, VS orders in examples such as (7) mark “entscheidende Phasen des Handlungsverlaufs: Einsatz, Wendepunkte oder Schluß

238 Ludwig Fesenmeier [crucial moments in the plot: starting point, turning points, or conclusion]” (1980: 125), while VS orders such as the one documented in (8) can be found “an den Angelpunkten zwischen Erzählung und Beschreibung [at the transition from narration to description]” (1980: 126).20 Unfortunately, Blumenthal speaks of postverbal “topics-subjects” (thematisches Subjekt) in both cases, thus robbing himself of the possibility of finer analysis on the pragmatic level. In a more recent article, Sornicola, too, has stated that “[t]he VSO type often appears in Italian prose of the XIIIth and XIVth centuries […], with a stylistic value of ‘turning point in the narration’ [emphasis ours]” (2000: 108), citing the examples repeated here in (9)–(11), equally provided with a slightly larger context:21 (9)

Uno signore di Grecia, lo quale possedea grandissimo reame et avea nome Aulix, avea uno suo giovane figliuolo, il quale facea nodrire et insegnarli le sette liberali arti, e faceali insegnare vita morale, cioè di be’ costumi. Un giorno tolse questo re molto oro e diello a questo suo figliuolo e disse: […]. (Anonymous, Il Novellino, 1300) ‘A Greek gentleman, who possessed a very great realm, and was called Aulix, had a young son whom he had nurtured and instructed in the seven liberal arts, and taught to lead a moral life, that is, to be well behaved. One day this king withdrew a great deal of gold and gave it to his son, and said’ (Consoli [ed.] 1997: 29)

(10) Il che come Giosefo ebbe udito, così si ricordò delle parole di Salamone e disse verso Melisso: “[…]” Quindi, dopo alquanti dì divenuti a Antiocia, ritenne Giosefo Melisso seco a riposarsi alcun dì; […]. (Boccaccio, Decameron) ‘When Giosefo heard this, he quickly remembered Solomon’s words and said to Melisso: “[…]” Then, after a few days, they reached Antioch, and Giosefo invited Melisso to stay with him for several days to rest.’ (Musa and Bondanella 1983: 594) (11) […] quasi pentuta del non avere alle lusinghe di Pericone assentito, senza attendere d’essere a così dolci notti invitata, spesse volte se stessa invitava non con le parole, ché non si sapea fare intendere, ma co’ fatti. A questo gran piacere di Pericone e di lei, non essendo la fortuna contenta d’averla di moglie d’un re fatta divenire amica d’un castel-

On two functions of verb-subject order in Old Italian

239

lano, le si parò davanti più crudele amistà. Aveva Pericone un fratello d’età di venticinque anni, bello e fresco come una rosa, il cui nome era Marato; il quale, avendo costei veduta e essendogli sommamente piaciuta, […] cadde in un crudel pensiero: e al pensiero seguì senza indugio lo scellerato effetto. (Boccaccio, Decamerone) ‘she repented of having rejected Pericone’s previous advances; and not waiting a second time to be beckoned to such sweet nights again, she often invited herself – not with words, since she did not know how to make herself understood, but with actions. While she and Pericone enjoyed each other, Fortune, not content to have made the wife of a king the mistress of a lord, prepared an even cruel[l]er love for the lady. Pericone had a brother named Marato who was twenty-five years old, handsome, and as fresh as a rose; when he saw the lady, he was immensely attracted to her, […] and he devised a cruel plan whose evil effects followed quickly upon its inception.’ (Musa and Bondanella 1983: 113) According to the author, “[t]hese structures […] have the usual function of turning point in the narration, as shown by the fact that they have the pragmatic property of being ‘all-in-FOCUS’ [emphasis ours] (note that both S and O are [+GIVEN] in the Italian […] examples quoted)” (2000: 108).22 It is obvious that the examples (9) and (10) parallel that in (7), while (11) is analogous to (8). Thus, in a sense, (9)–(11) can well be subsumed under a category ‘turning point in the narration’, but such a description certainly blurs the differentiation between ‘crucial moment of the plot’ and ‘transition from narration to description’, already drawn, as we have seen, in Blumenthal (1980). However, what appears particularly noteworthy in Sornicola’s analysis is her clear distinction between contextual features (“[+GIVEN]”) and the pragmatic level (“‘all-in-FOCUS’”, i.e. absence of any topic-comment structure), even if the latter analysis is not quite convincing in the case of (11). In the Grammatica dell’Italiano Antico, more precisely in the chapters dedicated to constituent order,23 a situation analogous to that found in Sornicola (2000) can be discovered. Benincà (n. d.) rightly points out that in the case of a lexical subject, there have to be good reasons for it to appear after the verb, and then goes on to list inside the same set of examples those repeated here in (12) and (13):

240 Ludwig Fesenmeier (12) Dim[m]i, verace maestra, in che modo Avarizia fa le sue operazioni per le dette vie? – Ed ella disse: – Fa Avarizia l’operazioni sue per Simonia […]. (Bono Giamboni, Il Trattato di Virtù e di Vizi, 1292; cf. Benincà [n. d.], ex. [32d]) ‘Tell me, veritable mistress, how does Covetousness precisely carry out her acts in the aforementioned ways? – And she said: Covetousness carries out her acts by Simony’ (13) Li ambasciadori fecero la dimanda loro […]. Andar li ambasciadori… (Novellino; cf. Benincà [n. d.], ex. [32e]) ‘The ambassadors asked their question […]. The ambassadors left…’ According to the author, the “good reason” for this equal treatment consists in the fact that in these cases the postverbal subjects “riprendono […] un tema già introdotto, sono cioè una sorta di ripresa del tema [resume a topic already introduced, i.e. they are a kind of resumption of the topic]” (n. d.: § 3.1). However, besides the fact that the distances between the subjects in (12) and (13) amount to 11 and 38 graphical words respectively, neither is givenness a sufficient condition for an element to assume the topic function, nor do given subjects always appear in postverbal position, as we shall see below.24 Finally, one can easily observe that (12) and (13) differ manifestly from each other as far as their contents (exchange of information vs. story) and their textual structures (dialogue vs. narration) are concerned, but such differences are equally disregarded by the author. In order to account for the verb’s utterance-initial position in (13),25 Benincà further argues that there is a “elemento tematico sottinteso [che] potrebbe essere un indicatore di causa (perciò) o di tempo (dopo di ciò […]) [implicit topic element (which) could be an indicator of cause (‘therefore’) or time (‘after that’)]” which fills the structural position left empty so far.26 However, as we shall see in section 3.2, the presumable appearance of adverbs such as dunque ‘then’ and dopo di ciò ‘after that’ is precisely the modern “reflex” of what in older language stages could be expressed by mere topological means. In the end, the three approaches presented so far testify to a rather intricate situation as far as Old Italian VS ordering with definite subject referents is concerned. Thus, in what follows, we shall try to clear away at least some of the fog.

On two functions of verb-subject order in Old Italian

241

3.1. Categorical VS order In spite of the controversial analysis of the type of VS ordering exemplified in (8) and (11), in these cases, the referent of the subject is most likely to be considered (a part of) the topic. In (8), the sentence Era Pericone uomo di fiera vista e robusto molto ‘Pericone was a very robust, bold-looking man’ can be paraphrased by ‘as far as Pericone’s appearance is concerned, he was …’. In the case of the passage Aveva Pericone un fratello d’età di venticinque anni ‘Pericone had a brother who was twenty-five years old’ of the example (11), on the other hand, a paraphrase with “as far as” turns out less natural.27 Examining more closely the context of the example, one discovers that, on the content level, the utterance under scrutiny follows a kind of “vaticinal parenthesis”, but at this point of the narration, there is no reason not to assume that both Pericone and the lady are still a matter of current concern. Therefore, it seems completely in line with the content structure of the passage to paraphrase Aveva Pericone … by ‘as far as X is concerned’, adding, however, the causal conjunction ‘for’: ‘For, as far as Pericone is concerned, he had a brother …’. Nevertheless, one also finds instances of VS ordering where it is less difficult to prove the topic status of the subject referent;28 consider the following examples: (14) a. La battaglia sì s’inkomincioe forte e dura intra l’una parte e·ll’altra, e dura la battaglia per grande ora del die. (Anonymous, Il Tristano Riccardiano, 1300) ‘The battle started cruel and violent between the two parties, and the battle lasts many hours of the day.’ b. Dela seta que i nostri di Lonbardia ci mandaro nela fiera di Lagnino, a Provino sì no n’avemo venduta neiente, ed è ogi in grande viltà la seta sì chome v’abo divisato per altra letera. (Letter of Andrea de’Tolomei, 1269) ‘As for the silk that our people in Lombardy sent us to the fair at Lagnino, in Provino we have not sold anything, and now the silk is in a bad state, as I have told you in another letter.’ As can be seen from the structurally analogous examples given in (15), in Old Italian also an anaphoric zero can be used “in contexts of maximal referential and thematic continuity” (Givón 22001: 418):

242 Ludwig Fesenmeier (15) a. […] e ine fu ismisurata e crudele battaglia, e durò più dì. (Anonymous, Chiose Selmiane alla Commedia di Dante, 1337) ‘and therefore rose an excessive and cruel battle, and it lasted several days.’ b. E tornando in Gostantinopoli, lo imperadore Michele gli volle donare molti grandissimi tesori, nulla volle prendere, se non il legno de la santa croce e ’l chiovo di Cristo, lo quale in Francia ne recò, ed è oggi in Parigi. (Villani, Nuova Cronica [ed. Porta]) ‘And when he returned to Constantinople, albeit the Emperor Michael desired to give him many very great treasures, yet would he take nothing, save the wood of the holy cross and the nail of Christ, which he took in France, and it is in Paris to this day.’ (translation adapted from Wicksteed [ed.] 21906: II, 12) Against the background of (15a) and (15b) one feels thus entitled to assume that the explicit repetition of the subject in (14a) and (14b) would not have been necessary; consider the slightly modified versions of (14) given in (16): (16) a. La battaglia sì s’inkomincioe forte e dura intra l’una parte e·ll’altra, e dura ___ per grande ora del die. ‘The battle started cruel and violent between the two parties, and lasts many hours of the day.’ b. Dela seta que i nostri di Lonbardia ci mandaro nela fiera di Lagnino, a Provino sì no n’avemo venduta neiente, ed è ogi in grande viltà ___ sì chome v’abo divisato per altra letera. ‘As for the silk that our people in Lombardy sent us to the fair at Lagnino, in Provino we have not sold anything, and now it is in a bad state, as I have told you in another letter.’ It appears to be beyond doubt, thus, that in the face of the situation illustrated in (15) and (16), la battaglia ‘the battle’ in (14a) and la seta ‘the silk’ in (14b) can be considered to form (part of) the topics of the respective utterances.29 Following the ideas put forward in Blumenthal (1980),30 we have elsewhere offered an explanation of the VS ordering in cases such as (1b), (8), (11), (12), and (14),31 which can be summed up as follows: the function of categorical VS utterances with the subject being the topic concerns exclusively the level of text organisation. Such utterances can be used, among others,

On two functions of verb-subject order in Old Italian

243

– to grant the cohesion/coherence32 of a text (cf. (1b) where the speaker returns to his main subject after a short digression, or (12) where the repetition of Avarizia simply marks topic continuity); – to signalize the passage from, among others, the level of narration to that of description (cf. (8), (11), and (14)), or from narration to a comment by the author.33 While the further details of this account are of no particular relevance here, we would still like to stress the following: if in Old Italian the topic expression is syntactically coded as the subject of the sentence, its position before or after the finite verb does by no means affect the pragmatic value of its referent. What has not been treated yet is the question why the subject-topic sometimes appears after the finite verb, the basic constituent order of Old Italian being SVO.34 In primarily syntactically oriented studies, it has been claimed that certain cases of VS, at least, are due to the verb-second constraint which is often supposed to hold in Old Romance: […] la posizione fissa del verbo come secondo costituente della frase fa sì che un solo costituente possa trovarsi in posizione preverbale: se questo costituente è diverso dal soggetto, questo deve trovarsi necessariamente in posizione postverbale […]. [the fixed position of the verb as the second constituent of the sentence leads to only one constituent appearing in preverbal position: if this constituent is not the subject, the subject necessarily appears after the verb.] (Vanelli 1999: 234)35

Vanelli adduces the example already proposed in (5), repeated for convenience in (17): (17) Dopo ’l pranzo parlò Socrate alli ambasciadori ‘After dinner, Socrates talked to the ambassadors’ Nevertheless, there are also cases where a constituent different from the subject does not cause the subject to appear in postverbal position: (18) […] e dopo molti parlamenti i detti ambasciadori si partiro inn accordo col re […]. (Villani, Nuova Cronica) ‘and after many discussions the aforementioned ambassadors left with the permission of the king’

244 Ludwig Fesenmeier The type of VS order under discussion, i.e. with the subject referent forming (part of) the topic, has repeatedly been accounted for in mainly pragmatically oriented studies as well: – with regard to Latin historiographical texts, Selig (1992: 208) has pointed out that die VS-Stellung […] zur Kodierung von unbetont thematischen Subjekten und die SV-Stellung zu der von betont thematischen genutzt wurde. Das Verfahren der Subjektinversion zur Kodierung unbetont thematischer Subjekte ist auch im heutigen Spanischen und in älteren Sprachstufen des Italienischen verbreitet. [all emphasis ours] [VS ordering was used for marking unstressed topical subjects and SV ordering for marking stressed topical subjects. The procedure of subject inversion in order to mark topical subjects explicitly as being unstressed can also be found in Modern Spanish and in older stages of Italian.]

– especially with regard to Modern Spanish, already Wandruszka (1982: 56) had suggested that “[d]as thematische Subjekt […] seine prominente satzeröffnende Position [emphasis ours] deshalb verlieren [kann], weil es unmittelbar vorerwähnt ist aber dennoch in nominaler Form expliziert werden soll [a subject-topic can lose its prominent sentence-inital position, because, albeit being mentioned in the immediately preceding context, it has to be expressed in nominal form]”.36 It has often been claimed that the sentence-initial position is a comparatively “prominent”, cognitively salient one,37 but as far as Old Italian is concerned, this can also be seen from the fact that, in order to assign the function of topic to (the referent of) a non-subject constituent, it was indeed sufficient to collocate the latter in sentence-initial position: (19) Avemo chomperato […] da meser Lambertuccio […] X braccia di tereno […]. Questo tereno rivendemo ad Asinello Bremençone selaio […]. (Memoriale d’un proprietario terriero fiorentino dei primi del Trecento, 1318) ‘We have bought […] from Sir Lambertuccio ten cubits of land. This piece of land, we resold it to Asinello Bremençone, the saddler’ As can be seen from the functionally equivalent construction with left dislocation in (20) below, questo tereno is also the topic of the utterance in (19), clitical resumption being by no means mandatory: collocating the direct object in sentence-initial position was completely sufficient.

On two functions of verb-subject order in Old Italian

245

(20) Avemo comperato […] da meser Lambertuccio […] XIJ bracia di tereno […]. […] Questo tereno rivendemolo a’ Maci. (Memoriale d’un proprietario terriero fiorentino dei primi del Trecento) ‘We have bought from Sir Lambertuccio twelve cubits of land. This piece of land, we resold it to the Maci.’ In order to illustrate the difference ‘stressed vs. unstressed topic’, let us start by considering the following example: (21) Il quarto confine di Toscana di verso settentrione sono le dette alpi Apennine, le quali confinano e partono la provincia di Toscana da Lombardia e Bologna e parte di Romagna; e gira la detta provincia di Toscana VIIc miglia. Questa provincia di Toscana ha più fiumi: […]. (Villani, Nuova Cronica) ‘The fourth border of Tuscany in the northern direction is constituted by the aforementioned Apennines, which separate the province of Tuscany from the Lombardy and Bologna and a part of the Romagna; and the aforementioned province of Tuscany has a circumference of seven hundred miles. This province of Tuscany has several rivers’ In (21), though being already explicitly mentioned in the first sentence, the provincia di Toscana appears again as a full NP in the following sentence. This is due to the different syntactic function of the respective NPs, being first the direct object and then the subject. Note furthermore that in the paragraph these utterances belong to the discourse topic38 consists in the geographic extent of the province. In turn, what is of interest in the next section is its hydrologic facts, the reader being thus confronted with a new discourse topic. In the utterance introducing the new discourse topic, the provincia di Toscana obviously still fulfills the topic function, it is again coded as full NP and functions as the syntactical subject of the sentence, preceding, however, the finite verb: according to what has been suggested by Selig and Wandruzska, it is, then, an instance of “stressed topic”, but what might be the purpose of such a “stressing”? As has been pointed out by Weigand (1979: 173–174), a contextually given referent can be assigned a certain emphasis when it is inserted in a thematic relation or framework different from the preceding one. Since this description fits well the case in (21), we would suggest that the SV ordering in the second utterance serves precisely the function of signalizing the beginning of a new thematic frame-

246 Ludwig Fesenmeier work, where, of course, the provincia di Toscana once again plays an important, yet different role with respect to the preceding framework. The postverbal subject la detta provincia di Toscana is on the contrary an “unstressed” topic, since in this utterance, the reader learns about the circumference of the same province whose borders he is informed about in the preceding context, all these informations being part of the same discourse topic, i.e. the province’s geographical extent. We would like to briefly turn to another aspect, as yet unconsidered, regarding the subject’s shift from sentence-initial to postverbal position. It seems correct to assume that with regard to the type of VS order under scrutiny, the subject is shifted from sentence-initial and not only preverbal position, given the striking rarity of examples such as (22), where the verb is preceded by an adverbial phrase specifying the “temporal […] framework within which the main predication holds” (Chafe 1976: 50):39 (22) E Tristano lo serviva bene e gentile mente, e facevasi molto amare. E allora era Tristano di quindici anni, e schermiva e armeggiava […]. (Anonymous, La Tavola Ritonda o l’Istoria di Tristano, 1350) ‘And Tristram attended to him well and politely, and endeared himself much. And at this time Tristram was fifteen years old, and he fenced and practised the use of arms’ The utterances analyzed so far rather denote ‘states’ (the verbs appear primarily in the present or past tense), but ‘events’, though being rare, are not excluded on a priori grounds: (23) Monna Margherita […] fu moglie di Giovanni de’ Cocchi; della quale nacque una fanciulla: la quale fu […] moglie di Iacopo di messer Vieri de’ Bardi, e di lei nacquono più figliuoli, maschi e femmine. Morì il detto Iacopo in Fiandra in pregione. La detta donna, quand’era fanciulla, stette certo tempo nel palagio di Via Maggio co’ zii […]. (Donato Velluti, Cronica Domestica, 1370) ‘Lady Margherita was the wife of Giovanni de’ Cocchi; she gave birth to a girl, who became the wife of Iacopo son of messer Vieri de’ Bardi, and who gave birth to several children, boys and girls. The aforementioned Iacopo died in Flanders in prison. When she was still a girl, the aforementioned lady spent some time in the house in Via Maggio, together with her uncle and aunt’

On two functions of verb-subject order in Old Italian

247

Note, however, that the discourse topic of the passage quoted in (23) is Margherita’s daughter, the content of the sentence emphasized being nothing but a short “parenthesis”, which can be paraphrased by ‘(As regards the aforementioned Iacopo, let me add that he died in Flanders in prison.)’.40 Some final remarks deal with the text types where categorical VS ordering can be found: we have quoted from rather elaborate literary texts (romances of chivalry, the Novellino tales, Boccaccio’s Decameron) and stylistically more or less ambitious historiographic ones (Villani’s Nuova Cronica, Velluti’s private Cronica Domestica), but also from rather simple commercial correspondence. Instances of this type of VS ordering can also be found in bookkeeping records: (24) It. prestai a messer Attaviano […] fior. d’oro sesanta, i quali gli portò Arigho suo fante a Cholle per chonperare le mulina. Venero questi dr. da Monpuslieri. (Il Quaderno di Ricordi di Messer Filippo de’ Cavalcanti, 1324) ‘Also I lent to Sir Attaviano sixty gold florentines, which brought him his child Arigho to Cholle in order to buy the mills. This money came from Monpuslieri’ Let us briefly sum up what has been claimed so far with regard to the “categorical” type of VS order: – in the instances of this type, the subject referent forms (part of) the topic of the utterance in which it appears; – in some cases, the subject is explicitly coded as a full NP in order to grant the cohesion/coherence of the text; – the VS order can be considered the conscious choice by the author, since, as far as the topic-comment structure of the utterance is concerned, SV ordering would not have brought about any difference, but – would have marked the subject-topic as “stressed”, while VS ordering marks it as “unstressed”; – taking into consideration the linguistic contexts in which this type of VS order occurs, it appears that VS serves a function which pertains to the level of discourse structure, e.g. that of signalizing the shift from the level of narration to that of description,41 or to a comment by the author: (25) E questo medesimo dicea il podestà, che venía da casa messer Carlo, che gliele avea udito giurare di sua bocca che farebbe impiccare messer Corso Donati. Il quale […] era entrato in Firenze la mattina con

248 Ludwig Fesenmeier compagni […]. Non giurò messer Carlo il vero, perchè di sua saputa venne. (Dino Compagni, Cronica delle Cose Occorrenti ne’ Tempi Suoi, 1312) XII

‘The podestà said the same thing: he had come from the house of messer Charles and had heard him swear with his own mouth that he would hang messer Corso Donati. Messer Corso entered Florence that morning with twelve companions. What messer Charles had sworn was not the truth, for messer Corso had come with his knowledge.’ (Bornstein 1986: 46) – instances of categorical VS order can be found in texts pertaining to the most diversified text types, ranging from very simple ones (bookkeeping records) up to elaborate literary texts (Boccaccio’s Decameron).

3.2. Thetic VS order As indicates the title of the present section, we shall in the following argue in favor of the hypothesis that a second type of VS ordering can be identified, in the case of which not only the subject referent must not be interpreted as the topic of the utterance in question, but, what is more, the latter lacks any topic element at all – in the sense of what the utterance is pragmatically about. According to the approach to be defended, we would be dealing here with what Sasse (1987: 558) has called a “pragmatically unanalyzed state of affairs [emphasis ours]”. Let us start by reconsidering some examples, in part already referred to in the previous sections (and repeated here for convenience), in part frequently adduced in the relevant literature: 42 (26) a. Li ambasciadori fecero la dimanda loro, e videro li costumi e la corte; poi, dopo pochi giorni, adomandaro commiato. Lo ’mperadore diede loro risposta e disse: “[…]”. Andar li ambasciadori e rinunziaro e raccontaro ciò ch’aveano veduto et udito […]. (Novellino) ‘The ambassadors asked their question and observed the customs and the court; then, after a few days, they asked for permission to take their leave. The emperor answered them by saying: “[…]”. The ambassadors left and referred and narrated what they had seen and heard’ (adapted from Consoli [ed.] 1997: 19)

On two functions of verb-subject order in Old Italian

249

b. Quel re ragionò con Balaam profeta e disse: “Dimmi, Balaam: che è ciò, che li miei nemici sono assai meno poderosi di me, e io non posso fare loro nullo danno?” E Balaam rispuose: “Messere, però ch’e’ sono populo di Dio. Ma io farò sì che tu potrai sopra loro: ch’io andrò e maladicerolli, e tu darai la battaglia e averai sopra loro vittoria”. Salìo questo Balaam in su uno asino e andò su per uno monte; e ’l popolo era quasi là giù al piano, e quelli andava per maladirli di sul monte. (Novellino) ‘That king discussed the situation with the prophet Balaam and said: “Tell me, Balaam: what is it, that my enemies are so much less powerful than me, and yet I cannot inflict any damage upon them?” And Balaam replied: “Sire, it is because they are the people of God. But I will do in such a way that victory will be yours: I will go and curse them, and you engage them in battle and victory will be yours.” This Balaam mounted an ass and journeyed up a mountain; and the people were almost below him on the plain, and he was going in order to curse them from the mountain top.’ (adapted from Consoli [ed.] 1997: 59) c. E la damigiella chiamoe uno suo nano e dissegli: “Domane mi farai uno messaggio a T., […]”. E allora disse lo nano che questo messaggio farà egli e sarae molto volontieri. Al matino si leva lo nano e montoe a ccavallo e venne ala corte der ree Marco. (Tristano Riccardiano) ‘And the noble lady called one of her dwarfs and said to him: “Tomorrow you will deliver a message to Tristram”. And thereupon the dwarf said that he would deliver the message and that he would be into doing so. In the morning the dwarf gets up and mounted and reached the court of the king Marco.’ d. Uno signore di Grecia, lo quale possedea grandissimo reame et avea nome Aulix, avea uno suo giovane figliuolo, il quale facea nodrire et insegnarli le sette liberali arti, e faceali insegnare vita morale, cioè di be’ costumi. Un giorno tolse questo re molto oro e diello a questo suo figliuolo e disse: “Dispendilo come ti piace”; e comandò a’ baroni che neuno non li insegnasse spendere questo oro, ma sollicitamente avisassero il suo portamento e ’l modo ch’e’ ne tenesse. (Novellino; to be continued in (26f)) ‘A Greek gentleman, who possessed a very great realm, and was called Aulix, had a young son whom he had nurtured and instructed in the seven liberal arts, and taught to lead a moral life, that is, to be

250 Ludwig Fesenmeier well behaved. One day this king withdrew a great deal of gold and gave it to his son, and said: “Distribute it as you please”; and he charged his barons not to instruct the boy in its allocation, but only to observe his comportment and the manner in which he conducted himself.’ (Consoli [ed.] 1997: 29) e. E Socrate parlò alli ambasciadori e disse: “Voi pranzerete inanzi, e poi intenderemo alle nostre bisogne”. Tennero lo ’nvito e pranzar assai cattivamente, con non molto rilievo. Dopo ’l pranzo parlò Socrate alli ambasciadori e disse: […]. (Novellino) ‘And Socrates spoke to the ambassadors and said: ‘First, you will dine, then we will address our tasks’. They accepted the invitation and ate an exceedingly unappetising meal of meagre portions. After dinner, Socrates spoke to the ambassadors and said: […].’ (adapted from Consoli [ed.] 1997: 81) f. [Continued from (26d):] I baroni seguitando questo giovane, un giorno stavano con lui alle finestre del palazzo. Il giovane stava pensoso. Vide passare per lo cammino gente che parea assai nobile secondo li arnesi e secondo le persone. Il camino correa a piè del palagio. Comandò questo giovane che fossero tutte quelle genti menate dinanzi da llui. (Novellino) ‘Following this youth, one day the barons found themselves with him at the windows of the palace. The young man was deep in thought. He saw people pass by on the road below who, from their dress and persons, seemed very noble indeed. The road ran to the foot of the palace. The young man demanded that all these people be brought before him.’ (Consoli [ed.] 1997: 29) Although the examples quoted so far are all taken from Novellino tales, similar ones can easily be found in other texts, too, among others in historiographic ones (27), but also in bookkeeping records (28): (27) [The preceding chapter ends with a detailed description of what preparatory measures are taken (which route would be the best one to take, the allies gathered); the next chapter starts as follows: ] Mossono le insegne al giorno ordinato i Fiorentini, per andare in terra di nimici: e passarono per Casentino per male vie […]. (Compagni, Cronica) ‘On the appointed day the Florentines unfurled their banners to go into enemy land: they passed through the Casentino by bad routes’ (adapted from Bornstein 1986: 12)

On two functions of verb-subject order in Old Italian

251

(28) Del mese di dicenbre nel CCCXIIJ venne a me a Siena donno Fedele […] e volle vedere tutte le ragioni d’ongni chosa ch’io avesse a fare cho·l’abate di San Salvi e de’ poderi e de’ CC fior. d’oro; e andonne a Firenze chon queste ragioni, ed io lo feci mio prochuratore a finire tutto ciò ch’io potesse adomandare a l’abate e al monestero. Poi tornò da Firenze il detto donno Fedele e rechommi una charta ch’elgli avea ricevuta […]. (Quaderno di Ricordi di Messer Filippo de’ Cavalcanti) ‘In december 1313 came to me to Siena sir Fedele and wanted to see all records concerning any affair I and the abbot of San Salvi are involved in, and with regard to the parcels and to the 200 gold florentines; and he left for Florence with these records, and I made him my representative in order to decide all what I could demand to the abbot and the abbey. Later on the aforementioned sir Fedele returned from Florence and brought me a charter which he had received’ With regard to the foregoing examples, the following can be immediately stated: – the subject referents are all contextually given (and hence definite), i.e. they have already been mentioned in the preceding context; – the utterances appear, quite regularly indeed, after passages of direct speech (26 a, c), and/or are introduced by an adverbial phrase specifying the temporal circumstances of the event ((26c,d), and (28); also (2b) and (10));43 – with view to their preceding contexts, the utterances in (26)–(28) turn out to fit quite naturally as answers to questions such as “what happened thereupon (26 a,b)/ in the morning (26c)/ one day (26d)/ after dinner (26e)/ else (26f)/ then (27, 28)?”,44 i.e. they express ‘events’ to the effect that, as regards their content, they form part of a temporal sequence, i.e. they “take us forward in time”, as Hatcher (1956a: 238) has put it; their only presupposition is that “something happened”, such a presupposition being, however, “merely situationally implied, not lexicogrammatically evoked in the sentence” (Lambrecht 1994: 233); they can thus be interpreted as instances of what Lambrecht (1994: 233) calls “sentencefocus structure”.45 The same analysis holds true for (29), to which equally applies the underlying question “what happened then?”: 46 (29) Nelle parti di Grecia ebbe un signore che portava corona di re et avea grande reame et avea nome Filippo; […]. […] Avenne un giorno che a questo signore fu appresentato, delle parti di Spagna, un nobile des-

252 Ludwig Fesenmeier triere di gran podere e di bella guisa. Adomandò lo signore mariscalchi per sapere la bontà del destriere; fuli detto che in sua pregione avea lo sovrano maestro intendente di tutte le cose. (Novellino) ‘Somewhere in Greece there was a lord who wore the crown of a king and had a great realm and he was called Philip. […] One day it happened that a noble steed of great strength and beauty, from somewhere in Spain, was presented to this lord. The lord asked for marshals in order to determine the worth of this steed; he was told that in his prison he had the consummate teacher, knowledgeable in all things.’ (adapted from Consoli [ed.] 1997: 19, 21) The fact that the examples considered so far form part of a temporal sequence, that they “take us forward in time”, deserves, however, closer attention. Consider to this end the following example, also discussed in Blumenthal (1980: 125): (30) Uno borgese di Bari andò in romeaggio e lasciò trecento bisanti a un suo amico con queste condizioni e patti: “[…]”. Andò il pellegrino in suo romeaggio; rivenne al termine ordinato e radomandò i bisanti suoi. (Novellino) ‘A citizen of Bari went on pilgrimage and left three hundred bezants with a friend under the following conditions and agreement: “[…]”. The pilgrim went on pilgrimage; he returned within the prescribed time and asked for the return of his bezants.’ (adapted from Consoli [ed.] 1997: 33) As Blumenthal (1980: 125) rightly points out, the state of affairs expressed by the emphasized utterance is known to the reader from the very beginning of the tale, whose first sentence tells us exactly the same state of affairs: somebody went on pilgrimage. Nevertheless, a literal reading of Uno borgese di Bari andò in romeaggio e lasciò trecento bisanti … [A citizen of Bari went on pilgrimage and left three hundred bezants] would violate the Gricean submaxim ‘Be orderly’,47 since, obviously enough, the citizen could not leave Bari and then leave the money with his friend staying home. Despite its linguistic form, the first sentence of the tale, thus, is not to be interpreted as the expression of the pilgrim’s factual departure, but only as that of his intention to do so: it is not until after the conversation with his friend that the pilgrim actually leaves Bari, and it is the event of the factual departure that the reader is told by the emphasized utterance.

On two functions of verb-subject order in Old Italian

253

Against the background of what has just been said, let us now consider more closely the content structure of the passages quoted so far: – in the examples (26a)–(26c), (26e), (27), and (28) (and also (7) and (10)), the utterances emphasized express the beginning of a new “episode” inside the overall plot,48 the chronological connection to the preceding one sometimes being explicitly stated (26e); – in (26a)–(26e), (27), (28), and (30) (and also (7)), the VS utterances only open a new episode whose protagonist is the subject referent, who then becomes also the subject-topic of the following sentences (e.g. (26a): “e rinunziaro e raccontaro [and referred and narrated]”); – in (29), the sentence emphasized informs the reader about the next “step” after the presentation of the steed; – in the examples (26d), (26f), and (28) in contrast, we find the passage from mere description to narration of an event: – in (26d), the tale starts by describing how the king trained his son in general and then the author goes on to give a concrete example; – the beginning of (26f) describes a certain scene, i.e. the young man standing there and observing what happens on the road, and then goes on to the young man’s sudden reaction to what he has observed; – as far as (28) is concerned, note that the succession of the sentences andonne [he left] – feci [I made] – tornò [he returned] is, again, not orderly in the Gricean sense, for, logically, Fedele had to be made the speaker’s representative before departing; the sentence starting by ed io lo feci [and I made him] thus delivers some background information which does not fit in the narration of the mere temporal sequence ‘Fedele came to me – wanted to see all records – left for Florence – returned from Florence – brought me’. As we have already pointed out, the common denominator of the examples (26)–(30) consists in the fact that the utterances under analysis can be interpreted as an answer to the underlying question “what happened (next)?”, such a result supplying good evidence that they can count as instances of sentence-focus structures in the sense of Lambrecht (1994: 233). The subject referents being definite, however, we must still verify that they cannot be considered, alternatively, the topics of their respective utterances, or that, at least, such an interpretation is a lot less plausible.49 Let us start with the assumption that the subject referents of the examples discussed in this section are the topics of their respective utterances. If so, taking into account what has been pointed out in 3.1, especially as far as the

254 Ludwig Fesenmeier comparison of the positions of “stressed” and “unstressed” subjects-topics is concerned, one would indeed expect the subject to appear in utteranceinitial position or, at least, in the first slot of the core sentence in the sense of Salvi (2004: 14–15): 50 the utterances under analysis here all reinsert the subject referents in a “new” episode, and hence in a “new” thematic relation in the sense of Weigand (1979: 173–174).51 Such an “emphasis” is even more justified when one bears in mind that in all the passages cited so far, it is not to be expected that the “protagonist” of the new episode will be he whom he actually turns out to be: there are dialogue scenes ((26a–c) and (30)) or, more generally, several possible “future” discourse topics, e.g. the son of the king in (26d), the ambassadors in (26e), the people in (26f), the speaker in (28), or the stead in (29). Nevertheless the subjects follow the finite verbs. However, let us reconsider the suggestion made by Arcangeli (2004: 58– 59) with regard to the examples (26a) and (29), i.e. the fact that in these cases the author has simply decided to reserve the sentence-initial position for the verbs, in order to emphasize the events described. One might thus feel entitled to consider (at least) these examples as instances of “predicatefocus structures” in the sense of Lambrecht (1994: 226). As a consequence, the presuppositions evoked by the utterances would be ‘the ambassadors / the king are/is the topic for the assertion to be made’, but precisely this presupposition is not supported in any way by the preceding discourse context, the most plausible one being indeed, as we have tried to show above, ‘then something (else) happened’. With this specification in mind, one can reasonably ask why the author might have wished to emphasize the events described. From a topological point of view, the highlighted character of the verb is undoubtedly due to its being in the cognitively salient utterance-initial position.52 Thus, Arcangeli’s suggestion is actually in line with what has already been observed: as far as the state of affairs described is concerned, these VS utterances open a new “thematic section” inside the overall plot, and on that account it seems rather “useful” that, on the syntactic level, such a section starts by the constituent expressing first and foremost what happens. One must not forget, however, that the utterance- or at least sentenceinitial position of the verb is to be considered rather the consequence than the motive. Let us consider what has been pointed out with regard to nonsubject constituents: as has been shown in (19), the simple appearance in the initial position of, e.g., a definite direct object is sufficient to turn it into the topic of the utterance. Mutatis mutandis, the same holds true for (definite) subjects, and so, in order to avoid (at least) the (stressed-) topic reading

On two functions of verb-subject order in Old Italian

255

resulting from SV ordering, it can only appear after the verb. Whether a postverbal subject actually plays the topic role (categorical VS order) or not (thetic VS order), depends on parameters such as the state of affairs expressed by the utterance (basically, ‘state vs. event’) and/or the relation the content of the latter has to the overall context (‘[not] taking the reader forward in time’). A last remark concerns the appearance of both post- and preverbal subjects together with the demonstrative adjective questo in the examples (1b), (24), (26b), (26d), and (26f) on the one hand, and (21) on the other hand. We cannot go into details here,53 but while the intrinsically text-anaphoric element detto seems to signalize nothing but simple referential identity/continuity, questo appears to sometimes have an “added value”. Considering the copresence of both detto and questo in (21) together with the difference between unstressed and stressed topics, questo, as an “added value”, can be assigned a kind of highlighting-function. This interpretation is confirmed by Selig’s (1992: 145) observation with regard to Late Latin stories of the saints, where determiners can also appear as means of text structuring: Der neue Protagonist ist schon in den Text eingeführt und bereits im “Personeninventar” fest verankert. Durch die explizite Kennzeichnung wird nochmals der anaphorische Bezug verstärkt, gleichzeitig aber auch der Neubeginn einer Episode angezeigt […]. [emphasis ours] [The actual protagonist has already been introduced in the text and is insofar well integrated in the “stock of characters”. The explicit marking reinforces once again the anaphoric relation, but at the same time also signalizes the opening of a new episode.]

In the above mentioned cases of cooccurrence with postverbal subjects in thetic utterances (26b,d,f), questo may thus allow the subject referent to receive some emphasis in spite of the latter’s comparatively unstressed position. In the categorical instances of VS order in (1b) and (24), in contrast, questo would merely reinforce the anaphoric relationship with the respective antecendents, since both terza parte and denari have themselves anaphoric character. We have on various occasions insisted on the episode-opening character of the VS type discussed in this section, which seems to be a distinctive feature in comparison to that analyzed in 3.1: while the function fulfilled by categorical, as we have termed it, instances of VS order concerns the level of discourse structuring (among others: signalizing background information, a comment by the author, mere expression of topic continuity), we have argued that the thetic instances of VS order examined here are linked to the

256 Ludwig Fesenmeier circumstances in the narrated, but nevertheless extra-linguistic (even if sometimes fictitious) reality.54 It might be of some interest here to recall that Hatcher (1956a: 238) has already addressed the question of ‘what is the relationship, if any, between a given predication and the one that has preceded?’; her answer reads as follows: I find myself able to see clearly and sharply only the distinction between ‘temporal sequence’ (whether narrative: and then he left, or instructional: e.g. cookbook recipes), and those predications which do not take us forward in time. Among the latter, however, I have been able to note, in quite a confused fashion: descriptions, explanations, illustrations, corrections, objections, corroborations, deductions, conclusions, personal appreciations – and what I can call only ‘information proper’ (or ‘further information’).55

Note, however, that in stylistically less elaborated texts, it may become difficult to distinguish clearly the level of discourse structuring from that of merely reproducing a temporal sequence; this is clearly the case in (31): (31) [The author is currently talking about his brother Lottieri:] […] stette a studio a Pisa, e Napoli. Fecesi prete novello e cantò messa, poi tornai per giudice in Firenze, e innanzi morisse nostro padre, con grande onore. Morì nostro padre, e Filippo e Piccio tornarono in Firenze, e dividemmoci; e per la divisa fu sentenziato, di dovere io contentare fra Lottieri di fiorini cento d’oro […], e Filippo dargli ogni anno fiorini X per sue spese […]. (Velluti, Cronica Domestica) ‘he studied at Pisa and Naples. He was ordained to the priesthood and he ministered, then I returned in Florence with great honours as a judge, before our father died. Our father died, and Filippo and Piccio returned in Florence, and we divided up the legacy among ourselves; and it was resolved that I had to satisfy Lottieri with 100 gold florentines and that I had to give Filippo every year ten florentines for his expenses’ In (31) Velluti starts by talking about his brother, then passes abruptly to inform about himself, and finally goes on to a new discourse topic, i.e. the details of the distribution of the estate, the “turning point” being the death of the siblings’ father. The father’s death thus, opens a new “episode” in the author’s life and text, the signalizing of the textual “turning point” being, however, rather a derivative effect of that of “episode-opening”.56

On two functions of verb-subject order in Old Italian

257

4. Conclusion In this paper, a distinction has been roughly sketched between to two functional types of VS order with definite subject referents. As van Dijk points out, “[a] serious pragmatics should account for the function of utterances with underlying textual structures” (quoted in Fleischman 1990: 33). Such a demand is valid all the more when one is engaged in the analysis of a “text language” like Old Italian, “for which all evidence derives from texts” (Fleischman 1991: 252, note 1). Against this general background it has proved useful, as we think, to consider “Texte zunächst einmal als Texte – mit ihren charakteristischen Notwendigkeiten und ihrer charakteristischen Tradition – […], nicht nur gewissermaßen als ‘Steinbrüche’ für die Beobachtung von Einzelerscheinungen [texts first of all as texts, with their characteristic necessities and their characteristic tradition – not only, in a way, as a quarry for observing isolated phenomena]” (Raible 1985: 67). As becomes evident from the discussion in the preceding section, occurrences of VS order in Old Italian with subject referents definite in the sense of Chafe (1976) can receive pragmatically different interpretations, depending on the relations in which these utterances stand with the context. We have proposed to distinguish between the “categorical” (3.1) and the “thetic” types (3.2), which, according to our proposal, differ from each other in so far as in categorical utterances, the subject referent is what the utterance is pragmatically about, though appearing as an unstressed topic; as for the thetic ones, however, one deals with “a pragmatically unanalyzed state of affairs [emphasis ours]” (Sasse 1987: 558), i.e. there is no topic element at all. As far as the functions of these two utterance-types are concerned, categorical VS utterances play a role on the level of text structuring, while the thetic ones bear essentially upon the extra-linguistic reality. As far as the distinction ‘categorical vs. thetic’ is concerned, it should be obvious that throughout the analysis, we have used these expressions without an extensive aspiration on the theoretical level.57 Let us, however, point out that, if the thetic examples of VS order previously analyzed were assigned to the “episode-opening function” (Sasse 1995: 16), this result is merely due to our choice of the examples, principally governed by which were referred to in former studies.58 Furthermore, we have argued that as for the subject referent, his functioning as such as the topic of an utterance must be considered independent from the position relative to the verb: if the subject referent is the topic, it is a stressed (SV) or an unstressed one (VS), for whatever reason the author has decided to emphasize this element or not. Yet, while a definite preverbal

258 Ludwig Fesenmeier subject is by default interpreted as the topic, the interpretation of a postverbal one as topic or not, depends on the various parameters we have referred to (state of affairs expressed by the utterance, its relation to the overall context, etc.).59 The starting point for the distinction proposed consisted in the rather simple statement that the feature of definiteness is a necessary, but by no means a sufficient condition for the referent of a given constituent (here: the subject) to function as the topic of the utterance. Once allowing for such elements not to be “automatically” interpreted this way, examples of such various kinds as those quoted in (9) and (10) on the one hand, and in (11) on the other hand need not be put any longer in Procrustes’ Bed.60 The reason for the problems we have encountered in the earlier studies quoted appears to be the lack of, or at least insufficient advertence to, making two important differentiations: that between ‘(contextually) given’ and ‘topic’ and that between ‘textual structure’ and ‘extralinguistic reality referred to in the text’. For, as concerns both the facts (the examples) considered here and the interpretations proposed, they can easily be retrieved in the different contributions we have been referring to. It is just by paying the appropriate attention to such essential differentiations that the situation becomes more transparent. With regard to VS order in Old Italian, we are thus faced with a situation similar to that described by Lombardi Vallauri (2004: 296) with reference to certain types of OV ordering: […] la situazione si presenta in un certo senso “schizofrenica”, o se si preferisce, contradittoria. Non è usuale, e si spiega piuttosto male, che in una lingua gli stessi strumenti formali servano categorie opposte. [the situation is, in a sense, “schizophrenic”, or, if one prefers, contradictory. It is unusual, and quite difficult to explain, that in a language the same formal means can be used for opposite categories.]

The apparent contradiction is resolved, then, by saying that the different functions of OV order identified do not appear in the same types of texts. 61 The examples discussed here testify to a similarly “contradictory situation”, but we have suggested a different explanation. It is certainly true that both types of VS order appear with greater frequency in more elaborate texts, characterized by communicative distance. However, since both categorical and thetic VS structures can be found in documents such as bookkeeping records,62 it seems justified to conclude that both patterns were equally current in Old Italian.

On two functions of verb-subject order in Old Italian

259

Acknowledgements I am indebted to Edith Szlezák for the stylistic revision of the text (of course, all remaining errors are mine). I am equally grateful to the anonymous reviewers for their truly helpful comments on a previous version of this paper; I did my best to be receptive when revising it. Notes 1. 2.

3. 4. 5.

6.

Following current terminology, by “Old Italian” we refer to the Tuscan vernacular of the 13th and 14th centuries. All examples are quoted from the textual database of the Tesoro della Lingua Italiana delle Origini (release “02/2006 [r1]”; the database is accessible at http://tlio.ovi.cnr.it/TLIO, for further information cf. Dupont 2001); where necessary, we have specified the edition chosen. Cf. also Givón (22001: 418), who speaks of “the classical definition of definiteness as identifiability”. Cf. also Givón (22001: 418, (61a)), where the sun is classified under “Globallyaccessible (‘generic’) definite referents”. The same situation (SV vs. VS ordering) can be observed in the case of indefinite subject referents: (i) a. […] uno savio villano e antico si levò, e disse: […]. (Villani, Nuova Cronica [ed. Porta]) ‘a wise and ancient inhabitant rose and said’ (Wicksteed [ed.] 21906: VII, 26) b. […] si levò un barone e disse ad Argo: […]. (Anonymous, Il Milione di Marco Polo [Versione Toscana del Trecento], 1310) ‘it rose a lord and said to Argo’ Such cases, however, will not be treated here; likewise instances of VS order with the subject referent pertaining to the comment, as is the case in (ii), will not be considered: (ii) Diciesette dì di febraio avemmo due lettere che nne mandaste, […]: recollene il primo corriere di Langnino; […]. (Letter of messer Consiglio de’ Cerchi to Giachetto Rinucci, 1291) ‘February 17, we received two letters that you sent: it was the first carrier from Langnino who brought them’ However, in Vanelli (1999: 235) the topic function is considered a pragmaticsemantic one, while in Berretta (1995: 129) topic (“tema”) and comment (“rema”) are explicitly assigned to the “livello semantico, che riguarda la struttura della proposizione [semantic level, which concerns the structure of the proposition]”.

260 Ludwig Fesenmeier

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

21. 22. 23. 24. 25. 26.

27.

28. 29.

We are well aware of the fact that, strictly speaking, the conceptual pairs ‘topic-comment’ and ‘theme-rheme’ belong to different frameworks; it is only for the sake of terminological uniformity that, wherever possible, “theme”, “rheme”, etc. have been replaced by “topic”, “comment”, etc. Cf. also Lambrecht (1994: 119). Cf. also Lambrecht (1994: 131), Stark (1997: 39), and the differentiation between “topic” and “topic expression” below. For a brief summary of their basic ideas cf. Oesterreicher (1997: 191–196). We shall return to these examples in a more detailed manner below, cf. ex. (26a) and (29). Cf. in particular the synopsis in Stark (1997: 42–44). We shall return to this example in a more detailed fashion below, cf. ex. (26e). Cf. the discussion in Ulrich (1985: 44–54). Cf. also Primus (1993: 880): “There are sentences without a topic, and these will be called thetic sentences (as opposed to categorical sentences).” Cf., e.g., ne la passione in (2b) and Dopo ‘l pranzo in (5); cf. however Vanelli (1986: 260) for an analysis of such elements in terms of topics. And which all can turn out to be difficult in their concrete application, cf. Schlobinski and Schütze-Coburn (1992). Cf. also Hatcher (1956b); for a positive evaluation of Hatcher’s approach cf. Bossong (1984). For a brief summary of previous research on VS order in Italian cf. Blumenthal (1980: 119–123) and Fesenmeier (2004: 105–107). Cf. Blumenthal (1980: 126, ex. (35) and (36)). The question of ‘what is the relationship, if any, between a given predication and the one that has preceded it?’ has already been addressed as an important one in Hatcher (1956a: 238), cf. below, section 3.2. Cf. Sornicola (2000: 108, ex. (18), (20), and (19)). As is obvious, it is only in (10) that the object referent is “[+GIVEN]”. Cf. Benincà (n. d.) and Poletto (n.d.). Cf. below, ex. (21) and (25). And also in her example (32f), repeated here below as (29). Note that with regard to ex. (30) quoted below, already Blumenthal (1980: 125) had observed that in the modern language one would expect an adverb such as dunque (‘then’). Given the overall context, the “underlying question”-test (“what had Pericone?”) does not seem natural either. These difficulties should explain why Sornicola (2000: 108) accounted for this example in terms of the “pragmatic property of being ‘all-in-FOCUS’”. The topic status of the subject referent is particularly evident in (12), where the “underlying question” is even explicitly given. Moreover, the paraphrase ‘as far as X is concerned’ applies perfectly well to both cases: “as far as the duration of the battle /the actual state of the silk is concerned”.

On two functions of verb-subject order in Old Italian

261

30. Cf. also what has been reported above, section 3. 31. Cf. Fesenmeier (2004), where numerous examples of this kind have been taken into consideration. 32. For the difference between these two concepts cf. Stark (2001). 33. Cf. below, ex. (24) and (25). Admittedly, these descriptions are completely “untechnical”, rather being the results of “hermeneutic efforts” than of a precise linguistic analysis. However, even if one might discuss the “categorisations” proposed, there seems to be no essential difficulty to justify such an approach, since what we are concerned with are the content relations the VS utterances under scrutiny bear to their preceding contexts (cf. also above, note 20; we shall return on this point once again, cf. below, section 3.2). 34. Cf., e.g., Vanelli (1999: 229), where further bibliographical references are given. 35. However, Sornicola (2000: 107) and Kaiser (2002: 165–168) prove to be still more apprehensive in this respect. More recently, Salvi (2004: 14–15) has proposed to distinguish between what he calls the “periferia [periphery]” of a sentence and its core, assigning to the periphery for example elements such as those specifying the temporal framework of the state of affairs denoted. However, this approach still leaves open the question of the reason for the alternative exemplified in (17) vs. (18). Cf. also the analysis of the “Left Periphery” proposed in Benincà (2006) and Benincà (n. d.). 36. For a similar account cf. already David (1887: 38–39). 37. Cf., e.g., Lambrecht (1994: 201) and, as regards Old Italian, Vanelli (1999), where the assumption of the prominence/cognitive saliency of the sentenceinitial position delivers the methodological framework of the analysis. 38. Cf. Barnes (1985: 28–31), where the notion of ‘discourse topic’ is defined as “roughly, that thing which a segment of discourse larger than the sentence is about, i.e. about which it supplies information” (1985: 28). 39. Recall that we do not consider such elements to play the topic role (cf. what has been said on this in section 2 [note 15]). 40. As far as the “unstressed” character of postverbal subjects-topics is concerned, in the case of the topic expression Iacopo this is due to the fact that its referent is actually extraneous to the current discourse topic ‘Margherita’s daughter’; Iacopo’s death constitutes rather what Hatcher (1956a: 238) has called “‘information proper’ (or ‘further information’)”, given only for the sake of completeness. 41. Cf. above, ex. (8), (11), and (14). 42. To mention but a few: (26a) is discussed in Arcangeli (2004: 58–59) and Benincà (n. d.: ex. (32e)), cf. above, ex. (4a) and (13); as for (26c), cf. Vanelli (1986: 260); (26d) is considered in Vanelli (1986: 260; 1999: 234) and Sornicola (2000: 108); (26e) is treated in Lombardi Vallauri (2004: 295) and Renzi (1988: 130, note 5), cf. above, ex. (5), and also in Vanelli (1986: 260, 1999: 234). 43. More rarely, however, such specifications can also occur in postverbal position, as is attested in (7).

262 Ludwig Fesenmeier 44. Imagine a person X reading out the text and stopping before the utterances highlighted; a hearer Y might well ask such questions in order to make X continue his reading. As Hatcher (1956a: 240, note 10) has rightly pointed out, “[t]he presence of a given adverbial element may, in some cases, be of great importance.” 45. Cf. also Sornicola’s (2000: 108) qualification of (9) (= 26d) and (10) as structures having “the pragmatic property of being ‘all-in-FOCUS’”. 46. This example is also considered in Arcangeli (2004: 58–59), cf. above, ex. (4b), and in Benincà (n. d.: ex. (32f)). 47. This submaxim pertains to the category of ‘manner’, which Grice understands “as relating not […] to what is said, but, rather, to how what is said is to be said” (Grice 1975: 46). 48. Note that as far as (27) is concerned, the attack of Arezzo is the discourse topic of both the current chapter and the preceding one. 49. Although the subject referents are the most promising candidates for fulfilling the topic function, the referents of other constituents (including what is denoted by the verbs) cannot be excluded from such a function on a priori grounds; nevertheless, it seems far from difficult to assure oneself of the fact that in the examples as yet considered, the subject referents are indeed the only plausible candidates. Cf., however, what is discussed below with regard to a possible explanation in terms of “predicate-focus” in the sense of Lambrecht (1994: 226– 228). 50. Cf. above, note 35. 51. Cf. above, the detailed discussion of example (21). 52. Even if there is an adverbial expression which specifies the temporal, etc. framework within which the event takes place, and the verb therefore is “only” in sentence-initial position, this does not seem to interfere with the verb being emphasized. 53. Vanelli (n.d.) delivers a synchronic overview of deictic elements in Old Tuscan. For an analysis of demonstratives in different diatopic varieties of Mediaeval Italian, cf. Stavinschi and Irsara (2004); the situation in Late Latin is analyzed in Selig (1992). 54. This has been observed as early as in Blumenthal (1980: 126). In her study of the Novellino tales, Iulianella (1994) appears to be aware of this difference when she assigns certain instances of VS order to the function of “recall[ing] an entity that is momentarily dropped” (1994: 68, with regard to our example (26f)), others to that of being “indicative of a switch from one scenario to another” (1994: 67), or of indicating “the consequence of the preceding context” (1994: 71). 55. Cf. what we have pointed out with regard to the examples (8), (11), (14), (23), and (25). 56. It is presumably the lack of this distinction between “extra-linguistic world” and discourse structure that has prompted Sornicola (2000: 108) to speak of a

On two functions of verb-subject order in Old Italian

57. 58.

59. 60.

61.

62.

263

“turning point of the narration”, not distinguishing between the levels of respectively what and how is narrated. For a more detailed discussion and further bibliography, cf. especially Sasse (1987, 1995, and 1996), but also Venier (2002). As Sasse (1995) points out, other functions can be identified as well, e.g., the “reactive/consequential function” (Sasse 1995: 17), to which (2b) can be attributed; in (2a), on the contrary, the orbs are the topic of the utterance in question, and what the reader is told about them is the way in which they showed their sadness. Cf. also Sasse (1987: 570), especially his “algorithm” concerning “the choice of thetic and categorical expressions [emphasis ours]”. Cf. also the following remark in Iulianella (1994: 68), which does not only apply to the Novellino tales: “there is not a one to one correspondence with inversion and a specific pragmatic function”. Cf. Lombardi Vallauri (2004: 320): “La prima [funzione] prevale nei testi letterari, la seconda in quelli pratici [The first function prevails in literary texts, the second in texts related to practical purposes]”. Cf. ex. (24) and (28).

References Arcangeli, Massimo 2004 Strutture tematizzanti e ordine delle parole nella prosa narrativa toscana. Dal Novellino al Decameron: prove tecniche di variazione. In SintAnt. La Sintassi dell’Italiano Antico. Atti del Convegno Internazionale di Studi (Università “Roma Tre”, 18–21 Settembre 2002), Maurizio Dardano and Gianluca Frenguelli (eds.), 33–63. (Studi linguistici e di storia della lingua italiana 5.) Rome: Aracne. Barnes, Betsy K. 1985 The Pragmatics of Left Detachment in Spoken Standard French. (Pragmatics & Beyond VI: 3.) Amsterdam /Philadelphia: John Benjamins. Benincà, Paola 2006 A detailed map of the left periphery of Medieval Romance. In Crosslinguistic Research in Syntax and Semantics: Negation, Tense, and Clausal Architecture, Raffaella Zanuttini, Héctor Campos, Elena Herburger, and Paul Portner (eds.), 53–86. Washington D.C.: Georgetown University Press. n. d. La struttura della periferia sinistra. In Grammatica dell’Italiano Antico, Giampaolo Salvi and Lorenzo Renzi (eds.). [www.geocities.com/ gpsalvi/konyv/index.html; last access: 02/12/2008.]

264 Ludwig Fesenmeier Berretta, Monica 1995 Ordini marcati dei costituenti maggiori di frase: una rassegna. Linguistica e Filologia 1: 125–170. Blumenthal, Peter 1980 Die Stilistik der Subjektinversion im Italienischen. Italienische Studien 3: 119–131. Bornstein, Daniel E. 1986 Dino Compagni’s Chronicle of Florence. Translated, with an Introduction and Notes, by Daniel E. Bornstein. Philadelphia: University of Pennsylvania Press. Bossong, Georg 1984 Diachronie und Pragmatik der spanischen Wortstellung. Zeitschrift für Romanische Philologie 100: 92–111. Chafe, Wallace L. 1976 Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View. In Subject and Topic, Charles N. Li (ed.), 25–55. New York / San Francisco /London: Academic Press. Consoli, Joseph P. (ed.) 1997 The Novellino or One Hundred Ancient Tales. An Edition and Translation based on the 1525 Gualteruzzi editio princeps. (Garland Library of Medieval Literature 105A.) New York /London: Garland Publishing. David, Rudolf 1887 Über die Syntax des Italienischen im Trecento. Genf: Pfeffer. Dupont, Christian 2001 The Opera del Vocabolario Italiano Database: Full-Text Searching Early Italian Vernacular Sources on the Web. Italica 78: 526–539. Fesenmeier, Ludwig 2003 L’Ordine dei Costituenti in Toscano Antico. Padova: Unipress. 2004 Inversione del soggetto e strutturazione del testo nell’italiano antico. In Storia della lingua e filologia. Per Alfredo Stussi nel suo sessantacinquesimo compleanno, Michelangelo Zaccarello and Lorenzo Tomasin (eds.), 101–120. Florence: Edizioni del Galluzzo. Fleischman, Suzanne 1990 Philology, Linguistics, and the Discourse of the Medieval Text. Speculum 65: 19–37. 1991 Discourse Pragmatics and the Grammar of Old French: A Functional Reinterpretation of si and the Personal Pronouns. Romance Philology 44: 251–283. Givón, Talmy 2 2001 Syntax. An Introduction. Vol. I. Amsterdam /Philadelphia: John Benjamins.

On two functions of verb-subject order in Old Italian

265

Grice, H. Paul 1975 Logic and Conversation. In Speech Acts, Peter Cole and Jerry L. Morgan (eds.), 41–58 (Syntax and Semantics 3). New York /San Francisco/London: Academic Press. Hatcher, Anna Granville 1956a Syntax and the Sentence. Word 12: 234–250. 1956b Theme and Underlying Question. Two Studies of Spanish Word Order (Supplement to Word 3). New York: The Linguistic Circle of New York. Iulianella, Claudia 1994 The Nominative Pronoun and Word Order in 13th Century Italian: A Case Study of the Novellino. MA thesis, Department of Linguistics, University of Ottawa. Kaiser, Georg 2002 Verbstellung und Verbstellungswandel in den Romanischen Sprachen (Linguistische Arbeiten 465). Tübingen: Max Niemeyer. Koch, Peter and Wulf Oesterreicher 1985 Sprache der Nähe – Sprache der Distanz. Mündlichkeit und Schriftlichkeit im Spannungsfeld zwischen Sprachtheorie und Sprachgeschichte. Romanistisches Jahrbuch 36: 15–43. Lambrecht, Knud 1994 Information Structure and Sentence Form. Topic, Focus, and the Mental Representations of Discourse Referents (Cambridge Studies in Linguistics 71). Cambridge/New York / Melbourne: Cambridge University Press. Lombardi Vallauri, Edoardo 2004 Sintassi e informazione nell’italiano antico: l’oggetto preverbale. In SintAnt. La Sintassi dell’Italiano Antico. Atti del Convegno Internazionale di Studi (Università “Roma Tre”, 18–21 Settembre 2002), Maurizio Dardano and Gianluca Frenguelli (eds.), 293–321. (Studi linguistici e di storia della lingua italiana 5.) Rome: Aracne. Morgan, Jerry L. 1975 Some remarks on the nature of sentences. In Papers from the Parasession on Functionalism, Robin E. Grossman, L. James San, and Timothy Vance (eds.), 433–449. Chicago: Chicago Linguistic Society. Musa, Mark and Peter Bondanella 1983 Giovanni Boccaccio. The Decameron. Translated by Mark Musa and Peter Bondanella. New York /London: W. W. Norton & Company. Oesterreicher, Wulf 1991 Verbvalenz und Informationsstruktur. In Connexiones Romanicae: Dependenz und Valenz in den Romanischen Sprachen, Peter Koch and Thomas Krefeld (eds.), 349–384. (Linguistische Arbeiten 268.) Tübingen: Max Niemeyer.

266 Ludwig Fesenmeier 1997

Types of Orality in Text. In Written Voices, Spoken Signs: Tradition, Performance, and the Epic Text, Egbert Bakker and Ahuvia Kahane (eds.), 190–214. Cambridge, MA / London: Harvard University Press. Poletto, Cecilia n. d. La struttura del corpo della frase. In Grammatica dell’Italiano Antico, Giampaolo Salvi and Lorenzo Renzi (eds.). [www.geocities.com/ gpsalvi/konyv/index.html; last access: 02/12/2008.] Primus, Beatrice 1993 Word Order and Information Structure: A Performance-Based Account of Topic Positions and Focus Positions /Wortstellung und Informationsstruktur: Eine performanzbezogene Erklärung von Topikund Fokus-Positionen. In Syntax. Ein Internationales Handbuch Zeitgenössischer Forschung /An International Handbook of Contemporary Research. 1. Halbband / Volume 1, Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld, and Theo Vennemann (eds.), 880– 896. (Handbücher zur Sprach- und Kommunikationswissenschaft / Handbooks of Linguistics and Communication Science 9.1.) Berlin / New York: Walter de Gruyter. Raible, Wolfgang 1985 Nominale Spezifikatoren (“Artikel”) in der Tradition lateinischer Juristen oder Vom Nutzen einer ganzheitlichen Textbetrachtung für die Sprachgeschichte. Romanistisches Jahrbuch 36: 44–67. Reinhart, Tanya 1981 Pragmatics and Linguistics: An Analysis of Sentence Topics. Philosophica 27: 53–94. Renzi, Lorenzo 1988 Italiano antico e moderno: aspetti pragmatici a confronto. In Structure Thème-Rhème dans les Langues Romanes et Slaves, Wiesaw Bany and Stanisaw Karolak (eds.), 123–138. (Prace Slawistyczne 65.) Wrocaw/Warszawa/Kraków/Gdask/Lód: Zaklád Narodowy Imienia Ossoliskich Wydawnisctwo Polskiej Akademii Nauk. Salvi, Giampaolo 2004 La Formazione della Struttura di Frase Romanza. Ordine delle Parole e Clitici dal Latino alle Lingue Romanze Antiche. (Beihefte zur Zeitschrift für Romanische Philologie 323.) Tübingen: Max Niemeyer. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25: 511–580. 1995 “Theticity” and VS order: a case study. Sprachtypologie und Universalienforschung 48: 3–31. 1996 Theticity. (Arbeitspapiere – Neue Folge 27.) Köln: Institut für Sprachwissenschaft der Universität zu Köln. Schlobinski, Peter and Stephan Schütze-Coburn 1992 On the topic of topic and topic continuity. Linguistics 30: 89–121.

On two functions of verb-subject order in Old Italian

267

Selig, Maria 1992 Die Entwicklung der Nominaldeterminanten im Spätlatein. Romanischer Sprachwandel und Lateinische Schriftlichkeit. (ScriptOralia 26.) Tübingen: Gunter Narr. Sornicola, Rosanna 2000 Stability, Variation and Change in Word Order: Some Evidence from the Romance Languages. In Stability, Variation and Change of WordOrder Patterns over Time, Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds.), 101–115. (Current Issues in Linguistic Theory 213.) Amsterdam/ Philadelphia: John Benjamins. Stark, Elisabeth 1997 Voranstellungsstrukturen und topic-Markierung im Französischen. Mit einem Ausblick auf das Italienische. (Romanica Monacensia 51.) Tübingen: Gunter Narr. 2001 Textkohäsion und Textkohärenz. In Language Typology and Language Universals /Sprachtypologie und Sprachliche Universalien / La Typologie des Langues et les Universaux Linguistiques. An International Handbook/ Ein Internationales Handbuch /Manuel International. Volume 1/1. Halbband /Tome 1, Martin Haspelmath, Ekkehard König, Wulf Oesterreicher, and Wolfgang Raible (eds.), 634–656. (Handbücher zur Sprach- und Kommunikationswissenschaft/ Handbooks of Linguistics and Communiction Science/Manuels de Linguistique et des Sciences de Communication 20.1.) Berlin /New York: Walter de Gruyter. Stavinschi, Alexandra Corina and Martina Irsara 2004 Il sistema dimostrativo in alcune varietà italiane medievali: punti di riferimento e marcatezza. In SintAnt. La Sintassi dell’Italiano Antico. Atti del Convegno Internazionale di Studi (Università “Roma Tre”, 18 –21 Settembre 2002), Maurizio Dardano and Gianluca Frenguelli (eds.), 609–629. (Studi linguistici e di storia della lingua italiana 5.) Rome: Aracne. Strawson, Peter Frederick 1964 Identifying reference and truth-values. Theoria 30: 96 –118. Ulrich, Miorita 1985 Thetisch und Kategorisch. Funktionen der Anordnung von Satzkonstituenten am Beispiel des Rumänischen und Anderer Sprachen. (Romanica Monacensia 24.) Tübingen: Gunter Narr. Vanelli, Laura 1986 Strutture tematiche in italiano antico. In Tema-Rema in Italiano / Theme-Rheme in Italian /Thema-Rhema im Italienischen. Symposium, Frankfurt am Main, 26 /27-4-1985, Harro Stammerjohann (ed.), 249– 273. (Tübinger Beiträge zur Linguistik 287.) Tübingen: Gunter Narr.

268 Ludwig Fesenmeier 1999

Ordine delle parole e articolazione pragmatica nell’italiano antico: la “prominenza” della prima posizione nella frase. Medioevo Romanzo 23: 229–246. n. d. La deissi. In Grammatica dell’Italiano Antico, Giampaolo Salvi and Lorenzo Renzi (eds.). [www.geocities.com/gpsalvi/konyv/index.html; last access: 02/12/2008.] Venier, Federica 2002 La Presentatività. Sulle Tracce di una Nozione. (Gli argomenti umani 6.) Alessandria: Edizioni dell’Orso. Wandruszka, Ulrich 1982 Studien zur Italienischen Wortstellung. Wortstellung – Semantik – Informationsstruktur. (Tübinger Beiträge zur Linguistik 193.) Tübingen: Gunter Narr. Weigand, Edda 1979 Zum Zusammenhang von Thema/Rhema und Subjekt /Prädikat. Zeitschrift für Germanistische Linguistik 7: 167–189. Wicksteed, Philip H. (ed.) 2 1906 Villani’s Chronicle: being Selections from the First Nine Books of the Chroniche Fiorentine of Giovanni Villani. Translated by Rose E. Selfe and Edited by Philip H. Wicksteed M.A. Second Edition, Carefully Revised. London: Archibald Constable & Co. [www.elfinspell.com/ VillaniContents.html; last access: 02/16/2008.]

Between thematicity and grammaticalisation: The diachronic rearrangement of information structure and the position of clitic pronouns in Italian Shingo Suzuki

1. Introduction According to Bossong, the range of Italian clitics goes, in order of grammaticalisation, from strong pronouns to affixes: “Les clitiques se situent quelque part entre les ‘vrais’ pronoms et les ‘vrais’ morphèmes conjugationnels. Il ne sert à rien d’accumuler les arguments pour défendre leur appartenance exclusive à l’une ou à l’autre de ces catégories”. (Bossong 1998: 12) [Clitics stand somewhere between “real” pronouns and “real” conjugational morphemes. There is no point in building up arguments for or against their belonging to one or the other of these categories.]

Rather than putting clitics in the category of “real” pronouns or of “real” conjugational morphemes, what we have is a gradual scale from one prototype category to the opposite category, represented respectively by strong pronouns and by affixes. This process is thus understood as an evolving continuum. Bossong (1998: 12) further states that it would be useless to try to analyse a question like that of the clitics without recourse to diachrony and the presence of diachronic elements within synchrony. In this paper, we will investigate the rearrangement of the information structure within the declarative main clause which took place in the course of the history of the Italian language, with reference to the changes in the position of clitics and with particular emphasis on the position of the thematic subject (S) in relation to the verb. We will first examine the characteristics of the pronouns, bearing in mind that they evolve towards becoming conjugational morphemes, and highlight some of the similarities that exist between clitics and the implied S (section 2). Then, we will see that the syntactic mechanism which regulates the position of clitics in Old Italian is different from the mechanism in Modern Italian (section 3). Subsequently, we will verify that in Old Italian the S that stays behind inflected verbs is highly predictable on the basis of its context when it incorporates a

270 Shingo Suzuki continuous, wide-ranging theme (section 4). We will therefore demonstrate that there is a relation between the position of such a S and the assumed position of the S implied (section 5). Finally, we will express the view that the enclisis in medieval Italian is connected to the post-verbal position of strongly predictable S, whether phonically realised or not (section 6). Our conclusion will be that the diachronic rearrangement of information structure has contributed to the development of a tendency for clitics to be placed mainly before verbs, and at the same time to their gradual evolution towards the grammatical category of conjugational morphemes (section 7). 2. Two aspects of clitics: as pronouns and as affixes Before starting to investigate the diachronic process of rearrangement of the information structure, we need to verify if, as we may suppose, the clitics themselves have thematicity, from the point of view of their properties as pronouns. It is interesting to note that the Padua Group (Gruppo di Padova 1974: 156) had already suggested such a possibility in the 1970s, as clitics can be found at the beginning of a sentence even when normal intonation is maintained. Clitics were still considered to be full elements from a syntactic point of view, but, as far as I know, the Padua Group was the first to suggest the idea of their thematicity in the Italian language. In Suzuki (2001: 67), I have stated that, as mentioned in Givón (1989: 222–226), the less predictable the theme is, the more it tends to be placed on the left. Thus, for instance, both Modern Italian sentences in (2) contain a thematised element within the left-dislocated syntactic construction and are more or less appropriate answers to question (1): (1)

A chi dà questi fiori Giovanni? to whom gives these flowers G. ‘To whom does Giovanni give these flowers?’

(2)

a. Questi fiori, li dà a Maria. these flowers them.CL gives to M. ‘These flowers he gives to Maria.’ b. Il mazzo di rose lo dà a Maria e quello di gigli and that of lilies the bouquet of roses it.CL gives to M. a sua madre. to his mother ‘The bouquet of roses he gives to Maria and that of lilies to his mother.’

Between thematicity and grammaticalisation

271

Sentence (2a) is less appropriate than (2b) as an answer to the question in (1), as the direct object (O) questi fiori ‘these flowers’, in spite of being fully predictable, is in a redundant left-dislocated position. Sentence (2b), on the other hand, is more suitable as an answer because the O il mazzo di rose ‘the bouquet of roses’, being less predictable on account of the presence of the other term quello di gigli ‘that of lilies’, is correctly left-dislocated. In other words, when the theme is not easily predictable, it becomes more urgent to stabilise it. Conversely, the need to stabilise the theme diminishes markedly if it is totally predictable. This appears clearly in both answers (2), where the S Giovanni, being at the maximum level of predictability, remains unstated, or rather coded with zero theme. We must note that the zero degree can also concern the O. If we omit the predictable, and therefore redundant, left-dislocated O questi fiori ‘these flowers’ in sentence (2a), the answer becomes more appropriate to question (1): (3)

– A chi dà questi fiori Giovanni? [= (1)] – Li dà a Maria. them.CL gives to M. ‘– To whom does Giovanni give these flowers? – He gives them to Maria.’

We may consider that in the above answer, the O questi fiori is understood in the same way as the S Giovanni. In this case, we could also say the answer in (3) is coded with two types of zero themes, one for the S, the other for the O. However, in both cases the coding is not unconnected with the morpheme associated in some way with its “antecedent”. For the S, the associated morpheme is the ending of the conjugated verb dare ‘give’; for the O, it is the clitic li ‘them’, which, as we have already said, has properties that place it between conjugational morpheme and strong pronoun. In other words, we can compare the clitic either to the S that is implicit or to the conjugational morpheme which agrees with the S. In fact, when we put the clitic on the same plane as the unstated S, we tend to treat the clitic as a pronoun; if we compare it to the conjugation in agreement with the S, we are inclined to consider it as a conjugational morpheme. If we attempt to sum up the pronominal properties of clitics, from a semantic point of view, we see that “i clitici sono dotati di referenza, che può essere testuale in presenza di anafora o extratestuale in presenza di deissi [clitics are endowed with a reference which can be textual in the presence of anaphora or extra-textual in the presence of a deixis]” (Nocentini 2003:

272 Shingo Suzuki 274). We can therefore posit that thematicity, whether based on anaphora or on deixis, is absorbed in some way within these clitics. Keeping in mind their conjugational properties, from a syntactic point of view, we note that “i clitici italiani ricorrono in stretta contiguità col verbo o col gruppo verbale [Italian clitics recur in strict contiguity with the verb or verbal group]” (Nocentini 2003: 274). In most cases nowadays the clitics are found to the left of the verb, in proclitic position, and cannot therefore be detached from it, whereas full elements can be placed even further to the left, so much so that they may even become dislocated, or “marginalised”. We have posited that the less predictable the theme, the further to the left it is found. A counterproof of this is the fact that clitics, which must always be placed next to the verb, cannot be moved further to the left. As a matter of fact, when the theme becomes more predictable, it becomes less urgent to stabilise it. Although clitics are found on the left of the verb, they can only be adjacent to it, all the more so that the thematicity they absorb is of the highest grade of predictability. 3. The syntactic mechanism in the positioning of clitics in Old Italian Romance languages in the medieval period present a syntactic mechanism to regulate the position of clitics. In many respects, this mechanism differs from that which we find nowadays. Thus, in Old Italian, although clitics are placed adjacent to the verb, the proclitic position is not as dominant as it is today (see details in Mussafia [1886] 1983), as is evidenced in Florentine texts of the 13th century and the beginning of the 14th century. We can see this in the following examples from Il Novellino, written during the last decades of the 13th century, and from the Tuscan version of Il Milione, which dates from the first decade of the 14th century: (4)

Li cavalieri lo cercavano, […] e lo re Marco the knights him.CL sought … and the king M. n’andò in sul pino (N, p. 854 = LXV) LOC.CL-went up on-the pine.tree ‘The knights searched for him. […] King Mark climbed up the pine tree.’

Between thematicity and grammaticalisation

(5)

273

a. Ragunârsi le donne e andârne a’ gathered-REFL.CL the women and went-LOC.CL to-the senatori (N, pp. 856–857 = LXVII) senators ‘The women came together and went to the senators.’ b. Quando li due frategli videro che papa non si facea, when the two brothers saw that Pope NEG PASS.CL made mossersi per andarne al Grande Cane (P, p. 12 = X) moved-REFL.CL for to.go-LOC.CL to-the Great Khan ‘When the two brothers saw that no Pope was being made, they decided to return to the Great Khan.’

When the verb follows another full element, i.e. is not placed at the beginning of the sentence, as is the case for cercavano ‘(they) sought’ or andò ‘(he) went’ in (4) above, the clitics are found before the verb. This is similar to Modern Italian. However, when the verb is at the beginning of the sentence, the clitics are placed after the verb in enclitic position, as we can see in (5). In (5a) the verbs ragunârsi ‘(they) gathered’ and andâr ‘(they) went’ are found either at the very beginning of the sentence or after the coordinate conjuction e ‘and’ (this also applies to ma ‘but’ and o ‘or’); in (5b) mossersi ‘(they) moved’ is placed after a subordinate clause.1 All the examples have in common the fact that the verb appears at the beginning of a clause. In other words, in Old Italian, the choice of enclisis and proclisis is dictated by the position of the verb, whereas in Modern Italian it depends on the mood of the verb. According to Tobler-Mussafia Law, the main points of which we have just started to outline, in Old Italian clitics are not placed predominantly to the left of verbs in proclitic position as is the case in Modern Italian. At this point, we must ask ourselves why the alternative of an enclisis with a verb in initial position, which we have seen in (5) above, has been abandoned. Bossong calls it Alternative B in the following quotation: “La raison de l’abandon de l’alternative B) est indépendante de la clitisation; elle tient aux problèmes intrinsèques de la position initiale du verbe, variante positionnelle minoritaire dans une perspective typologique. La position initiale coïncide le plus souvent avec la position thématique; elle a tendance à se développer en position subjectale”. (Bossong 1998: 23) [The reason why Alternative B was abandoned has nothing to do with clitisation. It has to do with intrinsic problems connected to the initial position of

274 Shingo Suzuki the verb, which is a minority position in a typological perspective. The initial position coincides most of the time with the thematic position. It tends to develop into the position of the S.]

So, the loss of the enclisis has to do with the fact that the initial position tends to become the position of the S. In the following sections, we will examine how the position at the beginning of the clause has come to coincide with the position of the S in the course of the development of the Italian language, with reference to the changes in the position of clitics. Before that, however, we must verify another coincidence of the thematic S in Old Italian, viz. its seemingly optional coincidence with the post-verbal position of the main declarative clause. 4.

The subject in the post-verbal position in Old Italian

4.1. The high predictability of the subject in post-verbal position in Old Italian 4.1.1. From Bossong’s opinion quoted at the end of the previous section we can draw the following conclusion: clauses with the verb in initial position – with so-called “thetic” judgement2 – are certainly in the minority, even in Old Italian. In these clauses the word order is #V+S, which deprives them of an articulation in two parts (theme and rheme): (6)

Fue uno filosofo molto savio, lo quale avea nome Diogene was a philosopher very wise who had name D. (N, p. 856 = LXVI) ‘There was a very wise philosopher whose name was Diogenes.’

(7)

e cerca l’uomo la ruga per li piue netti mangiari and seeks the-man the street for the most fine foodstuffs e più dilicati (N, p. 808 = IX) and most delicate ‘And people seek out the street where the finest and most delicate foodstuffs are to be found.’

The word order is V+S in (6) and V+S+O in (7). In (6) the figure of Diogenes is introduced by the S uno filosofo molto savio ‘a very wise philosopher’ in the post-verbal position, whereas in (7) the generic S l’uomo ‘the

Between thematicity and grammaticalisation

275

man’ has become one with the preceding verb and forms a clause that presents a certain fact. Both (6) and (7) have in common a non thematic S, hence the theticity of the clauses. I have argued in Suzuki (2005: 242) that in medieval Romance languages, the pattern V+S…, represented in thetic clauses, constitutes a key structure for the main clause. Furthermore, I have posited that, in the main clause, the word order which is unmarked from the pragmatic point of view is also S+V+O, with the S in initial position, which, as Bossong said, coincides mostly with the thematic position. In contrast to the cases above, in Old Italian there are also clauses which have indeed the pattern #V+S but can only be considered as categorical rather than thetic. In the following examples, the S has full thematicity even though it is in a post-verbal position (see Arcangeli 2004: 58–59, who speaks of a mise en relief of the verb in (8a) and (9a) below): (8)

a. Andâr li ambasciadori went the ambassadors ‘The ambassadors went away.’

(N, p. 789 = II)

b. Ed è questa contrada molto di lungi d’India (P, p. 170 = CLVII) and is this country very far from-I. ‘And this country is a long way from India.’ (9)

a. Adomandò lo signore mariscalchi called.for the lord farriers ‘The king called for his farriers.’

(N, p. 800 = III)

b. Comandò questo giovane che fossero tutte quelle ordered this young.man that were all those genti menate dinanzi a lui (N, p. 807 = VIII) people brought before him ‘The young man ordered that all those folk should be brought before him.’ The order of the words in (8) and (9) is #V+S(…) and #V+S+O respectively, as it was in (6) and (7). Despite this, in all the sentences the S constitutes a theme, hence their non-theticity. In the (b) series, the continuity of the theme in the text is obvious since the S is accompanied by the demonstrative questo ‘this’, which reveals the presence of anaphora, without contrastive element. In the (a) series, the continuity is understood from the context:

276 Shingo Suzuki (10) Li ambasciadori fecero la dimanda loro, e videro li costumi e la corte. Poi, dopo pochi giorni, adomandaro commiato. Lo ’mperadore diede loro risposta, e disse: – Ditemi al signore vostro, che la migliore cosa di questo mondo si è misura. – Andâr li ambasciadori [= (8a)], e rinunziaro e raccontaro ciò ch’aveano veduto e udito […]. ‘The ambassadors asked their questions, and beheld the court and its customs. Then after a few days, they asked permission to return. The Emperor gave them his answer and said: tell your master that the best thing in this world is moderation. The ambassadors went away and related to their master what they had seen and heard.’ (11) Avenne un giorno che a questo signore fu appresentato delle parti di Spagna un nobile destriere di gran podere e di bella guisa. Adomandò lo signore mariscalchi [= (9a)], per sapere la bontà del destriere […]. ‘It happened one day that the king received from Spain the gift of a noble courser of great strength and perfect form. And the king called for his farriers that he might learn of the worth of the steed.’ The S li ambasciadori ‘the ambassadors’ (8a) and lo signore ‘the lord’ (9a), as well as questa contrada ‘this country’ (8b) and questo giovane ‘this young man’ (9b), are easily predictable since they have a fairly wide thematic range in each text, and as such have no need to be strongly thematised. It seems that, in Old Italian, the thematic S can remain behind the finite verb3 when the thematic continuity is stable and wide-ranging, and the predictability is relevantly high, as in (8) and (9). On the other hand, we postulate that the less predictable the thematic S, the more common it is to find it before the verb. Indeed, in such case it is more important to stabilise it as theme. Thus, in the following example, the two S are placed on the left, before the verb: (12) Li due fratelli li donarono delle gioie ch’egli aveano di the two brothers him.CL gave some jewels which-he had of gran quantità, e Barca re le prese volentieri (P, p. 6 = III) great quantity and B. king them.CL took willingly ‘The two brothers gave him some of the many jewels they had, and King Barka took them willingly.’ Here the two terms li due fratelli ‘the two brothers’ and Barca re ‘King Barca’ interfere with each other, being in contrast to each other. The reader

Between thematicity and grammaticalisation

277

then is unable to predict who gives the jewels and who takes them. In a similar way, the S li cavalieri ‘the knights’ and lo re Marco ‘King Mark’ in (4) are before the verb because of their minor predictability caused by a potential reciprocal interference. The situation does not change when the less predictable S is a pronoun: (13) a. Io andrò alle nozze, e tu al morto I will.go to-the wedding and you to-the dead.person (N, p. 862 = LXXV) ‘I will go to the wedding, and you to the funeral.’ b. Ma io farò così, ch’io andrò sopra loro e but I will.do in.this.way that-I will.go above them and maladicerolli; e tu darai la battaglia will.curse-them.CL and you will.give the battle (N, p. 829 = XXXVI) ‘But I will do in this way, that I will go unto them and will curse them, and you shall attack them.’ The two S io ‘I’ and tu ‘you’ in each example in (13) are in contrast to each other. Therefore each one, being less predictable, is urgently stabilised as contrastive theme to the left of the corresponding verb: andrò ‘(I) will go’ and andrai ‘(you) shall go’ in (a) (with an obvious ellipsis of the second verb), farò ‘(I) will do’ and darai ‘(you) will give’ in (b). However, we have not forgotten that we have earlier posited that from a pragmatic point of view the unmarked pattern is S+V+O even in Old Italian. If this is correct, as Benincà mentions in Grande grammatica italiana di consultazione (Renzi, Salvi & Cardinaletti [eds.], 129), the S placed before the verb in Old Italian must adapt to a much higher number of linguistic contexts or situations, and not just to a context of thematic contrast.4 This question will be examined in section 6. 4.1.2. At the beginning of the previous sub-section we have seen, in connection with examples (6) and (7), that in Old Italian we can find examples of pattern #V+S which are not articulated in theme and rheme. Among such sentences, from a thetic judgement, those that introduce a new S, as in (6) above, have an unpredictable S, as a matter of course. Such unpredictability of the S is kept even when another element X (“X” refers to any lexical element other than S) is placed before V+S, i.e. when the verb is in second

278 Shingo Suzuki position in the clause, with the pattern X+V+S. In the following examples, this element preceding the verb is an adverbial complement, al tempo di re Giovanni d’Acri ‘in the days of King John of Acre’, in (14), and an O, la terra ‘the land’, in (15): (14) Al tempo di re Giovanni d’Acri fue ordinata una campana at-the time of king G. of-A. was put a bell ‘In the days of King John of Acre a bell was hung.’ (N, p. 389 = LII) (15) La terra tengono li cristiani the land hold the Christians ‘The Christians govern the country.’

(P, p. 75 = LXXIII)

In these examples the first elements are thematic, whereas the S, shown in roman type, is part of the rheme in both sentences. Each S is introduced for the first time in the text, hence its complete unpredictability. Indeed, the sentence in (14) is the beginning of a new section, whereas the sentence in (15) is the beginning of a new paragraph. On the other hand, still on the S of sentences with the verb in initial position in Old Italian, we have mentioned in sub-section 4.1.1 that the S stays after the verb when, as in (8) and (9), its predictability is such that it does not require urgent thematisation. Furthermore, such a thematic S stays behind the verb even when the verb is in second position, i.e. when, in sentences somewhat similar to those in (14) and (15) (see, however, note 3), a lexical element different from the S is placed before the verb: (16) a. Ciò tenne il re a grande maraviglia this held the king at great surprise ‘The king was greatly surprised by this.’

(N, p. 800 = III)

b. E queste tavole dona egli a li 3 grandi baroni and these tablets gave he to the 3 great barons (P, p. 87 = LXXX) ‘And these tablets are given to the three great barons.’ (17) a. Così sostiene lo Grande Sire sua gente in.this.way supports the Great Lord his people ‘Thus the Great Lord supports his people.’ (P, p. 114 = XCVIII)

Between thematicity and grammaticalisation

279

b. e in questa battaglia avea egli la croce di Cristo and in this battle had he the cross of C. sulla sua insegna (P, p. 85 = LXXVIII) on-the his standard ‘And in this battle he had the cross of Christ on his standard.’ In (16), the elements at the beginning of the sentence are noun phrases which function as O, ciò ‘this’ in (a) and queste tavole ‘these tablets’ in (b); in (17), they are adverbs or adverbial phrases, così ‘in this way’ in (a) and in questa battaglia ‘in this battle’ in (b). The previous context makes it clear that the S in these examples are easily predictable (here we are dealing only with the context of (a); the predictability of the S egli ‘he’ in (b) is obvious from the fact that they are non-contrastive anaphoric elements): (18) Lo re mandò in Ispagna ad invenire come fu nodrito, e invenero che la destriera era morta, e il puledro fu nutricato a latte d’asina. Ciò tenne il re a grande maraviglia [= (16a)…]. ‘The king sent into Spain to learn how the horse had been reared, and heard that its dam having died, the foal had been reared on ass’s milk. The king was greatly surprised by this.’ (19) E questo fa [lo Grande Sire] la state. Lo verno fa cercare se ad alcuna gente muore sue besti’, e fae lo somigliante. Così sostiene lo Grande Sire sua gente [= (17a)] ‘This he (= “the Great Lord”) does in the summer. In winter if he finds any man whose cattle have died, he gives him some of his own. Thus the Great Lord supports his people.’ As we have said, in all examples in (16) and (17), all S placed after the verb are thematic; but this is not enough to determine thematicity. In fact, we can see that other thematic elements are also present. Among pre-verbal elements, even the O in (16) are thematic, and so is at least the adverbial phrase in questa battaglia ‘in this battle’ in (17b), which turns it into the “frame” of the sentence. Nevertheless, there appears to be a quantitative difference of predictability between these two groups, with their different positions. Pre-verbal elements, including così in (17a), are anaphoric only to a limited extent, and their thematicity – if they are indeed thematic – is only transitory, which as a result decreases the predictability. On the other hand, the post-verbal S constitute a more continuous theme throughout an

280 Shingo Suzuki entire section (in the case of (16a)) or through more than one chapter (in the case of (16b) and (17a–b)) and thus offer a more consistent predictability.

4.2. The thematic subject in the post-verbal position in the spoken language of the 13th century 4.2.1. Sornicola (2007) demonstrates that, from the point of view of textual functions, in Old Italian, the pattern #V+S and the pattern X+V+S with a clearly thematic and unfocalised S show a definite legacy of macro-textual function linked to Latin. Regarding such patterns, certain properties, rather than others, of textual functions of classical Latin or Biblical Latin have been selected according to cultural and rhetorical factors throughout time and space in diverse parts of the Romance world. Even in the Tuscan area we find certain stylistic properties inherited from Latin – more precisely from historical Latin texts, according to Sornicola. The thematic S stays in the post-verbal position, for the purpose of stylistic effects, when it has a textually durable predictability, with a fairly long thematic range. We can therefore estimate that the examples of #V+S in (8) and (9), and of X+V+S in (16) and (17), all of them from narrative texts, have the stylistic properties of the literary tradition of the written language. In this section, we will turn our attention from the narrative, which represents the written language, to the dialogues in direct speech, which simulate the spoken language and thus allow us, albeit indirectly, to see what it was like in Old Italian.5 In our examination of the dialogues, we will continue to limit ourselves to the main declarative clauses and will leave aside interrogative, volitive and exclamative sentences.6 4.2.2. There are few examples of direct speech in Il Milione, and I have found only one of the pattern X+V+S with thematic S: (20) Signor re, aguale puo’ tu bene vedere che tu non Sir king, now can you well see that you NEG se’ da guerregiare meco (P, p. 119 = CVIII) are to fight me-against ‘Sir King, now you can perceive that you are not the man to presume to make war against me.’

Between thematicity and grammaticalisation

281

Here, in the main clause, the S tu ‘you’ follows the inflected verb puo’ ‘(you) can’, which is in the second position in the clause. The dialogue starts with this remark addressed to the king. Therefore we cannot say that tu is anaphorically recoverable from the linguistic context, at least for the listener (the king). Nevertheless, the 1st and 2nd persons, singular and plural pronouns, belong to the category of personal deixis (see Salvi & Vanelli 2004: 323) and their referents, speaker and listener, must always be present in – or be recoverable from – the context of the statement. The speaker can therefore consider these pronouns as predictable from the extra-linguistic context, unless they are contrasted with another interchangeable element, even if this element is a personal pronoun (see examples in (13)). In other words, the post-verbal S tu is not only thematic but also predictable for the king, given its extra-textual presence. Il Novellino, on the other hand, contains several passages in direct style. I have found seven examples of the X+V+S pattern, but none of the #V+S pattern. Here are the seven examples: (21) a. Cotesto farò io voluntieri (N, p. 864 = LXXVII) that will.do I gladly ‘That will I do right gladly.’ b. del vino ti do io volentieri (N, p. 819 = XXIII) some wine you.CL give I gladly ‘I will gladly give you some wine.’ c. E così andomando io al Comune di Bologna che […] and so ask I of-the commune of B. that… ‘And so I ask of the commune of Bologna that…’ (N, p. 838 = L) d. Di costui avremo noi grande mercato (N, p. 847 = LXI) of this.man will.have we great bargain ‘This man will be an easy bargain for us.’ e. A questo nappo non porrai tu bocca7 (N, p. 819 = XXIII) to this cup NEG shall.put you mouth ‘You shall not set your lips to this cup.’ f. E sì se’ tu troppo migliore e più savio di me and so are you too better and more wise than me ‘And yet you are far superior to me and wiser.’ (N, p. 835 = XLIV) g. Dunque se’ tu la più forte [cosa] ch’io mai trovasse therefore are you the most strong thing that-I ever found (N, p. 858 = LXX) ‘You are therefore the strongest thing I have ever encountered.’

282 Shingo Suzuki Here too all post-verbal S are personal deictic pronouns: first person singular io in (a) to (c), first person plural noi in (d), second person singular tu in (e) to (g). Pre-verbal elements X have diverse syntactic functions but if we leave aside those in (c), (f) and (g), which are either adverbs or coordinate conjunctions, all are recoverable from the linguistic or extra-linguistic context. Those in (a) and (b) are recoverable from the linguistic context since on the one hand, cotesto ‘that’ in (a) is an anaphoric demonstrative, and on the other hand, del vino ‘some wine’ in (b) comes after the following context: (22) e avea [il poltrone] suo tamerice con vino, e suo mazzero molto pulito. Lo ’mperadore giunse e chieseli bere. El poltrone rispuose: – Con che ti dare’ io bere? A questo nappo non porrai tu bocca [= (21e)]. Se tu hai corno, del vino ti do io volentieri [= (21b)]. ‘(The countryman) had a tamarisk flask full of wine, and a nice clean loaf of bread. The Emperor came up and asked leave to drink. The countryman replied: with what should I give you to drink? You shall not set your lips to this cup. If you have a drinking horn, I will gladly give you some wine.’ As we can see, the recipient in (21b), i.e. the emperor himself, starts referring to ‘wine’ (vino). In both (21a) and (21b) an anaphoric reference to what has been referred to previously appears at the beginning of the sentence and becomes a theme, which it was not before. The remaining preverbal elements X, costui ‘this man’ in (d) and questo nappo ‘this cup’ in (e), are recoverable mainly from the extra-linguistic context. Naturally, ‘this man’ (costui) can be said to be the subject of discussion between the speakers in (d), and we can deduce the presence of a ‘cup’ (nappo) from the action of drinking which is mentioned in the passage that precedes (e) (see the quotation in (22)). But such anaphoric references are mere additions since ‘this man’ is in sight of the speaker in (21d) and the ‘cup’ exists in the scene in which it is named in (21e), with the help, perhaps, of a strategy of “riferimento ostensivo [demonstrative reference]” (Salvi & Vanelli 2004: 328–329). We therefore have good reason to consider that costui in (21d) and questo nappo in (21e) are interpreted mainly in the presence of deixis. Let’s go back to post-verbal S in (21) for a moment. We have seen that they come within the personal deixis, i.e. they belong to the category represented by first and second persons singular and plural pronouns. Now, these pronouns always refer to the participants in the act of communication; for this reason, they are predictable from the beginning of the statement

Between thematicity and grammaticalisation

283

and, especially when they have the syntactic function of S, they are destined to be themes without being indicated as such. This is why these S pronouns remain in post-verbal position without being further thematised each time another element X is placed before the verb. We have seen, through the examples in (a), (b), (d) and (e) of (21), that (pro)nominal elements X are pushed to the first position in the clause as they need to be marked as themes. If we turn our attention to the speaker of each statement, of particular interest are examples in (21b) and (21e), as the speaker is a poltrone (see (22)), i.e. a person of low birth and condition. Now, if we suppose that the person who uses the pattern under study here is, presumably, uncultured and has little knowledge of Latin, we can assume that Old spoken Italian – in other words, a language bereft of any connection with Latin literary tradition – had at least the pattern X+V+S with S as continuous theme, even though its use was limited to personal deictic pronouns. Besides Il Novellino and Il Milione, I have also conducted the same research on selected chapters of Tristano Riccardiano, where we can find a large number of dialogues (and occasionally monologues) spoken generally by high ranking characters. I have not been able to find even one example of the pattern #V+S with S as continuous theme, but I have found about forty examples with the pattern X+V+S where the S belongs, again, to the category of personal deixis.8 Here are some of them: (23) a. Ree Marco, questo non saprete voi da mee in King M., this NEG will.know you from me in nessuna maniera (T, p. 612 = LXXXV) no way ‘King Mark, there is no way you will ever find that out from me.’ b. Ora possiamo noi essere al sicuro da ogne cavaliere now can we be safe from every knight ‘Now we will be safe from every knight.’ (T, p. 633 = CXXXIV) c. Anzi voglio io combattere, che io non faccia mio messaggio rather want I to.fight than I NEG fulfil my errand ‘I would sooner fight than not fulfil my errand.’ (T, p. 594 = LXXV) d. Re Marco, in questo campo m’avete voi messo King M., in this camp me.CL -have you put (T, p. 659 = CCXXXIV) ‘King Mark, you have put me in this camp.’

284 Shingo Suzuki e. ma per altrui sono io venuto in questo dolore but for others am I come into this sorrow (T, p. 613 = LXXXV) ‘But because of other people I am led into this sorrow.’ Besides the S pronouns which belong to the category of personal deixis as in (23), I have found several occurrences of other types of pronouns (see (24a) and (24b)), as well as nominal expressions (see (24c)), which are undoubtedly thematic S: (24) a. Bene sono questi de’ colpi di Tristano truly are these some blows of T. ‘Truly these are Tristan’s blows.’

(T, p. 601 = LXXIX)

b. Dunqua averrà elli, se Dio piace (T, p. 660 = CCXXXVI) then will.happen it if God pleases ‘Then it will happen, if it pleases God.’ c. S’egli èe fatto oggi cavaliere novello, e domane sarae if-he is made today knight new and tomorrow will.be morto lo cavaliere novello (T, p. 566 = XVII) dead the knight new ‘If today he was newly made a knight, then tomorrow the newmade knight will be dead.’ In these examples, the reference to the S is based on deixis in (24a), on anaphora in (24b) and (24c).9 It is worth noting the pre-verbal element X domane ‘tomorrow’ in (24c), because it is contrasted with oggi ‘today’ in the subordinate clause and is placed as “frame” for the main clause. Sentence (24c) is said after the speaker has been told of the existence of Tristan, who has just been made a knight. The referent of lo cavaliere novello ‘the new-made knight’ is therefore already designated when the words in (24c) are pronounced. Thus the most predictable element, cavaliere, which is antonomastic to Tristan, leaves the first position to the less predictable element, domane, in order to stabilise it as the “frame” of the sentence. As we have noted previously, the thematic S follows the inflected verb, but in doing this it will come between an inflected verb and a possible nonfinite verbal form (see note 3). Indeed, the examples in (23b–e) follow this pattern with the verbs in compound forms (see also (20)): the inflected verbs – which are followed by the S – are either modals (23b–c) or auxiliaries (23d–e), and the non-finite forms – which follow the S – are either the infinitive (essere ‘be’ in (23b) and combattere ‘fight’ in (23c)) or the past

Between thematicity and grammaticalisation

285

participle (messo ‘put’ in (23d) and venuto ‘come’ in (23e)). However, this syntactic rule is not absolute as we also have examples where the thematic S does not precede but follows the non-finite verbal form, as in sarae morto lo cavaliere novello ‘the new-made knight will be dead’ in (24c).10 4.2.3. To conclude this section, we can say that even in Old spoken Italian the thematic S of a declarative sentence can be found after the verb, even if there are syntactic and lexical limitations seemingly depending on registers linked to the social environment. From the results of lexical observations, it appears that several types of expressions, either nominal or pronominal, were used for the S (see (24)), at least in educated speech which imitates the literary tradition of the written language. However, it seems that the closer the register was to the level of popular speech, the more restricted the sphere of lexical choice became, to the point that only pronouns of personal deixis were used, i.e. first and second persons, singular and plural pronouns (see (20), (21) and (23)). As for the syntax, I have found several examples of the pattern X+V+S, but I have not found even one example of #V+S in any of the dialogues I have examined. Considering that this pattern is definitely present in the narrative passages (see (8) and (9)), it is conceivable that it was used in educated speech and that its absence from the texts is pure coincidence. It must be noted, however, that even in narrative passages I have not come across one example of the pattern #V+S with a personal pronoun for S. From this we can infer that in both written and spoken Italian of the 13th century, when a declarative clause starts with a verb, the S pronouns – including first and second persons – do not appear in post-verbal position as long as they are thematic.11 Furthermore, we have seen that in the case of ordinary speech, the S pronouns used with the pattern X+V+S are mainly – or perhaps even exclusively – first and second person pronouns, even by speakers of fairly high socio-cultural rank. It would therefore be quite legitimate to suppose that had the #V+S pattern existed in ordinary and popular speech, the S would also have been represented by first and second person pronouns. However, it seems that, as we have noted, such cases were not found even in the written form of 13th century Italian. We should therefore accept that, at least as far as ordinary and popular speech is concerned, the above-mentioned absence of #V+S in the texts under study is not without reason.

286 Shingo Suzuki 5. The thematic subject in the post-verbal position and the unexpressed subject in Old Italian As we have seen earlier, in the ordinary speech of the 13th century, the S pronouns used with the pattern X+V+S are mainly pronouns of personal deixis, i.e. first and second person pronouns (25). In such pattern – the socalled “V2” syntactic structure, i.e. with the verb in second position – where the first position is occupied by an element other than the S, the unexpressed S pronoun can be said to be zero theme coded (Ø) and, although obviously not visible, would logically be in post-verbal position (26): (25) E dunqua ti diroe io lo mio volere (T, p. 609 = LXXX) and therefore you.CL will.tell I the my will ‘And therefore I will tell you my will.’ (26) Or ti dico [Ø] che se noi andiamo […] (T, p. 609 = LXXX) now you.CL tell.1SG Ø that if we go… ‘I tell you that whether we go…’ It is true that the predictability of S pronouns of personal deixis is generally expected because their referents are always present in the context of the statement. But it can happen that, in typical and special cases, the superficial position of such pronouns, whether expressed or not, reveals their degree of thematic continuity. Thus, if we examine the context that precedes sentences (25) and (26) – and is given here in (27): (27) Tristano, io so bene che ciascuno di noi èe cambiato, de l’amore che porta l’uno all’altro. E dunqua ti diroe io lo mio volere [= (25)]. Or ti dico [Ø] che se noi andiamo [= (26)] indelo reame di Leonois od in altra parte, là ove cavalieri od altra buona gente sappia nostri convenentri, egli diranno di noi tutta villania. ‘Tristan, I know well that we both love each other equally, and therefore I will tell you my will. I tell you that whether we go to the kingdom of Leonis or anywhere else, wherever knights or other good people know our history, they will say base things about us.’ we realise that the term io ‘I’ continues to be the theme from the beginning of the statement. In the first sentence, it is placed before the inflected verb so ‘(I) know’ and is thus stabilised as the theme. Its thematicity continues in the sentences that follow (25)–(26); this is why the first person singular S,

Between thematicity and grammaticalisation

287

whether expressed or not, remains in the post-verbal position. In fact, in declarative sentences with the pattern X+V+S of the ordinary and popular speech, the S as repeated theme – and therefore of high predictablity –, when different from pronouns of personal deixis, can be omitted rather than being expressed in pronominal form after the inflected verb (cf. Salvi, website, section 1.1, in particular the comment regarding his example (16): Allora lodò Ø il greco d’oltremirabile senno ‘Then praised [he] the Greek for his marvellous science’, though quoted from a narrative section of Il Novellino). On the other hand, with the #V+S pattern, even pronouns of personal deixis are excluded from the post-verbal position. As we shall see shortly, in spite of this, the written language as represented in the narrative sections provides us with enough evidence to formulate the hypothesis that even in declarative sentences with verbs in the initial position, the position of the unexpressed S pronoun is post-verbal and not pre-verbal. Let us then return to the narratives. In declarative sentences with the pattern #V+S where the S is a noun, as in (8) and (9), we shall hypothesise that if the context ensures the highest level of predictability for the S, this S will usually be omitted, being coded as degree zero (provided that sentences beginning with a verb are declarative sentences).12 The example in (28) below, as well as that in (5a) above, are convincing evidence in favour of our hypothesis: (28) Salio questo Barlam in su uno asino, e andò su a mounted this B. up on a donkey and went on to un monte (N, p. 829 = XXXVI) a mountain ‘This Balaam mounted his ass, and went up on to a mountain.’ Here, the predictability of the S questo Barlam ‘this Barlaam’ is made clear by the presence of the anaphora without a contrastive element; in (5a), the predictability of the S le donne ‘the women’ as the continuous theme is verified by the context: (29) La madre, che li avea promesso di tenere credenza, il manifestò a un’altra donna, e quella [a] un’altra. Tanto andò d’una in altra, che tutta Roma il sentì. Ragunârsi le donne e andârne a’ senatori [= (5a)], e doleansi molto. ‘The mother, who had promised him to keep the matter a secret, told the thing to another woman, who told it to yet another. Thus it went from one to another until all Rome knew of it. The women came together and went to the senators, and made great complaint.’

288 Shingo Suzuki The first clauses with the verb in initial position in (5a) (= ragunârsi le donne ‘the women gathered’) and in (28) (= salio questo Barlam… ‘this Balaam mounted’) both have a thematic S which stays after the verb because its predictability is assured by long-range thematic continuity as in (8) and (9). However, in the second clauses in (5a) (= andârne a’ senatori ‘[they] went to the senators’) and in (28) (= andò su a un monte ‘[he] went up on to a mountain’), the S is unexpressed because it is fully predictable from the anaphoric reference to the S expressed in the first clause. The S of the second clause in (5a) and (28) should be behind the verb, as can be inferred from the position of the S in the first clause. But being made even more predictable by the repeated reference, it loses its phonic identity, thus achieving zero theme coding: (30) a. Ragunârsi le donne e andârne [Ø] a’ senatori, gathered-REFL.CL the women and went-LOC.CL Ø to-the senators e doleansi [Ø] molto (N, pp. 856–857 = LXVII) and complained-REFL.CL Ø much ‘The women came together and went to the senators, and made great complaint.’ b. Salio questo Barlam in su uno asino, e andò [Ø] mounted this B. up on a donkey and went Ø su a un monte (N, p. 829 = XXXVI) on to a mountain ‘This Balaam mounted his ass, and went up on to a mountain.’ In other words, in Old Italian, every time the S in a declarative sentence is omitted, whether the sentence contains a pre-verbal element or not, this S is understood to be unexpressed in the post-verbal position (also mentioned in Vanelli, Renzi & Benincà 1985: 166–169), as shown by Ø in the following examples: (31) a. Diròvvi [Ø] de la provincia la quale à nome Lacca will.tell.1SG-you.CL Ø of the province which has name L. ‘I will tell you of a province whose name is Lac.’ (P, p. 220 = CCVI) b. Or vi dirò [Ø] d’una usanza ch’è in questa isola now you.CL will.tell.1SG Ø of-a custom that-is in this island (P, p. 169 = CLVII) ‘Now I will tell you of a custom that is in this island.’

Between thematicity and grammaticalisation

289

To sum up, when a declarative sentence starts with a verb, the S pronoun does not appear in the post-verbal position; in such case the fully predictable S is usually unexpressed, as in (31a). However, if any element is placed before the verb, the S pronoun can be found after the verb, as we have seen several times (see (16b) and (17b) for narrative texts, (20), (21), (23) and (24b) for dialogues), but it can also be unexpressed as in (31b) or in (26). When in a declarative sentence the S is totally predictable, it either can or must be zero theme coded. We have put forward the hypothesis that such coding is carried out in post-verbal position, and not pre-verbal position as would be the case in Modern Italian. 6. The position of the highly predictable subject in relation to the collocation of clitics during diachronic mutation We have seen several times that, in a main declarative clause, the S which constitutes a continuous theme, whether expressed phonically or not expressed due to zero theme coding, stays after the verb. We have therefore either a #V+S pattern or a X+V+S pattern. Based on Tobler-Mussafia Law, we can assert that in the first case we have an enclisis, as in diròvvi (31a), in the second case a proclisis, as in or vi dirò (31b). In Modern Italian, unlike Old Italian, the thematic S is always placed before the verb (S+V…), whatever its degree of predictability, provided that the sentence is pronounced with normal intonation (see Suzuki 2005). As a consequence, the unexpressed S is also considered to be placed before the verb. We must therefore admit, from a diachronic point of view, that at a certain time between Old Italian and Modern Italian a rearrangement of information structure took place, and this is probably the case in all other Romance languages too (see Renzi 1994: 267–275). We have seen that the more predictable and the more referentially continuous the thematic S was in Old Italian, the more likely it was to stay behind the inflected verb. Given that in Modern Italian the S is placed before the verb whatever its degree of predictability and thematic continuity, as long as it is a thematic S, there must have been a change in the position of a certain type of S from the right of the verb to the left at a certain point in linguistic chronology. This change is primarily connected to the position of the thematic S whose predictability is relevantly high due to a lack of intermingled referents: (32) a. Old Italian:

# dirovvi Ø

b. Modern Italian: # Ø vi dirò

290 Shingo Suzuki (33) a. Old Italian:

or vi dirò Ø

b. Modern Italian: ora Ø vi dirò If we limit ourselves to a comparison between the two patterns in (32), it would seem that the diachronic change of the position of the thematic S, shown here as zero theme, is connected to another important change, which is the position of the clitic. While the initial position in the sentence begins to coincide with the position of the thematic S, including the zero theme coded S, the frequency of such usage of the enclisis loses ground at the same rate. Today there is complete coincidence between the initial position and the position of the thematic S, whereas in Old Italian such coincidence was not absolute. It must be said that the process towards pre-verbal coincidence of the thematic S, which led it to occupy the first position in the sentence, irrespective of its referential continuity, has occurred over a fairly long period.13 On the other hand, it is also true that the process of the disappearance of the enclisis in declarative clauses has been long and gradual.14 In short, we can consider that both processes took a long time to develop and occurred in parallel. It is therefore possible to envisage a bilateral, rather than unilateral, influence of one process on the other. We can therefore conclude that this mutual influence has made each process both cause and effect of the evolution of the other.15 Let us now turn our attention to the patterns in (33). If we compare the pattern in (33a) with that in (32a), both of them from Old Italian, we notice immediately a difference in the position of clitic pronouns which fits ToblerMussafia Law. While in (32a) the clitic occurs just in front of the position of the thematic S shown as Ø, in (33a), on the other hand, it is detached from that position. In fact, when the pre-verbal element X in the pattern X+V+S acts as the theme (or "frame"), as del vino ‘some wine’ and in questo campo ‘in this camp’ in the examples shown in (34) – which we have already seen in (21b) and (23d) –, we can see that the clitic finds itself close to the position of a less continuous and less predictable theme rather than adjacent to the post-verbal S: (34) a. del vino ti do io volontieri some wine you.CL give I gladly ‘I will gladly give you some wine.’ b. in questo campo m’avete voi messo in this camp me.CL -have you put ‘You have put me in this camp.’

(N, p. 819 = XXIII)

(T, p. 659 = CCXXXIV)

Between thematicity and grammaticalisation

291

In other words, for the declarative sentence in Old Italian, the assignment of the theme is systematically split into two parts, one in the pre-verbal position, the other in the post-verbal position, depending on the semantic-pragmatic functions of each constituent based on its syntactic functions. We can then think that the alternative between enclitic and proclitic pronouns attached to the verb is connected to these two positions of the theme. In the case of the pattern X+V+S in (34), clitics ti ‘you’ and mi ‘me’ are placed closer to the thematised constituent X and, as a consequence, are distinguished from the post-verbal S io ‘I’ and voi ‘you’ by their position. As we can see, on the other hand, in example (33a), where it is difficult to decide if X (or ‘now’) acts as a theme (although we cannot exclude the possibility that it might be interpreted as a “frame”), should any element X be present in pre-verbal position, such element – even if not thematic – can substitute for the thematised constituent, and this irrespective of whether the thematic S in post-verbal position is expressed or not. Furthermore, the proclisis generally appears when the pre-verbal constituent coincides with an expressed S:16 (35) o io v’ucciderò tutt[i], o voi vi farete saracin[i] or I you.CL-will.kill all or you REFL.CL will.make Saracens (P, p. 29 = XXVI) ‘I will put you all to death; or else you will become Muslims.’ (36) noi il ci avemo guadagnato, non ci hai we it.CL REFL.CL have earned NEG LOC.CL have-2SG che fare (N, p. 838 = L) what to.do ‘We have earned this: it has naught to do with you.’ The S io ‘I’ and voi ‘you’ in (35) are themes, and the S noi ‘we’ in (36) is a rheme, but all are far from being predictable from the linguistic context. In the case of (35), the S appear unpredictable because they are contrasted with each other. It is this type of referential interference that pushes them to become stabilised as contrastive themes, and therefore to place themselves in pre-verbal position. We are dealing here with a “compound type” contrast similar to what occurs in Modern Italian (see Suzuki, 2001). In the case of (36), however, the unpredictability of the S depends on its quality of contrastive focus, which quality is only recoverable from the previous passage, since there is no way to have recourse to evidence from the intonation:

292 Shingo Suzuki (37) E quelli […] ritornò in sua terra; trovò li figliuoli ricchi. Adomandoe a’ suoi figliuoli che ’l rimettessero in su le possessioni, sì come padre e signore. I figliuoli negaro, dicendo così: “Padre, noi il ci avemo guadagnato, non ci hai che fare” [= (36)…] ‘He […] returned home. He found his sons rich. He asked them to reinstate him in his possessions as their father and lord. The sons refused, saying: father we have earned this: it has naught to do with you.’ In (36) we note a “syntagmatic contrast” (in the sense of the definition of Lombardi Vallauri 1998: 211) established between the referents of two similar types, or between the ‘sons’ (figliuoli) designated by noi ‘we’ and the ‘father’ (padre) designated by unexpressed tu ‘you’. Here, the contrasting structure is of the “simple type” (see Suzuki 2001), to which would today correspond the construction marked by “focalising anteposition”, with the contrasted element placed before and thus endowed with emphatic intonation (see Salvi & Vanelli 2004: 310, who call this construction “contrastive anteposition”; in the example given in (36’), which is “translated” into Modern Italian, the emphasis is shown in upper case): (36’) NOI, ce lo siamo guadagnato, non tu. we REFL.CL it.CL are earned NEG you ‘WE have earned it, not you.’ Unlike such syntactic markedness in Modern Italian, the construction in Old Italian which we see in (36) is formed on the basis of a structure with the verb in second position, the so-called “V2” structure, thus utilised in various contexts. Furthermore, we also find examples where the pre-verbal S constitutes a continuous and predictable theme just like the post-verbal S, but in such case the pre-verbal S produces a proclisis: (38) ma io vi dirò di tutte spezie insieme (P, p. 159 = CXLIX) but I you.CL will.tell of all spices together ‘But I will tell you about all the various spices.’ We can verify also that the S io ‘I’ in (38) is a long range, continuous theme by checking the context:

Between thematicity and grammaticalisation

293

(39) Or vi dirò de l’altre cose. In questa contrada nasce e favisi più zucchero che in tutto l’altro mondo; e questo è ’ncora grandissima rendita; ma io vi dirò di tutte spezie insieme [= (38)]. ‘Now I will tell you another thing. This province produces more sugar than all the rest of the world put together, and this too is the source of an immense revenue; but I will tell you about all the various spices.’ As we can see, at the beginning of the quotation is pattern (33a): or vi dirò. The S io in (38) has therefore a fairly long-range theme and as such need not have been urgently thematised. Thus, for sentence (38), it would have been possible to have pattern (32a): dirovvi, to the detriment of the phonic realisation of the S. In fact, in the Italian of the 13th century, when no element X appeared in first position, the S pronoun either was unexpressed (31a) or was placed before the verb (38), as it could not appear immediately after the verb. We will recall having stated that from a pragmatic point of view, the S+V+O pattern was already unmarked in Old Italian because it seemed to be suited to a larger number of linguistic contexts. Indeed, as we have seen in (38), even a thematic and fully predictable S could be placed before the verb rather than after (this was obligatory for S pronouns when no other element was in pre-verbal position). In fact, pre-positioning was permissible for various types of S, either thematic S, whether predictable or unpredictable (see examples in (35) as well as in (4), (12) and (13) for the second case), or rhematic S (see (36). On the other hand, we do not seem to be able to find examples of unpredictable thematic S that would be placed immediately after an inflected verb (and this is specially in evidence when the verb is in a compound form). We have stated that there must have been a rearrangement of information structure in Romance languages between the medieval period and the modern period. Seeing that with the passing of time, the S with a certain degree of predictability, or even extreme predictability, has come to be placed easily in pre-verbal position, we can say that this diachronic rearrangement is oriented towards the disappearance of the separation (illustrated in (34), which is just a realisation of the pattern we have already seen in (33a)) of the setting of the theme before and after the verb, or is oriented towards the pattern in (33b). We can therefore easily imagine that in the course of such rearrangement, a construction similar to the one seen in (38) may have served as model. We can formulate this idea in the following manner: (40) Old and Modern Italian:

io vi dirò

294 Shingo Suzuki In fact, we find that in Old Italian the assignation of the theme is unified in a single position and not split into two. This is due to the fact that the coding of the S in (40) is realised phonically through its move to a pre-verbal position, irrespective of the degree of continuity and predictability of its thematic properties. We therefore consider that it was the above pattern that acted as a model for the diachronic rearrangement which brought about a gradual but radical mutation in the construction of the main declarative clause from the patterns of Old Italian found in (32a) and (33a), in such a way that in the end the modern patterns in (32b) and (33b) both adopt the same word order: Ø vi dirò. Thus in these two patterns the thematic S (including zero theme) as well as the clitic come to be placed to the left of the verb, exactly as in the pattern in (40). In Modern Italian, theme assignment is no longer done on the right of the verb, and the enclisis cannot be placed inside the declarative sentence.

7. Conclusion: The rearrangement of information structure and the grammaticalisation of clitics We must now return to the words of Bossong which we have quoted in section 3: “La position initiale coïncide le plus souvent avec la position thématique; elle a tendance à se développer en position subjectale [The initial position coincides most of the time with the thematic position. It tends to develop into the position of the S]” (Bossong 1998: 23). The diachronic observation of the Italian language, the main points of which could perhaps be extended to all Romance languages, confirms Bossong’s assertion. In Modern Italian, we have total coincidence between the initial position and the position of the thematic S, including when the S is coded as zero theme. We can therefore say that the diachronic rearrangement of information structure in Italian was completed when the thematic S, including zero theme coded S, started being obligatorily placed before the verb irrespective of its predictability or referential continuity, as long as the sentence was spoken in a normal intonation. The predominance of the proclisis in Modern Italian is probably linked to the achievement of this rearrangement of information structure. In this paper we have seen that when, in Old Italian, constituent X placed before the verb is different from the S, the position of these two constituents, X and S, is split in two, one before and the other after the inflected verb. Here are two examples from our corpus, already shown in (23a) and (21b):

Between thematicity and grammaticalisation

295

Old Italian: (41) a.

Ree Marco, questo non saprete voi da mee in NEG will.know you from me in King M., this nessuna maniera [= (23a)] no way ‘King Mark, there is no way you will ever find that out from me.’

b. Se tu hai corno, del vino ti do io volentieri if you have horn some wine you.CL give I gladly [= (21b)] ‘If you have a drinking horn, I will gladly give you some wine.’ In each example in (41) both the O (= X) (questo ‘this’ in (a), del vino ‘some wine’ in (b)) and the S (voi ‘you’ in (a), io ‘I’ in (b)) are thematic and, at the same time, more or less predictable, since they are all anaphoric or deictic without referential interference. The first – with a fairly transitory and therefore less predictable theme – is placed before the verb, whereas the second – with a longer range and therefore more predictable theme – is placed after the verb. In Modern Italian, the S is placed preferably in the first position in the sentence. As a consequence, the O, which is less predictable and less continuous as a theme, will be placed even further to the left in peripheral position, being left-dislocated and as such being obligatorily resumed by a clitic: Modern Italian: (41’) a. Questoi, voi non loi saprete da me in nessuna maniera. you NEG it.CLi will.know from me in no way thisi b. Se tu hai un corno [per bere], del vinoi io te if you have a horn for to.drink some winei I you.CL loi do volentieri. it.CLi give gladly D’Achille (1990: 91–203) reports that the left dislocation, which appeared sporadically from the very beginning of the history of the Italian language, shows a steadily increasing development in the linguistic chronology of the Italian language, except in high level texts written after the beginning of the 16th century.17 In other words, the clitic has come to be used more and more frequently as a mark of grammatical agreement in the spoken language. Indeed, we have shown that, based on this type of grammaticalisation, the clitic has evolved towards the category of conjugational morpheme.

296 Shingo Suzuki Acknowledgements I wish to thank Professors Rosanna Sornicola and Edoardo Lombardi Vallauri, without whom the present paper could not have been completed. I especially would like to thank Prof. Sanda Reinheimer Rîpeanu, who meticulously read the manuscript and offered many valuable suggestions. I am also grateful to Prof. Alain Kregine for his help in bringing this paper to its present form in English.

Notes 1.

2.

3.

Renzi (1994: 279) states however that after a subordinate clause, the opposite order, where the clitics precede the verb, not only can be found as early as the 13th century but is also the more frequent order by far: (i) Quando voi togliete, si vuole sapere perché when you take.away IMPERS.CL wants to.know why ‘When you take away, it should be known why.’ (N, p. 820 = XXIV) Clauses without an articulation in two parts – which correspond to our notion of theme and rheme – are called “thetic”, whereas those that do feature such an articulation are called “categorical”. These terms were applied to linguistics by Kuroda (1972), who, recalling the theory of human judgement proposed by Franz Brentano and elaborated by Anton Marty, demonstrated a reflection of such a logical dichotomy in the structure of Japanese grammar. Poletto (webiste) asserts that in the case of “inversion” of the thematic S in Old Italian, the S follows the inflected verb but precedes the participle; however, when the S is rhematic, it follows the participle as well. In our examples below, the highly predictable S precedes the participle (as in fu Azzolino preso ‘Ezzelino was taken’, in (i)) but the unpredictable S follows it (as in fue ordinata una campana ‘a bell was hung’, in (ii)): (i) Poi fu Azzolino preso in battaglia (N, p. 871 = LXXXIV) then was A. taken in battle ‘Then Ezzelino was taken in battle’ (ii) Al tempo di re Giovanni d’Acri fue ordinata una campana at-the time of king G. of-A. was put a bell ‘In the days of King John of Acre a bell was hung’ (N, p. 839 = LII) In each example the verb is preceded by an adverbial (poi ‘then’ in (i), al tempo di re Giovanni d’Acri ‘in the days of King John of Acre’ in (ii)), but here the situation remains the same (see Marcantonio 1976: 61–66, 74, notes 7 and 8, where the author mentions the information values of the S in various sentences with the pattern [X+]V+S). In fact, in the text which comes before

Between thematicity and grammaticalisation

4.

5.

6.

297

the sentence in (i), the character ‘Ezzelino’ is already recorded as a continuous theme on several occasions; in (ii), however, the idea of ‘bell’ (campana) is introduced here for the first time. Sornicola (2004: 204) has compiled statistics of the relative order of the V and the S on samples of main clauses in ancient novels. The figures give the percentages of the occurrence of the order S+V and V+S with one-argument verbs and two-argument verbs. In the case of two-argument verbs found in Il Novellino and Il Decameron, the frequency of S+V was 80 %, with 20 % for V+S. Lombardi Vallauri (2004: 310), in an investigation on the relative order of the V and the O, has noted a definite variation according to the text under study. The frequency of pre-positioned O ranges from about 10 % in Il Novellino and Tristano Riccardiano to more than 50 % in Conti di antichi cavalieri. If we consider all these data on the basis of the so-called “V2” structure (i.e. with the verb in second position, as is usually the case in Medieval Romance languages), we can deduce that the word order O+V+S is less frequent than S+V+O, which lets us infer the absence of pragmatic markedness for the second pattern in Old Italian. In the same way, Lombardi Vallauri (2004), by making a distinction between literary and “practical” ancient texts, has obtained significant results regarding the thematic and the rhematic aspects of the pre-verbal O in 13th century Italian. Furthermore, we will omit cases of rhematic S as in the following example: (i) E lo ’mperadore rispuose e disse: – Io ti sodisfarò, quand’io tornerò. – Ed ella disse: – Se tu non torni? – Ed elli rispuose: – Sodisfaratti lo mio successore (N, p. 857 = LXIX) will.satisfy-you.CL the my successor ‘And the Emperor answered: I will give you satisfaction when I return. And she said: and if you do not return? To which he replied: my successor will give you satisfaction.’

7.

8.

9.

The post-verbal S of the last sentence in (i), shown in roman type, is unpredictable for the listener (here a widow) and rhematic at the same time. We will not take such cases into consideration. We consider that the future tense always belongs to the indicative mood (see Serianni 19912: 527). This is why I have included in my lists sentences with verbs in the future tense even though these verbs could have a jussive function as in (21e). Of these forty odd examples, a dozen constitute a variation of the stereotype, which can be represented by the following sentence (cf. (21a)): (i) Questo farò io volontieri (T, p. 568 = XVIII) this will.do I gladly ‘That I will gladly do.’ The predictability of the S in (24a) and in (24c) is obvious. As for the predictability of elli ‘it’ in (24b), see the statements in (i) which precede it (note as

298 Shingo Suzuki well that in the sentence that immediately precedes it, we find the same X+V+S pattern – or sono io ‘now am I…’ – with thematic io): (i) – Mia dolce dama, – disse Tristano – vorreste voi morire con meco? – Amico, – disse ella – sì m’aiuti Idio, unqua cosa nulla mai tanto disiderai. – Or, – disse elli – or sono io troppo lieto. Dunqua averrà elli, se Dio piace [= (24b)…] ‘ “My sweet lady”, Tristan said, “would you die with me?” “Friend”, she said, “with God’s help, there was nothing else I wished so much.” “Now”, he said, “I am very happy. Then it will happen, if it pleases God.”’ Here the S elli refers to the idea, already mentioned in the previous chapter, that the ‘sweet lady’ (dolce dama), namely Isolde, should die together with Tristan. 10. The same rule, with similar exceptions, is also valid for direct interrogative sentences (see Munaro, website, section 1.1.1). Generally speaking, the S precedes a possible non-finite verbal form, either in case of yes/no-questions (see (a) below) or WH-questions (see (b)): (i) a. Ed hami tue ferito? (T, p. 568 = XVIII) and have-me.CL you wounded ‘Have you wounded me?’ b. Per che cagione hai tue ispegnati tutti i lumi? for what reason have you put.out all the lights ‘For what reason have you put out all the lights?’ (T, p. 592 = LXVI) Nevertheless, it is not excluded that it could follow the non-finite verb as in the following example: (ii) Unde èe venuto questo sangue, ch’èe cosie fresco? from.where is come this blood that-is so fresh ‘Where did this blood come from, that is so fresh?’ (T, p. 583 = XLIV) As Munaro (website, section 1.1.1) explains, one of the reasons why the S follows the whole of the verb group can be found in the “heaviness” of the S. Indeed, the S in example (ii) is characterised by its relative length. We can therefore imagine that in the case quoted in (24c) too, use of a figure of speech, i.e. of antonomasia for Tristan, makes the S all the more “heavy”, despite its total predictability. 11. Such distribution exists also in other old languages outside the Tuscan region. See, for example, the following statement by Salvi concerning the French language: Era stato notato già da Franzén (1939) […] che, in francese antico, nelle frasi dichiarative a verbo iniziale (senz’altro introduttore o introdotte da et/ne) non compare mai un pronome soggetto postverbale. (Salvi 2004: 80) [Franzén (1939) has already noticed that, in Old French, a S pronoun never appears in the post-verbal position in declarative sentences starting with a verb (without any other introductory word or introduced by et/ne).]

Between thematicity and grammaticalisation

299

This distribution in the Italian language has been carefully researched by Palermo (1997: see especially 150–161), even if only in relation to the enclisis. The author has discovered several “eccezioni trascurabili [negligible exceptions]” (Palermo 1997: 157) which, among others, belong to texts written after Il Decameron, i.e. after the middle of the 14th century. We can quote from Il Decameron: (i) e maravigliomi io come egli non è ora qui (B, p. 349 = III, 3) and surprise-REFL.CL I how he NEG is now here ‘And I wonder why he is not yet here.’ Before the middle of the 14th century, we can find one exception in the Divine Comedy (I have found this example thanks to the research carried out by Vanelli, Renzi & Benincà 1985: 169): (ii) Poi mi rivolsi a loro e parla’ io (D, p. 67 = Inf., V, 115) then REFL.CL turned.1SG to them and spoke I ‘Then I turned towards them and spoke.’ This word order is however more probably due to poetic licence and the need for a rhyme. Indeed, the S io ‘I’ is clearly under rhythmic stress, although semantically thematic. 12. This hypothesis only concerns declarative sentences. In interrogative or volitive sentences, where the verb is at the beginning, the S pronoun can stay after the verb without being omitted even when it is completely predictable, as is the case with tu ‘you’ in the following examples: (i) Conosceresti tu tuo barlione? (N, p. 820 = XXIII) would.recognise you your flask ‘Would you recognise your flask?’ (ii) Va tu con l’oste tua (N, p. 830 = XXXVII) go you with the-army your ‘Go you with your army.’ 13. In fact, after the middle of the 15th century, i.e. when proclisis starts appearing also in first position (see note 14), we can still find, albeit in fairly small numbers, patterns which can be considered as remnants of an ancient pattern of the main clause: (i) Fu costui da mio padre onorato (M, p. 147 = I, 1) was this.man by my father honoured ‘This man was honoured by my father.’ The example in (i) is taken from a play by Niccolò Machiavelli (1469–1527). Here the S costui ‘this man’ is an uncontrasted anaphoric element and as such is predictable. The sentence starts with a verb; we can therefore consider that the pattern in (i) is similar to the old pattern of high level style which we have seen in (8), (9) and others. 14. We can see for instance the following explanation given by Antinucci and Marcantonio:

300 Shingo Suzuki Dunque, esaminando il complesso dei dati relativi alla posizione del clitico in frase principale, abbiamo un quadro abbastanza articolato, caratterizzato da un gradiente di prevalenza quantitativo e temporale della posizione preverbale a seconda dei contesti esaminati. Massimo nel caso in cui la principale è preceduta da una subordinata, seguito dal caso in cui la principale è introdotta da una congiunzione coordinante, seguito infine dal caso in cui il clitico è in inizio assoluto di frase. Se escludiamo […] quest’ultimo caso, va osservato che questo passaggio di posizione sembra effettuarsi nel corso del ’300. È solo alla fine del ’300 che troviamo infatti, da una parte, ormai generalizzata, la posizione preverbale nel contesto più avanzato, e dall’altra, numericamente significativa nel contesto costituito dalla congiunzione coordinante. (Antinucci & Marcantonio 1980: 31) [Thus, if we examine a collection of data on the position of clitics in the main clause, we find a fairly articulate frame, characterised by a quantitative and temporal gradation of the prevalence of the pre-verbal position depending on the contexts under consideration: maximal prevalence when a subordinate clause precedes the main clause, followed by cases where the main clause is introduced by a coordinate conjunction, and finally cases where the clitic is at the very beginning of the sentence. Excluding the last case, we see that the change in position seems to have occurred during the 14th century. It is only at the end of the 14th century that we find that the pre-verbal position is, on the one hand, widely used in the more advanced contexts, and, on the other hand, numerically significant in contexts constituted by coordinate conjunctions.]

To this, we wish to add that we find evidence, towards the middle of the 15th century, that clitics start being able to occupy the very first position in a sentence. Here is an early example from the letters of Alessandra Macinghi Strozzi (1407–1471): (i) Ne darò libbre cinquanta alla Ginevra PART.CL will.give.1 SG pounds fifty to-the G. ‘I will give Ginevra 50 pounds.’ (S, p. 90 = Lettera VII) 15. Palermo (1997: 159–160) identifies the same type of mutual influence between the loss of enclisis and the gradual increase in the frequency of S personal pronouns during the length of time that runs from the early period to the beginning of the 16th century. The increase in the number of S pronouns in declarative clauses is postulated in the pre-verbal position. On the other hand, as we have surmised by quoting as evidence several examples starting from (28), when, in declarative sentences in Old Italian, the fully predictable S pronoun is not expressed, the zero-coded S is always assumed to be in post-verbal position. The increase in the use of S pronouns logically implies a decrease of zero-coding for the S. Consequently, these two phenomena, which in reality are the two sides of a single phenomenon, are connected to the diachronic change in the position of the S from post-verbal to pre-verbal. This increase in S pronouns is therefore closely linked to the coincidence of the position of the S with the initial position, i.e. pre-verbal position. Consequently, we believe that the mutual influence of two changing processes which we have mentioned earlier – one towards the loss of enclisis, the other towards a pre-verbal coincidence of the thematic S – is broadly similar to the one presented by Palermo.

Between thematicity and grammaticalisation

301

16. In reality, there are extremely rare cases of unexpected enclisis but, as Cardinaletti says (website, see the comment on her example (92a) in section. 2.12), several such cases can be explained by referring to the peripheral position of the element in question (in our case, the S), which is “marginalised” to the left of the sentence: (i) Le gentii ch’erano intorno a ser Frulli, domandârlo [Øi] com’era the peoplei that-were around Mr. F. asked-him.CL Øi how-was.3SG ‘The folk who were standing around Messer Frulli asked him what it was about.’ (N, p. 877 = XCVI) Here the “heavy” S can be interpreted as being left-dislocated, which leads us to conclude that the enclisis depends on the syntactic properties of the rest of the sentence which retains its verb in initial position. 17. D’Achille’s analysis deals with Tuscan and non-Tuscan texts dating from the early period to the end of the 18th century. The left dislocations he studies include not only those connected with clitic copies but also with hanging topics (this distinction we owe to Cinque 1977).

Sources Giovanni Boccaccio, Decameron. Ed. by Vittore Branca. Torino: Einaudi, 19916. Dante Alighieri, La Divina Commedia. Ed. by Natalino Sapegno. Milano: Ricciardi, 1957. M Niccolò Machiavelli, Clizia, in Id., Teatro, 141–201. Ed. by Guido Davico Bonino. Torino: Einaudi, 1979. N Il “Novellino”, in La prosa del Duecento, 793–881. Ed. by Cesare Segre & Mario Marti. Milano: Ricciardi, 1959. [Translation by Edward Storer: Il Novellino: the hundred old tales. London: George Routledge & Sons Ltd, 1925.] P M. Polo, Il Milione. Ed. by Antonio Lanza. Roma: Editori Riuniti, 1981. [Translation by Ronald Latham: The Travels of Marco Polo. London: Penguin Books, 1958.] S Alessandra Macinghi Strozzi, Tempo di affetti e di mercanti: lettere ai figli esuli. Milano: Garzanti, 1987. T Tristano Riccardiano, in La prosa del Duecento, 555–661. Ed. by Cesare Segre & Mario Marti. Milano: Ricciardi, 1959. [English translation: Italian literature, vol. II: Tristano Riccardiano. Ed. by F. Regina Psaki. Cambridge: D. S. Brewer, 2006.] B D

References Antinucci, Francesco and Angela Marcantonio 1980 I meccanismi del mutamento diacronico: il cambiamento d’ordine dei pronomi clitici in italiano. Rivista di grammatica generativa 5: 3–50.

302 Shingo Suzuki Arcangeli, Massimo 2004 Strutture tematizzanti e ordine delle parole nella prosa narrativa toscana. Dal Novellino al Decameron: prove tecniche di variazione. In SintAnt: La sintassi dell’italiano antico, Maurizio Dardano and Gianluca Frenguelli (eds.), 33–63. Roma: Aracne. Bossong, Georg 1998 Vers une typologie des indices actanciels: les clitiques romans dans une perspective comparative. In Sintassi storica (SLI 39), Paolo Ramat and Elisa Roma (eds.), 9–43. Roma: Bulzoni. Cardinaletti, Anna (website) Il pronome personale obliquo. In Salvi and Renzi (eds.), part 2, IV/2. Accessed on 6th September 2006. Cinque, Guglielmo 1977 The movement nature of left dislocation. Linguistic Inquiry 8: 397– 412. D’Achille, Paolo 1990 Sintassi del parlato e tradizione scritta della lingua italiana: analisi di testi dalle origini al secolo XVIII. Roma: Bonacci. Franzén, Torsten 1939 Étude sur la syntaxe des pronoms personnels sujets en ancien français. Uppsala: Almqvist & Wiksell. Givón, Talmy 1989 Mind, Code and Context: Essays in Pragmatics. Hillsdale, NJ: Lawrence Erlbaum. Gruppo di Padova 1974 L’ordine dei sintagmi nella frase. In Fenomeni morfologici e sintattici nell’italiano contemporaneo (SLI 7), Vol. 1, tomo 1, Mario Medici and Antonella Sangregorio (eds.), 147–161. Roma: Bulzoni. Kuroda, S.-Y. 1972 The categorical and the thetic judgment: Evidence from Japanese syntax. Foundations of Language 9: 153–185. Lombardi Vallauri, Edoardo 1998 Focus esteso, ristretto e contrastivo. Lingua e stile 33 (2): 197–216. 2004 Sintassi e informazione nell’italiano antico: l’oggetto preverbale. In SintAnt: La sintassi dell’italiano antico, Maurizio Dardano and Gianluca Frenguelli (eds.), 293–321. Roma: Aracne. Marcantonio, Angela 1976 Un aspetto dell’ordine delle parole nell’italiano del Due-Trecento. Rivista di grammatica generativa 1 (2): 57–77. Munaro, Nicola (website) La frase interrogativa. In Salvi & Renzi (eds.), part 6, I. Accessed on 31st August 2006.

Between thematicity and grammaticalisation

303

Mussafia, Adolfo 1983 Una particolarità sintattica della lingua italiana dei primi secoli. In Scritti di filologia e linguistica, Antonio Daniele and Lorenzo Renzi (eds.), 290–301. Padova: Antenore. Original edition: Firenze, 1886. Nocentini, Alberto 2003 Evoluzione e struttura dei pronomi clitici in italiano. In Italia linguistica anno Mille, Italia linguistica anno Duemila (SLI 45), Nicoletta Maraschio and Teresa Poggi Salani (eds.), 273–284. Rome: Bulzoni. Palermo, Massimo 1997 L’espressione del pronome personale soggetto nella storia dell’italiano. Roma: Bulzoni. Poletto, Cecilia (website) La struttura della frase. In Salvi & Renzi (eds.), part 1, I/2. Accessed on 1st April 2006. Renzi, Lorenzo (in cooperation with Giampaolo Salvi) 1994 Nuova introduzione alla filologia romanza. Bologna: Il Mulino. Renzi, Lorenzo, Giampaolo Salvi, and Anna Cardinaletti (eds.) 2001 Grande grammatica italiana di consultazione, Vol. 1. Bologna: Il Mulino. Salvi, Giampaolo 2004 La formazione della struttura di frase romanza: ordine delle parole e clitici dal latino alle lingue romanze antiche. Tübingen: Max Niemeyer. (website) La realizzazione sintattica della struttura argomentale. In Salvi and Renzi (eds.), part 1, III. Accessed on 9th July 2006. Salvi, Giampaolo and Lorenzo Renzi (eds.) (website) Grammatica dell’italiano antico (http://geocities.com/gpsalvi/konyv). Salvi, Giampaolo and Laura Vanelli 2004 Nuova grammatica italiana. Bologna: Il Mulino. Serianni, Luca (in cooperation with Alberto Castelvecchi) 19912 Grammatica italiana: italiano comune e lingua letteraria. Torino: UTET. Sornicola, Rosanna 2004 Tendenze di lunga durata delle strutture mono-argomentali tra scritto e parlato: gli schemi di ordine VS nelle frasi principali del latino e delle lingue romanze. In La variabilité en langue, I: Langue parlée et langue écrite dans le présent et dans le passé, Rika Van Deyck, Rosanna Sornicola and Johannes Kabatek (eds.), 177–230. Gand: Communication & Cognition. 2007 Continuità e discontinuità degli ordini Verbo – Soggetto e loro permanenza nel genere storico tra latino e lingue romanze. In Actes du XXIVe Congès International de Linguistique et de Philologie Romanes, Aberystwyth, August 1–6, 2004, David Trotter (ed.). Tübingen: Max Niemeyer.

304 Shingo Suzuki Suzuki, Shingo 2001 I costituenti a sinistra e la contrastività in italiano antico e moderno. Archivio glottologico italiano 86: 57–78. 2005 L’ordine delle parole in italiano e romeno: ricerca della non marcatezza dal punto di vista pragmatico. In Gengo Joho-gaku Kenkyu Hokoku [Research on linguistic informatics] 7: 211–249. Tokyo: 21st Century COE, Center of Usage-Based Linguistic Informatics, Tokyo University of Foreign Studies. Vanelli, Laura, Lorenzo Renzi, and Paola Benincà 1985 Typologie des pronoms sujets dans les langues romanes. In Actes du XVIIe Congrès International de Linguistique et Philologie Romanes, Aix-en-Provence, 1983, Vol. 3: 161–176. Aix-en-Provence: Université de Provence.

Section 4 Pragmatically oriented languages

Information structure in Slavic languages Lucyna Gebert

1. Introduction The aim of this paper is to present the main strategies by means of which Slavic languages convey information, illustrating the distinction between the group of inflected languages such as Polish and Russian, on the one hand, and, on the other, Balkan Sprachbund languages such as Bulgarian and Macedonian, which have lost case declension. I will first briefly illustrate the interaction between pragmatics and word order in the two northern Slavic languages mentioned above, which have very similar grammatical systems. I should add that intonation also plays an important role within the information structure, but the prosodic dimension will be examined only marginally. I will then present segmental exponents of pragmatic function in Polish and Russian such as the deictics to/eto, which occur in sentences that may be regarded as a functional equivalent of cleft sentences. The subsequent section deals with a Russian construction known as the “nominative of the theme”, which is a rather unusual construction in languages with noun inflection. Indeed, similar constructions also occur in Bulgarian and require the reduplication of the dislocated argument by means of a clitic. The final part is devoted to the pragmatic strategies of the two Balkan Slavic languages, and especially to the interaction between clitic doubling and word order. 2. Word order in Polish and Russian There is a widespread notion that free word order in Slavic languages is a product of noun inflection, in so far as the latter guarantees that syntactic relations be maintained within the sentence.1 It is a known fact that word order in these languages is governed by pragmatics. Thus, in Russian and Polish, all the constituents of a declarative sentence with a two-argument verb can move freely, without any consequences on its syntactic structure; this gives rise to six combinations, as the following examples from Polish show:

308 Lucyna Gebert (1)

a. b. c. d. e. f.

SVO SOV OVS OSV VSO VOS

Jan spotka Mari Jan Mari spotka Mari spotka Jan Mari Jan spotka Spotka Jan Mari Spotka Mari Jan ‘Jan met Maria’

The pragmatic value of these sentences, if pronounced with an unmarked, descendent intonation, reflects the linearization of old and new information as formulated in the principle of “progression from old to new information” by Antinucci and Cinque (1977). The first of the examples above illustrates the basic word order, whereby either the first element alone or the first two elements are presented as old information. It is thus suitable for contexts in which the entire sentence conveys new information, namely at the beginning of a discourse. In Polish (but the same also holds for Russian), when word order is pragmatically marked, the topical constituents which represent old information occupy the left part of the sentence string. This is illustrated in (2), where the topical constituents are linearized first as SO and then as OS: (2)

Co si dzieje z Jankiem i z Mari ? what REFL happens with Janek and Maria ‘What’s up with Janek and Maria?’ a. Janek Mari rzuci Janek Maria.ACC has-left ‘Janek has left Maria’

SOV

b. Mari Janek rzuci Maria.ACC Janek has-left ‘Maria was left by Janek’

OSV

As for the other marked word orders, OVS realizes an information structure in which the topic is O; V and S represent new information and are positioned to the right of O, as in (3a); alternatively, S alone may represent new information as in (3b): (3)

a. Co z Mari? what with Maria ‘What about Maria?’

Information structure in Slavic languages 309

Mari odprowadzi Janek Maria.ACC accompanies Janek.NOM ‘Maria will be acompanied by Janek’

OVS

b. Kto odprowadzi Mari? ‘Who will accompany Maria?’ Mari odprowadzi Janek Maria.ACC accompanies Janek.NOM ‘Maria will be acompanied by Janek’ Finally, the two VSO sequences illustrated in (4) and (5) are appropriate in contexts in which V alone expresses old information (as in 5), or in which the sentence as a whole conveys new information (as in 4), as in answers to a question of the ‘What’s up?’ type: (4)

Co nowego? what new.GEN ‘What’s up?’

Wygra Janek rower has-won Janek bike ‘Janek has won a bike’

(5)

a. Czy kto co przyniós ? Q somebody something has-brought ‘Has anybody brought anything?’ Przynios a Maria pyszny tort has-brought Maria delicious cake ‘Maria has brought a delicious cake’

VSO

VSO

The answer to the question in (5a), where only V is old information, may also take the form of a VOS sentence, in accordance with the above mentioned “new information progression principle”: (5)

b. Czy kto co przyniós ? Q somebody something has-brought ‘Has anybody brought anything?’ Przynios a pyszny tort Maria has-brought delicious cake Maria ‘Maria has brought a delicious cake’

VOS

As far as VOS is concerned, both V and O (or V alone) can represent old information:

310 Lucyna Gebert (6)

Kto widzia Janka? who has-seen Janek.ACC ‘Who has seen Janek?

Widzia a Janka Maria has-seen Janek.ACC Maria Maria has seen Janek’

Verb-inital orders2 are probably less frequent in Polish than the other orders, although Siewierska and Uhliová’s survey of written Polish (1997: 123) shows that VSO is 6.5% of the total and VOS is even more frequent: 9.5%. In any case, they appear to be more frequent than both SOV (2.4%) and OSV, which only represents 1.5% of the total. SVO is by far the most frequent order (72,5%), and OVS ( 7,4%) comes third on the scale of frequency. The above data is particularly interesting in that it offers a representation of intonation-free sentences from the written language.3 New information in Slavic languages can be also marked by contrastive intonation. The example in (7) shows that the focused element can receive intonation prominence both in situ (as in 7a) and when it is postposed to the right of V (as in 7b): (7)

Kto dzwoni do drzwi? ‘Who rang the bell?’ a. MARIA dzwoni a, ale potem przypomnia a sobie e ma klucze. Maria rang but then remembered REFL that has keys ‘MARIA rang, but then she remembered she had the keys’ b. Dzwoni a MARIA, ale potem przypomnia a sobie, e ma klucze rang Maria but then remembered REFL that has keys ‘MARIA rang, but then she remembered she had the keys’

3. Segmental markers. “Embryonic” clefts in Russian and Polish Traditionally, Slavic languages are described as languages which make no use of segmental means of information distribution.4 Huszcza (1980, 2000) drew attention to the fact that colloquial Polish does employ focalizing and topicalizing segmental means. Indeed, both Russian and Polish use the invariable deictics (PART) eto and to as shown in (8-10), in which the elements positioned to the right of the deictics carry the focal accent. Such sentences constitute a functional equivalent of the cleft constructions that are found in several Western-European languages: (8)

Russian Eto Ivan pri el ko mne i rasskazal vse PART Ivan has-come to me and has-told everything ‘It is Ivan who has come and has told me everything’

Information structure in Slavic languages 311

(9)

Polish To Maria nie ma racji PART Maria NEG has right ‘It is Maria who is not right

In Russian the appropriate answer to question a. is sentence b.: (10) a. Marija ne prava ili Anna? Marija NEG right or Anna ‘Is it Marija who is not right or is it Anna? b. Eto

Marija ne prava PART Marija NEG right ‘It is Marija who is not right’

To/eto may be followed by NPs, as in (8–10), but also by adverbial constituents: (11) Polish To dzisiaj Jan mia zdawa egzamin PART today Jan had to-take exam ‘It is today that Jan should have taken the exam’ Russian Eto segodnja Ivan dolen byl sdavat’ ekzamen PART today Ivan should AUX to-take exam ‘It is today that Ivan should have taken the exam’ as well as by infinitive verbs: (12) a. Polish To zdawa egzaminów Jan nie lubi a nie uczy si do nich PART to-take exams-GEN Jan NEG like and NEG to-study for them ‘It is taking exams that Jan does not like, not preparing them’ b. Russian Eto sdavat’ ekzameny Ivan ne ljubit a ne gotovit’sja k nim PART to-take exams.ACC Ivan NEG like and NEG to-prepare to them ‘It is taking exams that Ivan does not like, not preparing them’ In all of the constructions reported in (8 –12), the constituents which follow to/eto are under contrastive focus.

312 Lucyna Gebert In line with the Slavic descriptive tradition, Miller (2007: 203) points out that whereas cleft constructions are typical of Western European languages, they are instead absent in the languages of Eastern Europe. He does however recognize a sort of “embryonic” cleft status for Polish and spoken Russian (Miller 2007: 205) and advances a tentative hypothesis for assigning a twoclause structure (characteristic of clefts) to sentences such as (8) above: 1. eto + zero copula and 2. Ivan pri el ( Miller 2007: 188). Accordingly, to/eto may be interpreted as an expletive subject in a copular construction, analogous to the underlying structure of “regular” clefts,5 although what is still lacking is the relative clause that follows the so-called ‘Small Clause’, which is typical of clefts. A further analogy with clefts lies in the fact that the deictics to/eto cannot occur in front of a finite form of the verb, as widziski (1978) and Huszcza (2000) have pointed out. This can be seen in (13): (13) Polish *To nie ma racji Maria PART NEG has right Maria *To mia zdawa dzisiaj Jan egzamin PART had to-take today Jan exam *To spotkalimy ich w zesz ym roku na wakacjach PART we-met them in last year on holiday Russian *Eto ne PART NEG

pravà Marija right Marija

Such syntactic behaviour, is in line with the behaviour exhibited by clefts in Western European languages, where it follows naturally from the analysis in terms of a copular clause followed by a relative clause (see, among others, Frascarelli 2000). The same segmental marker, followed by a pause (graphically signalled by a hyphen), can be preceded by the topical part of the sentence and can obviously be followed by new information6: (14) Russian Kto takaja Marija? ‘Who is Marija?’ Marija – eto (byla) ena moego brata Marija PART (was) wife my.GEN brother.GEN ‘Marija is (was) my brother’s wife’

Information structure in Slavic languages 313

(15) Polish Maria to (by a) ona mojego brata Maria PART (was) wife my.GEN brother.GEN ‘Maria is (was) my brother’s wife’ In Russian, these constructions exclusively occur in sentences with a nominal predicate, whereas in Polish any kind of constituent may be placed to the right of to when it is preceded by topical elements, as (15), above, and as the following very common examples of colloquial Polish show (see Huszcza 2000): (16) a. Jan to kupi samochód Jan PART has-bought car ‘As for Jan, he has bought a car’7 b. Samochód to mam swój PART I-have own car ‘As for the car, I have my own’ For this very reason Huszcza regards the two possible occurrences of to (the focus to and the one that separates old and new information) as two distinct homophonous particles. He offers syntactic evidence in favour of this hypothesis showing that the focus to cannot be followed by a verb, as in (13), unlike the separator to, as can be seen in (16a–b); in these sentences, to separates the topic from the broad focus that bears on the VP. This poses a problem in terms of treating the two constructions in a unitary way. 4. Russian ‘nominative of the theme’ One of the distinguishing characteristics between Russian and other northern Slavic languages consists in its strong tendency to analytism. This property of Russian has been pointed out by a number of linguists (Glovinskaja 1996; Il’ina 1996; Comrie, Stone and Polinsky 1996) and it also affects the information structure level. It is manifested, among other things, by a considerable weakening of the case system in favour of constructions built on prepositions. According to some linguists (see, for instance Lapteva 1976, 1997), this phenomenon has led to the expansion of the construction known as “imenitel’nyj temy” (‘nominative of the theme’), which is particularly frequent in the colloquial variety of spoken Russian: the nominative case, typical of subject NPs, also marks topical NPs which are not subjects. This is illustrated by the following examples:

314 Lucyna Gebert (17) a. Vy ne videli belaja sobaka? you NEG saw white.NOM dog.NOM ‘You haven’t seen a white dog, have you?’

(Lapteva 1976: 158)

b. Gde pota? Pota na avtobuse nado echat’ where post-office post-office.NOM by bus one-needs go ‘Where is the post office? To get to the post office you need to go by bus’ (Lapteva 1976: 161) c. Televizory, v etom magazine ich mnogo they.GEN many TVs.NOM in this shop ‘As for TVs, there are many in this shop’ (Lapteva 1976: 158) d. Ta kotoraja zdes’ stojala lampa, ja ee ne bral that which here stayed lamp.NOM I it.GEN NEG took ‘As for the lamp that was here, I haven’t taken it’ (Lapteva 1976: 144) The NPs in question are, as it were, ‘appended’ to the sentence, on its left (17b) and right (17a) side and have no syntactic link with the sentence itself, whereas in the examples (17c–d) a left-dislocated topic, separated by a pause (a comma), is syntactically linked with the sentence by resumptive pronouns. Examples such as (17c–d) thus exhibit the properties of the ‘hanging topic’ (nominativus pendens) that is common to both Romance languages constructions such as the Italian ‘tema sospeso’ (see, for instance, Mereu 2004: 76) and to Bulgarian, a Slavic language with no noun inflection (see Krapova and Cinque, in print).

5. Slavic languages of the Balkan Sprachbund It is an established fact that two Slavic languages of the Balkan Sprachbund, Bulgarian and Macedonian, underwent a convergent evolution with the other languages of that area, such as Romanian, Albanian and Greek8, so that these two Slavic languages display quite different grammatical features as compared with the other languages of the Slavic family: in particular, they innovated their nominal systems, losing case inflection and grammaticalizing the definite article. Given the widespread opinion that a pragmatically conditioned free word order in Slavic languages is connected to noun declension, it is particularly interesting to see what happens in Bulgarian and Macedonian, which have both lost it.

Information structure in Slavic languages 315

5.1. Bulgarian In Bulgarian, word order in simple declarative sentences is flexible and governed by pragmatics (see Guentchéva 1994; Moldovan et al. 2005). In sentences with a one argument verb, the basic order is SV, as in: (18) a. Vojnicite dojdoxa soldiers-the have-arrived ‘The soldiers arrived.’ although the postposition of the subject (VS) is also acceptable, as in (18 b): (18) b. Dojdoxa vojnicite have-arrived soldiers-the ‘The soldiers arrived.’

(Guentchéva 1994: 107)

Descriptions of Bulgarian tend to make the claim that the unmarked word order of this language is SVO and that such order has a grammatical function. It is, in fact, relevant for marking syntactic relations in sentences where neither verb agreement nor verb semantics allow for the recovery of such relations (see Siewierska and Uhliová 1997). Therefore a sentence such as the following: (19) Marija napisa pismoto Marija has-written letter-the ‘Marija has written a letter’ does not create any problem of interpretation, given the semantic nature of its arguments. However, the interpretation of the following SVO sentence: (20) a. Ivan obia Marija ‘Ivan loves Marija’ is problematic; if word order were inverted and became OVS, as in: (20) b. Marija obia Ivan Marija loves Ivan the interpretation of (20b) would necessarily be that ‘Maria loves Ivan’ (as in English), since in this kind of sentence the topical NP, placed in first position, is associated with S. Thus, if one wants to topicalize the object NP by

316 Lucyna Gebert placing Marija in the first position, such choice is manifested on the structural level by a resumptive clitic pronoun, coreferential with the dislocated NP, exactly as in Romance languages. Both in Romance and Balkan-Slavic, personal pronouns maintain case inflection. This is illustrated in (20 c): (20) c. Marija ja obia Ivan Marija her loves Ivan ‘As for Maria, Ivan loves her.’ There is a general agreement that the doubling of an argument by means of a clitic pronoun occurs in Bulgarian when the associate NP is both topical and referential (see Rudin 1997 in Franks & King 2000: 53). However, there are exceptions to the referentiality condition, as can be seen in (21), where the doubled NP has a generic value: (21) Uvaavat go edin u itel zaradi vseotdajnostta mu respect.3PL him.ACC one teacher for devotion.DEF his ‘A teacher is respected by his devotion.‘ (Franks & King 2000: 53) One more exception to the above rule of clitic doubling is constituted by sentences in which the associate NP is an interrogative pronoun and represents the focus of the sentence, as in (22): (22) Kogo kazva sa go uvolnili? who.ACC say.2SG aux.3PL him.ACC fired ‘Who do you say they fired?’ (Franks & King 2000: 53) According to Franks and King, these problematic and hardly accountable examples might be due to a process of language change that is still in progress.9 Another contradictory features of Bulgarian is the construction in which the doubling pronoun can be referred both to the topic, as in (23a): (23) a. Kruata ja risuva deteto pear-the it draws child-the ‘The pear is being drawn by the child’ and to the contrastive focus, as in (23b): 10 (23) b. KRUATA ja risuva deteto pear-the.F her is-drawing child-the ‘It is the pear the child is drawing.’

(Guentchéva 1994: 108)11

Information structure in Slavic languages 317

This data seems inconsistent with the rule of clitic doubling mentioned above. It could perhaps be accounted for in the light of Mereu’s observation (2004: 170) that a certain degree of equivalence does hold between contrastive topic and contrastive focus in languages.12 Obviously this phenomenon requires more in-depth study for what concerns Bulgarian, given also that interpretations of the available data (see footnote 12) are rather contradictory. In a recent paper, Krapova and Cinque (in print) show that clitic doubling in Bulgarian is not a unitary phenomenon. It applies to “five separate cases, with clearly distinct properties ((1) Hanging Topic, (2) Clitic Left Dislocation, (3) Focus Movement, (4) Clitic Right Dislocation, (5) Clitic Doubling proper)” (:1). These five constructions are exemplified under (24 a–e): (24) a. (1) Toj ne mogat da go prikrepjat kam nikogo he NEG can.3PL PART him attach.3PL to nobody ‘Him, they cannot attach him to anyone’ (Krapova & Cinque 2007: 2) b. (2) Marija nikoj ne ja obia Maria nobody NEG her loves ‘(As for Maria), nobody loves her’ (Krapova & Cinque 2007: 3) c. (3) Samo na Ivan ne mu se spe e only to Ivan NEG him.DAT REFL was-sleeping ‘Only Ivan didn’t feel like sleeping’ (Krapova & Cinque 2007: 6) d. (4) Poznavam go tova uvstvo I-know it this sentiment ‘I know this sentiment’

(Krapova & Cinque 2007: 10)

e. (5) Filmt *(mu) xaresa dori na Ivan film-the him.DAT appealed even to Ivan ‘Even Ivan liked the film’ (Krapova & Cinque 2007: 8) The authors specify that case (3) constructions are only acceptable with a certain number of predicates that are typical of case (5), namely with psychological and physical perception predicates. Furthermore, clitic doubling proper also occurs with tonic pronouns and is favoured by certain other parameters (such as indirect object status, animacy and definiteness). Given that in the constructions that fall under case (5), doubled NPs can also carry new information, as in (24e), obligatory reduplication in these cases represents a further exception to the topicality condition.

318 Lucyna Gebert 5.2. Macedonian In Macedonian, the other Slavic language of the Balkan Sprachbund to have lost case declension, pronominal doubling is not related to pragmatically marked word order. Word order variation in this language is governed by pragmatics (see Friedman 1994a) and the basic order is SVO. Clitic doubling is obligatory when argument NPs are referentially identified, independently of their informative status and the word order of the sentence. This can be seen in (25a), which displays an unmarked SVO order: (25) a. Marija go poznava uenikot / Vlado /toj u enik / nego Marija him knows student-the / Vlado / this student / him ‘Maria knows the student/Vlado/this student/him’ (Berent 1980: 151) whereas the sentence is not acceptable without clitic doubling of the object NPs, as in (25 b): (25) b. *Marija poznava u enikot / Vlado / toj uenik / nego Marija knows student-the / Vlado / this student / him Moreover, clitic doubling can also occur (although not very frequently) with indefinite NPs when the latter are referential, as illustrated in the following sentences in (26): (26) a. Ja vidov enata her I-have-seen woman-the ‘I have seen the woman’ b. (Ja) vidov edna ena her I-have-seen one woman ‘I have seen a (certain) woman’ c. Vidov edna ena I-have-seen one woman ‘I have seen a woman’

(Topoliska 1999: 47)

In Macedonian, pronoun reduplication appears to be related to pragmatics only in constructions such as (27), in which a personal pronoun doubles the strongly topicalized subject NP:

Information structure in Slavic languages 319

(27) Toa mom eto e dojdeno he boy-the is arrived ‘As for the boy, he has arrived’

(Topoliska 1999: 83)

Given that in Macedonian clitic doubling is not only grammaticalized when referred to definite NPs, but also to all the referentially identified NPs, it also occurs when such doubled NPs fall under the scope of contrastive focus, as in (28): (28) Mu i goi i dadov pismencetoi i nemu i , a ne nej e him.DAT it.ACC gave.1SG note.DEF him.DAT and NEG her.DAT ‘I gave the note to him, not to her’ (Frank & Kings 2000: 71) as well as when referred to wh-words, as in (29): (29) Koe maei najmnogu * ( goi ) saka? which kitten most it.ACC like.2SG ‘Which kitten do you like most?’

(Franks & King 2000: 72)

5.3. Balkan-Slavic clitic doubling Examples such as the sentence in (29) are reminiscent of the ones from Bulgarian in (22) and (23b). Together with the data of case (5) presented by Krapova and Cinque, the sentences in (22) and (23b) also show that the interpretation of clitics in Bulgarian as pronouns whose function is to resume topical arguments does not hold for all the occurences of this construction. As already mentioned (see the comments concerning (22)), the phenomena which this data illustrates can only be accounted for if it is framed as a process of desemantisation which is still in progress. Said process seems to be moving in the direction of a condition in which doubling clitics, deprived of the pronominal status they had in pragmatically motivated sentences, grammaticalize as loose morphemes with the function of conveying object inflection. The data above shows that Macedonian is closer to this final stage than Bulgarian 13, and the same may be claimed of the other Balkan languages such as Albanian and Romanian (see Friedman 1994 b: 109 and Heine & Kuteva 2005: 192–194). For this reason, there is a scholarly tradition which treats Balkan clitic doubling morphemes in terms of grammatical agreement rather than pronominal clitics (Cychun 1969; Rudin 1996, 1997; Franks & King 2000,

320 Lucyna Gebert among others). As a matter of fact, there are languages in which it is not always easy to distinguish formally between inflexional affixes and clitics (see Mereu 1997). Mereu refers to Lehmann (1985), among others, and to his scalar notion of grammaticalization between the pronominal and inflectional nature of clitics. Her study shows that in some languages clitics may exhibit such a double nature as ‘ambiguous markers’ with agreement and pronominal functions that overlap (Mereu 1997: 5). Doubling clitics in Bulgarian, when not motivated pragmatically, could be regarded as representing such ambiguous markers, whereas in Macedonian they are almost fully grammaticalized as inflection markers, the only extrasyntactic requirement for their appearance being the referentiality of the object.14 Being one of the products of centuries of multilingualism in the Balkans, the emergence and subsequent desemantisation of object doubling should be seen in Balkan Sprachbund languages as process of convergence leading to stronger analicity and therefore greater transparency of linguistic structure (see Topoliska 1995). 6. Conclusions This survey of information structure strategies shows that all Slavic languages employ free word order as the principal syntactic means of pragmatic manipulaton. Interestingly, Russian (though not Polish), an inflective language which reveals its tendency towards analicity in a number of features, exhibits “hanging topic” constructions, which are reminiscent of Bulgarian pronominal reduplication, a well known construction in Romance languages and typically occurring in languages without noun inflection. In Macedonian, which has also lost noun inflection, referential argument NPs can be dislocated freely in pragmatic function with no consequences on the syntactic structure of the sentence, since obligatory doubling clitics can be regarded as analytical case markers (see Topoliska 1995). In Bulgarian, the analoguous process of desemantisation and grammaticalization of doubling clitics seems to be restricted mainly to + human arguments of psychological and physical state verbs. Therefore only sentences in which such predicates occur can be considered as having free, syntactically unconditioned word order. In all other instances, clitics operate as resumptive pronouns in pragmatically marked sentences, which is exactly what happens in Romance languages, where clitic doubling is associated to all topical (in)direct object NPs.

Information structure in Slavic languages 321

Acknowledgements I would like to thank the anonymous referees for their precious remarks on an earlier draft of this paper.

Notes 1.

Sornicola (2007), however, points out that this correlation does not always hold (2007: 391–392); the connection between case inflection and flexible word order is questioned, among others, by Mereu (2004: 133). 2. Siewierska and Uhliová (1997) claim that V-initial word orders are more frequent in eastern Slavic languages than in western or southern ones (1997: 124). 3. The role of prosody will be given marginal treatment in this paper for reasons of space. A thorough analysis of the interplay between prosody and focus in European languages is presented in Sornicola (2007). 4. See Dalewska-Gre (1997: 452):”Nie ma w jzykach sowiaskich, ani innych europejskich segmentalnych wykladników tematu i rematu, inaczej ni np. w japoskim, w którym wykadnikiem tematu (gównego) jest partykula – wa” [‘Neither Slavic languages nor other European languages make use of segmental exponents of the theme or rheme, unlike Japanese, for instance, in which the theme is marked by the particle–wa’]. See also Moldovan et al. (2005). 5. The problem of Slavic clefts still requires in-depth study, given the somewhat contradictory nature of the data in Miller (2007). 6. The verb ‘to be’ in these examples is Ø in the present tense; whereas this is typical of Russian, the same cannot be said of Polish, which makes regular use of the verb ‘to be’ in the present tense of other constructions. 7. Note that the only way to translate sentences of this kind into Russian is by using the periphrasis: to kasaetsja Ivana, tak on kupil ma inu ‘As for Ivan, he bought a car’. 8. Other Balkan languages, such as Romani, Serbo-Croation and Turkish, are not acknowledged as being full members of the Sprachbund, although they do share part of its features. 9. In this connection, they report Rudin’s remark about the problem of defining “topic” in Bulgarian that “some speakers allow D(iscourse)-linked wh- phrases to be doubled even though these are not traditional topics” (Franks and King 2000: 254). 10. In this respect, the difference between Bulgarian and Italian lies in the fact that Italian does not allow for a doubling of an element under focus with a pronominal clitic:the same argument would thus be expressed twice, since the focused NP belongs to the argument structure of the sentence: 11. According to Leafgren (2002) clitic reduplication in Bulgarian can be referred to focused NPs. The same opinion is also shared by the Bulgarian linguist Iskra

322 Lucyna Gebert Likomanova (personal communication). However, Franks and King (2000: 254), with whom Leafgren disagrees, claim that if an NP is focused in the preverbal position “then doubling is ungrammatical”. Another Bulgarian linguist, Iliyana Krapova (personal communication), would regard (23b) acceptable in a strongly contrastive context. 12. As evidence of such equivalence she reports acoustic data of Italian together with the Korean data presented by Choi (1997), who points out that wherever topic and focus have a contrastive character, Korean marks them segmentally in the same way. 13. See Topoliska (1995, 1999) for a similar view, further substantiated by very interesting data from Macedonian dialects. 14. This grammaticalization process reflects the history of the phenomenon which, according to Friedman (1994b: 105), originated in south-western Macedonia and expanded towards north-eastern Bulgaria.

References Antinucci, Francesco and Guglielmo Cinque 1977 Sull’ordine delle parole in italiano:l’emarginazione. Studi di grammatica italiana 6: 121–146. Berent, Gerald P. 1980 On the realization of trace:Macedonian clitic pronouns. In Morphosyntax in Slavic, Catherine V. Chvany and Richard D. Brecht (eds.), 150–186. Columbus, Ohio: Slavica Publishers. Comrie, Bernard, Gerald Stone, and Maria Polinsky 1996 The Russian Language in the 20th century. Oxford: Oxford University Press. Cychun, Gennadij A. 1969 Sintaksis mestoimennych klitik v junoslavjanskich jazykach [Syntax of pronominal clitics in Southern Slavic languages]. Minsk. Dalewska Gre, Hanna 1997 Jzyki s owiaskie [Slavic languages]. Warszawa: PWN. Franks, Steven and Tracy H. King 2000 A Handbook of Slavic Clitics. Oxford /New York: Oxford University Press. Frascarelli, Mara 2000 Frasi scisse e ‘Small Clauses’: un analisi dell’inglese. Lingua e Stile 35 (3): 417–446. Friedman,Victor 1994 a Macedonian. In The Slavonic Languages, Bernard Comrie and Greville Corbett (eds.), 249–305. London: Routledge.

Information structure in Slavic languages 323 Friedman,Victor 1994 b Variation and grammaticalization in the development of Balcanisms. Chicago Linguistic Society 30 (2): 101–115. Gebert, Lucyna 1978 L’ordine delle parole in polacco. Rivista di grammatica generativa II (2): 181–239. 1991 Il russo dal punto di vista tipologico. In Lingua russa: storia, struttura, tipologia, Francesca Fici, Lucyna Gebert and Simonetta Signorini (eds.), 295–332. Roma: La Nuova Italia Scientifica. Glovinskaja M. Ja. 1996 Aktivnye processy v grammatike [Active processes in grammar]. In Russkij jazyk konca XX stoletija [Russian language of the end of the XX century], E. A. Zemskaja (ed.), 237–304. Moskva: Jazyki russkoj kul’tury. Guentchéva, Zlatka 1994 Thématisation de l’objet en bulgare. Bern: Peter Lang. Heine, Bernd and Tania Kuteva 2005 Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Huszcza, Ryszard 1980 Tematyczno-rematyczna struktura zdania w jzyku polskim [Thematic-rhematic sentence structure in Polish]. Polonica 6: 57–71. 2000 Nie ma eby nie byo – o segmentalnych wykadnikach tematycznorematycznej struktury zdania w polszczynie [On segmental exponents of thematic-rhematic structure in Polish]. Poradnik jzykowy (8): 1–9. Il’ina, N. E. 1996 Rost analitizma v morfologii [Increase of analytism in morphology]. In Russkij jazyk konca XX stoletija [Russian language of the end of the XX century], Elena A. Zemskaja (ed.). Moskva: Jazyki russkoj kul’tury. Krapova, Iliyana and Guglielmo Cinque to appear Clitic reduplication constructions in Bulgarian. In Clitic Doubling in the Balkan Languages, Liliane Tasmowski and D. Kallulli (eds.). Amsterdam/Philadelphia: John Benjamins. Lapteva, Ol’ga A. 1976 Russkij razgovornyj sintaksis [Syntax of spoken Russian]. Moskva: Nauka. 1997 O nekodificirovannych sferach sovremennogo russkogo jazyka [On uncodifyied domains of the Russian language], Voprosy Jazykoznanija [Problems of linguistics] 2: 40–45. Leafgren, John 2002 Degrees of Explicitness. Information Structure and the Packaging of Bulgarian Subjects and Objects. Amsterdam /Philadelphia: John Benjamins.

324 Lucyna Gebert Lehmann, Christian 1982 Universal and typological aspects of agreement. In Apprehension: Das Sprachliche Erfassen von Gegenständen II, H. Seiler and F. J. Stachowiak (eds.), 201–267. Tübingen: Gunter Narr. Mereu, Lunella 1997 For a lexical-functional representation of agreement affixes and clitics. In Proceedings of the LFG 97 Conference, Miriam Butt and Tracy Halloway King (eds.). University of California, San Diego CSLI Publications. http://www-csli.stanford.edu/publications. 2004 La sintassi delle lingue del mondo. Bari: Editori Laterza. Miller, Jim 2006 Focus in the languages of Europe. In Pragmatic Organization of Discourse in the Languages of Europe. Giuliano Bernini and M. L. Schwartz (eds), 121–214. Berlin /New York: Mouton de Gruyter. Moldovan, A. M., S. S. Skorbid, A. A. Kibrik, N. V. Rogova, E. I. Jakukina, A. F. uravlev and S. M. Tolstaja (eds.) 2005 Slavjanskie jazyki [Slavic languages]. Moskva: Rossijskaja Akademia Nauk. Rudin, Catherine 1997 AgrO and Bulgarian pronominal clitics. In Annual Workshop on Formal Approaches to Slavic Linguistics. The Indiana Meeting 1996, M. Lindseth and S. Franks (eds.), 224–252. Ann Arbour: Michigan Slavic Publications. Siewierska, Anna and Ludmila Uhlíová 1997 An overview of word order in Slavic languages. In Constituent Order in the Languages of Europe, Anna Siewierska (ed.), 105–149. Berlin / New York: Mouton de Gruyter. Sornicola, Rosanna 2006 Interaction of syntactic and pragmatic factors on basic word order in the languages of Europe. In Pragmatic Organization of Discourse in the Languages of Europe. Giuliano Bernini and M. L. Schwartz (eds.), 357–544. Berlin / New York: Mouton de Gruyter. widziski, Marek 1978 Przykadowe transformacje w jzyku polskim [Examples of transformations in Polish]. Polonica IV: 15–37. Topoliska, Zuzanna 1995 Convergent evolution, creolization and referentiality. In Prague Linguistic Circle Papers. Vol. I, Eva Hajiova, Miroslav ervenka, Oldich Leka and Petr Sgall (eds.), 239–247. Amsterdam/ Philadelphia: John Benjamins. 1999 Jzyk, cz owiek, przestrze [Language, Man, Space]. Warszawa / Kraków: Towarzystwo Naukowe Warszawskie.

Position, function and interpretation of topics in Somali Mara Frascarelli and Annarita Puglielli

1. Introduction In recent analyses evidence has been provided for a systematic connection between the discourse functions of Topics and their formal properties, which is encoded in a strict hierarchy in the C-domain. In other words, not only are Topics not included in the prosodic domain of the sentence (cf. Frascarelli 2000), they are also made prominent by different intonational events according to their specific discourse role. This crucial connection, originally proposed in Frascarelli and Hinterhölzl (henceforth F&H 2007) on Italian and German data, has been supported by subsequent analyses on typologically diverse languages (see Frascarelli 2008 for a comparison between Romance and Bantu languages, Frascarelli in press on Tagalog and Malagasy and, finally, Puglielli and Frascarelli 2007 for an investigation on signed languages). The aim of this paper is to examine the intonational properties of Topics in Somali, peruse the interpretation of these constituents and consider the complex interplay between different levels of analysis (i.e., the syntaxprosody-discourse interface) in order to check the validity of the aforesaid connection in a polysynthetic language (in the sense of Baker 1996; cf. section 2.2 and, in particular, note 6), in which full DPs are obligatorily realized as Topics. Like in the works mentioned above, the relevant investigation is based on natural speech (unstructured conversations among native speakers), which we consider as a basic feature for a real understanding of information structure and its properties.

2. Basics for the analysis 2.1. The fine structure of the Topic field in the C-domain In a cartographic approach (cf. Rizzi ed. 2004), the original CP-node (a recursive phrase, targeted by different functional categories) has been reanalysed as an array of functional projections, each dedicated to a specific func-

326 Mara Frascarelli and Annarita Puglielli tion related to information structure. The C-domain thus provides an interface between the propositional content (the IP-node) and specific discourse roles. In particular, F&H (2007) provide intonational and syntactic evidence that different types of TopP projections must be posited in the left periphery of the sentence: 1 “Discourse properties have structural correlates both in phonology and in syntax. In other words, […] different types of Topics show different intonational properties and are realized in a specific order in the CP-system.” (F&H 2007: 89)

Specifically, according to the authors’ analysis (that we assume), the ShiftP projection is dedicated to the Aboutness-shift Topics, FamP is the location of Familiar Topics and, finally, ContrP is the position in which Contrastive Topics are interpreted. A ‘free recursion analysis’ (Rizzi 1997) is thus refuted and a hierarchy is proposed in which different functional projections are distinguished in terms of discourse, prosodic and syntactic properties:2 (1)

[ForceP

[ShiftP [GP L*+H

[ContrP H*

[FocP

[FamP*3 L*

[FinP

As we can see, Topic constituents are associated with three different tonal events which, according to the Autosegmental-metrical theory (cf. Goldsmith 1990), can be described as L*+H, L* and H*.4 Since the interpretation and formal properties of Aboutness-shift and Familiar Topics will play an important role in the following discussion, let us consider them in some detail (on Contrastive Topics, see also Kuno 1976, Büring 1999, Molnár 2002). The Aboutness-shift Topic represents “what the sentence is about” (cf. Givón 1976; Reinhart 1981; Chafe 1987), with particular reference to the element that qualifies as a shifting point in the conversation and is made prominent with respect to other (possible) Topics in the discourse. This type of Topic can be therefore defined as something that is “newly introduced, newly changed or newly returned to” (Givón 1983: 8) – hence, it is not necessarily given in the discourse. It is realized in the highest TopP projection in the C-domain and is signalled by a sharp rise in the F0. This rise is aligned with the tonic vowel in its full extension and reaches its peak on the post-tonic syllable. Familiar Topics, on the other hand, are always part of background information (i.e., they are ‘D(iscourse)-linked’ elements, cf. Pesetsky 1987) and

Position, function and interpretation of topics in Somali 327

are generally produced to obtain Topic ‘continuity’ (cf. Givón 1983). These Topics are realized in a low position in the C-domain and their tonic vowel is marked as L*. In order to illustrate the discourse and formal properties of both types of Topic, let us consider the sentence underlined in the following text and the relevant Figure (for more details, cf. the references cited in the Introduction): (2)

Era tutto molto nuovo nel senso che comunque la lingua inglese attraverso i programmi sul computer diciamo non l’avevo mai- […] comunque l’inglese risultava anche facendolo da solo più interessante […] io, inglese non- premetto non l’avevo mai fatto. ‘Everything was totally new to me in the sense that I had never studied English through computer programs […] and through self-learning English appeared more interesting to me […] I must say that I had never studied English before.’

Figure 1.

As we can see, inglese – a direct object (DO) resumed by the clitic lo – is realized as low in the left periphery of the sentence and qualifies as a Familiar (continuing) Topic. On the other hand, the subject-Topic io marks a shift in the conversation: the speaker is still talking about English but, at that point, she wants to comment on her personal relation to that language. Accordingly, io is characterized by an intonational rise that is aligned with the tonic vowel [i].5

328 Mara Frascarelli and Annarita Puglielli 2.2. Topic constructions in Somali Somali is a polysynthetic language (cf. Baker 1996). This condition entails that -roles are only assigned through incorporation onto the verbal head (the so-called ‘Morphological Visibility Condition’, MVC). Hence, argument structure is realized by means of clitic pronouns that are disposed in the Verbal Complex (VC) in a rigid SOV order (Puglielli 1981; Svolacchia and Puglielli 1999), while full DPs are merged in non-argument position and connected to the sentence by means of resumptive pronouns.6 Topics in Somali can be located in different positions, as is illustrated in (3):7,8 (3)

Shalay CALI baank (baa-aan) aniguk Øj-ku arkay yesterday Cali FM-SCL .1SG I.NOM OCL.3SG-at see.PST.1SG jaamacaddaj university.DET ‘Yesterday, I saw CALI at the university’

As we can see, Topics can be connected to argument positions through clitic resumption (see the coindexing between the SCL aan and the Topic anigu, showing NOM Case), they can be interpreted as oblique complements (and be connected with elements in the VC)9 or, finally, they have the typical interpretation of adverbial expressions (cf. the DP shalay). As for their position, shalay is realized as a left-hand Topic (before the Focus 10), anigu is an ‘internal’ Topic (located between the FM and the VC) and, finally, jaamacadda is a right-dislocated Topic. Frascarelli and Puglielli (2007b) provide morpho-syntactic evidence that different Topic positions correspond to different pragmatic functions in the discourse. Consider, for instance, the contrast offered by the following yes-no questions: (4)

a. hadiyad-da (*hadiyad) ma CALI baa keenáy ? present-DET QM Cali FM bring.PST.RED ‘As for the present, did CALI bring it?’ b. Ma CALI baa hadiyad (*?hadiyadda) keenáy? ‘Did CALI bring a present?’

As is shown, left-hand Topics (4a) are not included in the scope of the QM ma, they must be [+definite] and obtain an ‘Aboutness’ role. ‘Internal’ Topics on the other hand, are interpreted as background information that is

Position, function and interpretation of topics in Somali 329

part of the question (the existence of a ‘present’ in (4b) is evidently ‘given’ in the discourse) and, as such, should be indefinite (definiteness is not excluded by informants, but considered as very marginal). In wh-questions as well, it is clear that left-hand Topics are used to propose or introduce “what the sentence is about” (cf. (5A)). In the relevant answer, on the other hand, speakers would normally reproduce the Topic in a right-hand position (for continuity, cf. (5Ba)) and exclude an internal collocation, independently of the presence of clitic resumption (the symbol “#” indicates the inappropriateness of sentence (5Bb)): 11 (5)

A: Cali muxuu sameeyay? Cali QM.thing.3SGM do.PST.3SGM ‘As for Cali, what did he do?’ B: a. MARYAM buu dilay, Cali ‘Cali beat MARYAM.’ b. #MARYAM baa/buu Cali dilay

In this paper, further evidence will be provided in this direction, based on information structure from a discourse-intonation perspective.

3. Different types of Topics in Somali This research is based on a corpus of semi-spontaneous conversations12 (ca. 20 minutes) among three native speakers of Somali (Moqadishu and Northern variety), all men whose age ranges between 35 and 60. The acoustic analysis has been developed using the WinpitchPro program (1996–2004, version 1.60; cf. Martin 1978). On a total number of about 250 utterances, we have analysed 136 Topic constituents, whose position in the sentence is indicated in (6): (6)

LEFT-HAND TOPICS: INTERNAL TOPICS: RIGHT-HAND TOPICS:

84 26 26

Interestingly, we found only six cases of multiple realizations, all concerning the co-occurrence of two Topics located in the left periphery of the sentence. This means that, though Topics are very frequent in Somali, speakers prefer not to overload the sentence with more than two per sentence,13 especially using overt DPs to shift the current theme (as is shown by the majority of

330 Mara Frascarelli and Annarita Puglielli left-hand Topics) and leaving the realization of background information to clitic pronouns in the VC.14 As for their intonational properties, Topics show specific contours depending on their position, as expected. The relevant facts are summed up in Table 1 below and considered in details in the following sections. Table 1. Left-hand Topics tonal event

rising tone & independent prosodic unit

low tone

Internal Topics prosodically integrated in the sentence

Right-hand Topics low tone

low tone & independent prosodic unit

3.1. Left-hand Topics When a DP is realized in the left periphery of the sentence, it can be associated with two tonal events: either it is marked by a rising contour or it is completely low-toned. Text analysis shows that the rising tone is definitely related to the Aboutness-shift function (cf. §2.1): the relevant DP is proposed as the most prominent referent in the discourse and represents what the following sentence is about. Consider for instance the dialogue in (7) and Figure 2. While speaker A does not propose any topic, speaker B chooses Cali for the Aboutness role, which is clearly signalled by its intonational contour (the sentence given in the Figure is underlined in the text): (7)

A: War hadal Axmedow lambarka kowaad ka hadal EXCL speak.IMP Axmed.VOC number.DET first.GEN of speak.IMP ‘Come on Axmed, tell me about the first picture’ in-uuk CAJIIN QASAYAA la B: Calik Cali.NOM that-3SGM pasta make.3SGM.DEP .FM IMPERS moodaa seem.PRES.3SGM ‘It seems that Cali, he is making pasta’

Position, function and interpretation of topics in Somali 331

Figure 2.

As we can see, Cali is a rising Topic (like io in Figure 1) and forms an independent prosodic domain with respect to the rest of the sentence (which is new information, in the scope of the FM waxaa15). Also notice that after the fall on the post-tonic vowel of Cali a new rise appears that has no correspondence with the wave form. This is due to the presence of a tone, which is used in Somali to mark NOM Case in the absence of determiners (which are morphologically marked).16 It is therefore clear that the Topic Cali is connected with a subject position in the sentence (i.e., the SCL uu), transferring the relevant Case. This causes a ‘reduplication’ of the rising contour. On the other hand, when a left-hand Topic is not realized to propose a shift, but simply to repeat a previously introduced Topic for the sake of continuity, it is marked by a low-tone, as is shown in Figure 3: (8) A: Maryan ma Axmed bay u baaqaysaa mise Cali? Maryan QM Axmed FM.3SGF to give a sign.PRES.3SGF or rather Cali ‘Is Maryan giving a sign to Axmed or to Cali?’ B: Maryan Cali u baaqi mayso Cali Maryan Cali to give a sign NEG.PRES.3SGF Cali.NOM cunayaa CUNTuu food.DET.FM.3SGM eat.PRES.RED ‘Maryan is not giving a sign to Cali, Cali is eating his FOOD’

332 Mara Frascarelli and Annarita Puglielli

Figure 3.

As we can see, Cali is not (intonationally) prominent in (8B), though located in the left periphery of the sentence (like the Familiar Topic inglese in Figure 1).17 This is evidence that the presence of a rising contour on the lefthand Topic Cali in (7) is not a ‘physiological’ consequence of its location: Topics are marked as high or low independent of their position. What counts is their function in the discourse, as expected in the present approach. 3.2. Internal Topics When a Topic is realized within the VC (i.e., between the subject clitic and the clitic cluster [prepositional particle+object clitic] preceding the verb), it is always integrated in the prosodic domain containing the VC itself. In recent analyses Frascarelli and Puglielli (2007a) have shown that the VC in Somali is characterized by an initial pitch followed by a downgrading contour18 and, in the absence of an internal Topic, this pitch marks the clitic cluster preceding the verb. On the other hand, when an internal Topic is present, it is marked by the relevant pitch and the rest of the sentence follows with a downgrading contour, without either pauses or F0 resetting. This is shown by miiska in Figure 4 below.

Position, function and interpretation of topics in Somali 333

(9)

waxaan arkaa in cuntad-ii ay hayeen FM.1SG see.1SG that food-AN SCL3PL have.PST.3PL.DEP ay miis-ka la yimaadeen SCL.3PL table-DET with come.PST.3PL.DEP ‘I see that they brought to the table the food they had.’ [lit.: ‘what I see (is) that the food they had, they went to the table with it]

Figure 4.

It is clear that internal Topics do not form an independent prosodic unit and, in a syntax-phonology perspective, this means that they cannot be considered as extrasentential constituents (since they are included in the Intonational Phrase containing the rest of the sentence; cf. Frascarelli 2000). This leads us to conclude that VC-internal DPs are not ‘Topics proper’. Further evidence in this direction is given by the fact they are included in the scope of negation (which excludes Topics by definition). Consider the following sentence and the relevant Figure: (10) Maryan Cali u baaqi mayso Maryan Cali to give a sign NEG.PRES.3SGF ‘As for Maryan, she is not giving a sign to Cali’

334 Mara Frascarelli and Annarita Puglielli

Figure 5.

As we can see, after the Aboutness-shift Topic (clearly marked by a rise and followed by a prosodic boundary), the Comment follows with its typical downgrading contour, having the internal Topic Cali as its first element. Intonational properties thus show that internal Topics cannot be part of the C-domain. Furthermore, since DPs cannot be inserted as arguments in polysynthetic languages, we must exclude that these elements are merged in the VP (cf. Chomsky 2001). Moreover, it is impossible to realize as internal Topics either adverbial DPs (like shalay ‘yesterday’) or any type of ‘circumstantial’ information – hence, we must also exclude the possibility to analyze these constituents as merged in the Spec of some functional projection in the I-domain concerning tense, aspect, mood and so on (following Cinque’s 1999 proposal). The only possibility left is to consider internal Topics as adjuncts to the VP (an option that is still admitted in the Minimalist framework, cf. Chomsky 1995) or – following a kaynian approach – we might assume the existence of additional projections in the I-domain, hosting nonargument DPs with different syntactic functions (benefactive, locative, comitative, instrumental and so on).19 In this line of analysis, internal ‘Topics’ can be considered as non-argument, non-extrasentential constituents which, being merged in the I-domain, are somehow interpreted as part of new information. This is the explanation we assume for the moment. Further research is of course needed to fully understand their formal and semantic properties.

Position, function and interpretation of topics in Somali 335

3.3. Right-hand Topics As is shown in Table 1, the intonation of right-hand Topics shows two variants: either (a) they are simply low-toned or (b) they form independent prosodic units. The two contours correspond to different functions in the discourse. Let us start with the first option. In the following question the speaker proposes Cali as the Aboutnessshift Topic and asks a question consisting of a comparison between Cali and the ‘teacher’: (11) Cali miyuu ka dheeryahay macallin-ka? Cali QM.SCL.3SGM of be tall.PRES. 3SGM teacher-DET ‘As for Cali, is he taller than the teacher?’ As is clear from the presence of a definite article, the relevant teacher must have been mentioned in some previous moment. It is, therefore, a familiar element. However, since other referents had been also introduced and third person object clitics are null in Somali (cf. note 9), the speaker prefers to repeat it as a right-hand Topic, for the sake of continuity and to secure a correct interpretation. As a result, this constituent is produced after the (downgrading) sentential curve has reached its lowest point and its prosodic domain appears rather flat (like Familiar Topics in the left periphery of the sentence, cf. § 3.1 and Figure 3), even though no major break can be noticed before it:

Figure 6.

336 Mara Frascarelli and Annarita Puglielli In the second option, right-hand Topics are low-toned as well, but they are distinguished from the previous case for the presence of a strong prosodic break (showing that they form an independent Intonational Phrase; cf. Figure 7 below). Intonational diversity is related with a different pragmatic function, which is made clear in the following text: (12) Cali Maryan way Ø-u egtahay, waa Cali Maryan DECL.3SGF OCL.3SG-to be similar.PRES .3SGF DECL hooyadiis waayo […] wejigoodaa is-ku eg mother.POSS.3SG because [...] face.POSS.3PL.FM RIFL-to be similar.RED waa inankeed-ii marka AAD IYO (AAD) bay is-ku DECL son.POSS.3SG-AN so much and much FM.3SGF RIFL-to weji egyihiin, Cali iyo hooyadiis face be similar.PRES .3PL Cali and mother.POSS.3SG ‘As for Cali, Maryam looks like him, she is his mother because their faces are very similar, he is her son, so they look like each other a lot, Cali and his mother’ Undoubtedly, the right-hand Topic ‘Cali and his mother’ is not needed to guarantee interpretation: Cali was introduced at the beginning as the Aboutness-shift Topic and both referents assuredly represent continuing Topics throughout the whole piece of conversation. Their repetition at the end of the sentence thus serves as a typical ‘afterthought’, that is to say, a way to repeat the main Topic providing a ‘circular’ closure to the relevant utterance.

Figure 7.

Position, function and interpretation of topics in Somali 337

3.4. Multiple Topics After the examination of the different types of Topics in Somali, which proved the existence of a systematic correlation between formal properties and pragmatic functions, let us now consider the case of multiple realizations. As already mentioned (cf. § 3), multiple Topics constructions are not very frequent in Somali (like in other languages). Nevertheless, the few data we can consider completely confirm the generalization (hence, the hierarchy) argued for in F&H (2007), namely, rising (Aboutnees-shift) Topics precede low-toned (Familiar) Topics. Consider the following example and Figure 8: (13) aniga fandhaal i-ma horyaal PRO.S.1SG.ABS spoon OCL.1SG-NEG be in front.PRES.3SGM.NEG ‘I cannot see any spoon’ [lit: ‘As far as I am concerned, no spoon is in front of me’]

Figure 8.

Figure 8 shows that aniga forms an independent prosodic domain, while fandhaal marks a flat contour without breaks before the Comment. Hence, aniga is an Aboutness-shift Topic while fandhaal – like Cali in Figure 3 – is a Familiar Topic, both located in the C-domain. Indeed, fandhaal cannot be considered as an internal Topic, since the initial pitch of the downgrading contour is aligned with ima, the clitic cluster that initiates the VC. We could also find a sequence of two low-toned Topics in initial position. This was also expected: as is argued in F&H (2007), more than one

338 Mara Frascarelli and Annarita Puglielli familiar constituent can be repeated in the same sentence for the sake of continuity (cf. note 3). Consider the following example: (14) A: Cali gacmihiisu ma JEEB-KA bay Cali hand.POSS.3SG.DET.M QM pocket-DET FM.3PL u-gu-Ø jiraan? for-in-OCL.3SGM stay.PRES.3PL ‘As for Cali, does he keep his hand in his pocket?’ ku haystaa B: Cali gacan CALOSH-uu Cali hand stomach.DET-FM.3SGM on keep.PRES.3SGM ‘Cali keeps his hand on the stomach’

Figure 9.

The realization of two destressed constituents at the beginning of the sentence shows, once more, that the intonation of Topics relies on informationstructural requirements and is independent of other constraints (e.g., amount of air in the lungs, mechanisms of turn-taking and so on).

3.5. Rising Topics and Illocutionary Force In recent analyses authors like Haegeman (2002), Puglielli and Frascarelli (2008) have suggested that the left periphery of embedded clauses does not have the same structure as matrix clauses (illustrated in (1)). In particular subordinates are assumed to have a ‘reduced’ C-domain, since they are not

Position, function and interpretation of topics in Somali 339

directly connected with the illocutionary Force of the sentence (which is expressed in matrix clauses). In this line of analysis, Frascarelli (2007) proposes a connection between (information) Focus and Force: subordinate clauses lack the FocP projection and, for this reason, embedded C-domains cannot host a Focus constituent, even in a language showing the extra situm (i.e., ‘fronted’) strategy. Similarly, a connection can be traced between the Aboutness-shift Topic and the illocutionary Force of the sentence since – as is argued in F&H (2007) – rising Topics are completely absent in subordinate clauses, while Familiar Topics are not subject to this restriction. This asymmetry is confirmed by Somali data, in which embedded Topics are only marked by a low-toned event, as in the following case: 20 (15) CUNTA-da-naa ay Ø-la sugayaan food-DET-CONG.FM SCL.3PL OCL.3SG-with wait.PRES.3PL .DEP Axmed yare in-tuu ka imaanayo Axmed to be young.RED that-DET.SCL.3SGM from come.PRES.3SGM.DEP ‘And for the FOOD they wait for young Axmed, till he comes back’

Figure 10.

Finally, it is interesting to notice that Familiar Topics in embedded Cdomains can be located higher than the Complementizer (COMP) in (‘that’) while, according to Rizzi’s (1997) system, COMPs are located in the Force head, which is higher than any Topic projection (cf. (1)). Consider the following sentence:

340 Mara Frascarelli and Annarita Puglielli (16) Cali ADIGA baa lagaa rabaa su’aal Cali PRO.2SG FM IMPERS.OCL2SG.from want.PRES.3SGM question in aad Axmed weyddiiso that SCL.2SG Axmed ask.PRES.2SG.DEP ‘Cali, I want you to ask a question to Axmed’ [lit.: ‘Cali, it is wanted from you that you ask a question to Axmed’]

Figure 11.

On the other hand, Familiar Topics in Somali are also (and more frequently) found in a post-COMP position. This apparent discrepancy, can be provided an explanation considering the etymology of COMPs like in, in a cross-linguistic perspective. In many languages clausal subordination clearly originates from nominalized constructions (cf., among others, Korniflt 1997 for Turkish, Ho-Min 1999 for Korean), in which the COMP element is in fact a preposition (like infinitival subordination in Romance languages, cf. also Kayne 1999) or the head of a relative clause. This is the case of Cushitic languages like Somali or Afar (cf. Bliese 1982) in which subordinators diachronically derive from generic NPs like ‘thing’, ‘place’, ‘time’ and so on (this is especially evident in adverbial clauses). In particular, the COMP in in Somali corresponds to a noun meaning ‘thing, part’ and can be still used today as such (i.e., it can be modified by determiners, possessives and adjectives; cf. Antinucci 1981). In this line of analysis, a complement clause like the one proposed in (16) derives from (and can be paraphrased as) ‘I want from you the thing that you ask a question to Axmed’. It is thus feasible that speakers still maintain the actuality of this construction in their competence and that the grammaticalization of in as a COMP is not completed yet. Hence, when they

Position, function and interpretation of topics in Somali 341

produce in in a lower position with respect to the Familiar Topic, this is not counterevidence to the hierarchy proposed in (1), but simply the sign that in is still considered as the nominal head of a relative clause (and not a subordinating element). This means that the relevant NP-head belongs to a relative DP and the Familiar Topic to its left periphery (on relative clauses in Somali, cf. Frascarelli and Puglielli 2005b). Evidence for this analysis is provided, once more, by intonation: in is very often marked by a pitch, which is a typical feature of NP-heads in relative clauses (while COMPs are not prosodically prominent, crosslinguistically). Consider, in this respect, the realization of in in Figure 2, repeated below as Figure 12: (17) waxay i-la tahay Calik in-uuk CAJIIN FM.3SGF me-IMPERS seem.3SGF Cali.NOM that-3SGM pasta QASAYAA la moodaa make.3SGM.DEP IMPERS seem.PRES.3SGM ‘It seems to me that Cali, he is making pasta, I think’

Figure 12.

4. Conclusions As a result of this pioneering investigation, we can conclude that the interface analysis of Somali data fully confirms the validity of some crucial generalizations, which we consider as part of Universal Grammar for interpretation.

342 Mara Frascarelli and Annarita Puglielli In line with cross-linguistic research, Somali has shown that the functional projection hosting Topic constituents is not simply a ‘recursive’ TopP and different positions must be assumed to account for their specific formal and discourse properties. In particular, tonal events signal in a systematic way the pragmatic functions of extrasentential constituents, which are encoded in a rigid hierarchy in the C-domain. We have also seen that the formal properties of Topics are not defined on the basis of their pragmatic characterization as [±given] information, but on their discourse value and, in particular, on their shifting vs. continuing function. Finally, it is important to remind that rising (i.e., newly proposed) Topics are only allowed in the left periphery of matrix clauses. This restriction, combined with the fact that information Focus is also excluded from embedded C-domains (Frascarelli 2007), can be plausibly interpreted in the light of a crucial connection between ‘discourse new’ information and the illocutionary Force of the sentence, which is encoded in matrix clauses. This means that a reduced structure must be posited for embedded C-domains. Interface analysis thus plays a crucial role in the analysis of information structure, allows a deeper understanding of discourse phenomena and shows the advantages of an ‘integrated’ approach, which takes into consideration the complex interplay between pragmatics, phonology and the requirements imposed by the core grammar of languages.

Notes 1. The syntactic and pragmatic properties characterizing different types of Topics have been the subject of several works (cf., among others, Lambrecht 1994 and references given in this section), although no attempt, it would appear, has been made to connect intonational properties to syntactic structures. The originality of F&H’s (2007) analysis rests exactly in their showing the existence of a systematic correlation between the formal properties of Topics and their function in the discourse, which is encoded in a strict hierarchy in the C-domain. 2. ForceP and FinP represent the two ‘extremes’ of the C-domain, in which illocutionary Force and Finiteness are encoded (cf. Rizzi 1997). The ‘Ground Phrase’ (GP) projection, on the other hand, indicates a functional projection in the Cdomain that is targeted by presupposed information. It is, for instance, the landing site for IP (remnant) movement in right-Topic constructions (with the righthand Topic merged in the FamP projection; cf. Frascarelli 2004a). 3. The asterisk after FamP indicates that this projection can be recursive. This possibility is excluded for Aboutness-shift and Contrastive Topics. Indeed, while we can newly propose or contrast only one Topic per sentence, more

Position, function and interpretation of topics in Somali 343

4.

5.

6.

7. 8.

than one constituent can be part of background information and repeated for continuity. The description of intonational contours is based on Pierrehumbert’s (1980) system – generally known as ‘ToBI’ – in which tonal events are described as sequences of low (L) and high (H) tones, which determine the shape of the F0 contour. In particular, pitch accents (marked with a ‘star’) are aligned with stressed vowels and render prominent the linguistic material with which they are associated. As for the syntax-semantic interface, is important to notice that Topics are always realized through nominal (or nominalized) constituents (hence, DPs), which are connected with an argument or adjunct position for interpretation. For a cross-linguistic analysis of Topic-Comment and Focus-Presupposition structures in an interface perspective, cf. Puglielli and Frascarelli (2008). Major evidence supporting polysynthesis in Somali lies in the obligatoriness of clitics. Object clitics, in particular, can never be omitted, independently of the presence of coindexed full DPs and of its being either a Focus or a Topic. This is shown respectively in (i) and (ii): (i) a. Cali ADIGAk buu kuk arkay. Cali you FM- SCL3 SG OCL2SG see.PST.3 SGM ‘As for Cali, he saw YOU’ b. *Cali ADIGA buu arkay. arkay. (ii) a. (Adigak) CALI baa kuk you Cali FM OCL2SG see.PST.3 SGM ‘(As for you), it’s CALI who saw you’ b. *Adiga CALI baa arkay. Strong evidence in favour of a polysynthetic analysis is also provided by the absence of non-finite clauses. This restriction is a typical property of polysynthetic languages, owing to the fact that Subject agreement is obligatory to obtain argument visibility. (iii) Waxay doonaysaa in-ay bisha dambe thing.DET.SCL3 SGF want.PRES.SCL3SG that- SCL3 SG month.DET next seexdo leave.SUBJ.3SG ‘She wants to leave next month’ Focus is indicated in capital letters, as is standard use. The list of the abbreviations used in the glosses is the following: ABS = absolutive case, AN = anaphoric article, DECL = declarative (marker), DET = definite article, EXCL = exclamative particle, F feminine, FM = Focus Marker, GEN = genitive case, IMP = imperative, IMPERS = impersonal clitic, NOM = nominative case, M = masculine, OCL = object clitic, PL = plural, POSS = possessive, PRES = present, PROG = present progressive, PST = past tense, RED = reduced paradigm, QM = Question Marker, SCL = subject clitic, SG = singular, SUBJ = subjunctive, VOC = vocative case.

344 Mara Frascarelli and Annarita Puglielli 9.

10.

11. 12.

13.

14. 15.

16. 17.

18.

19. 20.

Note that the right-hand Topic jaamacadda is connected with a ‘zero pronoun’ governed by the prepositional particle ku in the VC. Indeed, 3rd person object clitics have no phonetic realization, a common feature in Cushitic languages (cf. Puglielli ed. 1981). Focus must be obligatorily present in declarative sentences and its realization is a major feature of Somali syntax (for details, cf. Puglielli ed. 1981, Frascarelli and Puglielli 2005a). As a matter of fact, internal ‘Topics’ are somehow part of new information (see discussion in section 3.2 below) and cannot be accepted in the relevant context. Specifically, the speakers had been given a sequence of pictures to comment. Only one of them was aware that the relevant pictures were slightly different for each speaker and was appointed to lead the conversation. Change of topics, background elements and contrasts were thus expected. This finding is consistent with cross-linguistic analysis. Based on the analysis of Italian corpora, Frascarelli (2000) also argues that multiple realizations are rather rare in spontaneous speech and sequences of three (or more) Topics preferably avoided (nonetheless, a profitable discussion on multiple Topics can be found in Mereu and Trecci’s (2004) analysis, based on Bonvino’s (2006) corpus). In particular, when two Topics are realized in the left periphery of the sentence, the highest is always connected with the subject position (cf. § 3.6 below). Differently from baa, the FM waxaa signals as new information the constituent located in postverbal position (cf. Svolacchia et al. 1995, Saeed 1987). Notice, in fact, that the focused VP in (7B), cajiin qasayaa, is marked by a pitch accent (H*) that forms a typical ‘hat contour’ (cf. Frascarelli 2004b) on the tonic vowel of cajiin. Though important, the complex interaction between tone and intonation is beyond the scope of this paper and will not be treated in detail. Indeed, the DP Cali is present already in the question (cf. (8A)) and realised as a Familiar Topic in the first occurrence of answer (8B) as well. It is therefore a background element, proposed as an Aboutness-shift Topic in a previous moment of the conversation. This intonational property is in line with cross-linguistic research: in a number of languages the predicational part of the sentence (the ‘Comment’) is typically marked by a downgrading contour (cf., among others, Hayes and Lahiri 1991; D’Imperio 2002; Frascarelli 2004b). In this respect, see the hierarchy proposed for non-argument PPs in Schweikert (2005), based on German data. Though the English translation might lead to consider the DP Axmed yare as a Focus, this interpretation is excluded by Somali grammar. Indeed, Focus in Somali always precedes the VC and must be immediately followed by the Focus Marker baa (or its contracted form -aa, as in the relevant case). Hence, cunta (‘the food’) is the Focus in (15) and Axmed yare necessarily a Topic

Position, function and interpretation of topics in Somali 345 (given the polysynthetic property, cf. section 2.2). Moreover, Somali excludes Focus in subordinate clauses (for details on the Topic /Focus structure in Somali, see Puglielli 1981; Svolacchia, Mereu and Puglielli 1995).

References Antinucci, Francesco 1981 Tipi di frase. In Studi Somali 2. Sintassi della lingua somala, Annarita Puglielli (ed.), 219–301. Roma: Ministero AA EE. Baker, Mark 1996 The Polysynthesis Parameter. Oxford: Oxford University Press. Bliese, Loren, F. 1981 A Generative Grammar of Afar. The Summer Institute of Linguistics and University of Texas, Dallas/Arlington. Bonvino, Elisabetta 2006 Le Sujet Postverbal en Italien Parlé: syntaxe, zones et intonation. Paris: Biblioteque des faits des langues, Ophrys. Büring, Daniel 1999 Topic. In Focus – Linguistic Cognitive and Computational Perspectives, Peter Bosch and Rob van der Sandt (eds.), 142–165. Cambridge: Cambridge University Press. Chafe, Wallace 1987 Cognitive constrains on information flow. In Coherence and grounding in discourse, R. Tomlin (ed.), 21–51. Amsterdam/ Philadelphia: John Benjamins. Chomsky, Noam 1995 The Minimalist Program. Cambridge, MA: MIT Press. 2001 Derivation by phase. In Ken Hale: A life in language, Michael Kenstowicz (ed.), 1–52. Cambridge, MA: MIT Press. Cinque, Gugliemo 1999 Adverbs and Functional Heads. A Cross-Linguistic Perspective. Oxford: Oxford University Press. D’Imperio, Mariapaola 2002 Italian Intonation: An overview and some questions. Probus 14: 37– 69. Frascarelli, Mara 2000 The Syntax-Phonology Interface in Focus and Topic Constructions in Italian. Dordrecht: Kluwer Academic Publishers. 2004a Dislocation, clitic resumption and minimality: A comparative analysis of left and right Topic constructions in Italian. In Romance languages and linguistic theory 2002, Reineke Bok-Bennema, Bart Hollebrandse, Brigitte Kampers-Manhe, and Petra Sleeman (eds.), 99 –118. Amsterdam /Philadelphia: John Benjamins.

346 Mara Frascarelli and Annarita Puglielli 2004b

L’interpretazione del Focus e la Portata degli Operatori Sintattici. In Il parlato italiano. Atti del Convegno Nazionale (13–15 febbraio 2003), Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino and Renata Savy (eds.), M. D’Auria Editore, CIRASS, Naples, (CD-ROM), B06. 2007 Narrow Focus, clefting and predicate inversion. Paper presented at GlowXXX (Workshop on ‘Selective Global Comparison’), University of Tromsø (April 11–14). 2008 The Fine Structure of the Topic Field. In The Bantu-Romance Connection, Cécile de Cat and Katherine Demuth (eds.), 261–292. Amsterdam /Philadelphia: John Benjamins. in press Intonation, Information Structure and the Derivation of Inverse VO Languages: An Interface Analysis of Tagalog and Malagasy. In Proceedings of AFLA XIV, Raphael Mercado (ed.). Amsterdam/Philadelphia: John Benjamins. Frascarelli, Mara and Annarita Puglielli 2005a The Focus System in Cushitic Languages. In Proceedings of the 10th Hamito-Semitic Linguistics Meeting (Quaderni di Semitistica, 25), Pelio Fronzaroli and Paolo Marrassini (eds.), 333–358. Firenze: Università degli Studi di Firenze. 2005b Restrictive vs. Appositive: A Comparative Analysis of Relative Clauses in Cushitic Languages. In Proceedings of the XXX Incontro di Grammatica Generativa, Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turano (eds.), 279–303. Venezia: Cafoscarina. 2007a Information Structure in Somali. Evidence from the Syntax-Prosody Interface. Paper presented at the Workshop on Minority Languages. Lexical Aspects and Discourse Grammar (Rome, 16 –17 June 2006). 2007b Focus in the Force-Fin system. Information Structure in Cushitic Languages. In Focus Strategies in African Languages. The Interaction of Focus and Grammar in Niger-Congo and Afro-Asiatic, Enoch Oladé Aboh, Katharina Hartmann, and Malte Zimmermann (eds.), 161–184. Berlin /New York: Mouton de Gruyter. Frascarelli, Mara and Roland Hinterhölzl 2007 Types of Topics in German and Italian. In On Information Structure, Meaning and Form, Susanne Winkler and Kerstin Schwabe (eds.), 86 –117. Amsterdam /Philadelphia: John Benjamins. Givón, Talmy 1976 Topic, Pronoun and Grammatical Agreement. In Subject and Topic, Charles Li (ed.), 149–188. New York: Academic Press. 1983 Topic Continuity in Discourse: An Introduction. In Topic Continuity in Discourse: A quantitative cross-language study, Talmy Givón (ed.), 5–41. Amsterdam /Philadelphia: John Benjamins.

Position, function and interpretation of topics in Somali 347 Goldsmith, John 1990 Autosegmental and Metrical Theory. London: Basil Blackwell. Haegeman, Liliane 2002 Anchoring to speaker, adverbial clauses and the structure of CP. Georgetown University Working Papers in Theoretical Linguistics 2: 117–180. Hayes, Bruce and Aditi Lahiri 1991 Bengali Intonational Phonology. Natural Language and Linguistic Theory 9: 47-96. Ho-Min, Sohn 1999 The Korean Language. Cambridge: Cambridge University Press. Kayne, Richard 1999 Prepositional Complementizers as Attractors. Probus 11: 39 –73. Kornfilt, Jaklin 1997 Turkish. New York; London: Routledge. Kuno, Susumo 1976 Subject, theme, and the speaker’s empathy – A re-examination of Relativization Phenomena. In Subject and Topic, Charles Li (ed.), 417–444. New York: Academic Press. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus and the mental representation of discourse referents. Cambridge: Cambridge University Press. Martin, Philip 1978 Intonation et syntaxe des langues romanes. In Macrosyntaxe et pragmatique: L’analyse de la langue orale, Massimo Moneglia (ed). Roma: Bulzoni. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. In Il Parlato Italiano. Atti del Convegno Nazionale, Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino and Renata Savy (eds.), B10, Napoli: M. D’Auria Editore – CIRASS (CDROM). Molnár, Valeria 2002 Contrast – from a Contrastive Perspective. In Information Structure in a Cross-linguistic Perspective, H. Hallelgard, S. Johansson, B. Behrens and C. Fabricius-Hansen (eds.), 147–162. Amsterdam /New York: Rodopi. Pesetsky, David 1987 Wh-in-situ: Movement and Unselective Binding. In The Representation of (In)Definiteness, E. Reuland and A. ter Meulen (eds.), 98–129. Cambridge, MA: MIT Press. Pierrehumbert, Janet 1980 The Phonology and Phonetics of English Intonation. Unpublished Ph.D. dissertation. Cambridge, MA: MIT.

348 Mara Frascarelli and Annarita Puglielli Puglielli, Annarita (ed.) 1981 Sintassi della Lingua Somala. Studi Somali 2. Roma: MinisteroAA EE. Puglielli, Annarita and Mara Frascarelli 2007 Interfaces: The Relation between Structure and Output. In Signed and Verbal Languages. Comparing Structures, Constructs and Methodologies, Elena Pizzuto, Raffaele Simone and Paola Pietrandrea (eds.), 133–169. Berlin /New York: Mouton de Gruyter. 2008 L’analisi linguistica. Dai dati alla Teoria. Cesena/ Roma: Caissa Italia Editore. Reinhart, Tanya 1981 Pragmatics and Linguistics: An Analysis of Sentence Topics. Philosophica 27: 53–94. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. In Elements of Grammar. Handbook in Generative Syntax, Liliane Haegeman (ed.), 281–337. Dortrecht: Kluwer. Rizzi, Luigi (ed.) 2004 The Structure of CP and IP: The Cartography of Syntactic Structures. Vol. 2. Oxford: Oxford University Press. Saeed, John Ibrahim 1987 Somali Reference Grammar. Maryland: Dunwood Press. Schweikert, Walter 2005 The order of prepositional phrases. In Proceedings of the XXX Incontro di Grammatica Generativa, Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turano (eds.), 129–148. Venezia: Cafoscarina. Svolacchia, Marco, Lunella Mereu, and Annarita Puglielli 1995 Aspects of Discourse Configurationality in Somali. In Discourse Configurational Languages, Katalin É. Kiss (ed.), 65–98. Oxford: Oxford University Press. Svolacchia, Marco and Annarita Puglielli 1999 Somali as a polysynthetic language. In Boundaries of Morphology and Syntax, Lunella Mereu (ed.), 97–120. Amsterdam/ Philadelphia: John Benjamins.

Section 5 Some more aspects of information structure

Acquiring the grammar of topicality in L2 Italian: A comparative approach Marina Chini

1. Introduction This study on topicality in L2 Italian adopts a functional approach to interlanguage, taking into account the results of linguistic typology and of comparative studies on textuality in several L1s and L2s.1 In these studies several language-specific preferences in information structure2 emerged, which seem to be correlated with the grammar of the specific language and to develop quite early in the first language, with more difficulty in L2. Languagespecific morphosyntactic structures possibly exert an influence on text planning both at the macrolevel and at the microlevel (Levelt 1996), i.e. both for the selection and the segmentation of the content in propositional units (macroplanning) and for the microplanning choices related to the form of propositional units and clauses (topic-comment structure, syntactic roles attributed to the referents, NP type, word order, verb aspect, use of subordination; cf. Carroll and Lambert 2003). Works by Carroll, von Stutterheim, Lambert and others show that “information structure is language dependent, i.e. it correlates with the specific system of grammaticised means in the language” (Carroll and Lambert 2005). As a consequence, in order to acquire a native speaker competence, L2 learners have to uncover the role of specific grammatical features and rules in information organisation and their implication for information structure (Carroll and Lambert 2003, 2005). Several analyses point out that at least some principles of information organisation in L2 (selection, linearization, connection) can trace back to the first language also in advanced learner varieties (Carroll et al. 2000; Carroll and Lambert 2003). On this basis and on the basis of systematic comparisons between parallel texts produced in different languages, some scholars propose a revised version of linguistic relativism (Gumperz and Levinson 1996; Slobin 1996) which claims the existence of a thinking for speaking (Slobin 1987), that is “a level of mental representation which may not be involved in perception and habitual behavior – as Whorf advocated – but which is nevertheless intimately involved with language, and, in a real sense, exists because of language” (Slobin 1987: 443).

352 Marina Chini Although we do not have access to this “thinking for speaking”, we can discover some superficial traces of it in real data. More specifically we are interested here in the correlates it has in the “information structure component” of the grammar of native speakers and learners of Italian, more precisely in that part of this component concerning topic treatment. We give here Lambrecht’s (1994) definition of information structure: (1) INFORMATION STRUCTURE: That component of sentence grammar in which propositions as conceptual representations of states of affairs are paired with lexicogrammatical structures in accordance with the mental states of interlocutors who use and interpret these structures as units of information in given discourse contexts. (Lambrecht 1994: 5) Every language has a set of linguistic means (prosodic, morphosyntactic and lexical means) to signal information structure: i.e. subject accentend vs. predicate accented sentences; active vs. passive sentences; canonical vs. topicalized sentences, clefted or dislocated sentences, topic markers; also the form chosen for NPs within a sentence or discourse has to do with information structure. In Lambrecht’s terms, learning a second language does not only mean to acquire the above-mentioned lexicogrammatical structures, but also to become able to use them with the information value they normally carry in a specific language. In the interface analyses 3, it means to acquire the specific grammaticalization of these pragmatic notions, their syntactic and prosodic constraints and requirements in the second language. In order to discover some features of this component related to topic in Italian, we are going to analyse some parallel narrative texts produced in Italian L2 by post-basic learners with two different language backgrounds (German and Spanish) and in Italian L1 by native speakers with the same age and cultural level. We know that the development of topic treatment in L2 has to do, more generally, with the acquisition of morphosyntactic and discourse competence in Italian L2 (cf. Giacalone Ramat 2003: ch. 3–6; Chini 1998, 2002, 2003a, 2003b), but we cannot deal here with all these aspects. Nevertheless some preceding analyses of the pragmatic structuring of interlanguages and of focus structures in Italian L2 will be recalled (Bernini 1995, Andorno 2000, Andorno et al. 2003; cf. infra §1). In the following sections, after introducing the Quaestio model, which serves as a useful framework for a text analysis dealing (among other things) with topic (§ 2), the notion of topic is outlined (mainly with reference to Lambrechts 1994); then, certain findings on information structure and topicality in L2 are

Acquiring the grammar of topicality in L2 Italian 353

summarised (§3); in §5 topic selection and referential forms employed by Italian native speakers and by Italian L2 learners during film retellings are analysed (the data and subjects are presented in § 4); finally, important aspects of the acquisition of the topic grammar and treatment in L2 are discussed (§6). 2. The Quaestio model Our data are not isolated sentences, but clauses taken from utterances as they occur in narratives. Therefore we need a simple, but clear framework for text analysis. Our approach to text refers to von Stutterheim and Klein’s (1989) Quaestio model, according to which every text can be seen as an answer to a specific question or quaestio. For instance a narrative text answers to a quaestio such as “What happened to X (in time t)?”, while a descriptive text can be seen as an answer to a question such as: “What does/ did X look like?”. In our opinion, the Quaestio model approach is useful to correlate text conceptualization with text formulation (Levelt 1989), to account for textual cohesion and coherence, and also to frame the notion of topicality in discourse. In fact the structure of a text can be seen as constrained by the nature of its quaestio. These constraints include: 1. the partitioning of the text in main structure (Hauptstruktur), which is the part of discourse which answers directly to the quaestio (the events happened to X in a narrative), and side structure (Nebenstruktur), which answers to other less central questions (for instance why and how X did something); at the beginning of (portions of) texts we can also have a setting section; 2. the assignment of specific meaning elements to the topic component or to the focus component of a main structure utterance (i.e. in narratives the protagonist, X, which is already given in the quaestio, typically belongs to the topic component, while events are typically focalised); 3. the referential movement within the domains of reference (time, space, person, mood) from one sentence of the main structure to the other: cf. the different referential devices (NPs, temporal adverbs and inflections, local adverbs, etc.) which convey information about the entities, time and space, their introduction, their maintenance and their shift in the main structure (von Stutterheim and Klein 2002: 71; 1989). In sum the quaestio imposes constraints on the macrostructure of the text, on topic/focus distribution and on the form of its single sentences, including its

354 Marina Chini referential elements. The quaestio of a text has to do with discourse representation (“conceptual structure” or “preverbal message”), which “prepares” it and is at the origin of its formulation. This discourse representation is characterised by features such as: selection, linearisation, addition and function assignment.4 According to von Stutterheim and Klein (2002: 66), “There is reason to assume that already before the decision on the final languagespecific form, various bits of information are marked for special linguistic features”, for instance the decision whether a discourse entity should be encoded in a text sentence as subject or as object, whether an active or passive form is to be chosen, and what belongs to the topic component and what belongs to the focus component of the utterance that is going to be produced. In our narratives, after their introduction, the entities considered often belong to the topic portion of the text (the X of the quaestio), while events and actions typically belong to its rhematic and focalized part. But a detailed analysis of the information structure of every sentence is necessary to or not to assign those elements to the topic component of the single clause. Before doing this and on the basis of the preceding considerations, we acknowledge that we are inclined to consider topic treatment in a specific text type as not independent from its quaestio (here “What happened in this film?”) and from its subjacent discourse representation, which has to do with the assignment of functions to the pieces of information codified, thus also with the choice of a specific information structure. Furthermore, while the Quaestio model (a cognitive and communicative model) can be applied as well to native speakers’ texts as to learners’, it is highly probable that its formal consequences, and eventually also the subjacent discourse representation diverge in different languages and interlanguages, also in relation to the level of linguistic and discourse competence of the speaker. An investigation on such divergences is part of our research goal. 3. Remarks on topic and topicality in languages and interlanguages Before analysing our data, we want to make our notion of topic explicit and to recall some results on information structure and topicality in L2. We are mainly interested here in the topic component of an utterance, especially in the topic entities about which the sentence asserts something, secondly we are interested in discourse topic. As for the topic of a sentence, we refer to Lambrecht’s (1994) definition and to other similar notions (Dik 1978; Gundel 1988; Molnár 1993; Berretta 1995); as for Italian information structure, we make reference to functional approaches by Berretta (1994, 1995)

Acquiring the grammar of topicality in L2 Italian 355

and Andorno (2003). These approaches speak of sentence topic (or theme) in terms of aboutness. Other criteria for topichood, which all have some motivations, but are not applied here, identify the topic in the first element of a sentence (ex. Halliday 1967), in the grammatical subject or, more generally, in a syntactic position (Rizzi 1997), in (temporal and spatial) framing elements (Chafe 1976; Stark 1999), in given or presupposed information, in elements with less communicative dynamism (Firbas; for these and other notions, cf. van Kuppevelt 1994; Sornicola 1994, 2006). Lambrecht’s definition of topic and topic expression in terms of “aboutness” (cf. also Vallduví 1992 and others) is the following5: (2)

TOPIC: “A referent is interpreted as the topic of a proposition if in a given situation the proposition is construed as being about this referent, i.e., as expressing information which is relevant to and which increases the addressee’s knowledge of this referent. TOPIC EXPRESSION: “A constituent is a topic expression if the proposition expressed by the clause with which it is associated is pragmatically construed about the referent of this constituent” (Lambrecht 1994: 131)

Topic elements are discourse referents (i.e. an entity, sometimes a proposition, never a predicate, according to Lambrecht 1994), about which we affirm something, for which what we assert is relevant (cf. Strawson 1964). Normally they are given information (cf. Chafe 1976; Givón 1976), and prosodically they are typically deaccented.6 According to many scholars (including Lambrecht 1994), topics are typically accessible, contextually relevant and pragmatically presupposed referents; their cognitive features are identifiability and a high activation state in memory (Chafe 1994).7 Furthermore there is a strong (though not obligatory) correlation between topic and subject, topic and sentence initial position (obligatory in Halliday’s vision; Chomsky [1965] characterizes a sentence topic as the leftmost NP immediately dominated by S in the surface structure), while the focus is often placed towards the end of an utterance8 and is more prominent in terms of information value and stress (indeed, it carries the utterance stress). In fact there are at least three types of topic, with different prosodic, syntactic and discourse features: aboutness-shift topic, familiar topic and contrastive topic (Frascarelli and Hinterhölzl 2007)9. In the following analysis we mainly consider familiar topics (in narratives), i.e. topics referring to given, already established information, “active referents”, often involved in topic continuity (Givón 1983); they are normally unstressed (Lambrecht

356 Marina Chini 1994), prosodically associated with a low (L*) tone (e.g. in Italian and German; cf. Frascarelli and Hinterhölzl 2007). As for the syntactic analysis of topic, within configurational models,10 there is no general agreement on whether a Topic constituent is base-generated in extrasentential position11, coindexed with a predicate internal gap or clitic (Cinque 1990) or extracted from its argument position by movement (cf. Frascarelli 2000: 2). Like Focus, Topic has also been accorded to a specific maximal projection in recent proposals: the Topic phrase (TopP), whose node is in a higher position than Focus Phrase, FP (Rizzi 1997). Going back to Lambrecht (1994), we can identify a topic only in one kind of focus structure, that is in predicate-focus structures, which have a topic-comment structure and where the topic can be left implicit (in these structures we can have a topic marker, as we have in Japanese wa; cf. 3a); there is no topic in argument-focus structures (3b) or in sentence-focus structures (3c). See the following examples in various languages: (3)

a. PREDICATE-FOCUS STRUCTURE (topic-comment sentences) What happened to your car? ENGLISH My car/It broke DOWN. ITALIAN (La mia macchina) si è ROTTA. FRENCH (Ma voiture) elle est en PANNE. JAPANESE (Kuruma wa) KOSHOO-shi-ta. b. ARGUMENT-FOCUS STRUCTURE (topicless) (identificational sentences) I heard your motorcycle broke down ? ENGLISH My CAR broke down. ITALIAN Si è rotta la mia MACCHINA. E’ la mia MACCHINA che si è rotta. FRENCH C’est ma VOITURE qui est en panne. JAPANESE KURUMA ga koshoo-shi-ta. c. SENTENCE-FOCUS STRUCTURE (topicless) (= event-reporting or thetic or presentational sentences) What happened? ENGLISH My CAR broke down. ITALIAN Mi si è rotta la MACCHINA. FRENCH J’ai ma VOITURE qui est en PANNE. JAPANESE KURUMA ga KOSHOO-shi-ta. (Lambrecht 1994: 223)

Acquiring the grammar of topicality in L2 Italian 357

In natural discourse the topics of canonical sentences with a predicate-focus structure (like 3a) are often not expressed by lexical NPs, but by unaccented pronominals (Lambrecht 1994: 188; cf. the Preferred Argument Structure in Du Bois 1987)12; moreover topic expressions are normally not totally indefinite referents and are not in the scope of negation. On the contrary, if we accept, differently from Lambrecht, that also argument-focus structures like 3b have a topic (a stressed “contrastive topic”), in this case the topic can be within the scope of negation. There is nevertheless a difference in information structure between a sentence with an unstressed topic as 3a and one with a so called “contrastive topic” as 3b: in the last case (for ex. in My CAR broke down. Mein AUTO ist kaputt) the statement is not really construed about the car, but its function is to identify a referent for the subject of the proposition “something broke down” which is pragmatically presupposed to this assertion (it is an identificational sentence). Thus according to Lambrecht (1994: 122) and to us the subject NP of 3b is not a real topic (in fact in Japanese in such cases the subject does not take the topic marker wa), but a particular type of focus expression (argument focus)13, as its prosodic prominence iconically marks (on the contrary the verb phrase in 3b is not prosodically prominent, not being part of the focus component). Let us now consider how the notions of topic and information structure are dealt with in acquisitional data. Some studies hypothesize a universal topic-comment stage in the first stages of language varieties (Fuller and Gundel 1987), which is often also considered basic and crosslinguistically widespread (Givón 1979; Dik 1989), typical of the pragmatic (pre-syntactic) mode (Givón 1979, 1984) and corresponding to the “Theme first” principle (Tomlin 1986). Other studies (Givón 1984) suggest that comment-topic orders have priority in the first L2 stages and in pidgins and that the development is the following: (4)

comment (zero-topic) > comment-topic > topic-comment > repeated topic (zero comment)

The underlying psychological principle, which is often observed in presyntactic modes and in early childhood, is to say first what is more urgent (cf. Givón 1984 and his Universals of discourse structure). But as a matter of fact normally two principles clash in languages: the GIVEN BEFORE NEW principle, which would lead to a topic-comment order, and the FIRST THINGS FIRST principle, according to which the most important information elements come first (cf. Gundel 1988). This happens especially in nonconfigurational languages and in non-configurational aspects of configura-

358 Marina Chini tional languages (Mereu 2004). It is highly probable that the same two principles compete also in interlanguages, with different possible solutions, that cannot necessarily be generalized, but that can be locally justified. In various L2 studies, for instance in the ESF Project on several European L2s of learners with different L1 (also non-Indoeuropean: i.e. Moroccan Arabic), a prominence of pragmatic principles (in general independent from the first language) is attested in pre-basic and basic-varieties (Klein and Perdue 1992, 1997); thereafter also semantic principles and at last syntactic language-specific principles are at work (see scale (5) below). The first pragmatic principle, according to which focalised elements are placed towards the end of the sentence, is Focus last; basic varieties resort also to the semantic principle Controller first (i.e. the controller of an action, not necessarily the subject, is placed towards the beginning of the sentence; Klein and Perdue 1992, 1997). These principles have also been attested in L2 Italian, where, initially, topicless sentences (only with comment/focus) appear and a Focus last principle is often applied (ex. (6) by HG, a young Eritrean, L1 Tigrinya, since few weeks in Italy): (5)

pre-basic varieties > basic varieties > post-basic varieties pragmatic organization > semantic organization > syntactic organization (Focus last) (Controller first) (SVO; lg-specific rules) (cf. also Andorno et al. 2003: 150)

(6)

I:

e i tuoi genitori tuo padre tua madre? quando sono venuti in Italia? ‘And your parents your father your mother? When did they come to Italy?’

HG: (T=0) cinq’ anni (=F) // mamma (=T) eh duodic’ anni (=F) five years // mum eh twelve years ‘(T=0, prob. dad) (came) five years (ago) (=F), mum (=T) (came) twelve years (=F) (ago)’ (Bernini 1995: 24) (T = topic element; F = Focus, focalized element) The occurrence of similar pragmatic-based structures induce some scholars to compare interlanguages to topic-prominent (TP) languages like Chinese. Fuller and Gundel (1987) for instance found more topic-prominent features in L2 English by learners with a topic-prominent L1 (Chinese, Japanese, Korean), but also with other L1s (Arabic, Persian, Spanish), than in native speakers’ English. But Jin’s (1994) more recent study on Chinese L2 by English learners challenges the supposed generalized topic-prominent char-

Acquiring the grammar of topicality in L2 Italian 359

acter of interlanguages, identifying rather a typological transfer from L1 to L2 in this field (cf. also Rutherford 1983): English-speaking learners of Chinese do not display the (supposed) universal topic-prominent stage, but begin with a subject-prominent stage [SP] (as in their L1), developing gradually topic-prominent [TP] features: (7)

SP > SP features and Null Subject > mixed stage SP-TP (increasing number of Zero Topics and Zero Objects) > TP (Jin 1994)

According to Jin (1994), Fuller and Gundel’s results are vitiated by the fact that the L1s considered are topic-prominent or Null Subject languages (as Spanish, Arabic, Persian), never subject-prominent non pro-drop languages like English. The question is still open. Nevertheless clear signs of the prominence of pragmatic factors over syntactic factors in the early interlanguage stages are recurrent and concern several L2s (Klein and Perdue 1992, 1997). We find creative topic markers often based on the copula (is/isa in L2 English of a learner with L1 Hmong [Huebner 1983]; is in L2 Dutch and English, Klein and Perdue 1992; c’est in L2 French of Spanish learners, Trévise 1986), or markers for focus or for the borderline between topic and focus (cf. it. che in L2 Italian by Hagos, a pre-basic learner with L1 Tigrinya, in Bernini 1995): (8)

I:

perché ti trovi bene? ‘ why do you feel well?’

HG: perché che papà- che mamma ‘because che father che mother [are here]’ (HG after 2 months and 24 days in Italy; Bernini 1995: 32) The utterance organisation in prebasic and basic learner varieties, being characterised by pragmatic or pre-syntactic mode (Givón 1979), is highly dependent on notions like topic. In those phases the information structure of the sentences in (Italian) L2 can be described as (Andorno et al. 2003: 131): (9)

Setting – TOPIC – Focus markers / TA markers / Polarity markers – COMMENT

The only obligatory part is the comment; the other elements, including the topic, can often be left implicit, especially if they are inferable from discourse, context or encyclopedic knowledge. We can suppose that during the subsequent stages of interlanguage the topic treatment gradually becomes more and more language-specific, assum-

360 Marina Chini ing also the (prosodic and morphosyntactic) codifications of the target language, both the obligatory ones and the preferential choices. We turn now to our data in order to discover some aspects of the development of topic treatment in L2 Italian. 4. Data, subjects and methodology Our corpus consists of oral narratives (film retellings) that were elicited after a videotape version of Charlie Chaplin’s film Modern Times (about 20 minutes long) was shown to three groups of young subjects, with similar age and cultural background (University students): 10 learners of Italian with German as L1, 10 learners of Italian with Spanish as L1, 10 native speakers of Italian (the control group).14 The learners, living and studying in Pavia for one or two semesters, display all post-basic varieties of Italian, although with different proficiency levels; these two groups can be compared with one another (and with the native speaker group), in that they have passed the basic stage (characterised by a pragmatic structure of the utterance) and show a certain morphosyntactic competence: use of inflected articles and of some pronominal forms, verb-subject agreement, some kind of subordination (cf. Chini 1998, 2002, 2005). In this occasion we take into consideration the narration of the first episode of the film, which was narrated by the subjects after having seen it with the interviewer (the author); we do not analyse here the narratives in Spanish and German L1, produced by our learners after the Italian L2 retelling. Transcriptions of the retellings (quite simplified as for prosody and interactional phenomena) were segmented into clauses (=cc.) and numbered; we analysed the expressions referring to human referents and to some major inanimate referents, virtually eligible to become topical entities in the narrative (we excluded places), with respect to several syntactic, semantic and textual parameters 15 (cf. Hickmann et al. 1990 [19942]; von Stutterheim and Klein 1989; more details on the methodology in Chini 1998, 1999, 2005). For the present study we will concentrate on the following parameters: I. Topical status of the referent in the clause; II. Referential movement: establishment of reference or referent introduction, first mention (= er); maintenance of reference (= mr); shifting reference (= sr); III. Syntactic function (essentially S = subject; O = object; IO = indirect object; OBL = oblique);

Acquiring the grammar of topicality in L2 Italian 361

IV. NP referential form: INDN = indefinite NP; NPRO = proper noun; DEFN = definite NP (introduced by a definite article); DEMN = definite NP introduced by a demonstrative; REL = relative pronoun; PSREL = pseudo-relative pronoun (after presentative and cleft clauses); PRO3 = 3rd person full pronoun; CLIT = clitic pronoun; ZP = zero (subject) pronoun (in a finite clause; verbal agreement codifies the subject); Z = zero anaphora (in a non-finite clause). In the description of our data we will reserve the label “zero anaphora” for the empty subject of non-finite clauses (Z), following Berretta’s (1990: 94–95) analysis of anaphoric chains in Italian, and distinguish it from the zero subject pronoun of finite clauses (ZP). As for the sometimes controversial identification of the topic, we adopt the following criteria: 1. our topic identification concerns here primarily the sentence level; sentence topic and discourse topic can coincide, but not necessarily do; 2. the topic is the referent about which information is being conveyed in a proposition/clause (aboutness criterion; cf. § 2); 3. argument-focus (like 3b) and sentence-focus structures (like 3c) are considered as topicless sentences (cf. Lambrecht 1994: 223 and supra, §2; Berretta 1995 for Italian L1); 4. the topic must be an active or at least accessible referent; brand-new referents (often expressed by post-verbal NPs in Italian) cannot be topics; 5. (familiar) topic is normally deaccented 16; from the prosodic point of view it is realized with a low tone (cf. L* in Frascarelli and Trecci 2006’s analysis); 6. in predicate-focus structures the subject is the topic if the criteria in 2 and 4 can be applied; 7. when two possible candidates for topic function are plausible in a clause (cf. Lambrecht 1994; Grobet 2002: 120), if the above conditions are met, we choose to count as (primary) topic only the referent coded by the subject constituent, so to say the grammaticalized topic, as in the following example: (10) He married Rosa (c. 1) (he refers to the topic John, already introduced before) but he didn’t really love her (c. 2) (we only counted he as topic of c. 2, even if also Rosa, her, can be considered a possible, secondary topic of c. 2)17

362 Marina Chini 8. usually a post-verbal subject in Italian (cf. Bonvino 2005) is not considered a topic, unless it appears in right-dislocations structures (deemphatic, as in Ha comperato una casa, Giovanni ‘John has bought a house’, lit. ‘Has bought a house, John’); 9. in Italian topic can also be marked by constructions like per quanto concerne X / quanto a X” ‘as for’ (that can be placed to the left periphery of the sentence or sometimes to the right one), by left or right dislocations (il caffé non lo bevo or: non lo bevo, il caffè ‘Coffee, I do not drink’) or by topicalizations without clitic pronouns (hanging topic or freies Thema: le mele, non è mica stagione adesso ‘apples, it is not the right season now’, Berretta 1995: 313); 10. in some cases topic identification is highly controversial, for instance in sequences like: (11) X walks Y sees X (or: X is seen by Y) Y calls X

(c. 1; here X = topic, if already introduced before) (c. 2; X = topic, given that, and if, Y is new, not accessible) (c. 3: here we chose to identify Y, an active subject, as the topic, and to consider X a secondary topic, unless we had: X is called by Y)18

In too ambiguous cases (where the preceding criteria are not sufficient or clash) we do not assign topical status to referents; often combined formal criteria (deaccented referents, subject role, light codification means) help us to take a reasonable, though not always undisputable decision. Discourse topic, which is not central here, often coincides with the sentence topic. When the sentence topic keeps constant, we find the so-called ‘topic continuity’, an important structural feature of narrative discourse and cohesion (Givón 1983; von Stutterheim and Klein 1989), which can be realized through different morphosyntactic devices in different languages. Especially, but non only in pro-drop languages like Italian and Spanish, we often have long light topic chains expressed by zero pronouns (ZP) or zero anaphors (Z; cf. also Du Bois’ 1987 Preferred Argument Structure), such as in the following text, produced by an Italian native speaker in Italian (the protagonist, whose references are in Roman type and underlined, is Charlie Chaplin; M= main; DEP = dependent; c. = clause; the numbers on the left refer to the number of the clause):

Acquiring the grammar of topicality in L2 Italian 363

(12) ROB:

180

185

190

a questo punto allora 0 (ZP) ha capito come: /come 0 (ZP) deve fare per 0 (Z) andare in prigione e quindi 0 (ZP) va in un ristorante 0 (ZP) si fa portare: un/ – anzi no è un self service 0 (ZP) prende: prende due vassoi 0 (ZP) li riempie veram/stracolmi + infatti il/il suo tavolo è pieno quando 0 (ZP) ha finito eh 0 (ZP) si alza 0 (ZP) prende il conto passa dalla vetrina un poliziotto

‘so (he) understands’ (M c.) ‘how (he) has to do’ (DEP c.) ‘to 0 go to prison’ (DEP c.) ‘so (he) goes to a restaurant’ (M c.) ‘(he) orders a/no’ (M c.) ‘it is a self service restaurant’ (M) ‘he takes two trays’ (M c.) ‘(he) fills them up’ (M c.) ‘his table is practically full’ (M c.) ‘when (he) has finished’ (DEP c.) ‘(he) stands up’ (M c.) ‘(he) takes the bill’ (M c.) ‘a policeman walks along the window outside’ (M c.) 0 (ZP) lo chiama dentro ‘(he) calls him in’ (M c.) ‘(he) goes to the cash desk’ (M c.) 0 (ZP) va alla cassa ‘at the cash desk (he) says’ (M c.) e: alla cassa 0 (ZP) di(ce) “no/non ho soldi” ‘I have no money’ (Rep. Speech) (M c.) allora il poliziotto lo (CLIT) prende ‘so the policeman takes him’ per/per (0) portarlo (CLIT) in prigione ‘to lead him to prison’ (DEP c.) (ROBERTA, Italian L1, cc. 175–193)

Such topic chains referred to the protagonist are highly cohesive devices that are very frequent in Italian native speakers’ narratives, but can also occur in German narratives in sequences of coordinated clauses with a constant subject. Other minor topics (in 12 the policeman) can appear in some passages and can be treated as local topics, being referred to by similar empty means (cf. the zero anaphora in c. 193). Appropriate usage of similar and other referential means in the topic entity domain requires a certain degree of L2 command and knowledge of L2 grammar and discourse patterns related to the distribution of given and new information in texts and is possibly influenced by L1 patterns. Cross-linguistic analysis of L1 and L2 (mainly narrative) texts by Berman and Slobin (1994), Carroll, Lambert, von Stutterheim, Watorek, Hendriks and colleagues (cf. parr. 0, 1) point out that in Italian, Spanish, French and English narratives the grammatical subject typically codes topic information (with pro-drop features in Italian and Spanish, with a non-pro-drop subject in a fixed position in French and English), while in German any constituent in Vorfeld (preverbal) position can code a topic information (not only the subject, but also objects, places etc.).

364 Marina Chini Since work done on advanced learner varieties (cf. Carroll and Lambert 2003) shows that structural and textual features typical for the L1 can have an influence also on advanced learner languages, it can be useful to take into consideration some structural divergences between the languages considered (under 13): (13) Comparison between some syntactic features of the three source languages ITALIAN

SPANISH

GERMAN

PRO-DROP

yes

yes

no

SUBJECT TYPICALLY CODES TOPIC

yes

ZERO-TOPIC

no

yes, but mainly Actor no

yes, but also focus yes19

WORD ORDER

free, mostly SVO/VS yes

free, mostly SVO/VS yes

V2; SOV in sub. no

CLITIC PRONOUNS

We can observe some structural differences between Italian and Spanish on the one side, and German on the other side. This could also imply some divergences in structural patterns and topic treatment in discourse produced by speakers of these three languages, especially between Romance speakers and German speakers. We can verify this in the data. 5. Analysis of topical referents In the following sections we examine topic selection and referential forms employed by our subjects in the first episode of the film retelling.20 The phenomena analysed are documented in examples (12) and (19) regard Italian L1, whereas examples (14–18) and (20) concern Italian L2; in brackets we provide the analysis of the relevant referential expressions (which are in Roman type) within the translation of the clause (cf. abbreviations in § 4). 5.1. Topic selection In the clauses dedicated to the first episode (in total 218 cc. for the Italian native speakers = ITA1, 188 cc. for the Spanish learners = SPA1-ITA2 and 228 cc. for the German learners = GER1-ITA2) Charlie is the most frequently selected topic, being also the global discourse topic of the whole retelling, but other (local) topics are possible, in slightly different proportions (cf. Table 1 21).

Acquiring the grammar of topicality in L2 Italian 365 Table 1. Modern Times – I episode: TOPICS (10 subjects per group) TOPICS

ITA1

GER1- ITA2

SPA1- ITA2

n.

% cc.

%

n.

% cc.

CHARLIE DEMONSTRATION POLICE LORRY FLAG

117 17 23 13 11

54.0 8.0 10.5 6.0 5.0

64.6 9.4 12.7 7.2 6.1

69 13 27 12 14

38.0 7.0 14.0 6.0 7.5

Sum

181

83.0 100.0

135

n. cc.

218

188

%

n. % cc.

51.1 108 9.6 23 20.0 26 8.9 21 10.4 7

72.0 100.0

185

47 10 11 9 3

% 58.4 12.4 14.0 11.4 3.8

81 100.0

228

The proportion of clauses having one of the above-mentioned topics is almost identical in Italians’ and German learners’ texts (81–83 %), while it is more reduced in Spanish L1 learners’ texts (72 %), as they use more thetic clauses (Sasse 1995), i.e. clauses without a topic. As far as topic selection is concerned, we see similar, but not identical, choices in native speakers and learners. Animated referents are the most frequently selected ones; the flag is the only really inanimate entity which is sometimes treated as the topic of a clause (the other one, the lorry, can only partially be considered inanimate, as it is driven by a man); the Spanish learners seem to be the relatively most prone to promote the flag to topic (10 % of the cc. with one of the selected topic, vs. 4–6 %). In general in native speaker retellings clauses seem to centre their attention on the discourse topic, Charlie, more frequently than learner clauses (54 % of the clauses vs. 38–47 %; 65 % of the cc. with one of the selected topic, vs. 58 % German learners, 51% Spanish learners), while the learners give relatively more space to other topics as well (especially to the police); less central referents are also mentioned in Italians’ retellings, but gain topical status to a smaller extent than they do in learners’ texts. Let us now look at the way topics are syntactically treated in the narratives considered. 5.2. Topical referents and the role of subject in the first episode We now look at the percentages of clauses where the referents of Table 1 (that are often topical in the clauses of this episode) play the role of syntactic subject in the three groups of narrations. We also take into consideration the number of clauses making explicit or implicit reference to the protagonist (row “Charlie Chaplin’s presence”).

366 Marina Chini Table 2. Modern Times – I episode: referents being subject (=S) ITA1 n. N. clauses (10 informants)

218

Charlie Chaplin’s presence CH=S POL=S DEM=S LORRY=S INA=S

135 119 30 21 16 20

% cc.

SPA1- ITA2 n.

% cc.

188 61.9 54.6 13.8 9.6 7.3 9.2

100 75 37 23 20 24

GER1- ITA2 n.

% cc.

228 53.2 39.8 19.7 12.2 10.6 12.8

134 112 36 29 26 13

58.8 49.1 15.8 12.7 11.4 5.7

Table 2 gives a picture similar to Table 1: the syntactic subject is often also the sentence topic (unless it is a first mentioned or postverbal subject). Furthermore, the most selected topic and subject coincides with the protagonist, Charlie, which is cited or alluded to in the majority of the clauses of the retellings, in particular in Italian native speakers’ retellings (62% of the cc. vs. 53–59%). Therefore the protagonist plays an important role in the episode, especially in native speaker texts; he is in fact the continuous topic warranting textual coherence. The difference between the amount of subjects coinciding with the protagonist, Charlie (CH = S), and the other ones is normally high, but it is especially evident in Italian native speakers (four times more than the second referent, the police, the antagonist), while in German learner retellings it is three times more and in those of the Spanish learners two times more: learners assign the subject role of the clauses to other minor referents more frequently than Italian native speakers do. This is particularly true for Spanish learners, who turn out to be, in a certain sense, the most dispersive ones as for the choice of the subject referent. Spanish learners are also the most prone to assign subject status to inanimate entities (13%), much more than German learners (6%) and relatively more than Italian native speakers (9%). A similar observation was also made by Carroll and Lambert (2003: 272) who found that reference (as a subject) to inanimate entities is much less frequent in German L1 (24.5%) than in Italian L1 (37.1%) and Spanish L1 (30.9%); on the contrary a stronger tendency in German narrators to focus attention on the protagonist is attested by Carroll and Lambert. This tendency seems to emerge also in our German learners, who are more concentrated on the protagonist than Spanish learners (although less than Italian native speakers).

Acquiring the grammar of topicality in L2 Italian 367 Table 3. Modern Times – Clauses with topical referents having the subject (S) role in the first episode ITA1

SPA1- ITA2

GER1- ITA2

Clauses with:

Approx. ratio

Approx. ratio

Approx. ratio

S protagonist / S antagonist S protagonist / S flag

119 /30 = 4 /1 119 /20 = 6 /1

75 /37 = 2 /1 75 /24 = 3 /1

112 /36 = 3 /1 112 /13 = 10 /1

Total cc. (10 informants)

218

188

228

To have a more complete overview, in the following table we present the syntactic role played by the most important referents in the given episode. These syntactic choices have to do with the “function assignment” (to information units, referents) above mentioned (§1). Table 4. Modern Times – I episode: syntactic role of the main referents ITA1 n. N. clauses (10 informants)

218

CH =S CH =O CH =IO CH =OBL POL =S POL =O POL =IO POL =OBL DEM= S DEM=O LOR =S LOR =O INA =S INA =O

119 10 1 1 30 0 0 6 21 12 16 8 20 28

% cc.

SPA1- ITA2 n.

% cc.

188 54.6 4.6 0.5 0.5 13.8 0.0 0.0 2.7 9.6 5.5 7.3 3.7 9.2 12.8

75 21 0 2 37 1 0 1 23 4 20 2 24 27

GER1- ITA2 n.

% cc.

228 39.9 11.2 0.0 1.1 19.7 0.5 0.0 0.5 12.2 2.1 10.6 1.1 12.8 14.4

112 14 0 4 36 0 0 2 29 4 26 3 13 31

49.1 6.1 0.0 1.7 15.8 0.0 0.0 0.9 12.7 1.8 11.4 1.3 5.7 13.6

All three groups of informants assign the syntactic function of subject more frequently than other functions (object, oblique, etc.) to the animate referents (including the lorry, controlled by a driver): this corresponds to the fact that

368 Marina Chini they often play the role of controller of actions and of sentence topic. On the contrary, as for inanimate entities, the object function typically predominates over the subject function, probably because of their frequent semantic role of patient or theme (inanimates are typically objects affected by actions carried out by animates). This is especially true for German learners and Italian native speakers, but the difference is slight for Spanish learners who, as we saw before, are more inclined to assign the subject role (also) to inanimate entities. If we consider other animate referents, that are sometimes topic, and compare their syntactic codification to the one of the protagonist, we see that also for them the subject role prevails, particularly over the object role (6 –7 to 1 for the demonstration, 9–10 to 1 for the lorry driver), while the Italian native speakers use for them more often non subject roles than for the protagonist (10 subject role to 1 non subject for the protagonist, 2 to 1 for the demonstration and the lorry driver). Table 5. Modern Times – Syntactic role played by some referents (S = subject, O = object) in the first episode ITA1 Clauses with:

n.

protagonist S / other roles demonstration S / O role Lorry S / O role

119 /12 21 /12 16 / 8

N. tot. cc. (10 s.)

218

SPA1- ITA2

ratio 10 /1 2 /1 2 /1

n. 75 /23 23 / 4 20 / 2 188

GER1- ITA2

ratio 3 /1 6 /1 10 /1

n. 112 /18 29 / 4 26 / 3

ratio 6 /1 7 /1 9 /1

228

The more frequent choice of syntactic roles different from the subject (i.e. object) for referents different from the protagonist iconically marks their less central role in native speakers’ narratives. This choice is definitely less clear in learners’ retellings. It is a subtle textual and syntactic difference, which does not give rise to proper syntactic or pragmatic errors, but to a discourse and information structure which is not completely native-like, in that it is less cohesive and less hierarchically structured around the protagonist than in native speakers’ retellings. In spite of their relatively grammaticalized interlanguages, our post-basic learners do not yet completely manage to organize their narratives according to the same principles applied by native speakers.

Acquiring the grammar of topicality in L2 Italian 369

5.3. Referential means used to codify the main topical referent After having considered, at the level of information selection and function assignment, the topics selected by our informants and the syntactic functions assigned to the topical entities, we turn now to the formal morphosyntactic devices used by them to codify the main discourse referent, Charlie, the discourse topic and topic of the most sentences in the first episode. Also in this respect we find some differences between the three groups of informants.22 Table 6. Modern Times – I episode (all informants): referential means used to codify the protagonist (cf. ex. 19 for ITA1; ex. 15, 17, 18, 20 for SPA1- ITA2; ex. 14, 16 for GER1-ITA2) ITA1

SPA1- ITA2

n. % cc. % CH N. clauses (10 informants) 218

100

CH =NPRO

18

9.2

6

CH =PRO3 CH =CLIT

GER1- ITA2

n. % cc. % CH 188

100

13.7

8

4.3

2.7

4.6

0

16

7.3

12.2

8

3.7

CH =ZP CH =Z

n. % cc. % CH 228

100

8.2

24

10.5

18.8

0

0

0

0

0

29

15.4

29.9

47 20.6

36.7

6.1

17

9.0

7.5

6

2.6

4.7

56 25.7

42.7

33

17.6

4.0

40

17.5

31.2

20

9.2

15.3

6

3.2

6.2

4

1.8

3.1

CH =REL

5

2.3

3.8

1

0.5

1.0

4

1.8

3.1

CH =PSREL

2

0.9

1.5

3

1.6

3.1

3

1.3

2.3

27

12.4

0

0

7

3.1

CH =DEFN

PASSIVES

23

The percentages of the referential means used by the three groups of informants are relatively different and reveal some of the formal features contributing to the divergent discourse style and topic codification choices of native speakers and learners with a different language background. At first we notice the (expected) important presence of the so-called zero pronouns (ZP) referred to the protagonist, the discourse topic, who is often also the topic of the single clauses of the narrative (cf. Table 1): about 18% of the learner clauses and 26% of the native speaker clauses have a ZP subject referring to Charlie, thus displaying the typical light cohesive topic chains (cf. the post-basic German learner Fritz in 14, the Spanish learner Leonor in

370 Marina Chini 15; ex. 12 for the Italian native speaker). This light means is more heavily represented in native speaker retellings (43% CH) than in the learner ones (31–34% CH), but in any case is part of learner discourse and grammatical competence. In the examples below we give a coded analysis of the referents considered (mainly Charlie). (14) FRI: e: ehm ma questo camion ha perso questa bandiera

10

‘but this lorry (LOR=S DEMN-mr) lost this flag (INA=O DEMN-mr)’ e Charlie Chaplin ha visto queso/questo ‘and Ch. Chaplin (CH=S NPRO-sr) saw this’ e ehm 0 ha/ voleva aiutare ‘and 0 (CH=S ZP-mr) wanted to help’ e 0 ha -mes/metto la bandiera [instead of: ha preso la b. ‘took the flag’] ‘and 0 (CH=S ZP-mr) put the flag (INA=O DEFN-sr)’ e 0 ha-ha seguito il camion ‘and 0 (CH=S ZP-mr) followed the lorry (LOR=O DEFN-sr)’

(15) LON: è caduta la bandiera 5

‘the flag (INA=S DEFN-mr VS) fell off’ e lui la prende ‘and he (CH=S PRO3-sr) takes it (INA=O CLIT-mr)’ y: 0 voleva: tornarla: eh a/all’uomo ‘and 0 (CH=S ZP-mr)wanted to give it (INA=O CLIT-mr) back to the man’ e 0 comincia a muoverla [laughs] ‘and 0 (CH=S ZP-mr) begins to move it (INA=O CLIT-mr)’ ma subito arriva: i manifestanti: – ‘but immediately the demonstrators (DEM=S DEFN-er VS FOC) come’

Also explicit pronominal means are used for Charlie (full and unstressed pronouns, relative pronouns), but they are used twice more frequently by learners (26% by Spanish learners, 28% by German learners) than by native speakers (14%), who prefer empty means to refer to the protagonist. The most explicit lexical means (NPRO, DEFN), usually employed at the beginning of the narrative or of an episode, are less frequent (11–12% by Italian and by German learners), especially in Spanish learners’ texts (4%). This datum confirms the results of some previous studies on anaphora in L1 and L2 which show that light referential means are preferred for accessible referents, in conformity with results of typological works24 (cf. Givón’s [1983] accessibility scale of referring expressions in relation to the accessibility of the topic).

Acquiring the grammar of topicality in L2 Italian 371

As for the other referential means used to codify the protagonist, these L2 narratives are divergent from L1 narratives especially as for: 1) the more reduced use of zero anaphors (= Z; we find them in only 2–3% of learners’ clauses, in 9% of native speakers’ clauses); 2) the more important use of stressed pronouns (often subject pronouns), also for highly accessible and active referents, especially in German learners (ex. 16, ALE), three times more than in native speakers, and a bit less (two times more) in Spanish learners (ex. 17, ASU): (16) ALE: c’era un/un: uomo eh – secondo me Charlie Chaplin ‘there was a man – according to me Charlie Chaplin (CH=S; INDN > NPRO-er; VS) S eh: lui ehm eh + ha fatto una passeggiata in una città ‘he (CH=S PRO3-mr) walked in a town’ 1P + è: : passato u: /un camion con una+/con una bandiera ‘a lorry (LOR=S INDN-er; VS FOC) passed by with a flag (INA=OBL INDN-er)’ S non so indietro sul una: cosa non so ‘I do not know behind a thing – I do not know’ S 5 come 0 si chiama ‘what is the name’ 1P + e: : lui ha visto questa bandiera ‘and he (CH=S PRO3-sr) saw this flag (INA=O DEMN-sr)’ 1P e la bandiera è caduta ‘and the flag (INA=S DEFN-mr) fell down’ 1P + e lui ha preso la bandiera ‘and he (CH=S PRO3) took the flag (INA=O DEFN-mr)’

(17) ASU: e- poi siccome lui si sente più importante, più forte (forse) per l’effetto

80

della- droga ‘and then since he (CH=S PRO3-mr-switch f.) feels more important, stronger (perhaps) because of (the) drug(s) (INA=OBL DEFN)’ eh lui prende il pane ‘he (CH=S PRO3-mr) takes (a piece of) bread (INA=O DEFN-mr)’ invece di 0 lasciarlo a questo- ra/ un uomo ‘instead of 0 (CH=S Z-mr) letting it (INA=O CLIT-mr) to this man (MAN=OI DEMN)’ e 0 si sente molto più forte vicino a lui/ di fronte a lui ‘and (he) (CH=S CH=ZP-mr) feels much stronger next to him/in front of him (=compared to him) (MAN=OBL PRO3-mr)’

372 Marina Chini The use of clitic pronouns, which is normally and globally (for other referents) higher in italophones and hispanophones, is reduced here not only for German learners, but also for Italian native speakers, while it is particularly high for hispanophones (2–3 times more25), also because they put Charlie in the object role more often than the other informants (11% of the cc. vs. 5–6% in Italian L1 and L2 by German learners; cf. Table 4). Another difference in syntactic choices and information structure between Spanish learners on the one side and German learners and Italian native speakers on the other has to do with the use of passive verb forms when Charlie is a subject with a semantically patient role. This choice is quite frequent in Italian native speaker narratives (12% of the cc.; ex. 19), is rare in German learners (3%), but is absent from Spanish learners’ texts, who in such contexts prefer active constructions, in which the topical referent is expressed by a clitic object pronoun (ex. 18), a possible option in Italian. On the contrary, some German learners sometimes try and manage (ex. 20) to resort to a passive construction, which is optimal from the point of view of topical cohesion, because it keeps the topical referent in subject position. (18) JCB (Spanish L1): poi la policia l’ha visto ‘then the police (POL=S DEFN) saw him (CH=O CLIT-sr)’ e- lui- correva ‘and he (CH=S PRO3-mr) ran (away)’ ma 0 l’hanno preso ‘but (they) (POL=S ZP-mr) caught him (CH=O CLIT-mr)’ (19) ALB (Italian L1): ehm + Charlie Chaplin si nasconde in un tombino ‘Charlie Chaplin (CH=S NPRO-sr) hides in a manhole cover’ tuttavia 0 viene scoperto ‘still (he) (CH=S ZP-mr PASS.) is detected’ 0 viene preso dai poliziotti ‘(he) (CH=S ZP-mr PASS) gets arrested by the policemen (POL=OBL DEFN)’ e 0 viene caricato su: una camionetta ‘and (he) (CH=S ZP-mr PASS) is loaded on a jeep’

Acquiring the grammar of topicality in L2 Italian 373

(20) CORnelia (German L1) e lui è per caso è avanti a questa/questa massa ‘and by chance he (CH=S PRO3-sr) is in front of this crowd (DEM=OBL DEMN-mr)’ e sempre eh 0 fa segni con/con questa: /eh con questa bandiera rossa ‘and (he) (CH=S ZP-mr) always beckons with this red flag (INA=OBL DEMN-sr)’ 20 e 0 grida ‘and (he) (CH=S ZP-mr) shouts’ + ehm e poi + ehm come sempre Ch. Chaplin non è molto fortunato ‘and then as usual Charlie Chaplin (CH=S NPRO-mr) is not very lucky’ co/e viene la polizia ‘and the police (POL=S POL=DEFN-er VS FOC) come’ e: ehm lui eh viene arrestato eh della polizia ‘and he (CH=S PRO3-mr PASS) gets arrested by the police (POL=OBL DEFN-mr)’

6. Conclusion Our comparative analysis on topic treatment and grammatical codification in Italian L1 and L2 narratives confirms some trends already pointed out in other works and adds new data on this subject: a. as for topic selection it turns out that the global thematic priority of the protagonist, the discourse topic, is also reflected in the fact that he is often also the sentence topic in the narratives analysed here. Nevertheless his thematic priority is much clearer in Italian native speaker retellings than in learners’ ones, as he is more coherently kept as topic of the single clauses by native speakers than by learners (Table 1), whose narratives turn out to be in a sense more dispersive. Thus it seems that, in order to become a competent narrator in Italian, a learner must learn to keep his attention more constantly on the protagonist and to assign topic status more often to him/her in the clauses of his/her narration. This cohesive choice can be found, yet it is not adequately handled by the learners considered here; b. as for the syntactic codification of the topic, the sentence topic often coincides with the subject of the clause in these narratives. Also the discourse topic is often codified as the subject of the sentence, but in different pro-

374 Marina Chini portions in the three groups of data: more often in Italian native speakers’ data, then in German learners’ data, least in Spanish learners’ data, where subject (and topic) status is also not rarely assigned to other, even inanimate, referents (Tables 1 and 4). In this respect textual and syntactic continuity of topic is tighter in German learners’ narratives than in Spanish learners’ ones; c. as for the referential means used for the central topic, i.e. the most accessible referent, light anaphoric means (mainly chains of zero pronouns, sometimes clitic pronouns) are generally attested, a bit more often in native speakers’ data than in learners’ data, where full pronouns are more frequent. On the one side this result conforms to typological trends (cf. Givón’s [1983] accessibility scale), on the other it confirms the tendency to a more explicit referential style often attested also in other L2 data (cf. Ahrenholz 1997; Muñoz 1995); d. at first sight the source language seems to have an impact on learner choices in the domain studied, in that Spanish learners (with a pro-drop L1) show a larger use of zero pronouns and less difficulty in the use of clitic pronouns than German learners (non pro-drop L1). Anyhow the L1 is not the whole story. Also Spanish learners sometimes overuse full pronouns (especially in subject role, ex. 17), not only German intermediate learners of Italian overuse them. In conclusion, our results show that language-specific preferences regarding topical strategies and cohesive patterns are only partially mastered by postbasic and (more) advanced learners of Italian and that they are strongly intertwined with language-specific grammatical means. They tend to be acquired rather late, especially if they imply the use of marked structures (i.e. passive) and forms (clitic pronouns). The acquisition of that part of the Italian information structure component concerning topic treatment is still in progress in these learner varieties and seems to be influenced not only by general typological trends and grammaticalization patterns (for instance: from lexical to grammatical referential means, cf. Chini 200526), but also, subtly, by L1 grammatical patterns and textual preferences, attested in works cited before (ex. Carroll and von Stutterheim 2003) and in the retellings in Spanish and German L1 produced by our learners after the retelling in L2 Italian we could not analyse here. Furthermore, discourse cohesion in L2 Italian turns out to be mainly based on topic and subject continuity in German learners’ narratives, like in Italian L1 narratives, although through sometimes too explicit means (full pronouns, lexical anaphors), while Spanish learners express it through light anaphors, like in Italian L1, less through topic continuity (unlike Italian),

Acquiring the grammar of topicality in L2 Italian 375

given that their attention is directed towards the agent more than towards the topic (cf. also Murcia Serra 2003), as it appears from the avoidance of passive structures referred to the protagonist (cf. ex. 18; Table 6). In sum, learning the “topic grammar” in L2 seems to imply the acquisition not only of the morphosyntactic means typical for topic codification in a specific L2, but also of some preferences in function and perspective assignment to textual information units. Further research is necessary to confirm the trends identified in these data.

Notes 1.

2.

3.

4.

Cf. inter alia Givón (1983, 1984), Huang (2000), Giacalone Ramat (2003a, 2003b, ch. 1) as for typological works; Berman and Slobin (1994), Hickmann et al. (1996), Hickmann (2003) as for text in several L1s (often not Italian); on text in L2, cf. Carroll et al. (2000), Carroll and Lambert (2003), Carroll and von Stutterheim (2003); one of the first Italian publications on text organisation both in L1 and in L2 is Chini and Giacalone Ramat (1998). By information structure we refer to various formal (mainly morphosyntactic, prosodic) means expressing the pragmatic structure of a proposition in discourse (Lambrecht 1994: 5), topic and focus relations, the mental representation of discourse referents as [+/–identifiable] and as [+/–active]. Cf. also infra. The syntax-phonology interface has been widely discussed, especially, but not only, in the Minimalist program (cf. Chomsky 1995); it is not possible to give a sufficiently representative account of it here. Works on this interface normally reject the hypothesis that the study of notions such as topic and focus should be confined to a ‘grammar of discourse’, based only on pragmatic appropriateness, and maintain that “there is a specific, basic connection between their formal structure and their use” (Frascarelli 2000: 196, 208). According to some formal analyses (cf. Rizzi 1997’s cartography of the left periphery) discourse grammar is not free from syntax, but is encoded in a hierarchy of functional projections and is dependent on syntactic constraints and requirements. In respect to Italian, Frascarelli (2000) has argued for a phonological grammaticalization of Focus and Topic in Italian (to which she adds cross-linguistic support), and for a systematic connection between the syntactic and the phonological realization of “subjects” and their discourse properties (Frascarelli and Trecci 2006). New interpretations of the interface syntax-phonology have recently been presented (e.g. by Mereu and Frascarelli 2005), which sometimes go beyond the Minimalist program. The features of discourse representation are: a) selection: in narrating the speaker selects only some pieces of information, some events and subevents, gives global accounts or more detailed reports;

376 Marina Chini

5.

6.

7.

8.

9.

b) linearisation: the conceptual units (in narratives events and subevents) must be sequentially ordered according to a certain criterion (often in everyday contexts it is natural order); c) addition: comparisons, explanations, comments, evaluations can be added to the simple main structure answering the quaestio; these added parts belong to the so-called “side structure”; d) function assignment: this has to do with the speaker’s decision of what to express and in which way. Other approaches (i.e. Alternative Semantics) suggest that the topic of a sentence is the set of alternatives induced by the focus part of the sentence; similarly Klein and von Stutterheim (1987) take as topic the ‘alternative’, that is “the choice between one or more possibilities” as an answer to the question posed; “for all the approaches in terms of questions, the criterion for a sentence to have a topic is that it answers a question” (van Kuppevelt 1994). With respect to Lambrecht’s (1994) definition, Sornicola’s (2006) one is more clearly related to the utterance level (not to the propositional one), but is still in terms of aboutness (cf. her test for topichood: test for X = topic, ‘As for X, I am telling you that X [+ comment]’; Sornicola 2006). The reality is more complicated: normally topics do not carry prosodic stress, but can be characterised by specific prosodic contours and prosodic salience (cf. infra). In a similar recent approach to topic in dialogues (Grobet 2002) topic is defined as an “immediate anchoring point” for minimal discourse units, which is characterised by identifiability, activation state, aboutness (à propos) relation; furthermore a topic identification process in several stages is proposed, as well as some linguistic (mainly lexical and syntactic) and discourse relevant factors (saliency, discourse structure, etc.). This is not obligatory. Focus is placed towards the end of an utterance mainly in cases of “new, completive focus” (Dik 1989); sentence-initial focus are attested for example in cases of contrastive focus (JOHN has come, not James). As for the notion of focus we refer to Lambrecht (1994: 207), according to whom the focus “is seen as the element of information whereby the presupposition and the assertion DIFFER from each other. The focus is that portion of a proposition which cannot be taken for granted at the time of speech. It is the UNPREDICTABLE or pragmatically NON-RECOVERABLE element in an utterance. The focus is what makes an utterance into an assertion” (cf. also Dik 1989: 277, 282–283). Aboutness-shift topics propose a matter of current interest and exhibit a L*+H contour (with rise in the F0 contour on the tonic syllable, underlined here: l’ultima unit la sto facendo ‘ I’m doing the last unit (now)’); contrastive topics are elements which create oppositional pairs with respect to other topics and have a H* contour (col francese benissimo ‘with French, I’m fine (as opposed to English)’; for familiar topic, see below (cf. Mereu and Trecci 2004; Frascarelli and Trecci 2006; Frascarelli and Hinterhölzl 2007).

Acquiring the grammar of topicality in L2 Italian 377 10. These models are not always satisfying, especially in the case of non configurational languages and of non configurational features of configurational languages (cf. alternative views in Mithun 1987; Mereu 2004, ch. 5–6). 11. Frascarelli argues for the extrasentential generation of Topic: “no featurechecking process is associated with it, hence no movement is necessary for its interpretation” (Frascarelli 2000: 14). As for focus information, it is contained in a strong feature, checked in overt syntax by the verbal Head and thus made visible at interfaces (Frascarelli 2000: 14, 208; cf. also chap. 3). 12. In some languages lexical NPs can never function as subject arguments (cf. Van Valin 1985; Lambrecht 1994: 191). Du Bois (1987) formulates the “non lexical agent constraint”: the agent is normally not codified lexically, rather the object or the subject of intransitives verbs are. 13. In Korean the opposite is true: the contrastive focus takes a topic marker (Mereu and Trecci 2004). 14. The three groups of informants are composed in the following way: I. Italian L2 group (learners = ITA2): Ia. with Spanish L1 (SPA1-ITA2): 10 Erasmus students with Spanish L1, age 21–26 years: IsaBEl = IBE, RAQuel = RAQ, JaCoB = JCB, ANUunciación = ANU, AdeLA = ALA, León = LEO, LeONor = LON, DAMián = DAM, ASUnción = ASU, MaRCelo = MRC. They are all post-basic learners of Italian, who had lived in Pavia for 5–9 months; they had never studied Italian before coming to Pavia; Ib. with German L1 (GER1-ITA2): 10 Erasmus students with German L1, age 22–25 years (ALExia, ANGela, ANTon, CHRistine, CLAra, DORsten, CORnelia, FRAnz, FRItz, GISela). On the basis of their decreasing morphosyntactic competence (assessed against the results of the Pavia Project on L2 Italian; cf. Giacalone Ramat 2003), we can identify some more advanced (ANG; FRA, ALE, COR), some intermediate (CHR, CLA, ANT) and some post-basic learners (GIS, DOR, FRI) of Italian. In this study we consider them together, without taking their differing levels of competence into consideration. These subjects are part of a larger corpus which was analysed in other studies (i.e. in Chini 2005). II. Italian L1 group (native speakers = ITA1): 10 Italian university students studying at Pavia University. 15. Syntactic status of the clauses (main or subordinate clause) and special constructions (presentational, passives, dislocations, etc.) are analysed in other recent works (Chini 1998, 1999, 2002a, 2002b). 16. We will not consider contrastive topics and newly introduced /reintroduced topics (aboutness-shift topics; cf. supra, §3) which are characterised by different prosodic features, respectively by high tones (H*) and by complex (L+H)* tones in Italian (Frascarelli and Trecci 2006; cf. Marotta’s 2000 notation proposal), but also in German (Frascarelli and Hinterhölzl 2007). 17. “Although the sentence primarily adds to our knowledge of John (John being the primary topic), it also has the effect of increasing our knowledge of Rosa,

378 Marina Chini

18.

19.

20.

21.

22.

23. 24.

25.

by informing us that she was not loved by her husband. Both John and Rosa are under discussion at the time the clause [c. 2] is uttered” (Lambrecht 1994: 148). In spontaneous usage, sequences like this in Italian would sound: X passeggia (c. 1), Y lox vede, or better: lox vede Y (c. 2), e 0y lox chiama (c. 3). The topic expression of the cc. 2 and 3 would correspond to the least explicit codifications (the clitic pronoun in c. 2 and the zero subject pronoun in c. 3). Before c. 2 some informants could opt for producing an additional clause, in order to introduce Y (ex. Arriva Y ‘Y arrives’); in that case the topic of the clause Y lox vede would be Y (often codified by a zero or a relative pronoun: Arriva Y / che y lox vede or: Arriva Y / e 0y lox vede). Cf. Lillo-Martin (1992) for a discussion. German is considered a zero-topic language in that it allows the omission of topics irrespective of whether they are subject or object, while Spanish and Italian (pro-drop languages) only allow subject pronoun dropping. During the first episode the protagonist, Charlie Chaplin, accidentally gets involved in a demonstration, because he is shaking a flag which fell from a lorry. So he gets arrested by the police as the chief of the strikers. The first percentage (% cc.) refers to the frequency of occurrence for given referents as topics in all the clauses considered; the second one (%) refers to the percentage of clauses having one of these referents as topic. We only considered retellings in Italian, L1 and L2. The sum in the second column is not 100 because clauses do not always have a topic and only a selection of topics is reported in the table (the most important are). In Table 6, besides the absolute figures, we give the percentages calculated on the total number of clauses (= % cc.), and the percentages referred to the total number of references to Charlie (= % CH: n. 131 for ITA1, n. 97 for SPA-ITA2, n. 128 for GER1-ITA2); we also give the number of passive verb forms we will briefly comment on later. We are not referring here to the regional (Northern Italian) construction ‘definite article + proper noun’ (ex. il Charlot), which is not attested in our data, but to the definite (standard) NP il protagonista ‘the protagonist’. Typological works have analysed the relationship between topic, topicality and referential means used to code topic in several types of languages. The problem of keeping track of referents is in fact a significant area of interaction between syntax and pragmatics, form and function (cf. Van Valin 1987). Languages have developed several lexical, grammatical and pragmatic means to solve it: grammatical gender, switch function and switch reference at the grammatical level; pragmatically constrained inferential systems (Haiman and Munro 1983), where referents are sometimes not mentioned (“pragmatic” zero anaphora), but have to be inferred on the basis of the hearer’s linguistic and socio-cultural knowledge (i.e. Japanese, Mandarin, Thai). We did not count as occurrences of clitic pronouns the cases of clitic doubling which are normal in Spanish and sometimes attested also in our Italian L2 data (ex. IBE: 0 le ha preso a Charlot ‘(the police) caught Charlie’).

Acquiring the grammar of topicality in L2 Italian 379 26. Cf. some details on the development of referential strategies, subordination and word order in German learners’ narratives in Chini (1999, 2002, 2003b, 2005).

References Ahrenholz, Bernt 1997 Modalità e movimento referenziale in istruzioni. Un confronto fra istruzioni orali prodotte in tedesco e in italiano L1. Studi Italiani di Linguistica Teorica e Applicata 26 (3): 449–478. Andorno, Cecilia 2000 Focalizzatori fra connessione e messa a fuoco. Il punto di vista delle varietà di apprendimento. Milano: Franco Angeli. 2003 Linguistica testuale. Un’introduzione. Roma: Carocci. Andorno, Cecilia, Giuliano Bernini, Anna Giacalone Ramat, and Ada Valentini 2003 Sintassi. In Verso l’italiano. Percorsi e strategie di acquisizione, Anna Giacalone Ramat (ed.), 116–178. Roma: Carocci. Ariel, Mira 1988 Referring and accessibility. Journal of Linguistics 24: 65–87. Berman, Ruth A. and Dan I. Slobin (eds.) 1994 Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Bernini, Giuliano 1995 Au début de l’apprentissage de l’italien. L’énoncé dans une variété prébasique. Acquisition et Interaction en Langue Etrangère 5: 15–45. Bernini, Giuliano and Marcia L. Schwartz (eds.) 2006 Pragmatic organization of Discourse in the Languages of Europe. Berlin /New York: Mouton de Gruyter. Berretta, Monica 1986 Per uno studio dell’apprendimento dell’italiano in contesto naturale: il caso dei pronomi personali atoni. In L’apprendimento spontaneo di una seconda lingua, Anna Giacalone Ramat (ed.), 329–352. Bologna: Il Mulino. 1990 Catene anaforiche in prospettiva funzionale: antecedenti difficili. Rivista di Linguistica 2 (1): 91–120. 1994 Ordini marcati dei costituenti di frase in italiano. La frase scissa. Vox Romanica 53: 79–105. 1995 Ordini marcati dei costituenti maggiori di frase: una rassegna. Linguistica e Filologia 1: 125–170. Bonvino, Elisabetta 2005 Le sujet postverbal: une étude sur l’italien parlé. Paris: Ophrys. Carroll, Mary and Monique Lambert 2003 Information structure in narratives and the role of grammaticised knowledge. In Dimroth and Starren (eds.), 267–287.

380 Marina Chini Carroll, Mary and Monique Lambert 2005 Problems in reorganising principles of information structure in advanced L2s: a study of French and German learners of English. Georgetown University Round Table on Languages and Linguistics. Carroll, Mary and Christiane von Stutterheim 1997 Relations entre grammaticalisation et conceptualisation et implications sur l’acquisition d’une langue étrangère. Acquisition et Interaction en Langue Étrangère 9: 82–115. 2003 Typology and information organisation: perspective taking and language-specific effects in the construal of events. In Typology and second language acquisition, Anna Giacalone Ramat (ed.), 365–402. Berlin /New York: Mouton de Gruyter. Carroll, Mary, Jorge Murcia-Serra, Marzena Watorek, and Alessandra Bendiscioli 2000 The relevance of information organization to SLA studies. The descriptive discourse of advanced learners of German. Studies in Second Language Acquisition 22: 441–466. Chafe, Wallace L. 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Subject and Topic, Charles N. Li (ed.), 27–55. New York: Academic Press. Chini, Marina 1998 La subordinazione in testi narrativi di apprendenti tedescofoni: forma e funzione. Linguistica e Filologia (Università di Bergamo) 7: 121– 159 1999 Riferimento personale e strutturazione di testi narrativi in italofoni e in apprendenti tedescofoni di italiano. In Grammatik und Diskurs/ Grammatica e discorso. Studien zum Erwerb des Deutschen und des Italienischen, Norbert Dittmar and Anna Giacalone Ramat (eds.), 213–243. Tübingen: Stauffenburg. 2002 Ordres marqués et perspective du locuteur en italien L2. Revue Française de Linguistique Appliquée 7 (2): 107–127. 2003a Aspetti della testualità (Introduzione. L’anafora). In Verso l’italiano. Percorsi e strategie di acquisizione, Anna Giacalone Ramat (ed.), 179–200. Roma: Carocci. 2003b Le phénomène de la jonction interpropositionnelle dans la narration en italien L2: entre agrégation et intégration. Acquisition et Interaction en Langue Etrangère 19, 71–106. 2005 Reference to person in learner discourse. In Henriëtte Hendriks (ed.), 65–110. Chini, Marina and Anna Giacalone Ramat (eds.) 1998 Strutture testuali e principi di organizzazione dell’informazione nell’ apprendimento linguistico. Studi Italiani di Linguistica Teorica e Applicata 27 (1).

Acquiring the grammar of topicality in L2 Italian 381 Chomsky, Noam 1965 Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Cinque, Guglielmo 1990 Types of A’ dependencies. Cambridge, MA: MIT Press. Conte, Maria-Elisabeth 1988 Condizioni di coerenza. Ricerche di linguistica testuale. Firenze: La Nuova Italia. Dik, Simon 1978 Functional Grammar. Amsterdam: North Holland. 1989 The Theory of Functional Grammar. Dordrecht: Foris. Dimroth, Christine and Marianne Starren (eds.) 2003 Information structure and the dynamics of language acquisition. Amsterdam: John Benjamins. Dittmar, Norbert (ed.) 1992 Topic – From Grammar to Discourse. Linguistics 30. Dittmar, Norbert and Anna Giacalone Ramat (eds.) 1999 Grammatik und Diskurs/Grammatica e discorso. Studien zum Erwerb des Deutschen und des Italienischen. Tübingen: Stauffenburg. Du Bois, John W. 1987 The discourse basis of ergativity. Language 63 (4): 805–855. Fox, Barbara (ed.) 1996 Studies in anaphora. Amsterdam: John Benjamins. Frascarelli, Mara 1997 The Phonology of Focus and Topic in Italian. The Linguistic Review 14: 221–248. 2000 The Syntax-Phonology Interface in Focus and Topic Constructions in Italian. Dordrecht: Kluwer. 2006 Subjects, Topics and the interpretation of pro. Ms., University of Roma 3. Frascarelli, Mara and Roland Hinterhölzl 2007 Types of Topics in German and Italian. In On Information Structure, Meaning and Form, Susanne Winkler and Kerstin Schwabe (eds.), 87–116. Amsterdam /Philadelphia: John Benjamins. Frascarelli, Mara and Alessandra Trecci 2006 Subjects in a pro-drop Language. Syntactic roles, discourse categories and the interpretation of pro. In La Comunicazione Parlata. Atti del Convegno Internazionale (23–25 febbraio 2006), Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.). Napoli: Liguori Editore (CD-ROM). Fretheim, Thorstein and Jeannette K. Gundel (eds.) 1996 Reference and referent accessibility. Amsterdam: John Benjamins. Fuller, Judith W. and Jeannette K. Gundel 1987 Topic-prominence in Interlanguage. Language Learning 37: 1–18.

382 Marina Chini Giacalone Ramat, Anna (ed.) 2003a Typology and Second Language Acquisition. Berlin /New York: Mouton de Gruyter. 2003b Verso l’italiano. Percorsi e strategie di acquisizione. Roma: Carocci. Givón, Talmy 1976 Topic, pronoun and grammatical agreement. In Subject and Topic, Charles Li (ed.), 149–188. New York: Academic Press. 1979 From discourse to syntax: grammar as a processing strategy. In Discourse and Syntax, Talmy Givón (ed.), 81–112. New York: Academic Press. 1983 Topic continuity in discourse: An introduction. In Topic continuity in discourse: A quantitative cross-language study, Talmy Givón (ed.), 1–41. Amsterdam: John Benjamins. 1984 Universals of discourse structure and second language acquisition. In Language Universals and second language acquisition, William E. Rutherford (ed.), 109–133. Amsterdam: John Benjamins. Grobet, Anne 2002 L’identification des topiques dans les dialogues. Bruxelles: DeBoeck-Duculot. Gullberg, Marianne 2003 Gestures, referents, and anaphoric linkage in learner varieties. In Dimroth and Starren, 311–328. Gundel, Jeannette 1988 Universals of topic-comment structure. In Studies in Syntactic Typology, Michael Hammond, Edith A. Moravcsik and Jessica R. Wirth (eds.), 209–239. Amsterdam: John Benjamins. Gumperz, John and Stephen C. Levinson 1996 Rethinking linguistic relativity. Cambridge: Cambridge University Press. Haiman, John and Pamela Munro 1983 Switch reference and universal grammar. Proceedings of a symposium on switch reference and universal grammar. Amsterdam: John Benjamins. Halliday, Michael A. K. 1967 Notes on transitivity and theme in English. Part II. Journal of Linguistics 3: 199–244. Hendriks, Henriëtte 1998 Reference to person and space in narrative discourse: a comparison of adult second language and child first language acquisition. In Marina Chini and Anna Giacalone Ramat (eds.), 67–86. 2000 The acquisition of topic marking in L1 Chinese and L1 and L2 French. Studies in Second Language Acquisition 22: 369–397.

Acquiring the grammar of topicality in L2 Italian 383 Hendriks, Henriëtte (ed.) 2005 The structure of learner varieties. Berlin /New York: Mouton de Gruyter. Hickmann, Maya 2003 Children’s Discourse: Person, Space and Time across Languages. Cambridge: Cambridge University Press. 2005 Determinants in first and second language acquisition: person, space, and time in discourse across languages. In Henriëtte Hendriks (ed.), 231–262. Hickmann, Maya, Henriëtte Hendriks, Françoise Roland, and James Liang 1990 [19942] The development of discourse cohesion: A Coding manual. Nijmegen, Max Planck Institut für Psycholinguistik. 1996 The marking of new information in children’s narratives: a comparison of English, French, German, and Mandarin Chinese. Journal of Child Language 23: 591– 619. Huang, Yan 2000 Anaphora. A cross-linguistic approach. Oxford: Oxford University Press. Huebner, Thom 1983 A Longitudinal Analysis of the Acquisition of English. Karoma: Ann Arbor. Jin, Hong G. 1994 Topic-Prominence and Subject-Prominence in L2 acquisition: Evidence of English-to-Chinese typological transfer. Language Learning 44: 101–122. Khorounjaia, Ekaterina and Liliana Tolchinsky 2004 Discursive constraints on the lexical realization of arguments in Spanish. In Language development across childhood and adolescence, Ruth A. Berman (ed.), 83–109. Amsterdam: John Benjamins. Klein, Wolfgang 2006 On finiteness. In Semantics meets Acquisition, Veerle Van Geenhoven (ed.), 245–272. Dordrecht: Kluwer. Klein, Wolfgang and Clive Perdue 1992 Utterance structure. Developing grammars again. Amsterdam: John Benjamins. 1997 The basic variety, or: Couldn’t natural language be much simpler. Second Language Research 13 (4): 301–347. van Kuppevelt, Jan 1994 Topic and Comment. In Encyclopedia of Language and Linguistics. Vol. 9, Ronald E. Asher and J.M.Y. Simpson (eds.), 4629–4633. Oxford: Pergamon Press.

384 Marina Chini Lambrecht, Knud 1994 Information structure and sentence form. Topic, focus, and the mental representations of discourse referents, Cambridge: Cambridge University Press. Lenart, Ewa and Clive Perdue 2004 L’approche fonctionnaliste: structure interne et mise en œuvre du syntagme nominal. Acquisition et interaction en langue étrangère 21: 85–121. Levelt, Willem 1989 Speaking. From intention to articulation. Cambridge, MA: MIT Press. 1996 Perspective taking and ellipsis in spatial descriptions. In Language and Space, Paul Bloom, Mary A. Peterson, Lynn Nadel, and Merrill F. Garrett (eds.), 77–107. Cambridge, MA: The MIT Press. Lillo-Martin, Diane 1992 Comments on Hyams and Weissenborn: on licensing and identification. In Theoretical issues in language acquisition, Jürgen Weissenborn, Helen Goodluck, and Thomas Roeper (eds.), 301–308. Hillsdale, NJ: Lawrence Erlbaum. Marotta, Giovanna 2000 Allineamento e trascrizione dei toni accentuali complessi: una proposta. In Atti delle Decime Giornate del Gruppo di Fonetica Sperimentale, 139–149. Napoli: A.I.A. Matras, Yaron and Hans-Jürgen Sasse (eds.) 1995 Verb-Subject order and theticity in European languages. Sprachtypologie and Universalienforschung 48 (1/2). Mereu, Lunella 2004 La sintassi delle lingue del mondo. Roma: Laterza. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. In Il parlato italiano. Atti del Convegno (Napoli, 13–15 febbraio 2003), Federico Albano Leoni, Francesco Cutugno, Massimo Pettorino, and Renata Savy (eds.). Napoli: M. D’Auria (CDROM) Mithun, Marianne 1987 Is basic word order universal? In Coherence and grounding in Discourse, Russell S. Tomlin (ed.), 281–328. Amsterdam: John Benjamins. Molnár, Valeria 1993 Zur Pragmatik und Grammatik des TOPIK-Begriffes. In Wortstellung und Informationsstruktur, Marga Reis (ed.), 155–202. Tübingen: Niemeyer. Muñoz, Carmen 1995 Markedness and the acquisition of referential forms: The case of zero anaphora. Studies in Second Language Acquisition 17: 517–527.

Acquiring the grammar of topicality in L2 Italian 385 Murcia Serra, Jorge 2003 Acquiring the linkage between syntactic, semantic and informational roles in narratives by Spanish learners of German. In Dimroth and Starren (eds.), 289–309. Rizzi, Luigi 1997 Sintassi: Le strutture. In Il linguaggio. Strutture linguistiche e processi cognitivi, Alessandro Laudanna and Miriam Voghera (eds.), 205–229. Bari: Laterza. Sansò, Andrea 2003 Degrees of event elaboration. Passive constructions in Italian and Spanish. Milano: Franco Angeli. Sasse, Hans-Jürgen 1995 “Theticity” and VS order. In Matras and Sasse (eds.), 3–31. Slobin, Dan I. 1987 Thinking for speaking. In Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistic Society, 435–445. 1996 From “thought and language” to thinking for speaking. In Rethinking linguistic relativity. John J. Gumperz and Stephen C. Levinson (eds.), 70–96. Cambridge: Cambridge University Press. Sornicola, Rosanna 1994 Topic, focus and word-order. In Encyclopedia of Language and Linguistics. Vol. 9, Ronald E. Asher and J. M.Y. Simpson (eds.), 4633– 4640. Oxford: Pergamon Press. 2006 Topic and Comment. In Encyclopedia of Language and Linguistics. 2nd Edition, Vol. 12, Keith Brown (ed.), 766–773. Oxford: Elsevier Stark, Elisabeth 1999 Antéposition et marquage du thème (topic) dans les dialogues spontanés. In La thématisation dans les langues. Actes du colloque de Caen, 9–11 octobre 1997, Claude Guimier (ed.), 337–358. Bern: Lang. Strawson, Peter F. 1964 Identifying reference and truth-values. Theoria 30: 96–118. von Stutterheim, Christiane 1997 Einige Prinzipien des Textaufbaus. Tübingen: Niemeyer. von Stutterheim, Christiane and Mary Carroll 2005 Subjektwahl und Topikkontinuität im Deutschen und Englischen. Zeitschrift für Literaturwissenschaft und Linguistik 139: 7–28. von Stutterheim, Christiane and Wolfgang Klein 1989 Referential movement in descriptive and narrative discourse. In Language processing in social context, Rainer Dietrich and Carl F. Graumann (eds.), 39–76. Amsterdam: Elsevier. 2002 Quaestio and L-perspectivation. In Perspective and perspectivation in discourse, Carl F. Graumann and Werner Kallmeyer (eds.), 59–88. Amsterdam: John Benjamins.

386 Marina Chini Tomlin, Russell 1986 Basic word order: Functional principles. London: Croom Helm. Trévise, Anne 1986 Is it transferable, topicalization? In Crosslinguistic Influence in Second Language Acquisition, Eric Kellerman and Michael Sharwood Smith (eds.), 186–206. New York: Pergamon. Valentini, Ada 1992 L’italiano dei cinesi. Questioni di sintassi. Milano: Guerini. 1994 Soggetti pronominali nell’italiano L2 di sinofoni. In Italiano: lingua seconda, lingua straniera, Anna Giacalone Ramat and Massimo Vedovelli (eds.), 297–318. Roma: Bulzoni. Van Valin, Robert D. 1985 Case marking and the structure of the Lakhota clause. In Grammar inside and outside the clause, Johanna Nichols and A. Woodbury (eds.), 363–413. Cambridge: Cambridge University Press. 1987 Aspects of the interaction of syntax and pragmatics: discourse coreference mechanisms and the typology of grammatical systems. In The pragmatic perspective, Jef Verschueren and Marcella BertuccelliPapi (eds.), 513–531. Amsterdam: John Benjamins. Vallduví, Enric 1992 The Informational Component. New York: Garland. Vater, Hans 19942 Einführung in die Textlinguistik: Struktur, Thema und Referenz in Texten. München: Fink. Zubizarreta, Maria L. 1998 Prosody, Focus, and Word Order. Cambridge, MA: MIT Press.

“Appendix” or “postposed Topic”: Where does the difference lie? Edoardo Lombardi Vallauri

1. The problem with Appendix Although the term Appendix is sometimes used by scholars worldwide to refer to certain units within Information Structure, it is probably in Italy that it has most currency, due to the fundamental work of E. Cresti, to which I refer.1 We will summarize Cresti’s definitions of Information Structure in Section 2. Before that, we will shortly try to show why we have a problem with the concept of “Appendix”. Following Cresti, Ferrari (2004: 15) distinguishes three fundamental categories of Information Structure: (…) considero che l’Unità Comunicativa sia composta da (almeno) tre tipi di Unità Informativa: il Rema, il Tema e l’Appendice. Queste unità vanno considerate – nello spirito di Halliday 1985 – come “contenitori”, come “spazi”, linguistico-informativi che organizzano il contenuto semantico-pragmatico dell’Unità Comunicativa, in modo da definirne il profilo gerarchico, e di conseguenza le modalità della sua inserzione nell’architettura del capoverso e del suo sfruttamento comunicativo. [I see the Communication Unit as being composed of (at least) three kinds of Information Units: the Rheme, the Theme and the Appendix. These units are to be considered – in the spirit of Halliday 1985 – as linguistic-informational “containers”, or “spaces” that organize the semantic-pragmatic content of the Communication Unit, so as to define its hierarchical profile and, consequently, the ways it may participate in the architecture of the paragraph and be exploited in communication.]

We will try to verify what the roles and functions of these units are within the information hierarchy of the utterance. Rossi (1999), explicitly referring to Cresti’s theoretical framework, proposes an analysis of (1) in terms of Comment and “Appendix”: (1)

un nodo / fagli a knot make-to.it C A

(make a knot, to it)

388 Edoardo Lombardi Vallauri Scholars not belonging to the same tradition would analyse it as RhemeTheme. So, the following question arises: If fagli is an Appendix, under what respect is it different from a Theme? In other words, what distinguishes (2) from (1), where fagli is a bona fide Theme, if both utterances are produced within the same context (say, as an answer to the question: Che cosa gli faccio? ‘What shall I do to it’)? (2)

fagli un nodo T C

The difference is not one of illocutionary force. Semantically (and textually), one can maintain that the two utterances have different functions in that the information encoded by the verb in (1) is introduced “out of place”, thus having a more marked effect, i.e., for instance, it has the flavour of a “wink” or “camaraderie” towards the addressee (Berruto 1986: 61), of more explicitly acknowledging that its content is already present in the linguistic context. But fagli still conveys information that, in the manner of a Theme, and differently from a Rheme, doesn’t convey any illocution. However, as we will see right away, in the kind of framework Cresti proposes, categories of information status such as Topic and Comment are usually defined in terms of illocutionary force. This raises the question about the level on which we should locate the distinction between what is usually called Topic (or Theme), and what some call Appendix. 2. Definitions of Appendix In the specific sense it has taken on within the Italian tradition, the term dates back to Cresti (1987). In this paper we will however refer to later formulations, namely to those in Cresti (2000), who on page 51 introduces what she calls the «classi funzionali di profili intonativi» [functional classes of intonation contours]: a) unità nucleari (root), cioè necessarie e bastanti a fare pattern, che abbiamo chiamato comment; b) unità con carattere di opzionalità e subordinazione melodica (prefix) e tuttavia posizione di antecedenza temporale rispetto ad unità nucleari, che abbiamo chiamato topic; c) Unità con carattere di opzionalità e un livello di subordinazione melodica ancora ulteriore (suffix), con occorrenza dopo un’unità di nucleo o addirittura dopo una di antecedenza, che abbiamo chiamato appendice;

“Appendix” or “postposed Topic”: Where does the difference lie? 389 d) unità diverse (ausilio dialogico, inciso, incipit, introduttore locutivo, ecc.) con livello di subordinazione melodica, ma con caratteri distributivi variati (…) [a) nuclear units (root), i.e. necessary and sufficient to build a pattern, which we have called comment; b) units that have the features of optionality and melodic subordination (prefix), but also a position of temporal precedency with respect to nuclear units, which we have called topic; c) units with features of optionality and a further level of melodic subordination (suffix), occurring after a nuclear unit or even after a unit of precedency, which we have called appendice; d) various units (dialogic aid, parenthesis, incipit, locutive introductor, etc.) with a level of melodic subordination, but with different distributional features (…)]

Ferrari (2004: 15–16), after defining the Rheme, much in Cresti’s fashion, as the bearer of the utterance’s illocution, says that: Le unità di Tema e di Appendice sono invece sprovviste di una funzione illocutiva autonoma e la loro connessione con il cotesto è mediata dal passaggio attraverso il Rema, a cui in modo diretto o indiretto sono sistematicamente funzionalizzate. Il Tema è l’Unità Informativa che esplicita il quadro della pertinenza illocutiva, semantico-denotativa o testuale dell’interpretazione del Rema. Quanto all’Appendice, essa è saturata da contenuti che stanno sullo sfondo informativo dell’Unità Comunicativa e che sono localmente funzionalizzati al Tema o al Rema: alla loro comprensione denotativa, alla loro elaborazione inferenziale, al loro sfruttamento nel cotesto successivo. [The units of Theme and Appendix instead lack an autonomous illocutionary function and their relation to the context is mediated by the Rheme, to which, directly or indirectly, they are systematically functionalized. The Theme is the Information Unit which manifests the frame of illocutionary, semanticdenotative or textual relevance for the interpretation of the Rheme. As for the Appendix, it is implemented by contents that lie in the informational background of the Communication Unit, and are locally functionalized to the Theme or Rheme: to their denotative understanding, their inferential elaboration, their exploitment in the subsequent co-text.]

Consequently, if we take Cresti and Ferrari together, the Theme (or Topic) and the Appendix are much the same under the following respects: – they are not autonomous (= they cannot constitute an utterance by themselves) – they do not convey the illocution of the utterance

390 Edoardo Lombardi Vallauri – there can be more than one of them within the same tone pattern (differently from the Comment) – they can co-occur (= a Topic can be followed by an Appendix, without a Comment in between) On the contrary, they differ as to the following: – they have different position with respect to the Comment – they have different intonation contours In Cresti (1999), where the problem of Functional Sentence Perspective is touched in a general way, only Topic and Comment are mentioned. But when she has to explain the difference between, for instance, (3) e (4), (3)

Carlo / va a ROMA // is going to Rome

(4)

CARLO / va a Roma

Cresti points out that they are informationally fit to answer different questions: Che cosa fa Carlo? what is Carlo doing? (3)

Carlo / va a ROMA // Chi va a Roma? Who’s going to Rome?

(4)

CARLO / va a Roma

Both questions select only a certain part of the answer as the informational purpose of the utterance (the Rheme), whereas the rest is given as known information, which consequently only has the function of semantically locating the Rheme. Now, alongside the Rheme, (3) would be said to contain a Topic and (4) an Appendix. Therefore, Topic and Appendix should be informationally equivalent. We can strengthen this conclusion by illustrating the equivalence of Topic and Appendix in reply to the same question:

“Appendix” or “postposed Topic”: Where does the difference lie? 391

Che cosa fa Carlo? What is Carlo doing? (5)

Carlo / va a ROMA // T R

(6)

va a ROMA / Carlo R A Chi va a Roma? Who is going to Rome?

(7)

CARLO / va a Roma R A

(8)

va a Roma / CARLO T R

However, according to Cresti (2000: 70), while the Comment is devoted to the «compimento della indispensabile funzione illocutiva» [accomplishment of the indispensable function of illocution], Topic and Appendix are «destinate al compimento di funzioni diverse» [devoted to the accomplishment of different functions]: Topic:

«l’espressione del campo di applicazione della forza illocutiva» [expressing the scope to which illocutionary force applies] Appendice: «l’integrazione testuale del Comment o del Topic» [textual integration of the Comment or the Topic] We may observe that the two definitions above concern different functions of an information unit, and are not incompatible. In principle, nothing prevents a certain portion of an utterance from doing both things: expressing the scope of illocutionary force, and textually integrating a Comment or a Topic. It is hard to deny that every Topic “textually integrates” a Comment (cf. examples (5) and (8) above), and sometimes another Topic. On the other hand, every Appendix “expresses the scope to which illocutionary force applies”, as one can see in (6) and (7) above. Starting from these considerations, it seems difficult not to formulate the following Informational Equivalence of Topic and Appendix: All Appendices (Comment-Appendices and Topic-Appendices) are Topics from an informational point of view.

392 Edoardo Lombardi Vallauri This may be verified with any actual utterance.2 In particular, as we have shown, in any context where two portions of a “P-Q” utterance are respectively Topic and Comment, it is possible to replace the utterance by another illocutionarily equivalent utterance “Q-P” where Q remains the Comment and P is something postposed and non-rhematic (i.e., what may be duly termed an Appendix). Still, in Cresti (2000) the difference between the two information units is also entrusted to a slightly more detailed definition: (p. 117) Topic: «quell’unità di informazione che svolge la funzione di campo di applicazione della forza illocutiva, con ciò costituendo a livello della costruzione testuale dell’enunciato la premessa semantica del contenuto locutivo del Comment» (italics mine). [Topic: the information unit which functions as the field of application of illocutionary force, thus constituting, on the level of the utterance’s textual construction, the semantic premise to the locutive content conveyed by the Comment.]

As we may see, this specification does not affect the illocutionary value of the Topic, but only the locutionary level, where the fact of appearing first actually entails certain consequences: (p. 120): «Il Topic, occorrendo come unità informativa prima del Comment, ma all’interno di una programmazione azionale unica, in realtà svolge una funzione di limite e condizionamento conoscitivo rispetto all’applicazione pragmatica della forza illocutiva. […] la locuzione del Topic può essere definita in negativo come quell’espressione che, non essendo dedicata al compimento dell’illocuzione, non è concepita secondo una prospettiva attitudinale verso l’interlocutore. In maniera positiva, quindi, possiamo apprezzare il suo carattere conoscitivo (non affettivo), che implica una forma di valutazione, conoscitiva appunto, diversa da quella del Comment». [The Topic, occurring as an information unit before the Comment but within the same actional planning, functions as a limit and a cognitive conditioning with respect to the pragmatic application of illocutionary force. […] the locution effected by the Topic can be negatively defined as the expression which, not being devoted to the accomplishment of illocution, is not conceived in an addressee-oriented perspective. Positively, then, we can appreciate its cognitive (not affective) character, which implies a form of evaluation, precisely cognitive, different from that of the Comment.]

“Appendix” or “postposed Topic”: Where does the difference lie? 393

Appendices are not redefined more extensively. According to Cresti (2000: 133) «tendono a riprodurre la forma dell’unità da cui dipendono, in minore, con medie di F0 più basse» [they tend to reproduce the intonational contour of the units they depend on, in minore, with lower F0.] One cannot avoid thinking that what is said about the functions of the Topic towards the Comment (and consequently, what defines the Topic) is also evidently true for every Comment Appendix. Therefore, Comment Appendices should be regarded as Topics. They are, simply, “second Topics” with a downgraded intonational contour. I think this situation is illustrated by the examples of Topic Appendices proposed by Cresti (2000: 132) (our glosses): (9)

sì / yes

la seconda / quella grossotta / l’è più giovane the second the plump one is younger

INCIPIT TOPIC

(10) però / still

TOPIC APPENDIX

COMMENT

tutta questa complessità / affermarsi come personaggio / all this complexity to be successful as a personality

INCIPIT TOPIC 1

TOPIC APPENDIX

affermarsi come attore / io proprio / non ce la vedo being successful as an actor I really cannot see her TOPIC APPENDIX

TOPIC 2

COMMENT

Prosodic prominence is similarly reduced for Comment Appendices, of which Cresti says, «hanno un andamento discendente, piatto, privo di movimento focale e spesso privo anche dell’allungamento proprio delle sillabe finali del comment» [their pitch is either falling or level, often lacking the lengthening which is typical of the final syllables of the Comment]. Semantically, they are completions of the Comment, sometimes with what (paraphrasing Ferrari 2003) we may call a “widening” of the horizon of knowledge, sometimes with its “specification”. These two possibilities seem illustrated by the examples in Cresti (2000: 131–132), here reported as (11–14). Thus, it is not incorrect to say that Appendices help define the scope of the Comment: but this is also, incidentally, the “traditional” definition of Topics. (11) parla della famiglia /il Signore // speaks of the family the Lord COMMENT

APPENDIX

394 Edoardo Lombardi Vallauri (12) fatto sta / eran tutte piccolette /queste figlioline / as a matter of fact they were all small those girls INCIPIT

COMMENT

COMMENT APPENDIX

no // no PHATIC

(13) po’ ‘un è freddo nemmeno fòri / stasera // then it is not even cold, outside tonight COMMENT

COMMENT APPENDIX

(14) son discorsi tutti maschili / questi // is all manly talk this COMMENT

COMMENT APPENDIX

That Appendices, as Topics, help define the scope of the Comment by “specifying” its content is shown even better by Cresti’s recent findings from the C-ORAL-ROM corpus, where Appendices are reported to systematically host “cataphoric relations for clitics placed in Comments”.3 We quote her examples here as (15): (15) f222 *BIA: però /INP non lo so /COM quale vuoi //APC ‘but / I don’t know (it) / which (one) you want’ f533 *CMA: perché non me lo potete dimostrare /COM il contrario //APC ‘why can’t you prove it / the contrary’ i1xx*LIA: no a Firenze [/] a Firenze /TOP ce n’era rimasti pochi /COM dell’Americani //APC ‘no in Firenze[/] in Firenze / there were few left (of them) / of the Americans’ f653*CUS: io questi nomi /TOP non intendo farli /COM questi nomi dei giornalisti //APC ‘these names I / have no intention of making them / these journalists’ names’ Cresti (this volume: § 8.2.) signals that 1) Topics may feature cross-utterance and intra-utterance heads of anaphoric relations, but no clitics with cataphoric intra-utterance relation; 2) Appendices only feature heads of intra-utterance cataphoric relations with Comments, and hardly any clitics.

“Appendix” or “postposed Topic”: Where does the difference lie? 395

and that (§10): a cataphora: a) never involves a clitic in Topic; b) typically occurs between a clitic in Comment and its head in Appendix;

According to Cresti (§10): The above distribution can be seen as a function of the strong divergent informational role of both Topic and Appendix. The Topic, whose role is to specify the domain of application of the Comment, requires identified entities, and therefore prevents back-word anaphora. On the contrary, the cataphoric interpretation of a clitic in Comment is coherent with the definition of an Appendix as a “textual integrator of the Comment”. Since these integrations don’t change either the syntactic configuration or the modal value of the Comment, they only constitute a lexical explication of the clitic in Comment.

I see here a risk of overestimating the role of the obvious differences in sequential order between the information units we are considering, while disregarding other, actually more significant features. The fact that clitic-head cataphoras are avoided between Topic and Comment while they are the typical pattern between Comment and Appendix is seen as distinguishing Topics from Appendices, while on the contrary it simply reveals that they behave in the same way towards the Comment. In fact, the situation described by Cresti clearly shows the perfect symmetry and similarity of Topics and Appendices towards the Comment even on the field of anaphorical relations involving clitics: in Cresti’s findings, both Topics and Appendices tend to contain heads for entities which are referred to by means of clitics within the Comment, and not vice versa. In other words, at least in Cresti’s corpus, the typical pattern for a Topic is to contain the head of a clitic which is in Comment, and the typical pattern for an Appendix is… to contain the head of a clitic which is in Comment. This pattern is shown in Scheme 1: TOPIC HEAD

COMMENT CLITIC

APPENDIX HEAD

Scheme 1. As for clitic reference, Topic and Appendix have the same relation to the Comment

In each single utterance, this Scheme usually applies either on the TopicComment side or on the Comment-Appendix side. However, nothing pre-

396 Edoardo Lombardi Vallauri vents it from being instantiated contemporarily on both sides of one utterance. This is what happens in Mereu and Trecci’s (2004) example (31), which we report here as (16): (16) la strada che porta all’aeroporto /Top so riuscita a beccarla /Com la Roma Fiumicino/App ‘the road that leads to the airport / I managed to get it / the Roma-Fiumicino’ As a conclusion of this section, we must add that it is also noteworthy that Appendices are clearly endowed with cognitive content: another feature they share with Topics and that differentiates them from those units such as those Cresti defines under the label Ausilii dialogici [Dialogic aids] (that is to say, Incipit [Incipit], Fatici [Phatics], Allocutivi [Allocutions], Conativi [Conatives]4). To sum up, it seems that even on the strictly locutionary level of utterance semantics and cohesion relations, there is no clear distinction between Topic and Appendix.

2.1. “Point of view” Still, yet another distinctive criterion has been proposed. According to Cresti (2000: 182n) the actual difference between Topic-Comment structures and Comment-Appendix structures may reside in the presence/absence of what she calls punto di vista: Nel caso dell’articolazione Topic-Comment, infatti, l’enunciato è costituito sulla composizione di due punti di vista diversi, mentre nel caso dell’articolazione Comment-Appendice esso non ha composizione in punti di vista diversi, perché solo la prima espressione in Comment ha un proprio punto di vista, mentre quella in Appendice lo sussume da essa. [In the case of Topic-Comment articulation, the utterance is built on the composition of two different points of view, whereas in the case of CommentAppendix articulation there is no such composition of different points of view, because only the expression in Comment has its own point of view, which the one in Appendix derives from it.]

All the same, the objective correlates of said “points of view” do remain partly unexpressed: it remains unclear what they actually are, and it is possibly also unclear whether they exist on the pragmatic-informational (that

“Appendix” or “postposed Topic”: Where does the difference lie? 397

is, illocutionary) level, or on the level of semantic content (that is, locutive). Cresti (p. 184) explains that the point of view has to do with being able to convey different modalities: the Topic, in sum, can have positive modality even if the Comment is then negative, and a number of analogous configurations. For example, this should be observed in (17) and (18): (17) di queste vaccine / ‘un ce n’è rimaste quasi punte of these cows there remains almost none TOPIC

COMMENT

(18) no // le ciliegie / non lo so dove sono no the cherries I don’t know where they are COM TOPIC

COMMENT

In addition, Nel caso invece che l’articolazione dell’enunciato comprenda CommentAppendice, quest’ultima unità d’informazione, che non ha una funzione informativa né un punto di vista diverso da quello del Comment, serve in ge5 nere ad aggiungere integrazioni locutive, precisazioni, correzioni. [When the articulation of the utterance includes Comment-Appendix, the latter information unit, which has neither informational function nor a different point of view with respect to that of the Comment, usually adds locutive integrations, specifications, corrections.]

The latter, however, appears not to be true in many instances. If we invert the elements in (17–18), the information given towards the end of the utterance (Appendix? Postposed Topic?) maintains its different point of view, at least in so far as it can remain not negated: (19) ‘un ce n’è rimaste quasi punte / di queste vaccine there remains almost none of these cows COMMENT

TOPIC

(20) no // non lo so dove sono / le ciliegie no I don’t know where they are the cherries COM COMMENT

TOPIC

The same occurs if we add the negation non to the Comment in a CommentAppendix utterance, as in Cresti’s example on page 187:

398 Edoardo Lombardi Vallauri (21) non sono queste / le due osservazioni che volevo fare not are these the two remarks that (I) wanted to do COMMENT

COMMENT APPENDIX

And yet, when the difference in point of view doesn’t take the positive vs. negative form, but is rather a difference in modalities (cf. Cresti 2000: 185, Ferrari 2003: 30), it is still true that what goes for the Topic may equally go for the Appendix, as the inversion (given in 23) of the following example by Cresti (2000: 185) shows: (22) se io sto interpretando un film / un personaggio / io / if I am performing in a film a character I TOPIC 1

TOPIC APPENDIX

realmente / really

TOPIC 2 INCISO

devo essere [//] devo essere / quel personaggio // must be must be that character COMMENT made of two intonation units (23) io devo essere quel personaggio / se sto interpretando un film I must be that character if (I) am performing in a film COMMENT

APPENDIX

Symmetrically, there are actual utterances in which the Appendices may be moved into preposed Topic position, because their discourse function is suitable for that. Cf. the example proposed by Cresti (2000: 136) and Ferrari (2003: 31): (24) […] quindi / lei rifletta /Top come mi posso sentire arrabbiato io /Comm then you consider how can feel angry I se si sente arrabbiato lei /App if feel angry you ‘Then consider how can I feel angry, if you feel angry.’ There seems to be no problem in obtaining the same pragmatic function with a reversed utterance: (25) […] quindi / lei rifletta /Top se si sente arrabbiato lei /Top then you consider if feel angry you come mi posso sentire arrabbiato io /Comm how can feel angry I ‘Then consider, if you feel angry, how can I feel angry.’

“Appendix” or “postposed Topic”: Where does the difference lie? 399

These last two examples clearly show that not only Topics, but also Appendices enjoy freedom of point of view, expressing a different modality with regard to the Comment. In sum, this criterion also fails to draw a boundary between two categories that only seem to be kept distinct by a very superficial fact, namely different sequential order.

3. Levels In the light of these comparisons, it is possible to establish that the differences between Topic and Appendix are not: – differences of information status; – differences of illocution. Rather, they may be: – differences of semantic relation to the Comment, but they need not be. In some cases (not necessarily the majority) Appendices provide “additional” information that could not be uttered before the Comment; but in many other cases they “complete” the Comment with information whose semantic relations, textual cohesion and coherence with regard to the Comment are the same that would hold if the same information were given as a preposed Topic. In linguistics (and in scientific language in general), categories are usually given different names when the things they group together are different in a respect that is substantial and strictly bound to their main function. In the case at hand, as a consequence, the difference between the categories we are considering should be found on the information/illocution level; but we have seen that it is not on this level that Appendix and Topic are different. Neither can we say that, semantically, the information that precedes and follows the Comment always has different functions. What remains clearly distinctive is the criterion of anteposition vs. postposition. The Appendix is simply a non-rhematic constituent that follows a Comment or a Topic. This criterion would be relevant to Halliday’s definition of Topic in terms of sequential order, which is, however, clearly superseded by Cresti’s definition in terms of illocutionary force. Indeed, Halliday mainly identifies the Topic with the first part of the utterance, which exposes him to rightful criticism (cf., for instance, Downing 19916).

400 Edoardo Lombardi Vallauri If we do maintain a functional point of view, based on the illocutionary force and information status held by the units we consider, we should not speak of Topic and Appendix, but rather, as many actually do, of pre- and postposed Topic. Referring to Topic and Appendix by different names would be comparable to giving different names to preposed and postposed temporal clauses, or subjects, instead of naming (as is usually done) temporal subordinate clauses according to the semantic relation they establish with the main clause, and subjects after the grammatical relation they have with the verb. As a matter of fact, adverbial clauses do enact slightly different functions depending on whether they are pre- or postposed, paralleling what happens with pre- and postposed Topics. As it has been shown,7 the scope of preposed adverbial clauses is broad and can extend as far as the content of the whole discourse, their function being to prepare and introduce the content of the main clause. The scope of postposed adverbial clauses, on the contrary, is restricted to the content of the main clause, of which they constitute an immediate completion. This does not prevent preposed and postposed conditional clauses from being both called “conditional”, because this remains the most relevant feature by which they are characterized. To call a Topic “Appendix” when it is postposed means giving up defining it after the main features by which it is actually identified: namely, the absence of illocutionary force and the function of semantically integrating the Comment. The name “Appendix” underlines a merely formal feature: namely, the fact of occurring to the right of the Comment, which, in all truth, does in many cases cause a postposed Topic to contain “completing” information, but not always and not necessarily. This is not a merely terminological concern. To understand the Appendix in terms of the (strictly interdependent) criteria of its most “typical” semantic content and its sequential ordering can give rise to actual drawbacks in the interpretation of facts, because it leads to disregarding the main feature related to Appendix (as to any Topic), namely its being devoid of illocutionary force. For example, it leads to treating as an Appendix something that – in illocutionary terms – is actually a Comment, just because on the semantic-locutionary level it exerts a similar function to that of an Appendix, although on the information level it is an autonomous speech act. The superficial criterion overrides the deep one. For example, Cresti (2000: 136) and Ferrari (2003: 31) propose the following information structure for the utterance in (25):

“Appendix” or “postposed Topic”: Where does the difference lie? 401

sennò / la mi fa morire / Comm (26) // perché quella lì / Top because that one there otherwise she me makes die perché la mi mette pensieri //App because she me puts thoughts ‘because she, otherwise, she makes me die, because she makes me worry’ But here the last information unit really seems to be a Comment, and this is confirmed by the fact that the subordinate clause is introduced by perché, which always introduces rhematic adverbials.8 Ferrari (2004: 16) defines as Appendices a number of constituents which, if considered on the level of their illocutionary force, should necessarily be classified as Rhemes: her reason for doing so is that, from the point of view of considering their denotational semantics, however, they comply with her definition of Appendix:9 (27) Un’idea di poesia, certo, la si può cavare con maggiore intensità e pertinenza dai Canti Orfici piuttosto che dalla vita di chi li scrisse, perché l’intera conoscenza e di un’opera letteraria e di una idea di poesia va ricavata dai testi soltanto, che costituiscono un sistema chiuso. // 1. IoTema la penso così,/Rema non ho mai amato le biografie,/App le ho sempre trovate noiose,/App spesso inutili.//App 2. Invece,/Tema questa volta che ne ho letto una d’un fiato,/Tema quella di Vassalli,/App mi ricredo./Rema 3. E ciò mi succede/Tema perché Vassalli è scrittore, è poeta.//Rema Non è uno che abbia fiducia nei fatti e meno nelle parole. S’è proposto di fare un romanzo il cui margine di arbitrarietà fosse limitato all’estremo, dai fatti, dai documenti, dalle ricerche d’archivio. [An idea of poetry, of course, can be drawn with greater intensity and relevance from the Canti Orfici rather than from the life of their writer, because the whole knowledge both of a piece of literature and of an idea of poetry must be taken only from the texts, which constitute a closed system. // 1. ITheme see it so,/Rheme I never liked biographies,/App I always found them boring,/App often unuseful.//App 2. Instead,/Theme this time that I have read one in one shot,/Theme that of Vassalli,/App I have to change my mind./Rheme 3. And this happens to me/Theme because Vassalli is a writer, and a poet.//Rheme He is not one who trusts the facts more than words. He has intended to produce a novel whose amount of arbitrariness should be extremely circumscribed by the facts, by the documents and by archive research.]

402 Edoardo Lombardi Vallauri It is hard to accept that the first two Appendices of Communication Unit 1 in (27), be informationally (i.e. illocutionarily) anything other than two Rhemes, i.e. something different from the Rheme immediately preceding them, or the one of Communication Unit 3. The same can be seen in the two examples reported here below from De Cesare (2004: 210), where two informationally analogous portions of text (“expansions” introduced by anche) are treated as an Appendix in the first instance and, no doubt correctly, as a Rheme in the second: (28) //E’ quasi una rivoluzione nelle tecniche di animazione,/Rema e anche nel

cinema./App MediaLab ha realizzato il sogno di ogni regista: dirigere i personaggi dei cartoni animati proprio come gli attori reali. [//It is almost a revolution in animation tecniques,/Rheme and also in cinema./App MediaLab has realized the dream of every director: to direct the characters of cartoons exactly as real actors.]

(29) //Quindi/ a sera/ ci si dispone davanti al cielo blu, in attesa di vedere una scia, una stella cadente,//Rema e anche una star in caduta.//Rema Ecco la stella Alberto Tomba, protagonista del film Alex l’ariete, che ha incassato un’elemosina.

[//Thus/ at night/ we set ourselves in front of the blue sky, waiting for a wake to appear, a shooting star,// Rheme and even a falling movie star.// Rheme There you have the star of Alberto Tomba, leading character of the film Alex l’ariete, which has collected an alms.]

It is apparent that the semantic feature of “completing” or “integrating” a Comment is not an exclusive privilege of Appendices, it being rather possible for any ensuing Comment. This gives rise to an obvious difficulty in using it as the basis for defining the Appendix. A problem of consistency between this approach and the “classical” one by Cresti also arises on the theoretical level, when it is said that (Ferrari 2004: 17): il segmento linguistico che esprime il Tema precede necessariamente il Rema e può essere preceduto solo da un’Unità Informativa dello stesso tipo (o eventualmente da altre micro-unità incipitarie); il segmento associato alla funzione di Appendice è sempre a destra dell’unità a cui si aggancia: Tema, Rema o Appendice. [the linguistic material that expresses the Theme necessarily precedes the Rheme and can be preceded only by an Information Unit of the same kind (or by other “opening micro-units”); the material associated to the function of Appendix is always at the right of the unit on which it is hanging: Theme, Rheme or Appendix.]

“Appendix” or “postposed Topic”: Where does the difference lie? 403

Rigidly linking the definition of the category “Theme” to a given sequential order excludes the relevance of its status on the level of illocutionary force as a defining feature of the category itself. However, this level is, at least starting from Cresti’s groundbreaking intuitions, the foundational criterion for distinguishing the categories of information structure, and in particular for distinguishing Topic and Comment. As for distinguishing Topic and Appendix, Cresti herself replaces this criterion with that of sequential order and its most typical semantic repercussions, such as the difference between “introductory” and “integrative” contents. As a consequence, it is implied that a postposed Topic cannot exist. This approach is confirmed by another definition, according to which, alongside the Rheme (Ferrari 2004: 23): Si riconoscono in particolare: (i) una funzionalizzazione informativa di tipo tematico, che consiste nell’esplicitare ab initio il quadro della pertinenza illocutiva, semantico-denotativa e testuale dell’interpretazione del Rema; e (ii) una funzionalizzazione informativa di tipo Appendice, il cui obiettivo sta nel precisare a posteriori il contenuto del Tema o del Rema, in modo da modularlo epistemicamente, da precisarlo, da concretizzarlo, da renderlo più comprensibile, ecc. [We recognize in particular: (i) a thematic informational function, which consists of manifesting ab initio the frame of the illocutive, denotative and textual relevance of the Rheme’s interpretation; and (ii) an informational function to be called Appendix, whose aim is to specify a posteriori the content of the Theme or Rheme, so as to epistemically modulate, specify, concretize it, make it more comprehensible, etc.]

Here the distinction between Theme and Appendix is set on the level of denotational semantics, not in an autonomous sense, but to the extent that semantics depends on sequential order. In other words, the difference between Theme and Appendix should entirely lie in the fact of appearing before or after the Rheme, which gives rise to different semantic interactions between that content and the content of the Rheme itself. But a problem is that the difference is often almost nil. For example, what is said of the behaviour of Appendices in the quotation above (except for the direct reference to sequential order: “a posteriori”) could be equally said of Themes: a non-rhematic chunk of information can modulate epistemically, specify, concretize, make more understandable the content of a Rheme, both if it precedes and if it follows it. This happens in many of the utterances we have examined so far, and also in (30–31):

404 Edoardo Lombardi Vallauri (30) le chiavi / le ho in TASCA the keys them (I)have in pocket T R (31) le ho in TASCA / le chiavi R T It is hard to maintain that such functions are held by the Theme only in one of its positions: on these grounds it seems difficult to justify that we use two completely different names for a preposed Theme (“Theme” or “Topic”) and a postposed one (“Appendix”).

4. Possible misunderstandings deriving from this approach A problem with this approach is that it can lead to disregarding crucial factors, such as intonation. It is true that written texts materially lack intonation, so that those who want to examine the categories of information structure in written language are forced to adapt themselves to a certain extent; but it is equally true that intonation is strictly bound to the expression of information structure (Cresti docet): consequently, it is not granted that one will be able to assess information structure where intonation is absent. Many written utterances potentially have different information structures, depending on how they are interpreted – and (mentally) intonated – by readers. For example, in the following passage (quoted by De Cesare 2004: 203): (32) // La rivoluzione /Tema sta lentamente approdando anche in Italia, /Rema malgrado la confusione del quadro normativo inviti gli operatori alla prudenza. //App ‘// The revolution /Theme is slowly landing also in Italy, /Rheme although the confusion of the regulatory framework invites operators to prudence. //App ’ clearly, information structure depends on the intonation that is given to the text. Still, the long Appendix is perhaps too long to lack any intonation prominence and consequently for being completely devoid of illocutionary force. On the contrary, it may well constitute a second Rheme. In fact, its information status may be comparable to that of a coordinate clause.10 Another problem connected to seeing possible Rhemes as Appendices just because they are located at the right end of the utterance is that of not

“Appendix” or “postposed Topic”: Where does the difference lie? 405

recognizing the focalizing function of certain linguistic elements. The issue would deserve to be treated in more detail: we can just sketch it by pointing to the case of anche utterances such as those examined by De Cesare (2004: 204): (33) // In mezzo alla retorica fascista,/ il romanzo/Tema fece scandalo,/Rema

anche per lo spazio dato al sesso: //App donde, pure per le dirette allusioni critiche al regime, la condanna anche a successivi romanzi di Moravia. [// In the midst of fascist rhetoric, / the novel/Theme caused a scandal,/Rheme also owing to the space given to sex: //App hence, in part because of the direct allusions to the regime, the condemnation also of the next novels by Moravia.]

(34) Al bazar domandano da dove veniamo e subito rammentano altri italiani: Celentano, Romina e Al Bano, Michele Placido, il commissario Cattani, la Piovra, mafia, mafia…// Ridono tutti, /Rema anche i venditori di tappeti. //App [At the bazar they ask where we are from and they immediately recall other Italians: Celentano, Romina and Al Bano, Michele Placido, the Commissario Cattani, La Piovra, mafia, mafia…// They all laugh, /Rheme also the carpet sellers.//App]

Within De Cesare’s framework, in these cases «l’avverbio anche inaugura un Atto Testuale autonomo, che si associa alla funzione di Appendice del Rema.» [the adverb anche inaugurates an autonomous Textual Act, which is associated to the function of Appendix of the Rheme.] A definition of the Appendix simply in terms of sequential order opens the way to also regarding portions of text that carry full illocutionary force (“autonomous textual acts”) as Appendices, provided that they are postposed to something. But this brings about a complete detachment from the original idea according to which the Appendix, defined informationally, is first of all linguistic material devoid of illocutionary force: that is to say, first of all a non-Rheme. As concerns the adverb anche, this approach leads to consistently claiming that authors such as Sabatini-Coletti (1997), Suomela (1996) and others wrongly identify its function, in that they consider it as a focalizing adverb. De Cesare says as much about (33–34): malgrado si collochi ad apertura dell’ultimo elemento dell’Enunciato, l’avverbio anche si pone dunque a destra del Fuoco Informativo dell’Enunciato, in un’Unità che fa per definizione da sfondo informativo all’Enunciato. Per questo motivo, non lo si può considerare focalizzante in senso informativo:

406 Edoardo Lombardi Vallauri il suo fuoco semantico non coincide con il Fuoco Informativo del Rema (…), di modo che non può, evidentemente, essere considerato come responsabile della messa in rilievo del Fuoco Informativo dell’Enunciato. [although it is located at the opening of the last element of the Utterance, the adverb anche thus lies at the right of the Information Focus of the Utterance, within a Unit that by definition acts as an information background of the Utterance. For this reason, it cannot be considered as focalizing in the informational sense: its semantic focus doesn’t coincide with the Information Focus of the Rheme (…), so that it cannot, evidently, be considered as responsible for the emphasis of the Information Focus of the Utterance.]

It is at least controversial that the informational units located to the right of the labelled utterances in (33) and (34) be Appendices in Cresti’s primitive sense, which was still bound to illocution; on the contrary, they are Appendices if defined in the way we have seen, which disregards information structure. But in this case one is no longer allowed to draw conclusions based on the primitive (informational) definition of Appendix as a “background” unit. This caution would allow granting anche its function as a focalizing adverb, which (at least in instances such as (33–34)) is the predominant interpretation in the literature. 5. A cross-linguistic confirmation A confirmation of the idea that non-rhematic information has the same linguistic value when preceding and when following the Rheme comes from Japanese. In that language the same particle (wa) marks not only preposed Topics, but also “Appendices” (alias postposed Topics). In other words, Japanese has no explicit markers to distinguish between those forms the literature we have been discussing would regard as different units: in that language, Appendices are (postposed) Topics. Hinds (1986: 161) points out that There is a structure in Japanese termed “postposing” […]. This structure allows elements to be placed after the sentence final verbal. This construction has often been termed an “after-thought” construction […], but there are more functions than merely the adddition of an after-thought. Frequently, wa marked noun phrases are “moved” to sentence final position with this structure.

Let us only consider a couple of examples of this situation, both taken from Hinds (1986: 161):

“Appendix” or “postposed Topic”: Where does the difference lie? 407

(35) anoo, koo iu fuu ni sagatta no ga kawaisoo da toka omowanai COP or think-NEG uh this say way to bowed NOM NM pitiful deshoo, nihonjin wa presumptive Japanese-people TOP ‘Don’t (they) think it’s a shame when (they) bow like this, the Japanese’ (36) nihon no daigaku ne, kore wa Japan LK university EM this TOP ‘(This) is a Japanese college, right, this’ 6. Conclusions Italian scholars who use the term Appendice use it essentially in two ways. The first, typically represented by Cresti, draws a distinction between Comment on the one hand and Topic and Appendix on the other, through the criterion of illocutionary force. Then, Topic and Appendix are seen as distinct according to the criterion of sequential order. We have tried to demonstrate that this criterion has no systematic correspondence with any difference of pragmatic status, and even of semantic function, since the semantic functions that are typical of Appendices are often held by preposed Topics, and even by Comments. Once this is clear, one may still wish to attribute a different label to constituents that are Topics both pragmatically and semantically, on the grounds of their being postposed and not preposed.11 The second concept of Appendix, represented for instance by Ferrari and De Cesare, originates in a tendency which is present in Cresti’s definition; it is more sensitive to semantic and textual criteria, and extends the definition of Appendix to any unit conveying information that is integrative or additional with respect to preceding linguistic material. This leads to overshadowing the criterion of illocution, and also to treating constituents that may be endowed with their own illocutionary force (such as to be able to constitute an independent speech act) as Appendices. A certain amount of ambiguity may arise in the literature, due to a partial lack of consistency between the two conceptions, which do not contradict each other explicitly. On the whole, the issue doesn’t seem to have been definitively settled. The aim of this intervention was to point out that there is a problem and suggest that it may need further analysis. It is less important whether the term Appendix should be abandoned because it cannot receive a fully distinct definition from that of Topic/Theme, or whether it should be maintained after it has been defined in a less ambiguous way, i.e. after it

408 Edoardo Lombardi Vallauri has been clearly admitted that its only regular difference from the Topic lies in sequential ordering, not in illocution or semantic function.

Acknowledgements This paper would not have existed without an invitation by Angela Ferrari and Anna Maria De Cesare to discuss about Lessico, Grammatica e Testualità in Basel, February 2006. Their theoretical positions and mine, which are often astonishingly similar, diverge on this particular topic. This allows me to underline even more openly my gratefulness to them for the opportunity of a dialectical exchange which is only at its beginning. Notes 1. 2.

3. 4.

Cf. Cresti (1999, 2000, and this volume) This was the opinion of Charles Bally, who (1944: 69), speaking of structures with a right dislocation, claimed that “le thème est comme l’écho” of the preceding Propos. He evidently identifies the linguistic material following the Rheme as a Theme, and attributes to it the same function as a preposed thème… Cresti (this volume: § 7.1.) We leave out of our discourse the so-called Incisi [Parentheses] and the Introduttori locutivi [Locutive Introductors], which in various respects seem a more complex category than the others, perhaps even possibly endowed with some illocutionary force. Cresti (2000: 179f.) puts it so: “Notiamo che: 1) il Comment, il Topic, e le Appendici, sono le unità sulle quali è costruito il testo dell’enunciato; 2) gli Introduttori locutivi e gli Incisi intervengono sul testo dell’enunciato, anche se non partecipano alla sua costruzione; 3) le diverse unità di Ausilio dialogico invece si collocano tutte fuori dalla composizione del testo dell’enunciato e non intervengono neppure su di esso, ma direttamente sull’interlocutore.” [We notice that: 1) The Comment, the Topic and the Appendices are the units on which the text of the utterance is constructed; 2) Locutive Introductors and Parentheses intervene on the text of the utterance, although they don’t take part in its construction;] 3) the different units of Dialogic Aid, on the contrary, are all located out of the composition of the utterance’s text and don’t even intervene on it, but directly on the addressee.]

5.

Cresti (2000: 186); italics mine.

“Appendix” or “postposed Topic”: Where does the difference lie? 409 6.

Probably the most common definition of Theme and Rheme is that “Theme is what the utterance is about, and Rheme is what is said of it”. This is for instance the definition adopted by Halliday (1985). But the applications of such a definition should not be oversimplified, as it happened to Halliday himself (at least until Halliday 1985, but to a certain extent also in Halliday 1989), when he identified the Theme in what occupies the left, and the Rheme in what occupies the right of the utterance. In doing so, he exposed himself to criticism, for instance by A. Downing (1991: 122), which we can reformulate as follows: in examples (ii) and (iii) below the articulation of Theme and Rheme made “à la Halliday” is in contrast with his definition: (i) you were to blame T

(ii) YOU T

R

were to blame R

(iii) It’s YOU who were to blame T

R

In (ii) and (iii) what the utterance is about (Theme) is not “you”, but “that someone was to blame”, while what is said about it (Rheme) is that the person to blame is “you”. The correct analysis of information structure will thus be: (ii) YOU were to blame R

T

(iii) It’s YOU who were to blame R

T

7. 8. 9.

Cf. among others Thompson (1985) and Diessel (2005). Cf. e.g. Schwarze (1986: 144f.) and Lombardi Vallauri (2000: 64–68). A similar example of information units classified as Appendices, though their information status is rhematic, is in Ferrari (2004: 24, ex. 19). 10. Cf. Lombardi Vallauri (1996a and 2000). 11. Provided that our point is understood, this is just a terminological issue.

References Bally, Charles 1944 Linguistique générale et linguistique française. Bern: Francke. Berretta, Monica 1995 Ordini marcati dei costituenti maggiori di frase: una rassegna. Linguistica e Filologia 1: 125–170. Berruto, Gaetano 1986 Le dislocazioni a destra in italiano. In Tema-Rema in italiano, Harro Stammerjohann (ed.), 55–69. Tübingen: Narr. Cresti, Emanuela 1987 L’articolazione dell’informazione nel parlato. In Gli italiani parlati, AA.VV., 27–90. Firenze: Accademia della Crusca.

410 Edoardo Lombardi Vallauri 1999

Force illocutoire, articulation topic/comment et contour prosodique en italien parlé. Faits de langue 13: 168–181. 2000 Corpus di italiano parlato. Firenze: Accademia della Crusca. this vol. Clitics and anaphoric relations in informational patterning: a corpusdriven research in spontaneous spoken Italian (C-ORAL-ROM). De Cesare, Anna-Maria 2004 L’avverbio anche e il rilievo informativo del testo. In Ferrari (ed.), 191–218. Diessel, Holger 2005 Competing motivations for the ordering of main and adverbial clauses. Linguistics 43 (3): 449–470. Downing, Angela 1991 An Alternative Approach to Theme: a Systemic-Functional Perspective. Word 42 (2): 119 –143. Ferrari, Angela 2003 Le ragioni del testo. Firenze: Accademia della Crusca. 2004 La lingua nel testo, il testo nella lingua. In Ferrari (ed.), 9–41. in press La fonction informationnelle d’Appendice. De la dislocation à l’apposition à travers la composante informationnelle. Cahiers Ferdinand de Saussure. Ferrari, Angela (ed.) 2004 La lingua nel testo, il testo nella lingua. Torino: Istituto dell’Atlante Linguistico Italiano. Halliday, Michael A. K. 1967–68 Notes on Transitivity and Theme in English 1, 2, 3. Journal of Linguistics 3 (1): 37–81 / 3 (2): 199 –244 / 4 (2): 179–215. 1985 An Introduction to Functional Grammar. London: Arnold. 1989 Spoken and Written Language. Oxford: Oxford University Press. Hinds, John 1986 Japanese. London / Sydney/Dover: Croom Helm. Höhle, Tilman 1992 Über Verum-Fokus im Deutschen. In Jacobs (ed.), 112–141. Jacobs, Joachim (ed.) 1992 Informationsstruktur und Grammatik. (Linguistische Berichte, Sonderheft 4). Lombardi Vallauri, Edoardo 1996a La sintassi dell’informazione. Roma: Bulzoni. 1996b A Simple Test for Theme and Rheme in Clause Complexes. Language Sciences 17 (4): 357–378. 1997 Relazioni informative fra clausole nel parlato e nello scritto. Rassegna Italiana di Linguistica Applicata (RILA) 29 (1): 41–58. 1998 Focus esteso, ristretto e contrastivo. Lingua e Stile 33 (2): 197–216. 2000 Grammatica funzionale delle avverbiali italiane. Roma: Carocci.

“Appendix” or “postposed Topic”: Where does the difference lie? 411 2001

La teoria come separatrice di fatti di livello diverso. L’esempio della struttura informativa dell’enunciato. In AA.VV., Dati empirici e teorie linguistiche, 151–173 (Atti del 33e Congresso della Società di Linguistica Italiana, Napoli 1999) Roma: Bulzoni. Mereu, Lunella and Alessandra Trecci 2004 Focus sul topic. In AA.VV., Il Parlato Italiano. Atti del Convegno Nazionale (Napoli, 13–15 febbraio 2003), F.Albano Leoni, F. Cutugno, M. Pettorino and R. Savy (eds.). Napoli: M. D’Auria Editore – CIRASS (CD-ROM). Rossi, Fabio 1999 Non lo sai che ora è? Alcune considerazioni sull’intonazione e sul valore pragmatico degli enunciati con dislocazione a destra. Studi di grammatica italiana 18: 144–193. Sabatini, Francesco and Vittorio Coletti 1997 DISC – Dizionario Italiano Sabatini Coletti. Firenze: Giunti. Schwarze, Christoph 1986 Tema e rema nella frase complessa. In Tema-Rema in italiano, Harro Stammerjohann (ed.), 141–155. Tübingen: Narr. Suomela-Härmä, Elina 1996 Anche. Analisi prammatico-linguistica. In Actes du 13e Congrès des Romanistes Scandinaves, Jyväskylä, 12–14 August 1996, 725–736. Jyväskylä: University of Jyväskylä. Thompson, Sandra A. 1985 Grammar and written Discourse: initial vs. final Purpose clauses in English. Text 5 (1–2): 55–84. Tomlin, Russell S. 1985 Foreground-Background Information and the Syntax of Subordination. Text 5 (1–2): 85–122.

Contributors

Giuliano Bernini Dipartimento di Scienze dei Linguaggi, della Comunicazione e degli Studi Culturali, Università di Bergamo (Italy) giuliano.bernini @ unibg.it Marina Chini Dipartimento di Linguistica Teorica e Applicata, Università di Pavia (Italy) marina.chini@ unipv.it

Edoardo Lombardi Vallauri Dipartimento di Linguistica Università Roma Tre (Italy) lombardi@ uniroma3.it Lunella Mereu Dipartimento di Linguistica Università Roma Tre (Italy) mereu @ uniroma3.it

Emanuela Cresti LABLITA, Dipartimento di Italianistica Università di Firenze (Italy) elicresti @ unifi.it

Ignazio Mauro Mirto DANAE – Dipartimento di Analisi dell’Espressione Università di Palermo (Italy) ignazio.mirto @ libero.it

Claudia Crocco Vakgroep Romaanse Talen Universiteit Gent (Belgium) claudia.crocco @ ugent.be

Annarita Puglielli Dipartimento di Linguistica Università Roma Tre (Italy) pugliell @ uniroma3.it

Ludwig Fesenmeier Romanisches Seminar Ruhr-Universität Bochum (Germany) ludwig.fesenmeier @ rub.de

Antonietta Scarano LABLITA, Dipartimento di Italianistica Università di Firenze (Italy) antonietta_scarano @ virgilio.it

Franck Floricic Département d’Italien et de Roumain Université Sorbonne Nouvelle – Paris 3 floricic @ univ-tlse2.fr

Shingo Suzuki Department of Foreign Languages Tokyo College of Music (Japan) suzukish @ a00.itscom.net

Mara Frascarelli Dipartimento di Linguistica Università Roma Tre (Italy) frascare @ uniroma3.it

Ida Tucci LABLITA, Dipartimento di Italianistica Università di Firenze (Italy) ida.tucci @ gmail.com

Lucyna Gebert Dipartimento di Studi Slavi e dell’Europa Centro-Orientale Università di Roma “La Sapienza” lucyna.gebert@ tiscali.it

Index of subjects

aboutness, 20, 38, 231, 328, 330, 355, 361, 376 accent, 4, 15, 17–27, 34–39, 40, 88, 234 A and B ~, 7, 17–19 focal ~, 310 melodic prominence, 15, 34, 53–54 pitch ~ (primary ~), 2, 16–26, 29, 33– 39, 81, 88, 93, 234, 343–344 post-nuclear ~, 24–26, 37, 39 pre-nuclear ~, 26 rise ~, 17, 19 topic ~, 4, 15, 17–19, 23, 33–36, 39 adverbial, 60, 66, 158–160, 166, 209, 211, 246, 251, 262, 278–279, 296, 311, 328, 334, 340, 400 adversative (discourse) context, 108, 121, 123, 254, 352 afterthought, 91, 126, 336 agreement, 6, 23, 83, 138, 173, 252, 271, 295, 315–316, 319–320, 343, 356, 360–361 anaphora (anaphoric), 5, 76, 80, 126, 130, 169–170, 175–195, 230, 241, 255, 271–272, 275, 279, 282, 284, 287– 288, 295, 299, 343, 361, 363, 370, 374, 378, 394–395 anaphoric relation, 169–170, 175–192, 195, 255, 394 anche, 118, 121, 181, 186, 189, 217–218, 233, 327, 393, 402–408 animacy, 317 appendix, 7, 23, 38, 63, 68–69, 97, 170, 173, 176–177, 184–185, 188–195, 205, 208, 213–214, 387–409 argument-focus structure, 356–357 Ausilii dialogici, 396, 408 background, 1, 15, 17–18, 20, 77, 97, 130, 153, 159–160, 210, 230, 234, 242, 253,

255, 257, 326, 328, 330, 343–344, 360, 369, 406 Balkan Sprachbund, 307, 314, 318–321 ‘cartographic’ approach, 6, 79, 96, 325 cataphoric relation, 177, 179, 180–186, 188–191, 195, 394 categorical, 235, 241–242, 247–248, 255, 257–258, 260, 263, 275, 296 causative clause, 156, 166 C-domain, 325–327, 334, 337–339, 342 CHAT (Codes Human Analysis of Transcripts), 52 cleft, 81, 87, 106, 108, 125, 130, 146, 155, 165–166, 307, 310, 312, 321, 361 clitic (form), 87, 142, 174–175, 186, 195, 272–273, 291 clitic doubling, 88, 307, 316–321, 378 comment, 16–18, 22–23, 26, 38, 52, 59– 69, 77, 84, 95, 97, 115–120, 123, 125– 126, 155, 158, 164, 170–179, 182– 195, 205, 208–223, 231, 233–235, 243, 247, 255, 259–260, 287, 301, 319, 327, 334, 337, 344, 357–359, 376, 378, 387–402, 407–408 concessivity, 121 configurational type, 79 conjugational morpheme, 269–271, 295 convergence, 320 coordination, 194, 212 copula, 109–113, 125, 145, 155–156, 209, 312, 359 declarative, 6, 25–27, 31, 33, 203, 269, 274, 280, 285, 287–291, 294, 298– 300, 307, 315, 343–344 declension, 307, 314, 318 definite, 6, 59, 78, 157, 230, 233, 240, 251–254, 257–259, 280, 297, 314, 317, 319, 328–329, 335, 343, 361, 378

416 Index of subjects defocusing, 146 deixis, 146, 172–173, 178–179, 184, 188, 192, 262, 271–272, 282–284, 295 personal ~, 281–287 desemantisation, 319–320 dialogical aid, 53, 396 allocutive, 64, 69, 173, 208, 213–214, 396 conative, 64, 173, 195, 208, 214, 396 incipit, 64, 69, 173, 208, 213, 232, 389, 393–394, 396 phatic, 64, 69, 173, 208, 213–214, 394, 396 Dictum, 203–205, 221 discourse ~ (-semantic) functions, 1, 4, 96, 105, 115, 121, 123–124, 325, 398 ~ grammar, 75, 94, 375 ~ patterning, 4, 363 dislocation left ~, 21, 39, 87, 89–90, 93, 119, 126, 244, 270–271, 295, 301, 314, 317 right ~, 7, 23, 86–87, 89, 93, 97, 119, 165, 317, 328, 362, 408 Do-support strategy, 153–154, 162, 163, 165 enclisis, 6, 142, 174–175, 195, 270, 273– 274, 289–291, 294, 299–301 exclamation, 125, 130, 143–145, 343 expletive, 312 f0 , 22, 28–33, 53, 64, 88–89, 91, 93, 326, 332, 343, 376, 393 factive predicate, 159 finite(ness), 5, 106, 108 –115, 120 –126, 229–230, 243, 245, 254, 276, 312, 342, 361 focalization, 5, 81, 108, 120–121, 123 focalization strategies, 81 focus, 2–6, 17–19, 25–26, 38–39, 53, 63, 77, 81–82, 85–96, 120, 126, 129, 137– 141, 145–148, 155, 169–173, 183, 192–195, 313, 316–317, 321, 328,

339, 342–345, 352–359, 364, 366, 375–377, 406 ~ clash, 129, 141 ~ markers, 81–82, 137–139, 148, 343– 344, 359 ~ phrase, 356 broad ~, 26, 78, 313 completive ~, 93, 376 contrastive ~, 2, 19, 26, 88, 91, 93–94, 291, 311, 316–319, 376–377 in situ ~, 88, 310 narrow ~, 21, 25, 33, 85, 92 prosodic ~, 16, 25–26, 31, 34, 38 fronting, 5, 76, 119, 125, 129, 131–134, 141, 146 constituent ~, 83, 86–88, 146 Functional Sentence Perspective, 1, 390 generic, 17, 171, 259, 274, 316, 340 given, 5, 16–19, 23, 27, 31, 37, 39, 51, 53, 59, 75, 77–80, 87, 94–96, 115, 129, 130–132, 141–142, 145–148, 153, 155, 160–162, 170, 176–177, 188, 192–193, 217, 220, 229, 233– 235, 239–242, 245–246, 251, 256– 262, 278, 281, 286, 289, 292, 299, 314–321, 326, 329–330, 333, 342, 344–345, 352–357, 362–363, 367, 375–378, 390, 397–399, 403–405 grammaticality judgements, 3 grammaticalization, 203, 320, 322, 340, 352, 374–375 identificational sentence, 356–357 illocution, 51, 58–61, 69, 204, 388–392, 399, 406–408 Illocutionary Force, 23, 61–62, 171–173, 191, 208, 217, 223, 338, 388, 391–392, 399–408 indefinite, 55, 78, 133, 259, 318, 329, 357, 361 infinitive, 5, 105–127, 153–156, 161–166, 311 tautological ~, 106–108

Index of subjects 417 information structure (‘information packaging’), 1–7, 16–20, 23, 35, 69, 75– 83, 86, 88, 93–97, 129, 153, 155, 157, 161–164, 208, 269–270, 289, 293– 294, 307–308, 313, 320, 325–326, 329, 342, 351–354, 357, 359, 368, 372, 374–375, 387, 400, 403–404, 406, 409 information structure pairs, 3, 76, 77, 85, 155, 352 background/foreground, 18, 20, 130, 145, 153, 165, 253–255, 326–328, 330, 343, 389, 406 given/new, 16, 75, 80, 155, 234, 363 presupposition/assertion, 17, 77, 117, 120, 123, 154–166, 211, 251, 254, 294, 357, 376 presupposition/focus, 343 theme/rheme, 5, 75, 155, 260, 274, 277, 296, 321, 388–389, 403, 409 topic/comment, 2, 5, 16, 18, 23, 26, 38, 63, 83, 86, 115, 123, 172, 194, 236, 239, 247, 260, 343, 351, 356–357, 388–392, 395–396, 403 topic/focus, 1–3, 6, 19, 75–78, 86–88, 93–96, 322, 359, 375 information unit, 5, 18, 154–155, 170, 204–205, 208–223, 367, 375, 387– 392, 395–397, 401–402, 409 appendix, 7, 23, 38, 63, 68–69, 97, 170, 173, 176–177, 184–185, 188–195, 205, 208, 213, 214, 387–409 comment, 16–18, 22–23, 26, 38, 52, 59, 60–69, 77, 84, 95, 97, 115–120, 123–126, 155, 158, 164, 170–179, 182–195, 205, 208–219, 221–223, 231–235, 243, 247, 255, 259–260, 287, 301, 319, 327, 334, 337, 344, 357–359, 376, 378, 387– 402, 407–408 locutive introducer, 170, 172, 176– 179, 182–188, 192–195, 408 parenthesis, 69, 170, 172, 176, 182– 195, 241, 247, 389, 408

topic, 1–7, 15–27, 31–40, 59, 62–63, 69, 75–78, 81–97, 108, 112–121, 126, 155, 170–173, 176–195, 204– 205, 208, 211–223, 230–235, 240– 248, 254–263, 308, 313–316, 321– 322, 325–345, 352–378, 387–400, 403–404, 407–408 interface, 2, 4, 76, 87, 94, 325–326, 341– 343, 352, 375 interlanguage, 351, 359 interrogative, 5, 129–130, 134, 141–148, 156, 165, 280, 298–299, 316 intonation, 1, 4, 16–17, 22–25, 35, 38, 53, 59, 61, 69, 107, 126, 129, 155, 270, 289–294, 307– 310, 335, 338, 341, 344, 388, 390, 398, 404 intonation contour, 388, 390 downgrading ~, 332, 334, 337, 344 flat ~, 89, 93, 337 rising ~, 330–332 intonation prominence, 310, 404 L1, 7, 352, 358–366, 370–378 L2, 7, 351–354, 357–360, 363–364, 370– 378 ‘laboratory’ sentences, 3 Language in Act Theory, 5, 170 languages ‘pragmatically based/oriented’ ~, 69, 80, 82–87, 94–96, 171, 244 ‘syntactically based /oriented’ ~, 80, 84, 87, 95, 243 discourse configurational ~, 96 focus-prominent ~, 2, 77, 84, 96 morphologically oriented ~, 80 subject-(/oriented) prominent ~, 2–3, 83, 359 topic-prominent/topic-oriented, 2–3, 83–84, 96, 358–359 learner variety, 351, 359, 364, 374 modal compositionality, 218, 221 modal indexes, 193, 204–205, 209, 211, 218–223

418 Index of subjects belief verbs, 209, 211–212 conditional, 127, 133, 210, 212, 223, 400 copular verbs, 209 evaluative adjectives, 210 future indicative, 210 modal adverbs, 5, 171–173, 191–196, 203–206, 209–223, 395 modals, 209–212, 284 periphrastic forms, 133, 153–154, 158– 164, 209, 212 modal scope, 173, 196, 209–223 modality alethic ~, 172, 205–206, 211, 215–217, 220–223 deontic ~, 194, 205–207, 210, 215–216, 219–220, 223 epistemic ~, 5, 117, 120, 123, 126–127, 172, 205–206, 211, 215–223 modalization, 204, 212–223 Modus, 203, 205, 221 negation, 5, 124, 129–130, 134–148, 158, 165–166, 212, 333, 357, 397 negative, 120, 123, 127, 130, 133–134, 137–148, 156–158, 165, 184, 186, 192, 397–398 Ø marking, 86, 286–290, 294, 321, 328, 336–339, p-entailment, 158 physical perception predicate, 317 post-nuclear region, 25–27, 31–35 pragmatic function, 1, 7, 15, 36, 78, 88, 108, 115, 230–233, 236, 263, 307, 320, 328, 336–337, 342, 352, 398 pragmatics, 4, 79, 87, 115, 153, 164, 204, 257, 307, 315, 318, 342, 378 Prague School, 1, 7, 76 predicate-focus structure, 254, 262, 356– 357, 361 pre-nuclear region, 25, 27, 33 presentational sentence, 356

presupposition, 17, 77, 154–166, 251, 254, 376 principles of information structure, 94– 95 principle of ‘newsworthiness’, 80–84 principle of progression from given to new information, 78, 94–95, 308 principle of prominence, 85 proclisis, 174–175, 272–273, 289–294, 299 proform proverb, 114 verbal ~, 110–112, 125–126 pronoun personal ~, 193, 281, 285, 300, 316, 318 strong ~, 269, 271 proposition, 132, 144, 203–208, 222– 234, 259, 352–357, 361, 375–376 prosodic ~ annotation, 37, 51, 55, 69 ~ marking, 15, 24, 27, 36, 88, 97 ~ parsing, 52–58, 69 ~ structure, 4, 15, 24, 35–38, 54–58, 65, 69, 124 ~ units, 4, 21, 51–52, 59, 107, 147, 171, 209, 330, 333, 335 prosodic focus, 15–18, 23–39, 59, 66–67, 137, 194 prosodic structure, 4, 15–40, 54, 56, 58, 65, 69, 124 phonological structure, 7, 35, 76, 81 prosody, 2, 4, 15–20, 23–26, 35–38, 51– 59, 65, 76, 97, 125, 207–208, 321, 360 pseudo-cleft, 5, 146, 153-157, 160–165 Quaestio (model), 352–354, 376 reduplication, 307, 317–321, 331 referential head referential means, 363, 369–371, 374, 378 referential movement, 353, 360 referentiality, 316, 320

Index of subjects 419 Relational Grammar, 95 relative clause, 155–156, 173, 312, 340– 341 resumptive pronouns, 81–82, 87–88, 314, 320, 328 scope, 111–112, 129, 132, 141, 145–146, 173, 184, 188, 193, 196, 203–205, 218– 222, 319, 328, 331–344, 357, 391–394, 400 segmental, 55, 307, 310, 312, 321 sentence adverbial, 157–160 sentence-focus structure, 251, 253, 356, 361 side structure, 353, 376 small clause, 312 stress, 6, 17, 22, 33–34, 36, 38–40, 107, 124–125, 132, 141, 146, 148, 155, 190, 244, 255, 299, 355, 376 metrical ~, 33–34 subordination, 53, 131, 212, 340, 351, 360, 379, 389 syntactic order, 75–79, 86, 94–95 basic word order, 308 free/flexible word order, 79–80, 86, 307, 314, 320–321 rigid word order, 342 word order, 2, 6, 16, 21–23, 38, 78, 80, 82, 131, 155, 160, 164, 166, 274– 275, 294, 297–299, 307–308, 315, 318–320, 351, 364, 379 verb-initial word order, 321 syntactically-marked phenomena, 83, 86, 87, 94 tail, 18, 93, 97 theme, 5, 18, 24, 75, 77, 91, 155, 161–164, 260, 270–279, 282–283, 286–297, 307, 313, 321, 329, 355, 357, 368, 387–409 thetic, 234–235, 248, 255–260, 263, 274– 277, 296, 356, 365 ~ sentence, 260 Tobler-Mussafia Law, 273, 289–290 tonal events, 4, 19–20, 23, 326, 330, 342– 343

tonic syllable, 89–90, 376 topic, 1–7, 15–40, 55, 59, 62–63, 69, 75– 78, 81–97, 108, 112–121, 126, 155, 170–195, 204–205, 208, 211–223, 230–235, 240–248, 253–263, 308, 313–316, 321–322, 325–345, 352– 378, 387–408 ~ assignment, 112 ~ chain, 362–363, 369 ~ construction, 108, 115, 328 ~ markers, 81, 352, 356–359, 377 ~ phrase, 6, 37, 356 ~ predicate, 113–114 ~ time, 112–117, 120–121 aboutness-shift ~, 355, 376–377 contrastive ~, 19, 20, 85, 90, 317, 326, 342, 355–357, 376–377 discourse ~, 245–247, 254–256, 261– 262, 354, 361–365, 369, 373 familiar ~, 20, 38, 326, 332, 335–341, 344–355, 376 hanging ~, 84, 87, 116, 301, 314, 317, 320, 362 infinitive ~, 116 internal ~, 329–332 left-hand ~, 329 multiple ~, 337 nominal ~, 115–119 non-nominal ~, 108 predicate ~, 113 right-hand ~, 7, 329, 397 sentence ~, 16, 19, 30, 355, 361–362, 366, 368, 373 topicalization strategies, 81 truth relation, 158 utterance compound ~, 54, 62–64, 204–205, 214, 223 simple ~, 54, 62, 204, 209–210, 219– 220, 223 verbal ~, 55, 169–170, 174, 176, 193, 212 verbless, 60, 67

420 Index of subjects V2, 286, 292, 297, 364 Verum-Fokus, 120

wa, 321, 356–357, 406–407 Wh-cleft, 155

Index of languages

Afar, 340 Afro-Asiatic, 106 Aghem (Kru), 137, 148 Albanian, 314, 319 Arabic, 358–359 Aramaic Babylonian Talmudic ~, 107 Bulgarian, 6, 307, 314–317, 319–322 Catalan, 97 Cayuga, 82 Chinese, 81–87, 96, 126, 358–359 Mandarin ~, 2, 26, 83 classical languages, 80, 280 Dutch, 359 English, 4, 15–19, 26, 34, 84, 90, 110– 113, 118, 122, 124–126, 146, 155– 156, 162–163, 166, 234, 296, 301, 315, 327, 344, 358–359, 363, 376 American ~, 37 European languages, 15, 106, 108, 312, 321 French, 15, 26, 52, 69, 87, 89, 117, 123, 130–131, 145, 148, 222, 298, 359, 363, 376 Gascon, 140 German, 15, 19–20, 110–115, 122–126, 234, 325, 344, 352, 356, 360, 363–379 Germanic, 120, 124 Grébo (Kru), 137–138 Greek, 15, 238, 249, 287, 314 Haitian, 125 Hebrew, 115, 126

Hmong, 359 Hungarian, 81, 86, 114–115, 126, 140, 147 Indo-European, 106, 111–113, 121, 124 Italian, 1, 4–7, 15, 20–27, 34–40, 52–53, 69, 77–78, 81, 86–88, 91–97, 105–126, 129, 148, 153–166, 169–178, 183, 193, 195, 203–204, 209, 215, 222, 230, 233, 236–239, 244, 260, 262, 269– 274, 277, 280, 283, 285, 289–295, 297–301, 314, 321–322, 325, 344, 351–378, 388, 407 ~ dialects, 1, 4, 129 Bari ~, 24, 252 Florentine ~, 23–24, 220, 272 Milanese ~, 20 Neapolitan ~, 4, 15, 24–25, 36 Old ~, 6, 229–230, 233, 236, 240–244, 257–259, 261, 269, 272–280, 286– 297, 300 Palermo ~, 24 Pisa ~, 21 Rome ~, 20–23 Japanese, 81, 165, 296, 321, 356–358, 378, 406–407 Korean, 2, 94, 97, 322, 340, 358, 377 Latin, 169, 244, 255, 262, 280, 283 Logudorese (Sardinian), 5, 129 Macedonian, 6, 307, 314, 318–322 Malagasy, 325 Paduan, 141–145 Persian, 358–359 Polish, 6, 307–308, 310–313, 320–321 Portuguese, 52, 69, 222

422 Index of languages Romance, 113, 117, 124, 129, 142, 169, 243, 272, 275, 280, 289, 293–294, 297, 314, 316, 320, 325, 340, 364 Romanian, 113, 124, 314, 319 Russian, 6, 107, 109, 112–113, 121, 138, 307–313, 320–321 Sardinian, 5, 129–135, 141–146 Semitic, 106 Sicilian, 5, 153–156, 158–165 Slavic, 6, 307, 310, 313–314, 318–321 Balkan ~, 307, 314–321 Somali, 6, 81, 86, 96, 325, 328–345 Spanish, 52, 69, 117, 222, 244, 352, 358– 378

Swedish, 114 Tagalog, 325 Tigrinya, 358–359 Turkic, 106, 124 Turkish, 113, 119–122, 321, 340 Tuscan, 6, 169, 174–175, 178, 193, 259, 262, 272, 280, 298, 301 Wambon, 2 Warlpiri, 79–80, 86, 95 Wobé (Kru), 137, 138 Yiddish, 105, 112–115, 121 Yukaghir, 138–139, 148

Index of persons

Abraham, Werner, 2, 77 Ahrenholz, Bernt, 374 Aissen, Judith, 2 Albano Leoni, Federico, 39 Allen, James F., 30 Amacker, René, 226 Andersen, Henning, 149 Anderson, Anne H., 27 Andorno, Cecilia, 352, 355, 358–359 Antinucci, Francesco, 299–300, 308, 340 Arcangeli, Massimo, 232, 254, 261–262, 275 Ariel, Mira, 379 Aronson, Howard I., 149 Austin, John L., 69, 170, 208 Avesani, Cinzia, 21, 25, 37, 39 Bacelar, Fernanda, 70 Bader, Miles, 40 Baker, Mark, 80, 82, 325, 328 Bally, Charles, 203, 408 Barnes, Betsy K., 261 Barry, William J., 52 Baumann, Stefan, 37 Beckman, Mary, 47 Belletti, Adriana, 79 Bendiscioli, Alessandra, 380 Benincà, Paola, 38–39, 86, 107–109, 118, 126, 165, 239–240, 260–262, 277–288, 299 Berent, Gerald P., 318 Berman, Ruth A., 363, 375 Bernini, Giuliano, 2, 5, 29, 105, 108, 115– 116, 125–127, 352, 358–359 Berretta, Monica, 121, 169, 175, 259, 354, 361–362 Berruto, Gaetano, 23, 91, 388 Bertinetto, Pier Marco, 24, 39 Bliese, Loren, F., 340

Blumenthal, Peter, 236, 238–239, 242, 252, 260, 262 Bolinger, Dwight. L., 17, 37, 55 Bondanella, Peter, 237–239 Bonvino, Elisabetta, 22, 87, 96, 344, 362 Bornstein, Daniel E., 248, 250 Bossong, Georg, 91, 260, 269, 273–275, 294 Boyle, Elizabeth H., 40 Brazil, David, 69 Bresnan, Joan, 96 Brody, Michael, 2 Brown, Gillian, 30, 37 Buommattei, Benedetto, 169 Büring, Daniel, 2, 16, 18, 20, 37, 96–97, 326 Burnard, Lou, 73 Burzio, Luigi, 96 Bussmann, Hadumod, 155, 157 Butler, Christopher S., 69 Bybee, Joan, 203, 209 Caldognetto, Emanuela M., 24 Canepari, Luciano, 24 Caputo, Maria R., 25, 38 Cardinaletti, Anna, 277, 301 Carletta, Jean, 27 Carroll, Mary, 351, 363–364, 366, 374– 375 Castelvetro, Lodovico, 169 Chafe, Wallace L., 1, 3, 37, 155, 230, 235, 246, 257, 326, 355 Chierchia, Gennaro, 159, 166 Chilin, Shih, 30 Chini, Marina, 6, 351–352, 360, 374–375, 377, 379 Choi, Hye-Won, 2, 93, 97, 322 Chomsky, Noam, 79, 334, 355, 375 Choucri, Kalid, 70

424 Index of persons Cid, Antonio J., 70 Cinque, Gugliemo, 79, 301, 308, 314, 317, 319, 334, 356 Cohen, Antonie, 23 Coletti, Vittorio, 405 Collier, René, 23, 53 Comrie, Bernard, 78, 138–139, 147–148, 313 Consoli, Joseph P., 238, 248–250, 252 Conte, Amedeo G., 207 Conte, Maria-Elisabeth, 169, 190, 207 Contini, Michel, 131, 147 Corblin, Francis, 196 Cornish, Francis, 196 Cresti, Emanuela, 4–5, 7, 16, 23, 38, 51– 55, 58–60, 63–64, 69, 97, 169–170, 191, 193–194, 204–205, 207–208, 221– 223, 387–400, 402–404, 406–408 Crocco, Claudia, 4, 7, 15, 23, 39, 96, 97 Cruttenden, Alan, 37 Crystal, David, 55 Culioli, Antoine, 145 Currie, Karen L., 41 Cutugno, Francesco, 39 Cychun, Gennadij A., 319 Dalewska Gre , Hanna, 322 Dane , Frantisek, 55 David, Rudolf, 261 De Cesare, Anna-Maria, 402, 404–405, 407, 408 De Jong, Jan R., 9, 98 De la Grasserie, Raoul, 148 De Mauro, Tullio, 149 De Meij, Sjaak, 2 De Vries, Lourens, 9, 98 Delais-Roussarie, Elisabeth, 26 Den Besten, Hans, 125 Dezs , Lazslo, 140, 147 Di Cristo, Albert, 26, 38, 193, 229, 242 Di Tullio, Ángela, 155, 165 Diessel, Holger, 409 Dik, Simon, 2, 81, 354, 357, 376 Dimroth, Christine, 381

Dittmar, Norbert, 381 Djiang, Sie Ing, 9, 98 Doetjes, Jenny, 43 Doherty-Sneddon, Gwyneth, 42 Downing, Angela, 399, 409 Dressler, Wolfgang U., 190 Du Bois, John W., 357, 362, 377 Dupont, Christian, 259 Endo, Reiko, 24 Engdhal, Elisabet, 48 Erteschik-Shir, Naomi, 2–3, 7, 97 Evans, K., 169 Falavigna, Daniele, 70 Fava, Elisabetta, 142, 148 Fenyvesi, Anna, 126 Ferrari, Angela, 387, 389, 393, 398, 400– 403, 407–409 Ferrari, Giacomo, 27 Féry, Caroline, 19 Fesenmeier, Ludwig, 6, 229, 260–261 Feuillet, Jack, 150 Firbas, Jan, 1, 355 Firenzuoli, Valentina, 16, 23, 58, 60, 64, 193–194, 222 Fivela, Barbara Gili, 20–21, 24–25, 38– 39 Fleischman, Suzanne, 203, 209, 257 Foulet, Lucien, 131 Fourcin, Adrian J., 52 Fox, Barbara, 381 Franks, Steven, 316, 319, 321–322 Franzén, Torsten, 298 Frascarelli, Mara, 2, 6, 16, 20–22, 34, 38, 59, 83, 87, 91, 96, 312, 325, 328, 332– 333, 338–339, 341–344, 355–356, 361, 375–377 Fretheim, Thorstein, 381 Friedman,Victor, 318–319, 322 Frison, Lorenza, 38–39, 86, 107–109, 118–119, 126 Frosali, Fabrizio, 64, 194, 222 Fuller, Judith W., 357–359

Index of persons 425 Galvan, Sergio, 207 Garrod, Simon C., 40 Gebert, Lucyna, 6, 307 Geluykens, Ronald, 30 Giacalone Ramat, Anna, 125, 352, 375, 377 Giani, Daniela, 64, 194 Gili Fivela, Barbara, 20–21, 24–25, 38– 39 Giordano, Rosa, 23, 39, 229 Givòn, Talmy, 59 Glovinskaja, M. Ja., 313 Goldenberg, Gideon, 106–109, 113–115, 120, 126 Goldsmith, John, 326 Gómez-Gonzalez, María A., 7 Graffi, Giorgio, 204 Greenberg, Joseph H., 78 Grice, H. Paul, 262 Grice, Martine, 24–26, 34, 37–39 Grobet, Anne, 361, 376 Gruppo di Padova, 270 Guentchéva, Zlatka, 315–316 Gullberg, Marianne, 382 Gumperz, John, 351 Gundel, Jeannette K., 16, 19, 38, 155, 166, 354, 357–359 Gurman Bard, Ellen, 40 Gussenhoven, Carlos, 17, 30, 35 Haegeman, Liliane, 338 Haiman, John, 378 Hale, Kenneth, 79, 80, 82, 86, 95 Halliday, Michael A. K., 1, 76, 190, 355, 387, 399, 409 Hare, Richard M., 207 Hasan, Ruqaiya, 190 Hatcher, Anna Granville, 235, 251, 256, 260–262 Hayes, Bruce, 344 Hedberg, Nancy, 16, 19, 34 Heine, Bernd, 319 Heldner, Mathias, 30 Hendriks, Henriëtte, 126, 363

Hickmann, Maya, 360, 375 Hinchliffe, Ian, 114 Hinds, John, 406 Hinterhölzl, Roland, 16, 20, 325, 355– 356, 376–377 Hirschberg, Julia, 17–19, 37 Hirst, Daniel, 193 Hoffman, Maria E., 9, 98 Höhle, Tilman N., 120 Holmes, Philipp, 114 Ho-Min, Sohn, 340 Horn, Laurence R., 159, 166 Horvarth, Julia, 9 Hoye, Leo, 206 Huang, Yan, 375 Huddleston, Rodney, 206 Huebner, Thom, 359 Huszcza, Ryszard, 310, 312–313 Hyman, Larry M., 137 Interlandi, Grazia, 39 Irsara, Martina, 262 Isard, Amy, 42 Isard, Stephen, 40, 42 Iulianella, Claudia, 262–263 Jackendoff, Ray, 7, 15, 17–19, 37 Jacobs, Joachim, 410 Jacobs, Neil G., 114 Jaku kina, E. I., 324 Jankowski, Ludovic, 26, 38 Jelinek, Elinor, 79–80 Jensen, Anne, 199 Jin, Hong G., 358, 359 Jinga, Laurenia, 113 Jones, Michael A., 131, 133, 145–146 Kaiser, Georg, 261 Karcevsky, Serge, 57 Kayne, Richard S., 79, 169, 340 Kenesei, István, 126 Kenworthy, Joanne, 41 Khorounjaia, Ekaterina, 383 Kibrik, A. A., 324

426 Index of persons Kiefer, Ferenc, 203, 206 King, Tracy H., 2, 237, 272, 276–278, 280, 283, 295–296, 316, 319, 321–322 Kiss, Katalin É., 2, 77, 81, 96 Kleiber, Georges, 169, 195 Klein, Wolfgang, 108, 111–113, 115, 123, 126–127, 353–354, 358–360, 362, 376 Koch, Peter, 232 Kornfilt, Jaklin, 347 Korzen, Iørn, 169, 193, 195 Kotschi, Thomas, 155–156, 162 Kowtko, Jacqueline, 40, 42 Krahmer, Emiel, 37 Krapova, Iliyana, 314, 317, 319, 322 Kuno, Susumo, 326 Kuroda, S.-Y., 296 Kuteva, Tania, 319 La Fauci, Nunzio, 156, 164 Ladd, D. Robert, 17, 24–25, 35, 37, 39, 55 Lahiri, Aditi, 344 Lambert, Monique, 351, 363–364, 366, 375 Lambrecht, Knud, 2, 19–20, 37–38, 59, 127, 194, 234–235, 251, 253–254, 260– 262, 342, 352, 354–357, 361, 375–378 LaPolla, Randy J., 76 Lapteva, Ol’ga A., 313–314 Lay, G., 222 Leafgren, John, 321–322 Lecarme, Jacqueline, 2, 96 Lehmann, Christian, 320 Lenart, Ewa, 384 Lepschy, Anna Laura, 193 Lepschy, Giulio C., 193 Levelt, Willem, 351, 353 Levinson, Stephen C., 158–159, 162, 166, 351 Li, Charles N., 1, 3, 77, 83–85, 194, Li, Ping, 126 Liang, James, 383 Lillo-Martin, Diane, 378 Llisterri, Joaquim, 52, 69 Lo Cascio, Vincenzo, 169

Lockwood, William B., 105, 114 Lombardi Vallauri, Edoardo, 2, 7, 35, 69, 95–97, 127, 155, 159, 194, 233, 258, 261, 263, 292, 296–297, 387, 409 Lyons, John, 203, 206, 222 MacWhinney, Brian, 52 Maiden, Martin, 150 Mambelli, Marcantonio, 199 Manzotti, Emilio, 159, 162 Mapelli, Valerie, 70 Marandin, Jean-Marie, 43 Marcantonio, Angela, 296, 299–300 Marchese, Lynell, 137, 148 Marotta, Giovanna, 39, 125, 377 Martin, Philippe, 193, 329 Maslova, Elena, 29, 108, 115–116, 125– 127, 138, 147 Mathesius, Vilem, 1 Matras, Yaron, 384 McAllister, Jan, 40 McArthur, Tom, 155 Meillet, Antoine, 80 Mereu, Lunella, 1–2, 4, 7, 16, 22–23, 59, 69, 75, 82–83, 87, 96–97, 194, 314, 317, 320–321, 344–345, 358, 375– 377, 396 Meyer-Lübke, Wilhelm, 131 Michaelis, Laura, 19 Miller, Jim, 312, 321 Milos, Elena, 27 Mirto, Ignazio M., 5, 153, 156, 162, 165 Mithun, Marianne, 3, 80, 82–84, 377 Moldovan, A. M., 315, 321 Molinu, Lucia, 133 Molnár, Valéria, 7, 97, 326, 354 Moneglia, Massimo, 4, 51–52, 69, 169– 170, 193–194, 204, 207, 222–223 Morgan, Jerry L., 232 Morris, S., 198 Mortara-Garavelli, Bice, 195 Muller, Claude, 169 Munaro, Nicola, 298 Muñoz, Carmen, 374

Index of persons 427 Munro, Pamela, 378 Murcia Serra, Jorge, 375 Musa, Mark, 237–239 Mussafia, Adolfo, 272 Nakajima, Shin’ya, 30 Nespor, Marina, 20, 27, 148 Nocentini, Alberto, 271–272 Nuyts, Jan, 206 Oesterreicher, Wulf, 232, 260 Ostendorf, Mori, 47 Palermo, Massimo, 24, 156, 160, 164, 299, 300 Palmer, Frank R., 203, 206, 209, 222 Papafragou, Anna, 206, 223 Parry, Mair M., 143 Payne, Doris, 2 Payne, Jonathan, 69 Perdue, Clive, 358–359 Perkins, Michael R., 206, 209 Perlmutter, David M., 96 Pesetsky, David, 326 Pettorino, Massimo, 97 Pierrehumbert, Janet B., 17–19, 21, 24, 37, 69, 343 Pietrandrea, Paola, 206 Pitrelli, John F., 47 Pittau, Massimo, 131 Plungian, Vladimir, 226 Poletto, Cecilia, 165, 260, 296 Polinsky, Maria, 313 Portner, Paul, 142–143, 145 Pottier, Bernard, 207 Price, Patti, 47 Primus, Beatrice, 231, 234, 260 Prince, Ellen F., 37 Puglielli, Annarita, 2, 6, 81, 96, 325, 328, 332, 338, 341, 343–345 Pullum, Geoffrey K., 206 Push, Claus D., 140 Quirk, Randolph, 55

Raible, Wolfgang, 257 Rainò, Nicola, 147 Ramat, Paolo, 113, 125, 352, 375, 377 Refice, Mario, 38 Reinach, Adolf, 137 Reinhart, Tanya, 2, 169, 231, 326 Renzi, Lorenzo, 193, 233–234, 261, 277, 288–289, 296, 299 Rialland, Annie, 43 Ricca, Davide, 165 Rigamonti, Alessandra, 159, 162 Rizzi, Luigi, 2, 6, 79, 96, 169, 325–326, 339, 342, 355–356, 375 Rochemont, Michael, 2 Rogova, N. V., 324 Rohlfs, Gerhard, 140, 148, 160 Roland, Françoise, 383 Rooth, Mats, 2 Rosen, Carol, 156 Rossi, Mario, 55, 387 Rudin, Catherine, 316, 319, 321 Sabatini, Francesco, 405 Saeed, John I., 158–159, 344 Salvi, Giampaolo, 38–39, 86, 107–109, 118–119, 126, 154–156, 251, 254, 261, 277, 281–282, 287, 292, 298 Sandoval, Antonio M., 70 Sansò, Andrea, 385 Sasse, Hans-Jürgen, 2, 248, 257, 263, 365 Savino, Michelina, 38, 39 Savy, Renata, 23, 39 Scarano, Antonietta, 4–5, 51, 193, 221– 222 Schlobinski, Peter, 231, 260 Schneider, Stefan, 195, 204 Schütze-Coburn, Stephan, 231, 260 Schwartz, Marcia L., 2 Schwarze, Christoph, 409 Schweikert, Walter, 344 Selig, Maria, 244–245, 255, 262 Selkirk, Elisabeth O., 20 Serianni, Luca, 193, 297 Seuren, Pieter A. M., 158, 169

428 Index of persons Sewell, Julie, 225 Siewierska, Anna, 310, 315, 321 Signorini, Sabrina, 60, 183, 194, 222 Silverman, Kim, 37 Simone, Raffaele, 91, 193, 217 Sinclair, John M., 69 Skorbid, S. S., 324 Slobin, Dan I., 351, 363, 375 Sluijter, Agaath, 30 Sobrero, Alberto A., 195 Soria, Claudia, 27 Sorianello, Patrizia, 39 Sornicola, Rosanna, 22, 96, 155, 238– 239, 260–262, 280, 296–297, 321, 355, 376 Sosa, Juan M., 16, 19, 34 Sotillo, Catherine, 40 Sperberg-McQueen, Michael C., 73 Stalnaker, Robert C., 158, 166 Stark, Elisabeth, 235, 260, 261, 355 Starren, Marianne, 381 Stavinschi, Alexandra C., 262 Steedman, Mark, 18–19 Steube, Anita, 7, 97 Stone, Gerald, 313 Strangert, Eva, 30 Strawson, Peter F., 231, 355 Stroomer, Harry, 9, 98 Sun, Xuejing, 26 Suomela-Härmä, Elina, 405 Suzuki, Shingo, 6, 269, 270, 275, 289, 291–292 Svolacchia, Marco, 96, 328, 344–345 Swerts, Marc, 30, 37 widzi ski, Marek, 312 Tamburini, Guido, 194 Terken, Jacques, 30 Tesnière, Lucien, 152 Thompson, Henry, 1, 3, 77, 83–85, 409 Thompson, Sandra A., 1, 3, 77, 83–85, 409 Tolchinsky, Liliana, 383 Tolstaja, S. M., 324

Tomlin, Russell S., 357 Topoli ska, Zuzanna, 318–320, 322 Trecci, Alessandra, 2, 16, 22–23, 59, 96– 97, 344, 361, 375,–377, 396 Trévise, Anne, 359 Tsujioka, Takae, 165 Tucci, Elena, 63, 194 Tucci, Ida, 5, 64, 69, 170, 193–194, 196, 203–204, 209, 217–218, 222–223 Uhlíová, Ludmila, 310, 315, 321 Ulrich, Miorita, 260 Umeda, Noriko, 30 Vago, Robert M., 126 Valentini, Ada, 379, 386 Vallduvì, Enric, 1, 7, 93 Van der Auwera, Johan, 127, 226 Van Kuppevelt, Jan, 355, 376 Van Valin, Robert D., 76, 159, 377–378 Vanelli, Laura, 231, 243, 259–260, 261– 262, 281–282, 288, 292, 299 Vater, Hans, 386 Vayra, Mario, 21, 37, 39 Veenstra, Tonjes, 125 Venier, Federica, 206, 263 Veronis, Jean, 70 Vilkuna, Maria, 19 Vincent, Nigel, 95 Virdis, Maurizio, 152 Vogel, Irene, 148 Von Stutterheim, Christiane, 351, 353– 354, 360, 362–363, 374–376 Von Wright, Georg H., 207 Wagner, Max Leopold, 131 Wandruszka, Ulrich, 244 Ward, Gregory, 17–18 Watorek, Marzena, 363 Weigand, Edda, 245, 254 Weinert, Regina, 40 Wicksteed, Philip H., 242, 259 Wietske, Vonk, 206 Wightman, Colin, 47

Index of persons 429 Winkler, Susanne, 2, 7, 97 Xu, Ching X., 26, 30, 34–35, 36, 40 Xu, Yi, 26, 30, 34–36, 40 Yule, George, 30

Zacharski, Ron, 16, 18–19 Zamboni, Alberto, 142 Zanuttini, Raffaella, 142–145 Zubizarreta, Maria L., 386 uravlev, A. F., 324