Discourse Functions at the Left and Right Periphery. Cross-linguistic Investigations of Language Use and Language Change 978-90-04-27480-8, 978-90-04-27482-2

Table of contents :
Contents......Page 5
List of Contributors......Page 7
Chapter 1 Introduction......Page 9
Chapter 2 Moi je ne sais pas vs. Je ne sais pas moi: French Disjoint Pronouns in the Left vs. Right Periphery......Page 32
Chapter 3 Motivations for Meaning Shift at the Left and Right Periphery: well, bon and hao......Page 55
Chapter 4 On the Function of the Epistemic Adverbs Surely and No Doubt at the Left and Right Peripheries of the Clause......Page 80
Chapter 5 Setting Up a Mental Space: A Function of Discourse Markers at the Left Periphery (LP) and Some Observations about LP and RP in Japanese......Page 100
Chapter 6 Italian guarda, prego, dai. Pragmatic Markers and the Left and Right Periphery......Page 125
Chapter 7 ‘So very fast then’ Discourse Markers at Left and Right Periphery in Spoken French......Page 159
Chapter 8 On the Development of Sentence Final Particles (and Utterance Tags) in Chinese......Page 187
Chapter 9 The Interplay of Discourse and Prosody at the Left and Right Periphery in Korean: An Analysis of kuntey ‘but’......Page 229
Author Index......Page 259
Subject Index......Page 263

Citation preview

Studies in Pragmatics Series Editors Maj-Britt Mosegaard Hansen (University of Manchester) Kerstin Fischer (University of Southern Denmark) Anne Barron (Leuphana University Lüneburg)

VOLUME 12

The titles published in this series are listed at brill.com/sip

Discourse Functions at the Left and Right Periphery Crosslinguistic Investigations of Language Use and Language Change

Edited by

Kate Beeching and Ulrich Detges

LEIDEN | BOSTON

Library of Congress Cataloging-in-Publication Data Discourse functions at the left and right periphery : crosslinguistic investigations of language use and language change / Edited by Kate Beeching and Ulrich Detges.   p. cm. — (Studies in pragmatics ; Volume 12)  Includes index.  ISBN 978-90-04-27480-8 (hardback : alk. paper) — ISBN 978-90-04-27482-2 (e-book) 1. Discourse markers. 2. Intercultural communication. 3. Cross-cultural orientation. 4. Linguistic change. 5. Pragmatics. I. Beeching, Kate, editor. II. Detges, Ulrich, editor. P302.35.D56 2014 401’.41—dc23 2014024215

This publication has been typeset in the multilingual ‘Brill’ typeface. With over 5,100 characters covering Latin, ipa, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.com/brill-typeface. issn 1750-368X isbn 978-90-04-27480-8 (hardback) isbn 978-90-04-27482-2 (e-book) Copyright 2014 by Koninklijke Brill nv, Leiden, The Netherlands. Koninklijke Brill nv incorporates the imprints Brill, Brill Nijhoff, Global Oriental and Hotei Publishing. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill nv provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, ma 01923, usa. Fees are subject to change. This book is printed on acid-free paper.

Contents List of Contributors  vii 1 Introduction  1 Kate Beeching and Ulrich Detges 2 Moi je ne sais pas vs. Je ne sais pas moi: French Disjoint Pronouns in the Left vs. Right Periphery  24 Ulrich Detges and Richard Waltereit 3 Motivations for Meaning Shift at the Left and Right Periphery: well, bon and hao  47 Kate Beeching and Yu-Fang Wang 4 On the Function of the Epistemic Adverbs Surely and No Doubt at the Left and Right Peripheries of the Clause  72 Elizabeth Closs Traugott 5 Setting Up a Mental Space: A Function of Discourse Markers at the Left Periphery (LP) and Some Observations about LP and RP in Japanese  92 Noriko O. Onodera 6 Italian guarda, prego, dai. Pragmatic Markers and the Left and Right Periphery  117 Chiara Ghezzi and Piera Molinelli 7 ‘So very fast then’ Discourse Markers at Left and Right Periphery in Spoken French  151 Liesbeth Degand 8 On the Development of Sentence Final Particles (and Utterance Tags) in Chinese  179 Foong Ha Yap, Ying Yang and Tak-Sum Wong

vi

contents

9 The Interplay of Discourse and Prosody at the Left and Right Periphery in Korean: An Analysis of kuntey ‘but’  221 Sung-Ock S. Sohn and Stephanie Hyeri Kim Author Index  251 Subject Index  255

List of Contributors Kate Beeching University of the West of England, Bristol, United Kingdom Liesbeth Degand Université Catholique de Louvain, Belgium Ulrich Detges Ludwig-Maximilians-Universität, Munich, Germany Chiara Ghezzi Università di Bergamo, Italy Stephanie Hyeri Kim California State University, Northridge, United States Piera Molinelli Università di Bergamo, Italy Noriko Onodera Aoyama Gakuin University, Japan Sung-Ock S. Sohn University of California, Los Angeles, United States Elizabeth Traugott University of Stanford, United States Richard Waltereit Newcastle University, United Kingdom Yu-Fang Wang National Kaohsiung Normal University, Taiwan, R.O.C. Tak-Sum Wong Hong Kong Polytechnic University, Hong Kong, S.A.R.

viii Foong Ha Yap Hong Kong Polytechnic University, Hong Kong, S.A.R. Ying Yang University of California, Los Angeles, United States

list of contributors

chapter 1

Introduction Kate Beeching and Ulrich Detges 1

Moving from Left to Right

A basic property of human language is that discourse unfolds in time—an insight whose significance we owe to Ferdinand de Saussure. This is represented in Western linguistics as progression from “left” to “right”. A logical implication of this is that the left and right margin of units of language have different functions. For example, in dialogical conversation, the left margin of the most basic unit, the turn, is the place where the speaker takes the right to speak, whereas at the right margin the floor is handed over to the hearer. The left and right margin of discourse units and of the sentence similarly fulfil very different functions and therefore do not behave in a symmetrical fashion. From a diachronic point of view, a hypothesis arising from Traugott (1982) and Traugott and Dasher (2002) is that lexical elements recruited from medial positions in the sentence or utterance can come to serve textual or subjective functions at LP (in English, at least). These elements may then gravitate to the RP, where they play a more intersubjective or modalising role. Functional approaches consider, as Croft (2000: 4) puts it, that “the real entities of language are utterances and speakers’ grammars. Language change occurs via replication of these entities not through inherent change of an abstract system.” Chronologically, however, linguistic scholars have concerned themselves firstly with the sentence, and the peripheries of the sentence, before moving to discourse-level and turn- and utterance-level considerations. In what follows, we trace the history of what has been referred to as ‘periphery’, provide some definitions of key terms used throughout the volume and an overview of the different ways that the notion of periphery has been modelled. This is followed by a list of the research questions which we aimed to address and summaries of the individual chapters in the volume. 2

The ‘Periphery’ of What?

The notion of ‘periphery’ may in principle be defined in relation to the argument structure, to the sentence, the turn or the utterance. It can also be used to

© koninklijke brill nv, leiden, ���4 | doi ��.��63/9789004274822_�02

2

Beeching and Detges

refer to discourse relations between two sentences, turns or utterances. Each of these conceptualisations will be examined below. 2.1 Periphery of the Sentence or Argument Structure A sentence may be made up of an argument structure or an argument structure along with additional adjuncts. The argument structure can be defined as the verb and obligatory arguments which pertain to the verb. The subject and, to a lesser degree, the object are the core arguments of verbal predicates, as we see in sentences such as Mary hugged Fred, Hugh ate (breakfast). Adjuncts might include adverbs such as furtively or quickly. Since argument structure as such is not a topic of this book, we will not further elaborate on this question. The left periphery of the sentence was extensively explored in functionalist linguistics in the 1970s and 1980s as well as in generative linguistics in the 1990s and 2000s. Themes that have been recurrently associated with the left periphery in language after language include: topicalization, topic shift, clefting and focalization. Attention has also been paid in such frameworks to ‘dislocation’, with both left- and right-dislocated elements. Cinque and Rizzi’s (2008) carto­ graphy project is an excellent example of work which endeavours to capture such elements in syntactic description. What we find at the peripheries of the sentence are essentially information-structuring devices. It is important to point out that elements which contribute to information-structuring are part of the argument-structure, often linked to it by co-reference, as we can see in examples (1) a.-d., cited from Traugott 2013. (1)  a. That guy over there we like. b. That gal over there, she’s smart. c. She’s smart, that gal over there. d. It’s smart she is, that gal over there.

Most attention in the literature has been paid to phenomena which occur on the left periphery of the sentence, such as the topicalization, focalization and clefting which we see exemplified in (1) a. b. and d. The right periphery, by contrast, has been much less investigated, from either a functionalist or a generative perspective, though right-dislocated elements such as ‘that gal over there’ in (1) c. and d. have been recognised as extremely common in Romance languages such as French (Lambrecht 1981, 1994, 2001; de Cat, 2007a, 2007b) and Italian (Cechetto 1999, Frascarelli 2004). Let us consider in more detail the right-dislocated element in Italian featured in example (2).

Introduction

3

(2) Spegnilo il registratore! (Italian) ‘Switch-off-it the recorder!’

Traditionally, it has been argued that the purpose of right dislocation is to address potential problems of reference tracking by proffering after the event the full NP corresponding to a previously mentioned but possibly ambiguous pronoun. However, many authors have recognised that right dislocation may have additional, more subtle, functions. For instance, in (2), as argued by Koch and Oesterreicher (2011: 67), the right dislocation, by removing focal material from the VP (i.e. ‘the recorder’), helps to bring the verb phrase itself into focus, thereby strengthening the speech act thus performed. In other words, in addition to playing an information-structuring role, right dislocation may also have a modal function. The pragmatic functions of right-dislocation in (varieties of) English have also been explored by Aijmer (1989) and Durham (2011). 2.2 Periphery above the Sentence Level At the level of discourse, that is to say, above sentence level, discourse markers to the left connect the upcoming stretch of discourse to what has already been said. For example, in (3) He is a Republican, but he is honest

the discourse marker but, at the left periphery of the proposition , serves to integrate the latter with the seemingly contradictory preceding proposition . A major function of many items in the left periphery of the sentence and of discourse segments seems to consist in helping to bring about coherence, at either a textual or propositional level. 2.3 Periphery of the Utterance or the Turn Everyday conversation is made up of turns, with one speaker taking a turn followed by another speaker taking a turn (often with overlap). The periphery of the turn can be simply defined as the start or end of a speaker’s phonation. A turn may be made up of one, or more than one, utterance. Utterances are recognised through predictable intonation patterns, or tone units. As we have seen, discourse markers can be used at LP to bridge the coherence gap between one sentence and the next—and this is also true of utterances and turns. According to Fraser (1999), it is exceptional for discourse markers to appear at the right periphery of the argument structure in ordinary everyday conversational English; however, if discourse markers or other similar items do appear in this position, they tend to have an interpersonal (i.e. intersubjective)

4

Beeching and Detges

function (cf. Traugott 2010), rather than an information-structuring one. They serve to confirm shared assumptions, check or express understanding, request confirmation, express deference or are used for face-saving (cf. Brinton 1996: 37). Cases in point are: tag-questions such as English isn’t it, items such as no, you know. Even though these elements are syntactically integrated in the RP of the sentence, they differ from the RP elements in (1) and (2) in that they are not part of the argument structure introduced by the main verb of the sentence. What right-peripheral elements at sentence-level and discourse-level seem to share is a modal function. Typically, they seem to reflect or invite attitudes towards the message or the situation rather than contributing to the message itself. 3

Terms Used in the Volume

3.1 Dialogual and Dialogic One of the objectives of the volume is to test theories about regularities in semantic change (Traugott and Dasher 2002), specifically the hypothesis that LP is inherently dialogual (two speakers) while RP is dialogic (two viewpoints): items at LP are often considered to have a dialogual (logical) connective role rather than a dialogic (interpersonal/modal) role. The distinction between the terms dialogual and dialogic was first made by Roulet (1984) and Ducrot (1984), and is further elaborated by Schwenter (2000), in Nølke (2006), in Schwenter and Traugott (2000), and in Traugott (2008, 2010). In dialogual contexts, in which speakers take turns to talk, LP would be the expected locus for speakers to take a turn, while RP is where speakers relinquish their turn, seek confirmation and so on. Waltereit (2006a) and Waltereit and Detges (2007) adduce convincing arguments to suggest that certain discourse markers, such as Italian guarda (‘look’) originally used dialogually, as part of turn-taking, are used for attention-seeking or for other dialogic purposes. Dialogic contexts by contrast with dialogual ones are characterised by perspectivisation. Much of day-to-day interaction is taken up in negotiating common ground and, to do this, different viewpoints (perspectives) may be explicitly evoked or implicitly hinted at. There are a number of mechanisms for doing this, for example, by including contrastive or adversative terms, often at RP, such as though and but (in Australian and Scottish varieties), quand même in French, buguo in Chinese or dakedo in Japanese. Schwenter and Traugott (2000) demonstrate the dialogicity of ‘in fact’.

Introduction

5

3.2 Discourse Marker The term ‘discourse marker’ has been used thus far in this introductory chapter to describe items like but and and which are procedural and which, in Schiffrin’s (1987) words, ‘bracket units of talk’ and ensure discourse coherence. Hansen (1998: 24) remarks that the items studied by Schiffrin constitute: a rather heterogeneous group, including coordinating and subordinating conjunctions such as and and because, parenthetical clauses such as you know and I mean, temporal and conjunctive adverbs such as now and so, and (not so easily categorised) particles like oh and well. The expressions which form the focus of most of the chapters of this volume have been given a variety of labels including discourse markers (Schiffrin 1987; Schourup 1999), discourse particles (Schourup 1985; Fischer 2006), modal particles (Abtönungspartikeln, Weydt 1969, 1979, 2001, 2006), punctors (Vincent and Sankoff 1992), connectives (Fraser 1988; Bazzanella 1990; Lamiroy 1994; Unger 1996; Degand 2000), pragmatic particles (Beeching 2002), pragmatic expressions (Erman, 1987), pragmatic markers (Watts 1988; Redeker 1990; Caron-Prague and Caron 1991; Brinton 1996, 1998; Andersen 1998, 2001; Erman 2001), hedges (Hyland 1998a), boosters (Hyland 1998b, 2000; Beeching 2009); fumbles (Edmondson 1981) and conversational greasers (Fillmore, cited in Kerbrat-Orecchioni 1992: 196), not to mention illocutionary force indicating devices (IFIDs) (see Verschueren, Östman and Blommaert 1995) and pragmatic force modifiers (PFMs, Nikula 1996: 43–45 and Lin 2010: 1174). Fraser (1996) used the term ‘pragmatic marker’ to englobe both discourse markers and pragmatic markers, considering ‘discourse markers’ as a subtype of pragmatic markers, and referring particularly to expressions which signal the relationship of the basic message to the discourse which precedes it. Fraser (1999: 931) defines discourse markers as signalling a relationship between the interpretation of the segment they introduce, S2, and the prior segment, S1. They have a core meaning which is procedural, not conceptual, and their more specific interpretation is ‘negotiated’ by the context, both linguistic and conceptual. There are two types: those that relate the explicit interpretation conveyed by S2 with some aspect associated with the segment, S1; and those that relate the topic of S2 to that of S1.

6

Beeching and Detges

He gives, as examples of discourse markers, so, and, furthermore, but, and after all and suggests that, according to the criteria he outlines, Schiffrin’s oh and y’know do not constitute discourse markers. Other, mainly Scandinavian, researchers, such as Erman (1986, 1987, 2001), Andersen (1998, 2000, 2001) and Aijmer and Simon-Vandenbergen (2006), have been particularly interested in the sociolinguistic, interactional and extra­ linguistic facets of these markers, rather than their logical connective qualities and have called them ‘pragmatic markers’, as does Brinton (1996) with her historical overview of ‘medieval mystery words’, such as hwæt and I gesse. Vincent (2005: 189) suggests that From a discursive point of view, markers are distributed around two functional poles: connectors—which ensure the articulation of utterances and, therefore, function like conjunctions and adverbs of liaison—and modal elements—which introduce a point of view on the discourse and take the form of adverbial expressions. Aijmer and Simon-Vandenbergen (2006: 2) opt for the term ‘pragmatic marker’ rather than ‘discourse marker’ and make the following distinction between the two: Discourse marker is the term which we use when we want to describe how a particular marker signals coherence relations. Pragmatic markers as we see them are not only associated with discourse and textual functions but are also signals in the communication situation guiding the addressee’s interpretation. The term as we are using it can also be defined negatively: if a word or a construction in an utterance does not contri­ bute to the propositional, truth-functional content, then we consider it a pragmatic marker. Some of the authors in the present volume (Beeching and Wang; Ghezzi and Molinelli; Yap, Yang and Wong) follow the Scandinavian conceptualization of markers, adopting the term ‘pragmatic marker’ rather than ‘discourse marker’, to highlight the items’ interpersonal rather than textual usages, though recognizing that pragmatic markers do have procedural meanings. Other authors (Degand; Onodera; Sohn and Kim) maintain Schiffrin’s term ‘discourse marker’ to cover a somewhat heterogeneous group of terms and/or terms which are polysemously discourse or pragmatic markers. In her chapter, Traugott sidesteps the issue of distinguishing between discourse markers and pragmatic markers by referring to surely and no doubt, the objects of her study, as

Introduction

7

‘epistemic stance adverbials (EPAs)’. The term ‘marker’ is, furthermore, prefaced by authors in the volume by a range of more precise and explanatory quali­ fiers, such as ‘adversative’, ‘agreement’, ‘evidential’, ‘hedging’, ‘hesitation’, ‘phatic’, ‘reformulation’, and ‘transition’ markers, to name but a few. 3.3 Subjective and Intersubjective Traugott (1982)’s seminal insight into the nature of semantic change, namely the tendency for lexical items to take a unidirectional route from propositional meanings to textual ones and thence to expressive ones, has since been extended to include hypotheses concerning the appearance of textual elements on the left periphery which can then move to the right as expressive items. This route has been confirmed in very many cases, with few counterexamples. Let us take Haselow (2012)’s study of then by way of an example. Schematizing its semantic evolution somewhat, then starts life as a temporal adverb, and takes on sequential meanings via a straightforward extension of the temporal meaning, and deductive ones via pragmatic inferencing from temporal succession to causal meaning. Once then is used at LP inferentially, it can be used at RP, both inferentially and interpersonally, to request confirmation of an inference. These uses are illustrated in 4a.–d. (4) a. It was then that he went to China. (deictic/temporal adverb = ‘at that time’). b. We went to the cinema then to a restaurant. (sequencer = ‘after that’) c. A: We went to the cinema then to a restaurant. B: Then you’ll be tired. (inferential = ‘I infer that, after you’ve done so many things, you will be tired’) d. A: I’ll see you at the beach? B: You’re coming with us then? (inferential + confirmation request = ‘I infer from what you have just said that, contrary to expectation, you are coming to the beach after all and I request confirmation of that’)

As it shifts from temporal to inferential uses, then might be said to become more ‘subjective’ in meaning, as the speaker is drawing a subjective inference from the context. At RP, when requesting confirmation, it could be said to be more ‘intersubjective’. The process by which these form-meaning pairings evolve are described as ‘subjectification’ and ‘intersubjectification’. Following Traugott (2011), Haselow (2012: 169–170) defines subjectification as a semantic process whereby meanings increasingly indicate or encode the speaker or writer’s subjective belief state or attitude toward the propositional content of an utterance

8

Beeching and Detges

whilst intersubjectification refers to a process where speakers come to encode their awareness of the ‘self’ of the interlocutor by means of an externalization of implicatures regarding the social relation, next to their subjective attitudes. One problem with applying the notions of subjectivity and intersubjectivity in analyses of everyday spoken interaction is that conversation is, by definition, both subjective and intersubjective. Speakers express their opinions (‘I’), engage with their interlocutors as a matter of course and draw on linguistic resources to negotiate meaning and to tend to their own and their addressee’s face. The question really is whether subjective and intersubjective usages lead to new coded meanings or whether the functions are contextual side-effects. I think it is fair to say that the notions of ‘subjectivity’ and ‘intersubjectivity’ have been difficult to pin down and apply (though see Davidse, Vandelanotte and Cuyckens (eds.), 2010, for some successful attempts at doing so). 4

Modelling Peripheries

Various models have been developed to map different conceptualisations of ‘periphery’ for different languages and are drawn upon by the authors of the chapters in the volume. Detges and Waltereit draw on Morel’s (2007) and Danon-Boileau et al.’s (1991) framework for French which includes a rhematic element at the core and pre- and post-rhematic elements at the periphery in the preamble and the post-script, respectively. Table 1.1

Detges and Waltereit’s preamble-rheme based model Preamble

Rheme

Binder

Viewpt.

Frame

Frame

Lexical topic

Tu vois

moi

hier

en classe

y avait un mec

‘See

I

yesterday in school there was a guy LP

(Detges and Waltereit, this volume).

qui me fait rire who makes me laugh center

P.S.

quoi like’ RP

9

Introduction

For Detges and Waltereit, the periphery includes everything which does not appear in the core propositional argument structure (the rheme). The model (see Table 1.1) combines syntactic and information-structure criteria with prosodic considerations. It highlights a very large number of different elements which are associated with the prosodic contours of French at the LP (preamble).

Table 1.2

Degand’s clause-based dependency-structure model

Ex. Turn Utterance initial initial

Utterance medial Clause initial

1

on

avait donné rendez-vous à un à un autre endroit

aux aux parents

well

we

had arranged to meet up somewhere else

with the parents

je I

voyais encore was still seeing

donc so

3

et donc

and so

4

we

it

entre euh / enfin tu vois midi et cinq heures had a slot between er I mean you see noon and five est surtout des Hollandais ça qui viennent chercher ’s mainly the Dutch who that come to get

dès qu’ il avait la tune as soon he had as Oui Yes

et and LP

(Degand, this volume).

Cédric Cédric

i/ on avait un trou

c’

5

Turnfinal

Clause final

alors

2

6

Clause medial

Utterance final

qu’ est-ce tu as fait what did you do clause

la tune the money

alors then quoi donc so to speak then alors then RP

toi toi

10

Beeching and Detges

Degand also provides a framework for French, which she illustrates for donc and alors (see Table 1.2). Degand takes the clause as the core propositional element and distinguishes between elements which occur at the left and right periphery of the clause, the utterance and the turn. This model facilitates the inclusion of response-markers turn-initially at LP and distinguishes between elements which are clause-final and utterance- or turn-final, thus permitting greater granularity at RP. One advantage of the preamble-rheme model as opposed to the clause-based model is that it makes it possible to show that the presentative y avait un mec is at the LP of the rheme qui me fait rire. The clause-based model integrates clefting within clause-medial position. The Japanese ‘onion-skin’ model, introduced by Onodera, nests A within B within C within D, (see Table 1.3), thus the core argument structure is made up of acts and events, which are couched within subjective items. These subjective items are, in turn, embedded within intersubjective items. This model thus Table 1.3

Onodera’s layered model (adapted from Shinzato, 2007: 177)1

D (dentatsu ‘communication’ = intersubjectivity) C (handan ‘judgment’ = subjectivity) B ( jitai ‘events’) A (doosa ‘acts’) Nee

Doomo

IP

Somehow

Yukiko ga SBJ

henji wo dasa-

Nakat-ta

yooda

Yo

reply OBJ send-

NEG-PST

Seem

I tell you

=Belief

=Assert

‘It seems that Yukiko didn’t send a reply, (I tell you)’ LP

RP

(Onodera, this volume).

1 Shinzato (2007) uses the abbreviations: IP (interjectional particle); NEG (negative); OBJ (object); PST (past); SBJ (subject). Shinzato’s layered structure model is cited from Minami (1974), Takubo (1987) and Noda (1997).

11

Introduction

proposes not only that there are LP and RP items outside the argument structure but that there is a symmetrical mirror arrangement of functional items at LP and RP. 5

Functional Asymmetry at LP and RP?

The main elements of the LP/RP functional asymmetry hypothesis are displayed in Table 1.4. The hypothesized asymmetry between left and right periphery may be viewed as a strong tendency, not a categorical rule. We find numerous counter-examples in the data; a case in point is frankly, as in frankly, I don’t give a damn, which serves to express stance in the left periphery. However, on closer inspection, it turns out that even for such expressions, fine-grained differences can be detected between left-peripheral and right-peripheral usages, as in I think he’ll be there and He’ll be there, I think. In the first the speaker’s subjective judgement is foregrounded whereas in the second it is backgrounded or added as an afterthought to mitigate the full strength of the assertion. Our hypothesis also seems to be confirmed by certain patterns of language change. Whereas the left periphery seems to be the locus for the grammaticalization of information-structure and argument-structure phenomena such as word-order and agreement, the right periphery is involved in the rise of modal elements such as negative particles out of resumptive negation (Jespersen 1917, see also Dowty 2008), modal particles (Waltereit 2006b) or modalising right dislocation constructions (see, e.g., Morel 2007 for French). The above are hypotheses building on arguably universal properties of human language. We would, therefore, expect them to be realised in a crosslinguistically consistent way. A corollary of this is that language change in the Table 1.4

Hypothesized usages of linguistic items on the left and right periphery

LP

Dialogual Turn-taking/attention-getting Link to previous discourse Response-marking Focalising, topicalising, framing Subjective

RP

Dialogic Turn-yielding/end-marking Anticipation of forthcoming discourse Response-inviting Modalising Intersubjective

12

Beeching and Detges

left and right peripheries should pattern in similar ways cross-linguistically. There has, however, been some evidence to suggest that language typology, particularly whether a language is SVO or SOV, might have a bearing on what appears at LP and RP, and on language change. Testing the cross-linguistic universality of LP/RP functional asymmetry is one of the main aims of our project and motivates the inclusion of chapters which introduce data from Chinese (both Mandarin and Cantonese), Korean and Japanese, side by side with the more commonly studied Indo-European languages, English, French and Italian. Of particular interest is whether verb-final languages, such as Korean and Japanese, have a richer and larger range of elements at RP. Historically, it appears that LP expressions in Japanese derive from RP expressions in a process of grammaticalisation (see Onodera, this volume), whereas most evidence from Indo-European languages suggests that the opposite occurs: medial particles shift to LP and only later to RP. By contrast with Japanese and Korean, matrix clauses in Modern Chinese are canonically SVO, while some of its subordinate clauses are head-final (e.g. relative clauses have VP + head noun structures). This leads to the possibility of a cluster of RP elements after the head noun. 6

Research Questions

The following questions served as general guidelines for the individual papers. Synchronic 1.

Are there items that can appear in the left periphery (LP), but not in the right periphery (RP)? Are there items that can appear in the RP, but not in the LP? Are there items that can appear in both? Are there items which are excluded from both? 2. Are there (grammatical or communicative) functions specific to the LP, i.e., not found in the RP? Are there functions specific to RP, i.e., not found in the LP? Are there functions found in both LP and RP? Are there functions excluded from both the LP and the RP? 3. Interactions of levels: How do the peripheries of the units of the different levels of speech and language interact with each other? More specifically: how do the LPs of the turn (conversational unit), the discourse unit and the sentence interact with each other? How do the RPs of the aforementioned levels interact with one another?

Introduction

4.

13

Subjectivity and intersubjectivity: How do LP and RP behave with respect to subjective and intersubjective functions? Are there preferred loci for each?

Diachronic 1. 2. 3. 4. 7

How do constructions of the LP/RP, respectively, arise diachronically? What are typical sources and targets for LP/RP constructions? What motivates the development of LP/RP constructions? What are the respective roles of LP and RP for the rise of dialogic items? Summaries of the Chapters in the Volume

Ulrich Detges and Richard Waltereit’s chapter on Moi je ne sais pas vs. Je ne sais pas moi examines the use of first-person strong pronouns in spoken French. The analysis shows that both in the left and the right periphery, strong first-person pronouns can in principle serve modal (subjective and intersubjective) as well as coherence-related functions. At first glance, this seems to refute a “strong” version of the working hypothesis outlined above. However, the authors show that modal and coherence-related functions are unevenly distributed between left and right periphery. In the left periphery, coherencerelated functions are statistically predominant, while being extremely rare in the right periphery. Conversely, modal (i.e. subjective and intersubjective) uses do occur in the left periphery, especially within parenthetical constructions of the type moi je trouve, ‘I think’ or moi je ne sais pas ‘I don’t know’, but are much more frequent in the right periphery, where they constitute more than 90% of uses of strong pronouns. As the authors argue, this striking asymmetry is a reflex of the fundamentally different functions of left and right periphery as such. Whereas the left periphery serves to anchor the upcoming utterance (cognitively, discourse-structurally and epistemically), the right periphery is a locus where aspects of the completed utterance can be re-negotiated. Kate Beeching and Yu-Fang Wang’s chapter looks at English well, French bon and Chinese hao. These expressions are canonically terms of positive evaluation. Furthermore, they can function as acceptance markers, and have developed hedging meanings. The cross-linguistic investigation presented here aims to gauge the extent to which left and right periphery have similar subjective and intersubjective roles in different (typologically divergent) languages and to what extent LP might be the locus of subjective and RP the locus of

14

Beeching and Detges

intersubjective functions. The literature on the history of well is outlined and the chapter breaks new ground by tracing the historical development of bon and hao. Synchronically, well is found exclusively at the LP, while bon can be used as an end-marker and to mark the stages in a narrative. Hao has developed from a verb form (‘to like’) via a predicative to an attributive adjective to its current LP usage as an acceptance marker and hedge. It may also be used in the compound form hao le at the RP as a backchannel device to elicit confirmation (see also Yap et al.’s chapter). The chapter argues that acceptance terms such as these are often used dialogually in second-turn LP positions—Speaker B provisionally pretends to accept what Speaker A has just said, in such a way as to mitigate an upcoming disagreement (‘OK, but . . .’). The demurral sense then becomes progressively established as a core meaning (as it has done with well). Well and bon can also be used monologually to introduce a concession (followed by mais, but), thus exploiting the ‘provisional’ dialogual acceptance usage rhetorically to introduce a rider to a previous comment. The chapter concludes with a cross-linguistic semantic map which charts the degree of overlap between the various functions of the terms. It seems that, in all three languages, it is the LP response-marking function (intersubjectivity) which leads to the possibility of a hedging usage of the term. Elizabeth Traugott’s chapter on surely and no doubt tests hypotheses concerning the differential impact of the LP and RP on rhetorical functions over time (data are presented from texts dating from 1500–1920). It is argued that surely has developed from a relatively objective interpretation of ‘certainly, in everyone’s view’ to a subjective epistemic evaluation ‘certainly, in my opinion’ and then to an intersubjective plea for confirmation, as in ‘surely, you tried to save it?’. No doubt by contrast emerged as an adverb with subjective epistemic meaning from c1400. It is generally found in medial positions but also at LP and is rare at RP. While no doubt is inherently inward-oriented and suggests the speaker’s/author’s private conclusion, surely is outward-oriented, subjective at both LP and RP and can be used intersubjectively as a plea for confirmation in both positions, especially with second person pronouns. Traugott concludes that where surely and no doubt are concerned, subjectivity or intersubjectivity appear to be correlated more closely with their inherent meaning and immediate context than with use at LP or RP. The extent to which this is unique to English and relates to the late development of the RP as a slot where interactional markers occur remains to be seen. Traugott suggests that a distinction needs to be made between turn-taking as an interactional practice and explicit marking of turn-taking. Noriko Onodera’s chapter looks specifically at the LP as a locus in which a “mental space” is set up, providing a prism which orients understanding of

Introduction

15

the upcoming part of the utterance. She also contends that there is not an automatic association between position and function in present-day Japanese, at least where subjectivity and intersubjectivity are concerned. The arguments concerning the emergence of left-peripheral intersubjective DM usages of concessives like German obwohl, Japanese dakedo and English though suggest that (barring morphological constraints) DMs which develop (inter)subjective functions on the LP can then (and only then) move to the RP. The chapter by Chiara Ghezzi and Piera Molinelli deals with verbbased pragmatic markers in Italian, mainly guarda ‘look’ (< guardare ‘to look, to watch’), prego ‘1. please, 2. you are welcome’ < (pregare ‘to request, to adjure, to pray’) and dai ‘come on!’ (< dare ‘to give’). These markers represent three different functional classes, namely a) “attention getters” (represented by guarda ‘look’), b) courtesy markers (instantiated by prego ‘please’ and ‘you are welcome’) and c) phatic or expressive markers (represented by dai ‘come on’). The article examines the interaction of the pragmatic functions of the three and the position they occupy in different discourse units. In an extremely intricate analysis of the aforementioned markers and their respective micro-functions, it is shown that their property to appear in the leftmost or rightmost peripheral position of a turn, utterance or phrase depends on their function. In particular, the authors argue that appearing in left- versus right-peripheral position can be meaningfully captured by an analysis along two dimensions, namely a) the distinction between dialogic and dialogual function and b) the distinction between utterance and enunciation. The question of whether the left and right peripheries attract specific meanings is discussed in Liesbeth Degand’s chapter on the basis of an analysis of alors ‘then, at that time, so’ and donc ‘so’ in modern spoken French. Specifically, this chapter examines whether both markers express different meanings when realized either in the left or in the right periphery. On the basis of an analysis of 50 occurrences of alors and of donc at left and right periphery of the utterance, it is argued that left periphery attracts subjective meanings while the right periphery mainly serves to realize intersubjective ones. For alors, this conclusion is supported more strongly than for donc. However, the fact that alors and donc share a number of functions that are specialized for LP (subjective argumentative and metadiscursive relations), on the one hand, and for RP (intersubjective argumentative and metadiscursive relations), on the other, suggests that such tendencies are also compatible with other discourse markers, and perhaps even for other kinds of linguistic expressions occurring in these positions. Foong Ha Yap, Ying Yang and Tak-Sum Wong observe that pragmatic markers are more often positioned on the right periphery in Chinese than they

16

Beeching and Detges

are in English. The chapter examines how RP markers emerge, and why they are favoured, in the Chinese language. Six RP strategies are posited, which capture the ways in which these particles tend to host (and in the process express) a wide range of speaker moods and stances, both subjective and intersubjective. In LP position, pragmatic markers often serve a discourse-organizing function of relating the current utterance to prior ones (e.g. Haole, ni qu bang ta ‘Okay, you go help him’). The findings further reveal, however, that some pragmatic markers in LP position can also convey a speaker’s subjective and intersubjective stance. In the case of haole, for example, this can range from expressions of approval to annoyance. In the case of pragmatic markers that originate from psych verbs, the subjective and intersubjective nuances often reflect differences in epistemic, attitudinal or evidential calibrations that can be traced to their etymological semantics (e.g. Kongpa ta bu hui lai ‘Probably (< ‘(I) fear’) he won’t come’ vs. Wo juede ta bu hui lai ‘Perhaps (< ‘I guess’) he won’t come’ vs. Tingshuo ta bu hui lai ‘Apparently (< ‘hearsay’) he won’t come.’). In this respect, the findings indicate that differences between LP and RP pragmatic markers are in some cases a matter of degree rather than kind. Finally, Sung-Ock S. Sohn and Stephanie Hyeri Kim’s chapter on the role of LP vs RP expressions in Korean focuses on the importance of prosodic factors in the emergence of final particles at RP. Previous studies of right peri­ pheral expressions in Korean have mainly focused on the development of clauseconnectives, which are morphologically attached to the predicate stem of the first clause in medial position, into utterance-final particles (e.g. Park 1999’s study of clause-connective nuntey ‘but’ in turn-final position in present-day Korean, where it has a contrastive or dispreferred-response-marking function). More recently, Kim and Sohn (2011) have highlighted the use of kuntey ‘but’, ‘by the way’ at RP. While kuntey is used for dispreferred responses and topic resumption at LP, at RP it is primarily used as a delay strategy to mitigate the speaker’s “disaffiliative stance” in socially sensitive contexts (cf. Heritage 1984). As final kuntey is downplayed prosodically, Kim and Sohn (2011) argue that it is on a grammaticalization pathway to an utterance-final particle. The present chapter draws on a fine-grained pitch track analysis of final kuntey, which demonstrates that the final particle forms a single prosodic unit with the preceding predicate. The study indicates that the RP represents the locus where speakers search for mutual understanding and negotiate for the subsequent turns. In contrast, the left periphery occupied by turn-initial discourse markers is used by speakers to signal actions such as dispreferred responses and topic-shift. The findings of the various chapters are synthesized in Table 1.5.

17

Introduction Table 1.5

Summary of the main findings Language(s)

Feature investigated

Synchronic or diachronic

LP/RP asymmetry?

Detges and Waltereit

French

tonic pronoun – moi ‘me’

synchronic

Beeching and Wang

English, French and Chinese

well, bon, hao ‘good’

diachronic

Traugott

English

surely, no doubt

diachronic

Onodera

Japanese

diachronic/ dakedo ‘though’ and synchronic other d-connectives

Ghezzi and Molinelli

Italian

guarda ‘look’, synchronic prego ‘please, you’re welcome’, dai ‘come on!’

Yes. Moi can appear on both LP and RP: functions are not periphery-exclusive, but 90% of RP items are modal. Yes. The LP responsemarking function (intersubjectivity) leads to the possibility of a hedging usage of the terms, crosslinguistically, with varying degrees of pragmaticalisation. No. (Inter)subjectivity related to core meaning to a greater extent than to position on LP/RP. No. Discourse coherence markers occur at LP but subjective and intersubjective functions can occur at both LP and RP. Yes. LP or RP determines function of the individual markers.

18

Beeching and Detges

table 1.5 (cont.) Language(s)

Feature investigated

Synchronic or diachronic

Degand

French

alors ‘then, at synchronic that time, so’, donc ‘so’

Yap, Yang and Wong

Chinese

Sohn

Korean

kongpa ‘I fear, diachronic probably’, wo juede ‘I guess’, tingshuo ‘hearsay, apparently’ kuntey ‘but, synchronic by the way’

LP/RP asymmetry?

Yes. LP favours subjective, RP favours intersubjective functions. Yes, but LP and RP interact with core meanings of items.

Yes, LP kuntey introduces dispreferred responses, topic shift and resumption while an incoming (and prosodically downplayed) RP usage mitigates a ‘disaffiliative stance’.

8 Conclusions Overall, the assembled chapters have gone some way to answering the complex questions posed by the editors of the volume. The picture that emerges from the different descriptions is, however, far from univocal. One of the reasons for this is the heterogeneous nature of the items analysed by the individual authors and their divergent theoretical approaches. However, in their entirety, the individual studies clearly indicate that the expressions and functions analysed here are not normally constrained to either left or right. Thus, the hypothesis that the left periphery is mainly concerned with discoursestructuring and the right periphery with modalising (stance, subjective and intersubjective) qualities cannot be upheld in a “strong” and exclusive way. On the other hand, many of the chapters—especially those based on synchronic

Introduction

19

corpus data—argue that some kind of asymmetry between left and right periphery does exist; in most cases it is shown that this asymmetry is a matter of frequency (and hence of degree) rather than being categorical and that the LP/RP position interacts with both the core meanings of the items and prosody. This result shows that exploring the working hypothesis outlined above was (and will continue to be) a useful enterprise. References Aijmer, Karin. 1989. Themes and tails: The discourse functions of dislocated elements. Nordic Journal of Linguistics 12: 137–53. Aijmer, Karen and Anne-Marie Simon-Vandenbergen (eds.). 2006. Pragmatic markers in contrast. Studies in Pragmatics 2. Amsterdam: Elsevier. Andersen, Gisle. 1998. “The pragmatic marker like from a relevance-theoretic perspective”. In A.H. Jucker and Y. Ziv (eds.), Discourse Markers: Description and Theory. Amsterdam: John Benjamins, 147–170. ———. 2000. “The role of the pragmatic marker like in utterance interpretation”. In G. Andersen and T. Fretheim (eds.), Pragmatic markers and Propositional Attitude. Amsterdam: John Benjamins, 17–38. ———. 2001. Pragmatic Markers and Sociolinguistic Variation: A relevance-theoretic approach to the language of adolescents. Pragmatics and Beyond series 84. Amsterdam & Philadelphia: Benjamins. Bazzanella, Carla. 1990. Phatic connectives as interactional cues in contemporary spoken Italian. Journal of Pragmatics 14: 629–47. Beeching, Kate. 2002. Gender, Politeness and Pragmatic Particles in French. Amsterdam: John Benjamins. ———. 2009. “Procatalepsis and the etymology of hedging/boosting particles”. In M.-B.M. Hansen and J. Visconti (eds.), Current Trends in Diachronic Semantics and Pragmatics. Bingley: Emerald, 81–106. Brinton, Laurel J. 1996. Pragmatic Markers in English. Grammaticalization and Discourse Functions. Topics in English Linguistics 19. Berlin & New York: Mouton de Gruyter. Brinton, Laurel. 1998. “‘The flowers are lovely; only they have no scent’. The evolution of a pragmatic marker in English”. In R. Borgmeister, H. Grabes, and A.H. Jucker (eds.), Anglistentag Giessen, Proceedings. Trier: Wissenschaftlicher Verlag, 9–33. Caron-Prague, Josiane & Jean Caron. 1991. “Psychopragmatics vs. sociopragmatics. The function of pragmatic markers in thinking-aloud protocols”. In J. Verschueren (ed.), Pragmatics at Issue: Selected Papers of the International Pragmatics Conference, Antwerp, August 17–22, 1987. Amsterdam: Benjamins, 29–36.

20

Beeching and Detges

Cechetto, C. 1999. A Comparative Analysis of Left and Right Dislocation in Romance. Studia Linguistica 53/1: 40–67. Cinque, Giulielmo & Luigi Rizzi. 2008. The cartography of syntactic structures. Studies in Linguistics 2: 42–58. Croft, William. 2000. Explaining language change. An evolutionary approach. Harlow: Longman. Danon-Boileau, Laurent, Annie Meunier, Mary-Annick Morel & Nicolas Tournandre. 1991. Intégration discursive et intégration syntaxique. Langages 104: 111–128. Davidse, Kristin, L. Vandelanotte and H. Cuyckens (eds.). 2010. Subjectification, Intersubjectification and Grammaticalization. Berlin: Mouton de Gruyter. De Cat, Cécile. 2007a. French dislocation: Interpretation, syntax, acquisition. Oxford: Oxford University Press. ———. 2007b. French dislocation without movement. Natural Language and Linguistic Theory 25: 485–534. Degand, Liesbeth. 2000. Causal connectives or causal prepositions? Discursive constraints. Journal of Pragmatics 32: 687–707. Dowty, David. 2008. “Resumptive negation as assertion revision” [http://www.ling .ohio-state.edu/~dowty/resumptive-negation.pdf, 10.10.2012]. Ducrot, Oswald. 1984. Le dire et le dit. Paris: Minuit. Durham, Mercedes. 2011. Right dislocation in Northern England: Frequency and use— perception meets reality. English World-Wide 32/3: 257–279. Edmondson, Willis. 1981. Spoken Discourse: A Model for Analysis. London: Longman. Erman, Britt. 1986. “Some pragmatic expressions in English conversation”. In G. Tottie & I. Backlund (eds.), English in Speech and Writing. A Symposium. Stockholm: Almqvist & Wiksell, 131–147. ———. 1987. Pragmatic Expressions in English: A study of you know, you see and I mean in Face-to-Face Communication. Stockholm Studies in English 69. Stockholm: Almqvist and Wiksell. ———. 2001. Pragmatic markers revisited with a focus on you know in adult and adolescent talk. Journal of Pragmatics 33: 1337–1359. Fischer, Kerstin (ed.). 2006 Approaches to Discourse Particles. Studies in Pragmatics 1. Oxford: Elsevier. Frascarelli, Mara. 2004. “Dislocation, Clitic Resumption and Minimality: A Comparative Analysis of Left and Right Topic Constructions in Italian”. In R. Bok-Bennema, B. Hollebrandse, B. Kampers-Manhe & P. Sleeman (eds.), Romance Languages and Linguistic Theory 2002. Amsterdam & Philadelphia: John Benjamins, 99–118. Fraser, Bruce. 1988. Types of English discourse markers. Acta Linguistica Hungarica 38: 19–33. ———. 1996. Pragmatic Markers. Pragmatics 6/2: 167–190. ———. 1999. What are discourse markers? Journal of Pragmatics 31: 931–952.

Introduction

21

Hansen, Maj-Britt Mosegaard. 1998. The function of discourse particles. Pragmatics and Beyond New Series 53. Amsterdam: Benjamins. Haselow, Alexander. 2012. “Discourse organization and the rise of final then in the history of English.” In I. Hegedűs & A. Fodor (eds.), English Historical Linguistics 2010. Selected Papers from the Sixteenth International Conference of English Historical Linguistics (ICEHL 16). Pécs, 23–17 August 2010. Amsterdam & Philadelphia: Benjamins, 153–175. Heritage, John. 1984. “A change-of-state token and aspects of its sequential placement”. In J.M. Atkinson & J. Heritage (eds.), Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press, 299–345. Hyland, Ken. 1998a. Hedging in Scientific Research Articles. Amsterdam: Benjamins. ———. 1998b. Boosting, hedging and the negotiation of academic knowledge. Text 18/3: 349–382. ———. 2000. Hedges, boosters and lexical invisibility. Language Awareness 9: 179–197. Jespersen, Otto. 1917. Negation in English and other Languages. Kopenhagen: Munksgaard. Kerbrat-Orecchioni, Cathérine. 1992. Les Interactions Verbales. Vol. 2. Paris: Armand Colin. Kim, Stephanie Hyeri & Sohn, Sung-Ock. (Under review). Grammar as an emergent response to interactional needs: a study of final kuntey ‘but’ in Korean conversation. Koch, Peter & Wulf Oesterreicher. 2011. Gesprochene Sprache in der Romania. Französisch, Italienisch, Spanisch. Romanistische Arbeitshefte 31. Berlin & New York: De Gruyter. Lambrecht, Knud. 1981. Topic, antitopic and verb agreement in non-standard French. Amsterdam: John Benjamins. ———. 1994. Information Structure and Sentence Form: Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: University Press. ———. 2001. “Dislocation”. In M. Haspelmath, E. König, W. Oesterreicher & W. Raible (eds.), Language Typology and Language Universals: An International Handbook. Handbücher zur Sprach- und Kommunikationswissenschaft 20. Vol. 2. Berlin & New York: De Gruyter, 1050–1078. Lamiroy, Béatrice. 1994. Pragmatic connectives and L2 acquisition: The case of French and Dutch. Pragmatics 4: 183–201. Lin, Chia-Yen. 2010. ‘. . . that’s actually sort of you know trying to get consultants in . . .’: Functions and multifunctionality of modifiers in academic lectures. Journal of Pragmatics 42: 1173–1183. Minami, Fujio. 1974. Gendai Nihongo no Koozoo. Tokyo: Taishukan Shoten. Morel, Mary-Annick. 2007. Le postrhème dans le dialogue oral en français. L’information grammaticale 113: 40–46.

22

Beeching and Detges

Nikula, Tarja. 1996. Pragmatic Force Modifiers: A Study in Interlanguage Pragmatics. University of Jyvaskyla, Jyvaskyla, Finland. Nølke, Henning. 2006. The semantics of polyphony (and the pragmatics of realization). Acta Linguistica Hafniensia 38: 137–160. Noda, Harumi. 1997. No (da) no Kinoo. Tokyo: Kurosio Shuppan. Park, Y.-Y. 1999. The Korean connective nuntey in conversational discourse. Journal of Pragmatics 31/2: 191–218. Redeker, Gisela. 1990. Ideational and pragmatic markers of discourse structure. Journal of Pragmatics 14: 367–381. Roulet, Eddy. 1984. Speech acts, discourse structure, and pragmatic connectives. Journal of Pragmatics 8: 31–47. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schourup, Laurence. 1985. Common Discourse Particles in English Conversation. New York: Garland. ———. 1999. Discourse markers. Tutorial overview. Lingua 107: 227–265. Schwenter, Scott A. 2000. “Viewpoints and polysemy: Linking adversative and causal meanings of discourse markers”. In E. Couper-Kuhlen & B. Kortmann (eds.), Cause—Condition—Concession—Contrast: Cognitive and Discourse Perspectives. Topics in English Linguistics 33. Berlin & New York: Mouton de Gruyter, 257–281. Schwenter, Scott A. & Elizabeth Closs Traugott. 2000. Invoking scalarity: The development of in fact. Journal of Historical Pragmatics 1: 7–25. Shinzato, Rumiko. 2007. “(Inter)subjectification, Japanese syntax and syntactic scope increase”. In N.O. Onodera and R. Suzuki. (eds.), Historical Changes in Japanese: Subjectivity and Intersubjectivity, Special Issue of Journal of Historical Pragmatics 8/2: 171–206. Traugott, Elizabeth Closs. 1982. “From propositional to textual and expressive meanings: Some semantic pragmatic aspects of grammaticalization”. In W.P. Lehmann & Y. Malkiel (eds.), Perspectives on Historical Linguistics. Current Issues in Linguistic Theory 24. Amsterdam & Philadelphia: John Benjamins, 245–271. ———. 2008. “‘All that he endeavoured to prove was . . .’: On the emergence of grammatical constructions in dialogual and dialogic contexts”. In R. Cooper & R. Kempson (eds.), Language in Flux: Dialogue Coordination, Language Variation, Change and Evolution, London: Kings College Publications, 143–177. ———. 2010. “Dialogic contexts as motivations for syntactic change”, in W. Kretschmar, A.-M. Hamilton-Brehm & R.A. Cloutier (eds.), Variation and Change in English Grammar and Lexicon. Berlin: Mouton de Gruyter, 11–27. ———. 2011. “‘He Withdrew, Disconcerted and Offended, No Doubt; but Surely It Was Not My Fault’. On the Function of Adverbs of Certainty at the Left and Right Periph­eries of the Clause”. Paper presented at the 12th International Pragmatics Conference, Manchester, 4 July 2011.

Introduction

23

———. 2013. “I must wait on myself, must I? On the rise of pragmatic markers at right periphery of the clause in English”. Talk given at Lund University, 4 Sept., 2013. Traugott, Elizabeth Closs & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge Studies in Linguistics 96. Cambridge: Cambridge University Press. Unger, Christopher. 1996. The scope of discourse connectives: implications for discourse organization. Journal of Linguistics 32: 402–438. Verschueren, Jef, Jan-Ola Östman & Jan Blommaert. 1995. The Handbook of Pragmatics. Amsterdam/Philadelphia: John Benjamins. Vincent, Diane. 2005. The journey of non-standard discourse markers in Quebec French. Networks based on exemplification. Journal of Historical Pragmatics 6/1: 188–210. Vincent, Diane & David Sankoff. 1992. “Punctors: A pragmatic variable”. Language Variation and Change, 4: 205–216. Waltereit, Richard. 2006a. The rise of discourse markers in Italian: A specific type of language change. In K. Fischer (ed.), 61–76. ———. 2006b. Abtönung. Zur Pragmatik und historischen Semantik von Modalpartikeln und ihren funktionalen Äquivalenten in romanischen Sprachen, Tübingen: Niemeyer. Beihefte zur Zeitschrift für Romanische Philologie 338. Waltereit, Richard & Ulrich Detges. 2007. “Different functions, different histories. Modal particles and discourse markers from a diachronic point of view”. In M.J. Cuenca (ed.), Contrastive Perspectives on Discourse Markers, Special issue of Journal of Catalan Linguistics, 61–80. Watts, Richard J. 1988. A relevance-theoretic approach to commentary pragmatic markers: the case of actually, really and basically. Acta Linguistica Hungarica 38:235–260. Weydt, Harald. 1969. Abtönungspartikel: Die deutschen Modalwörter und ihre französischen Entsprechungen. Bad Homburg: Gehlen. ———. (ed.). 1979. Die Partikeln der deutschen Sprache. Berlin: Mouton de Gruyter. ———. 2001. “Partikelforschung”. In G. Holtus, M. Metzeltin & C. Schmitt (eds.), Lexikon der Romanistischen Linguistik. Tübingen: Niemeyer. Vol. I/1, 782–801. ———. 2006. “What are particles good for? In K. Fischer. (ed.), 205–218.

chapter 2

Moi je ne sais pas vs. Je ne sais pas moi: French Disjoint Pronouns in the Left vs. Right Periphery Ulrich Detges and Richard Waltereit 1

Centre and Periphery

In this paper, we will discuss certain asymmetries in the behaviour of French strong pronouns when used either in the left or the right periphery of the sentence. Our analysis draws on a model of the French sentence proposed by Morel (2007) and Danon-Boileau et al. (1991). According to these authors, the core clause—the rheme—is an obligatory element which may optionally be preceded by the so-called preamble (see (1)) and followed by the post-script (originally termed post-rhème by Morel 2007 and Danon-Boileau et al. 1991). The preamble, in turn, may consist of smaller optional subunits. These include, among other things, “binding” elements with phatic function (tu vois ‘you see’ in (1)), expressions of “viewpoint” (moi ‘I’), frame elements (hier ‘yesterday’ and en classe ‘in school’) locating the rheme in time and space, and finally lexical topic expressions, i.e. expressions referring to the element which the respective rheme will be about. Spoken French has phrasal accent. Therefore, the different components of the phrase model in (1) all correspond to intonational units with conventional prosodic contours. Among these, the post-script has a characteristic prosodic contour of its own (see below, section 5.2.). By contrast, the various elements of the preamble are marked individually by a topic intonation contour (see Mertens 2011). Thus, the model sketched in (1) combines information-structural, syntactic, and prosodic criteria. (1)

Preamble

Rheme

Binder

Viewpt. Frame

Frame

Lexical topic

Tu vois

moi

hier

en classe

y avait un mec

‘See

I

yesterday in school there was a guy

qui me fait rire who makes me laugh

© koninklijke brill nv, leiden, ���4 | doi ��.��63/9789004274822_�03

P.S.

quoi you know’

French Disjoint Pronouns in the Left vs. Right Periphery

25

For our purposes, the model (1) offers two advantages. First, it allows us to define the rather vague notions of centre and periphery in more precise terms. The centre of the French sentence is the (obligatory and tightly structured) rheme. By contrast, preamble and post-script are optional and grammatically less structured (see Detges 2013). They can be identified as left and right periphery, respectively. Yet another advantage of this model is that it assigns a precise place to the strong pronouns of French. As can be seen from examples (2.a–c), where the respective strong pronoun is marked by the label S, their realization is mainly confined to the preamble and the P.S., i.e. to the left and right periphery. The only exceptions to this rule are cases where a strong pronoun is governed by a preposition—in this case it may be realized within the rheme (see 3.a, b). By virtue of these positional restrictions the strong pronouns are in complementary distribution with the weak (or clitic) pronouns of French (labelled as W in (2) and (3)), which, in turn, never appear outside the rheme and are never governed by prepositions. (2) Preamble Rheme a. ToiS, tuW m’as fait rire b. *TuW, tuW m’as fait rire c. ToiS, *toiS m’as fait rire ‘You you made me laugh’ (3) Preamble Rheme a. ToiS, jeW suis content de toiS b. ToiS, *jeW suis content de tuW ‘(You) I am pleased with you’ (4) Rheme a. TuW m’as fait rire b. *TuW m’as fait rire ‘You made me laugh

P.S. toiS tuW you’

These clear-cut rules apply only, though, to first- and second-person strong pronouns. By contrast, strong third-person pronouns may appear within the rheme (lui m’a fait rire ‘he made me laugh’). Moreover, for the remaining forms of the paradigm, homonymy between strong and weak pronouns makes it impossible to distinguish on independent grounds whether a given token is an instance of the strong or the weak series. However, for reasons which will become clearer in the following sections, we will concentrate on firstperson strong pronouns, where the distinction is unproblematic. Our paper is based on an analysis of the ELICOP version of the Corpus Orléans, a corpus of

26

Detges and Waltereit

spontaneous French conversations recorded in the 1960s, now freely accessible on the web. While more recent corpora of French are available, we chose this one because of its excellent search facilities. 2

Left vs. Right Periphery as Objects of Inquiry

The left periphery of the sentence was extensively explored in functionalist linguistics in the 1970s and 1980s and in generative linguistics in the 1990s and 2000s. Key construction-types included left dislocation, right dislocation, topicalization, and “hanging topic” (cf. Lambrecht 1981, Carroll 1982, Barnes 1985, to name just a few). Early research looked at the peripheries mostly from the point of view of syntactic representation and pragmatic function. The aim was an appropriate characterization of how those constructions differed from what was perceived as the canonical, unmarked, sentence. In addition to this, the matter was framed as a question of differences between spontaneous spoken discourse versus planned written style, often with the assumption that those constructions reflect the constraints of real-time speech production rather than being conventional grammatical objects in their own right. More recently, increased availability of acoustical analyses has led to a shift to the prosodic characteristics of those constructions (for French, see e.g. Doetjes et al. 2002). Furthermore, they now tend to be seen as conventional constructions of the language, not only as a last-resort option reflecting real-time formulation constraints. A key finding supporting the latter move was that the peripheries are not typically separated from the core by a pause, contrary to what was often previously assumed. Left and right periphery have enjoyed some attention as objects of inquiry in their own right. We are going to briefly review some of this work. From a minimalist perspective, De Cat (2007a) stresses that there are no genuine syntactic differences between left and right dislocation and that any difference found between the two constructions is to be located in general characteristics of the peripheries, not the syntax. A key feature of the left periphery identified as responsible for the characteristics of the left periphery (as opposed to the right one) is “prosodic salience” (2007a: 161–164). In his seminal study on French left and right dislocation, Lambrecht (1981) identified morpho-syntactic as well as pragmatic properties of left and right dislocation, called “topic” and “antitopic” respectively. These properties (and hence the asymmetry found between “topic” and “antitopic”) were motivated on the basis of information structure.

French Disjoint Pronouns in the Left vs. Right Periphery

27

Further to the observed differences between left and right periphery, Bossong (1981) and Nølke (1998) have claimed that right dislocation carries some emotive value. This resonates with Morel’s (2007: 42) claim that the post-rheme arises from an “unexpected judgment”. Thus, according to these authors, the right periphery is somehow associated with stance. The picture emerging from this overview is that the left periphery is associated with information structure functions and coherence-building, whereas the right periphery is reserved for subjective and interpersonal or modal functions. However, we do find those interpersonal/modal functions in the left periphery, too—the “binder” tu vois and “viewpoint” moi in example (1) clearly indicate stance. Moreover, the “binder’s” function is interpersonal in that tu vois expresses an expectation of convergence of speaker’s and hearer’s views. Thus, we need a larger model of the functions of the left vs. the right periphery that accommodates those apparent contradictions. We will attempt this in this paper, taking the French strong pronoun moi as a test case. 3

French Strong Pronouns

Usage of strong pronouns is characterized by striking frequency asymmetries. These do not only concern person, e.g. the distribution of first- and third-person pronouns, but—perhaps less expectedly—differences between the left and the right peripheries (as defined in section 1). The figures presented in table 2.1 are from the ELICOP corpus. Table 2.1

Frequency asymmetries in the distribution of strong pronouns

1s moi 3s lui

Left

Right

240 15

41 1

As in many other languages, the (strong) pronouns of French, especially the first-person singular forms, tend to occur as subjects of stance expressions as in (5). Normally, these are built around a “cognitive” verb (Fetzer & Johansson 2010), i.e. a verb expressing a belief or an attitude on behalf of the subject (normally the speaker him/herself). Due to their high frequency, these expressions have a more or less frozen form. This does not exclude, however, their

28

Detges and Waltereit

usage with or without strong pronouns (see below, section 6.1., especially table 2.8). In the literature, they are usually referred to as parentheticals. This notion reflects their syntactic status as phrases which are inserted in sentences from which they are structurally independent (Fetzer & Johansson 2010: 241). In the remainder of this article, we will refer to them as parenthetical expressions or simply as parentheticals, as opposed to free, non-formulaic uses of strong pronouns. (5) Parenthetical expression Moi je pense que ça doit être ça non? (ELICOP, Moi 113) Me I think that it must be that no ‘(I) I think that this must be it, no?’

Parentheticals are extremely frequent in conversation. In fact, they are among the quantitatively most important contexts for strong pronouns in French (for first person moi, see table 2.2). In principle, the pronoun can appear in the left periphery of the expression (moi, je pense) as well as to its right ( je pense, moi). As follows from table 2.2, the tendency of first person moi to appear in the right periphery is greater in parentheticals than in free use. The reason for this is the high frequency of one particular expression, namely je ne sais pas, moi (see section 6.2.). Table 2.2 First person moi in the left vs. right periphery of free vs. parenthetical expressions

Free uses Parentheticals

Left

Right

160 80

21 20

In the remainder of this article, we shall discuss four cases separately, namely: a) free left-peripheral uses of strong pronouns (section 4), b) free right-peripheral uses (section 5), c) left-peripheral uses in parentheticals (section 6.1.) and d) right-peripheral uses in parentheticals (sections 6.2. and 6.3.). It will become clear that in free uses as well as in parenthetical expressions, the different effects attached to right vs. left peripheral uses are basically of the same kind.

French Disjoint Pronouns in the Left vs. Right Periphery

4

29

Free Uses in the Left Periphery

In this section, we will discuss various functions of strong pronouns in the left periphery. As we shall see, all of these functions are relatively frequent. Given that they concern different levels of language (grammar as well as discourse), there is much overlap between them. 4.1 Referential Contrast The most basic function of strong pronouns is the expression of a referential contrast of some kind. Following Mayol (2010), we distinguish different types of contrasts. Thus, in (6), moi ‘I’ contrasts the person of the speaker with one particular, clearly identified entity. In (6), this entity is explicitly indicated by lui ‘he’. In other cases it may be inferable from the context. In all of these cases, we will speak of an exhaustive, definite or “strong” contrast. Indicating a contrast between the referent in question and other ones clearly serves the purpose of propositional coherence: (6) Exhaustive (“strong”) contrast Moi je en gagne des fois 90,000, lui il en gagne 400,000. (ELICOP, Moi 385) Me  I of.it earn indef times 90,000 he he of.it earns 400,000 ‘I sometimes earn 90,000, he earns 400,000.’

In other cases, the strong pronoun indicates a contrast of the respective referent with unidentified entities. In such cases, we speak of an uncertain, indefinite or “weak” contrast (see 7). This type of contrast is pragmatically interesting in that it lends itself to certain rhetorical manipulations. (7)

Uncertain (“weak”) contrast Enfin,  moi je me prends pas pour quelqu’un de supérieur (ELICOP, Moi 279) Finally me I me take neg for someone of   superior ‘At least, I don’t think of myself as someone superior.’ INFERENCE: ‘But there are people who do.’

The potential of indicating a referential contrast of some sort could be viewed as the lexical meaning of the strong pronouns. As shown in table 2.3, the cases in which moi signals a strong or weak contrast represent more than half of total occurrences. However, table 2.3 also shows that cases in which moi has lost its contrast-indicating function are far from rare.

30

Detges and Waltereit

Table 2.3 First-person moi indicating (strong and weak) referential contrast

contrast

others

strong

12%

weak

46%

58%

42%

4.2 Topic-Shift Topic-shift is a quantitatively important context for strong pronouns, particularly in the first person. Topic-shift operations, where some old discourse-topic is replaced by a new one (Stark 1999), may be conceived of as construing a contrast between new and old topic, i.e., as a contrast at discourse-level. As can be seen from example (8), use of a first-person moi ‘I’ does not necessarily imply that the speaker establishes herself as the new discourse-topic. In (8) reference to the speaker is a way of starting to elaborate the new topic, similar to the “viewpoint” in (1). In (8), the eventual new topic is the speaker’s father, i.e. an entity closely related to her. Reference to the speaker—the most salient real-world entity given in the situation—thus functions as a cognitive “anchor”, indicating that the current speaker’s universe is the overall domain from which the eventual new topic will be derived. For this reason, the first-person pronoun in example (8) finds itself in the leftmost position of the preamble, followed by the NPs mes parents ‘my parents’ and mon père ‘my father’ which serve to narrow down this domain in two successive metonymic shifts (again similar to Morel’s example in (1)). In such cases, first-person moi does not indicate a referential contrast in the strict sense of the term. Rather, it indicates a shift of interest which, in turn, is based on a contrast between the current speaker’s universe and the universe(s) of some other potential speaker(s). (8)  Elaboration of topic Moi mes parents mon père était sous-chef de gare (ELICOP, Moi 48) Me my parents my father was second-head of station ‘I my parents my father was second head of station.’

From what has been said so far, it follows that strong pronouns are used in topic-shift contexts to ensure discourse coherence. Thus, whereas marking of a referential contrast ensures textual coherence at the content-level (see 4.1.), topic-shift is a contrast at the level of discourse structure.

French Disjoint Pronouns in the Left vs. Right Periphery

31

As mentioned earlier, topic shift is a relatively important function for strong pronouns, especially for first-person singular moi. In the Corpus Orléans it represents exactly 50% of the latter’s overall uses. As can be seen from table 2.4, the degree of overlap between the uses of moi as a topic-shift device on the one hand and as a way of expressing referential contrast on the other hand is rather limited. Table 2.4 First person moi in topic-shift contexts

topic-shift with referential contrast without referential contrast

29% 21%

50%

others 50%

4.3 Turn-Taking It has often been observed (e.g. Stark 1997) that first-person singular strong pronouns play an important role in contexts of turn-taking (for a pragmatic explanation, see, e.g. Detges & Waltereit 2011). As can be seen in (9), they establish a contrast at the level of discourse roles, thereby marking a transition of the floor from one speaker to the next. Example (9) represents a situation of competition between two speakers (B and C) who simultaneously seek the floor.1 Thus, in such contexts, the pronouns organize the dialogical structure of the discourse. Moreover, as is shown in (9), they can also be considered as special instances of the topic-shift construction. It seems thus clear that the particular function under discussion here also has to do with textual coherence, albeit at the level of interactional coherence (allocation of speaker roles and their alignment), as opposed to transactional coherence (internal structure of turns at talk, cf. Hansen 1998). (9) First-person strong pronouns in turn-taking contexts (ELICOP, Moi 371–372). A: Vous pensez qu’on enseigne mieux dans l’école libre You think that-one teaches better in def-school free ‘Do you think that teaching is better in private schools?’

1 It should be mentioned, however, that the ELICOP version of the Corpus Orléans does not explicitly mark overlaps.

32

Detges and Waltereit

B: Moi je pense Me I think ‘I think that . . .’ C: Non moi il y a une chose au collège [. . .] qui se passe No me it there has indef thing at middle.school rel refl go.on ‘No I there’s something going on at the collège’

As can be seen in table 2.5, a still large share of 1s strong pronouns (roughly a third) is used in contexts of turn-taking. A further usage, metonymically derived from the role of first-person pronouns in turn-taking, is their function as quotation markers, as in (10). In such cases, moi indicates the beginning of the protagonist’s speech (as opposed to the beginning of his turn). By the same token, it signals a contrast of perspective, i.e., a shift from the narrator’s point of view to the protagonist’s perspective—in (10), both are identical with the speaker. Thus, moi ‘I’ has become, in such contexts, a device for structuring interactional coherence. (10) First-person strong pronouns as quotation markers On discutait ensemble, j’ai dit moi je sais pas (ELICOP, Moi 100) one debated together I have said me I know neg ‘We had a debate, and I said well I don’t know.’

As a quotation marker, moi is relatively rare. This function accounts for only 4% of the occurrences of moi (see table 2.5). Table 2.5 First person moi in contexts of turn-taking and quotation

Turn-taking contexts Quotation Others

29% 4% 67%

As we have seen so far, at least three major discourse functions of leftperipheral strong pronouns—referential contrast, topic shift and turn-taking— are related to the overall purpose of ensuring discourse coherence. These functions are located on different levels of discourse organisation. Thus, there is overlap between them. This, in turn, means that it makes no sense to simply add the numbers given in tables 2.3–2.5 in order to get an exact quantitative

French Disjoint Pronouns in the Left vs. Right Periphery

33

assessment of the overall importance of discourse coherence for left-peripheral strong pronouns. However, it does make sense to invert the perspective: Are there functions of left-peripheral strong pronouns which are clearly not related to discourse coherence? 4.4 Strong Speech Acts As noted in earlier work (Detges 2003), first-person strong pronouns are regularly used in order to give more illocutionary force to certain types of speech acts. This effect has been described in the literature on strong pronouns as a strategy of “pragmatic weight” (Davidson 1996: 551). This label covers, among other things, the traditional notion of ‘emphasis’. Such an effect can be observed in the following example. (11) Strong first-person pronoun in strong speech act (ELICOP, Moi 11) Je peux vous donner la certitude que  moi je vous les  ferai  les manches I can   you  give def  certainty that  me  I you   3pl make.fut def sleeves ‘I can assure you that I shall make them for you, these sleeves.’

In (11) moi is used in order to strengthen the commissive speech act ‘I shall make them for you, these sleeves’ by emphatically referring to speaker as the source of the illocution. Its function here is to “anchor” the speech act (and the ensuing self-obligation) in the speaker. This, it can be argued, is an inherently modal function. Crucially, in the context where (11) appears in the Corpus Orléans, moi is not denoting a contrast with respect to any other protagonist (e.g. another person who would have promised to tailor the sleeves in question but failed to do so). Moreover, it is used here in spite of straightforward topic-continuity (the topic in this and the preceding clauses being the speaker herself). This means that in (11) moi no longer expresses a referential contrast. Thus, rather than encoding information-structural meaning, the construction acquires expressive meaning which may often be expressed by the same form (cf. Lambrecht 1994: 239). No more than 22% of all uses of moi in the Corpus Orléans appear in the context of a strong speech act as in (11). Table 2.6 First-person moi in contexts of strong speech acts

Strong speech acts Others

22% 78%

34

Detges and Waltereit

4.5 Left-Peripheral First-Person Strong Pronouns are Anchors In table 2.7, we give the quantitative distribution of moi over the various functions discussed so far. Since there is considerable overlap between these functions, their percentages do not add up to 100%. Table 2.7 Quantitative distribution of moi over different discourse functions

Referential contrast Topic shift Turn-taking

58% 50% 29%

Coherence

Strong speech acts

22%

Modality

These figures call for two remarks. Firstly, the left periphery shows a pronounced bias for coherence-related functions. Secondly, however, the figures show that modal functions do occur in the left periphery, thereby contradicting Detges’ & Beeching’s working hypothesis. Thus, despite a clear preference for coherence-related functions, the left periphery does, in fact, allow strong pronouns with modal functions. However, it seems to us that all of the left-peripheral functions discussed so far, i.e. coherence-related as well as modal, have something in common. No matter what their precise function, all left-peripheral strong pronouns are anchors. When indicating a referential contrast (see (6), (7)), they are anchors for reference-tracking. As markers of a topic-shift they often act as anchors for the elaboration of the eventual topic as in moi, mes parents, mon père [. . .] (see (8)). In case the strong pronoun is identical with the topic, it functions as an anchor for the rheme by indicating the latter’s domain. As turn-taking devices, left-peripheral strong pronouns anchor the upcoming turn in a particular speaker. Finally, when introducing a strong speech act, the strong pronoun explicitly anchors the latter in the person of the speaker, thereby conferring the speech act greater illocutionary strength (see (11)). In this sense, anchoring the upcoming rheme either at the level of proposition, discourse or illocution (or at two or all three of these levels at the same time) is “the” function of the left periphery. 5

Free Uses in the Right Periphery

Let us now turn to the major right-peripheral functions of strong pronouns. As we have seen, strong pronouns are significantly less frequent in the right

French Disjoint Pronouns in the Left vs. Right Periphery

35

periphery than in the left one (see above, table 2.2). Moreover, their functions here do not simply mirror the uses discussed in section 4. 5.1 Right-Peripheral Strong Pronouns as Referential Anti-Topics As is widely known, the simplest function of right-peripheral NPs is to “repair” referential gaps by providing appropriate lexical information (see (10)). (12) Have you seen this one—(I mean:) that CAR?

Given that strong pronouns—as all pronouns—do not convey lexical information, we would expect them to be excluded from this use altogether. However, in the Corpus Orléans, there is an attestation of a right-peripheral (third person) strong pronoun with referential function. Its use here is motivated by an effect of “weak” contrast added to the rheme. (13) Uncertain (weak) contrast Michel a dû faire du latin, lui (ELICOP, Lui 51) M. has must.ptcp make of.masc  Latin he ‘Michael, by contrast, had to learn Latin.’

The function documented by (12) and (13) is clearly related to coherence. Even though it is attested for strong pronouns (thereby contradicting the second clause of Detges’ & Beeching’s working hypothesis), this case is very rare. In the entire Corpus Orléans, there is just one single attestation of a third-person pronoun of this kind (i.e. (13)), and first-person pronouns with this function are not documented at all. 5.2 The Post-Rheme Construction (Morel 2007) Apart from the function sketched in the foregoing section (5.1.), right-peripheral pronouns and NPs can serve yet another purpose. This is nicely documented by example (14), which we have already discussed as (11). (14) [. . .] je vous les  ferai les manches (Elicop, Moi 11)   I you 3pl make.fut def sleeves ‘I shall make them for you, those sleeves’

It has sometimes been noted in the literature that sentences such as (14) have an emotive flavour (Nølke 1998) brought about by the repetition of some element of the main clause. According to Morel (2007: 44), this results in a strong (sometimes even polemic) effect of validation of the illocution expressed by the main clause. As Morel (2007) points out, the pattern underlying (14) is a

36

Detges and Waltereit

construction, i.e. a conventional form-meaning pair. At the formal level, this construction—termed as post-rhème by Morel (2007)—combines syntactic with prosodic properties. Specifically, the peripheral element (in example (14) the NP les manches ‘those sleeves’) is not separated by an intonational break from the rheme (i.e., the main clause, see above, section 1). In spite of this, the end of the rheme is clearly identifiable by its terminal intonation contour. In French, focus is prosodically marked by a sharp fall or rise in pitch (see Mertens 2011): a rise in case of an interrogative intonation, a fall if the rheme is an assertive phrase. Since French has oxytonic phrasal accent, the focal intonation contour is always found at the end of a clause, mostly at the end of the rheme. By contrast, the right-peripheral element les manches ‘those sleeves’ in (14) is coded under a flat, low-key intonation contour (see (14’)). (14’) Rheme

Focus    P.S.



Je vous les ferai



The post-rheme construction (Morel 2007)

les manches (Elicop, Moi 11)

French is a strict SVO language. Therefore, in an unmarked French sentence, an object-NP such as les manches ‘those sleeves’ is following the verb (see (14”a)). According to Bossong (1981) and Simone (1997), the very point of the post-rheme construction is that it allows the removal of the NP in question from the core clause, thereby allowing the verb to occupy a phrase-final position under focal intonation (14”b). (14”) Rheme



Focus

(P.S.)

a. Je vous ferai les manches. ‘I will make you those sleeves’

Unmarked Sentence

b. Je vous les ferai, les manches. ‘I will make them for you, those sleeves’

Sentence with post-rheme

French Disjoint Pronouns in the Left vs. Right Periphery

37

This analysis has been further refined by Morel (2007). According to her, the terminal intonation contour in (14”) normally codes information considered as relevant, whereas the low-key intonation contour in the right periphery marks the respective chunk of information as uncontroversial. According to Morel, the effect of the post-rheme construction is a combination of both contours in that it presents the illocution expressed in the rheme as not negotiable. Under this analysis, the post-rheme construction speaks to the listener. In other words, it has an interpersonal, modal function. Unsurprisingly, all of the right-peripheral first and second-person strong pronouns are instances of the post-rheme construction, and among the thirdperson pronouns only the example given in (13) is an exception. Thus, virtually all strong pronouns in the right periphery instantiate the post-rheme construction. From what has been said so far, it follows that there are two ways of “upgrading” a strong speech act. Firstly, the speaker has the option of explicitly “anchoring” the speech act in his/her own person by using a left-peripheral strong pronoun, thereby giving it some extra weight. Secondly, s/he can use a strong first-person pronoun in the right periphery in order to present the speech act as non-negotiable. The following example, taken from the Corpus Orléans, is a combination of both effects. (15) Moi je suis d’ ici   moi (Elicop, Moi 316) Me I am of here me ‘I, I am from here I ’

6 Parentheticals Parenthetical constructions of the type (moi) je trouve que . . . ‘I think that . . .’ have a special discourse function. Normally, they introduce or follow key points of the speaker’s argumentation. In this sense, they are meta-pragmatic operators, and their main purpose is to fine-tune the following (or the foregoing) speech act (Fetzer & Johansson 2010). Given that their function is to inform the listener about how a given conversational contribution should be qualified, they are secondary speech acts with an inherently modal function (Fetzer & Johansson 2010: 242, see also Ducrot 1980). An example is given in (16). (16) C’est pas normal. Moi je trouve que c’est  pas normal (ELICOP, Moi 365) dem is neg normal.  Me I find that dem is neg normal ‘That’s not normal. I think that this is not normal.’

38

Detges and Waltereit

The function of these expressions explains their high frequency. Moreover, parentheticals are a preferred context for strong pronouns (see table 2.2, repeated here for convenience). Table 2.2 First person moi in the left vs. right periphery of free vs. parenthetical expressions Left

Free uses 160 Parentheticals 80

Right

21 20

While the parentheticals under discussion here have traditionally been analysed as mitigating devices (see, e.g., Benveniste 1966, Apothéloz 2003), Fetzer & Johansson (2010)—who do not explicitly discuss the role of strong pronouns— show that parentheticals can in fact function as both mitigators and boosters, depending on their co-occurrence with other discourse connectives and additional context factors (see also Simon-Vandenbergen 2000 for E. I think). In the following sections we will show that not only the strong pronouns play an important role in this issue, but also that the question of left and right periphery is crucial to this question. 6.1 Parenthetical Expressions with Left-Peripheral Subject Pronouns The variant moi je trouve que . . . with a strong pronoun is, despite its high frequency, still much less frequent than the unmarked variant with just the subject clitic je (see table 2.8). Table 2.8 Parentheticals with and without left-peripheral strong pronouns

Je trouve que Moi je trouve que

75% 25%

In principle, strong pronouns which are used within parentheticals have the same functions as in free uses elsewhere. A strong pronoun can be used to reinforce both the mitigating as well as the boosting effect of a given parenthetical. Thus, the strong pronoun in moi je trouve que X ‘I think that X’ could be used to imply a strong referential contrast (see section 4.1.) between speaker and hearer, thereby inviting an inference of the kind ‘My opinion is X, but you do

French Disjoint Pronouns in the Left vs. Right Periphery

39

not have to agree with me.’2 In this case, the use of the strong pronoun is motivated by a strategy of (negative) politeness, increasing the mitigating effect of the parenthetical. Unfortunately, such a case is not documented in the Corpus Orléans. In the following example, the pronoun is used to construe a strong contrast between the speaker’s personal belief and a generally accepted view. In such a case, it has a clearly boosting effect, because it portrays the speaker’s view as particularly original and noteworthy. (17) En principe la patronne coupe et essaie [. . .] en dehors de [. . .] l’apprentie ouvrière mais enfin moi j’ ai toujours trouvé qu’ il était nécessaire euh à l’apprentie de collaborer directement avec elle. (ELICOP, Moi 6)

‘In principle, the boss does the tailoring and the fitting without the apprentice, but I have always found that it was necessary uh for the apprentice to directly work together with her [i.e., with the boss, U.D. & R.W.].

However, pronouns indicating a referential contrast (strong or weak) are exceptional in our corpus. In fact, (17) is the only unequivocal example in our corpus. Normally, the function of the strong pronoun is to turn the parenthetical into a strong speech act (see section 4.4.), thereby giving it a boosting effect. This is shown in (18), where the viewpoint introduced by moi je trouve neither contrasts with a belief held by the addressee (who, in this case, is not an expert on fashion) nor by any of the protagonists mentioned in the previous context. (18) J’ai habillé la maman et je les ai habillées après pour le mariage du frère [. . .] alors euh moi je trouve que ça c’est c’est vraiment de la couture. (ELICOP, Moi 13)

‘I made the mother’s dresses and later, I made dresses for them at the occasion of the brother’s marriage. Now uh I think that’s truly couture!’

Thus, more often than not, parentheticals with left-peripheral strong pronouns have a boosting rather than a mitigating function. This means that leftperipheral strong pronouns in parentheticals mentioned in table 2.2 normally exhibit a modality-related function rather than a coherence-related one. Table 2.9 gives an overview of the individual parenthetical expression with left-peripheral strong pronouns found in the Corpus Orléans, ranked by frequency. In certain cases, our decision to lump two or more variants into one 2 We thank Kate Beeching for suggesting this case to us.

40

Detges and Waltereit

category may be debatable, in particular in the case of moi je dis ‘I say’ vs. moi je dirais ‘I would say’ and moi je vois ‘I see’ vs. moi je vois pas ‘I don’t see’. However, a discussion of the functional differences between these forms would require a separate investigation. Table 2.9 Parentheticals with left-peripheral strong first-person pronouns

Moi je trouve Moi je crois Moi je pense Moi je sais pas Moi je dis/dirais Moi je vois (pas) Moi je connais Moi je suis d’avis que Moi j’estime Moi j’ai l’impression Ʃ

‘I find, I think’ ‘I believe’ ‘I think’ ‘I don’t know’ ‘I (would) say’ ‘I (don’t) see’ ‘I know’ ‘I am of the opinion that’ ‘I guess’ ‘I have the impression’

25 14 9 9 9 8 3 1 1 1 80

31% 18% 11% 11% 11% 10% 5% 1% 1% 1% 100%

6.2 Parenthetical Expressions with Right-Peripheral Subject Pronouns The most frequent expression with a right-peripheral strong subject pronoun is, by some margin, je (ne) sais pas moi ‘I don’t know’. When used as a parenthetical expression, already its “weak” variant je (ne) sais pas functions as a hesitation marker and as a marker of irrelevance (see (19)), highlighting speaker’s lack of commitment to what follows. (19) Il faut connaître le détail il faut euh enfin il faut—je sais pas—il faut un don, oui, il faut savoir aussi parler enfin. (ELICOP, Je sais pas 2)

‘One has to know the details euh you know, one has to like one has to—I don’t know—one has to be gifted, yes, one needs to know how to talk, you know.’

Adding a strong first-person pronoun to the right periphery of je ne sais pas underscores the speaker’s disinclination to re-negotiate the effect expressed by this sequence, thereby turning it into a strong marker of irrelevance. (20) [V]ous avez des livres par exemple sur l’—je sais pas moi—la prononciation, l’art de parler, vous savez, l’ art d’écrire. (ELICOP, Je sais pas moi 2)

French Disjoint Pronouns in the Left vs. Right Periphery

41

‘Do you have books, for example, on the—I don’t know—the pronunciation, the art of speaking, you know, the art of writing.’

The ratio between the “strong” variant je sais pas moi and its weak counterpart je sais pas (see table 2.10) is quite similar to the ratio of the “weak” vs. the “strong” variant of the expression (moi) je trouve with a left-peripheral strong pronoun (see table 2.8). Table 2.10 Parentheticals with and without right-peripheral strong pronouns

Je sais pas Je sais pas moi

70% 30%

6.3 Parenthetical Expressions in the Left vs. Right Periphery Yet another case which we haven’t mentioned in section 3 is the option to place an entire expression in either the left or in the right periphery. This is illustrated by example (21) vs. (22). (21) Preamble Rheme Moi je trouve (que) cela ne va pas. ‘I think (that) this is not o.k.’

BOOSTER

(22) Rheme Cela ne va pas ‘This is not o.k.

MITIGATOR

P.S. moi je trouve. I think’

In cases where the parenthetical appears in the left periphery (21), its effect is a greater illocutionary force carried by the rheme, particularly when introduced by a strong pronoun as in that example. After all we have said about strong pronouns in the left periphery, this effect is expected. However, when the expression appears in the right periphery (22), it relativizes the illocution expressed by the rheme, i.e. it downgrades the latter’s illocutionary force. In other words, its function here is somewhat contrary to the effect observed for bare rightperipheral pronouns (for right-peripheral I think this effect has already been noted by Simon-Vandenbergen 2000: 50pp.). An explanation to this apparent paradox arises from the linear order of the respective elements. The centrepiece of the utterance is the rheme, which, as the locus of the illocution, represents the utterance’s climax. The respective

42

Detges and Waltereit

functions of the left and right periphery are best understood as being oriented towards a successful realization of the rheme. Thus, the expression moi je trouve que, as part of the preamble, has the function of preparing the illocution by explicitly indicating that the latter is grounded in the speaker’s belief. However, given that anchoring of the illocution in the speaker is obvious anyway, whether explicitly indicated or not, the real function of this expression is to make the hearer aware of the particular relevance of the upcoming illocution, thereby giving it extra weight within the organization of the discourse. If, on the other hand, the expression appears to the right of the rheme, its function with respect to the latter is profoundly changed. To understand this, one has to keep in mind how the utterance would be interpreted without a rightperipheral moi je trouve. In this (unmarked) case, the speaker will expect the hearer to ratify the illocution of her utterance by the time the rheme has been fully transmitted. If the illocution is accepted by the hearer, it will be added to the common ground, i.e. to the set of propositions agreed upon by speaker and hearer. If, however, an expression of the type moi je trouve ‘I think’ is added to the rheme after the latter’s completion, the speaker signals that she does not necessarily expect the hearer to accept the illocution. Thus, by explicitly tying the illocution to her own point of view, the speaker falls back behind a consensus that should have been reached at this point. This argument explains the relativizing effect of a right-peripheral moi je trouve. Moreover, it helps us to better understand the profound differences between left and right periphery. This will become clear in the following section. 7

Left Versus Right Periphery

The essential difference between left and right periphery seems to be this: while the elements of the left periphery serve to anchor the upcoming rheme and/ or the illocution expressed by it, the right periphery is a place where rheme and illocution can be re-negotiated. (As we have seen in section 5.2., bare firstperson pronouns can be used to underscore that the illocution is not negotiable, which in itself is a kind of re-negotiation.) Both functions, anchoring as well as re-negotiation, include modal as well as coherence-related aspects. This becomes clear as one examines one by one the elements of the phrase model sketched in section 1 (see (1)). Binding elements, i.e. expressions like tu vois at the outset of the preamble, are attention-catchers. Their function is to anchor the upcoming utterance at the interpersonal level by anticipating a consensus between speaker and hearer over the object of the upcoming utterance

French Disjoint Pronouns in the Left vs. Right Periphery

43

(Morel 2007: 41). Thus, intersubjectivity is not confined to the right periphery. Viewpoint elements, e.g. moi in (1), anchor the upcoming rheme in a general perspective, e.g. in the speaker’s experience. The various frame elements (in (1) hier ‘yesterday’ and en classe ‘at school’) define the temporal and spatial domains in which the rheme is situated. Finally, the lexical topic defines the object to which the rheme pertains. The binding element tu vois has a clearly intersubjective and hence modal function, whereas frames and lexical topic on the other hand are clearly coherence-oriented elements. Reconsidering the various discourse functions of strong first-person pronouns discussed in this article, we find broad confirmation for both hypotheses formulated in the last paragraph. As we have shown, both left and right periphery can have modal as well as coherence-related functions. In both left and right periphery, referential contrast (definite as well as indefinite, see section 4.1.) is a coherence-related function. The same is true for (left-peripheral) topic-shift contexts, where the first-person strong pronoun often serves as an easily available starting point for tracking down the eventual topic (see section 4.2.), thereby anchoring the rheme at the propositional level. In (leftperipheral) turn-taking contexts, the pronoun is used to anchor the following discourse in the respective speaker (see section 4.3.), thereby ensuring coherence at the level of discourse-organization. In strong speech acts, pronouns can strengthen the illocution of the unfolding utterance by anchoring it in the speaker. Contrary to the other context-types discussed so far, first-person strong pronouns have a clearly modal function only in strong speech acts (section 4.4.). In the right periphery, the anti-topic, designed to repair referential gaps, is directed towards propositional coherence. In view of the main hypothesis under discussion here, this could be interpreted as re-negotiating propositional coherence. By contrast, the post-rheme construction, where the validity of the illocution of the utterance is re-negotiated, has a modal (and—more precisely—an intersubjective) function. Thus, it could seem that there is no principled difference between left and right periphery with respect to their capacity to host either modal or coherence-related functions. However, this impression would be misleading. As depicted in fig. 2.1, there is a clear asymmetry between left and right periphery with respect to either type of function. Among the four function types identified for left-peripheral pronouns, only one (i.e. “strong speech acts”) is modal. As we have shown above (see tables 2.6 and 2.7), this function type accounts for only 22% of all left-peripheral strong pronoun tokens. However, this clear tendency is blurred by the high frequency of parentheticals (see table 2.2), where left-peripheral pronouns also have a modal function. In the right periphery, in turn, marking a genuine referential

44

Detges and Waltereit

Preamble Referential contrast Topic-shift Turn-taking Strong speech acts/ Stance

ANCHORING

Post-Script COHERENCE

MODALITY

Referential contrast (Anti-topic) Illocution un-negotiable (Post-rheme)

RE-NEGOTIATION

Figure 2.1 Left and right periphery and their respective functions

contrast via a pronominal anti-topic is extremely rare. Normally, strong pronouns realized in the right periphery are instances of the post-rheme construction and have a modal function. This asymmetry is capturedin fig. 2.1. What are the reasons for this asymmetry? Regarding the left periphery, it may be argued that coherence is a basic requirement for discourse. Any utterance needs to be linked to preceding discourse. However, stance is optional, and thus rarer than coherence-related functions. In the right periphery, however, the ratio between coherence- and modality-related functions is inverted. Anti-topic is a very particular form of conversational repair, and thus arguably rarer than the more varied forms of illocutionary re-negotiation. To conclude, our findings for French strong pronouns confirm Beeching & Detges’ working hypothesis. The asymmetry between left and right periphery is rooted in the linear unfolding of speech in time. The left periphery represents the beginning of a message. It occurs at a moment where the relevant parts of the message itself do not yet exist. Thus, it is suited to either linking the upcoming utterance to previous discourse or to the speaker her/himself—it anchors the utterance. The right periphery, by contrast, marks the end of a message at a moment where it is manifest to both speaker and hearer. Therefore the right periphery is the natural place to comment on a message, to express expectations pertaining to it towards the hearer, or, very rarely, to repair it after the fact. The coherence vs. modality/stance contrast sometimes associated with left vs. right periphery items is merely a reflection of the underlying anchoring vs. re-negotiation contrast. Given the linear unfolding of speech in time, this contrast is natural.

French Disjoint Pronouns in the Left vs. Right Periphery

45

References Apothéloz, Denis. 2003. La rection dite ‘faible’: Grammaticalisation ou degré de grammaticalité? Verbum 25/3: 241–262. Barnes, Betsy K. 1985. The pragmatics of left detachment in spoken Standard French. Amsterdam: Benjamins. Benveniste, Emile. 1966. Problèmes de linguistique générale. Vol. I. Paris: Gallimard. Bossong, Georg. 1981. Séquence et visée. L’expression positionnelle du thème et du rhème en français parlé. Folia Linguistica 15: 237–252. Carroll, Susanne. 1982. “Redoublement et dislocation en français”. In C. Lefebvre (ed.), La syntaxe comparée du français standard et populaire: approches formelle et fonctionnelle. Vol. 1. Québec: Editeur officiel du Québec, 291–357. Danon-Boileau & al. = Danon-Boileau, Laurent, Annie Meunier, Mary-Annick Morel & Nicolas Tournandre. 1991. Intégration discursive et intégration syntaxique. Langages 104: 111–128. Davidson, Brad. 1996. ‘Pragmatic weight’ and Spanish subject pronouns. The pragmatic and discourse uses of tú and yo in spoken Madrid Spanish. Journal of Pragmatics 26: 543–565. De Cat, Cécile. 2005. French subject clitics are not agreement markers. Lingua 115: 1195–1219. ———. 2007a. French dislocation: Interpretation, syntax, acquisition. Oxford: Oxford University Press. ———. 2007b. French dislocation without movement. Natural Language and Linguistic Theory 25: 485–534. Detges, Ulrich. 2003. “Du sujet parlant au sujet grammatical. L’obligatorisation des pronoms sujets en ancien français dans une perspective pragmatique”. In B. Combettes & C. Marchello-Nizia (eds.), Grammaticalisations en français, special issue of Verbum 25, 307–333. ———. 2013. “First person pronouns in spoken French. A case study in cliticization”. In K. Jeppesen Kragh & J. Lindschouw (eds.), Deixis and Pronouns in Romance Languages. Amsterdam: Benjamins, 33–47. Detges, Ulrich & Richard Waltereit. 2011. “Turn-taking as a trigger for language-change”. In S. Dessì Schmid, U. Detges, P. Gévaudan, W. Mihatsch & R. Waltereit (eds.), Rahmen des Sprechens. Beiträge zu Valenztheorie, Varietätenlinguistik, Kreolistik, Kognitiver und Historischer Semantik. Tübingen: Narr, 175–189. Doetjes, Jenny, Delais-Roussarie, Elizabeth & Petra Sleeman. 2002. “The prosody of left detached constituents in French”. In Proceedings of the Speech Prosody 2002 conference, Aix-en-Provence, 247–250. Ducrot, Oswald. 1980. “Je trouve que”. In O. Ducrot & al. (eds.), Les mots du discours. Paris: Minuit, 57–92.

46

Detges and Waltereit

ELICOP = Brosens, Veerle et al. Etude linguistique de la communication parlée. [http:// bach.arts.kuleuven.be/elicop] Fetzer, Anita & Marjut Johansson. 2010. Cognitive verbs in context. A contrastive analysis of English and French argumentative discourse. International Journal of Corpus Linguistics 14: 240–266. Hansen, Maj-Britt Mosegaard. 1998. The function of discourse markers. A study with special reference to Spoken Standard French. Amsterdam: John Benjamins. Lambrecht, Knud. 1981. Topic, antitopic and verb agreement in Non-Standard French. Amsterdam: John Benjamins. ———. 1987. “On the status of SVO sentences in French discourse”. In R.S. Tomlin (ed.), Coherence and Grounding in Discourse. Amsterdam: Benjamins, 217–261. ———. 1994. Information Structure and Sentence Form: Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: Cambridge University Press. Mayol, Laia. 2010. Contrastive pronouns in null-subject Romance languages. Lingua 120: 2497–2514. Mertens, Piet. 2011. “Prosodie, syntaxe et discours: autour d’une approche prédictive”. In H.-Y. Yoo & É. Delais-Roussarie (eds.), Actes d’IDP 2009. Paris, Septembre 2009 (=http://makino.linguist.jussieu.fr/idp09/actes_fr.html), 19–32. Morel, Mary-Annick. 2007. Le postrhème dans le dialogue oral en français. L’information grammaticale 113: 40–46. Nølke, Henning. 1998. “Il est beau, le lavabo, il est laid, le bidet. Pourquoi disloquer le sujet?”. In M. Forsgren, K. Jonasson & H. Kronning (eds.), Prédication, assertion, information. Actes du colloque d’Uppsala en linguistique française. Uppsala: Uppsala University, 537–543. Simone, Raffaele. 1997. Une interprétation diachronique de la dislocation à droite dans les langues romanes. Langue française 115: 48–61. Simon-Vandenbergen, Anne-Marie. 2000. The functions of I think in political discourse. International Journal of Applied Linguistics 10/1: 41–63. Stark. Elisabeth. 1997. Voranstellungsstrukturen und “topic”—Markierung im Französi­ schen. Tübin­gen: Narr. ———. 1999. “Antéposition et marquage du thème (topique) dans les dialogues spontanés”. In C. Guimier (ed.), La thématisation dans les langues. Actes du colloque de Caen, 9–11 octobre 1997. Sciences pour la communication 53. Frankfurt: Lang, 337–358.

chapter 3

Motivations for Meaning Shift at the Left and Right Periphery: well, bon and hao Kate Beeching and Yu-Fang Wang 1 Introduction The purpose of this chapter is to trace the historical evolution, crosslinguistically, of canonically left-peripheral pragmatic markers (PMs) which derive from—and remain in a polysemous relationship with—terms with the canonical meaning ‘good’ (such as well, bon, and hao). Each of these has developed a ‘good up to a point’ (qualified acceptance) sense, though well appears to be further along the pragmaticalisation scale than either bon or hao. Unlike bon and hao, PM well can no longer be interpreted to mean ‘acceptance’. The paper explores how and why such a process of semantic change occurs and attempts to further elucidate the role of left-peripheral items and the extent to which they can occur medially and at the right periphery. An underlying objective of the paper is to test current theories about regularities in semantic change (Traugott and Dasher 2002), specifically the hypothesis that LP is inherently dialogual (two speakers) while RP is dialogic (two viewpoints): items on LP are often considered to have a dialogual (logical) connective role rather than a dialogic (interpersonal/modal) role. The paper argues that language-external social factors such as face-management exert a strong functional pressure on the language system, in this case the semantic system; turn-taking and disagreeing with a previous speaker must be managed diplomatically, as well as efficiently. It is well-established that well begins a dispreferred second pair-part in an adjacency pair (Pomerantz 1984); in addition to flagging incoherence, well/bon and hao mitigate the upcoming message (e.g. Owen 1979, and Watts, 1987, take well, Brémond, 2004, takes bon and Y.F. Wang and Tsai 2005, take hao to be face threat minimisers). The existence of left-peripheral well/hao/bon appears to undermine a strong form of the hypothesis that LP is inherently connective and RP modal: PMs well, bon and hao immediately suggest a second view-point: they are dialogic/modal as well as being dialogual/logical (forging a coherent link with the preceding turn). Traugott (2010) makes a similar point concerning the dialogic use of Spanish bien. Where well and bon occur turn-medially (non-dialogually) in

© koninklijke brill nv, leiden, ���4 | doi ��.��63/9789004274822_�04

48

Beeching and Wang

synchronic data, they confer a modalising cloak on the surrounding discourse, through concession. Dialogicity appears to be a rhetorical use of dialoguality— by using well/bon a speaker evokes dialoguality (having a conversation with oneself, suggesting other viewpoints). The chapter starts in Section 2 with some general observations about both well and hao in diachrony and in synchrony (Schiffrin 1987; Finell 1989; Jucker 1995; 1997; L.Y. Wang 2004; H.L. Wang 2005; Wang and Tsai 2005; Wang et al. 2010) from the perspective of politeness-induced semantic change (see Beeching 2005 and 2007). Less work has been conducted on the historical development of PM bon. In Section 3, therefore, the paper reviews recent developments in the functions of bon in contemporary spoken French (Beeching 2009) and new data is presented on its diachronic development drawn from both the AngloNorman Corpus and from FRANTEXT. Section 4 presents diachronic data with respect to the development of hao. In Section 5, an attempt has been made to present a cross-linguistic semantic map (cf. Haspelmath 2003) showing the overlapping polysemies of well, bon and hao with an indication of their historical development and degree of pragmaticalisation. Some conclusions about terms on the left and right periphery are presented in Section 6. 2

Politeness-Induced Semantic Change

In English, French and Chinese, the typically left-peripheral response items well, bon and hao have developed ‘demurral’ meanings, whereby they flag that a respondent is not fully in agreement with the previous speaker’s utterance or background assumption. We argue that the ‘abusive’ use of a term originally meaning ‘good’ to mean ‘less than good’ or only provisionally good is motivated by reasons of politeness and face-management. As Wang and Tsai put it (2005: 231): (Pre-)closing signals such as hao may be regarded as a sub-variety of mitigating expressions used in conversation, that is, the desire to agree or appear to agree with the addressee, which leads to mechanisms for pretending to agree (Brown and Levinson, 1987 p. 113). Such expressions serve the twofold function of keeping the conversation going in a systematic manner while allowing the conversationalists to preserve either the reality or the appearance of cooperation. Having undergone this first semantic shift (from ‘good’ to ‘less than good’ or to flag some kind of demurral), the forms can go on to have further meanings and

Motivations for meaning shift at the left and right periphery

49

shift from a left peripheral to medial positions where they might be considered to be pause-fillers or modal particles. Our hypotheses are as follows: 1.

2.

There are dialogual and interpersonal motivations for the development of ‘less than good’ meanings of these expressions, that is, they develop as part of interactional turn-taking between two speakers. Well is acknowledged since Pomerantz to initiate a dispreferred sequence in the second pair part of an adjacency pair. Brémond (2004) also gives an excellent example of the way that bon is used to flag less than total agreement. There are dialogic motivations for the development of ‘less than good’ meanings of these expressions. ‘Dialogic’ generally refers to refutational sequences or sequences which involve more than one viewpoint. As Traugott points out (2010: 15), concession is one locus which indexes dialogicity.

The question arises as to whether the dialogual precedes the dialogic or vice versa. These two approaches are illustrated in the two main hypotheses proposed with respect to the development of well. Finell (1989) highlighted the concessive qualities of well and suggested that the hearer: looks at the matter differently than the speaker does. Nevertheless, the speaker is willing to admit that the interlocutor has got a point in his/ her argument; or, that the issue the interlocutor is presenting, either in the form of a statement or in the form of a question, is not totally out of the question, yet the speaker is not prepared to completely comply with the interlocutor. (p. 655) She traces this use of well back to the following use of well as a predicative adjective attested by the OED in 1560: And where as they saye that the Gospell must be taught after the interpretations approued by the churche (that is very well) but all the stryfe is, which is the trewe churche (OED, well) Jucker (1997) suggests that early uses of wel la (an attention getting device) in Old English and its use in Middle and Early Modern English preface direct speech and indicate acceptance. In the later Early Modern Period, well is beginning to have the face-threat mitigating senses we see in Modern English.

50

Beeching and Wang

Defour (2008: 79) concurs with both Finell and Jucker, concluding (about well) that: The inherently positive meaning of the manner adverb can be recruited by the speaker to serve as a marker of interpersonal agreement and acceptance when placed in utterance-initial position. More specifically, the evaluative meaning of well is transformed from its semantic origin into a semantically bleached means for the speaker to acknowledge the validity of the preceding speaker turn. When used in contexts where speaker and addressee have diverging points of view, well can introduce a subjective perspective without allowing the addressee to lose face. There seems to be thus some scope for identifying an early left-peripheral dialogual use of such expressions which can then be used rhetorically dialogically to flag concession in medial positions. In other words, we posit that interactional features, both dialogual and dialogic, underly the grammaticalisation from an adjectival or adverbial to a PM usage and the semantic change from ‘good’ to ‘only provisionally good’. 3

Bon

In contemporary spoken French, as demonstrated in Beeching (2009), bon is polysemous, retains a canonical adjectival form as in un bon vin blanc (‘a good white wine’), can be used to mark the successive stages in a narrative, to introduce repair or concession, and as a modal particle, hesitation marker or pausefiller. Most recently, it is often found after conjunctions and adverbs (mais bon, parce que bon, et puis bon) in positions where it cannot be translated as ‘good’ but is more naturally rendered as ‘well’, as in example (1). (1) mais si ça se passait bien à New York là en l’occurrence j’aimerais travailler euh bon pas pas faire toute ma carrière là-bas mais au moins faire cinq dix ans + * en en espérant que ça me plaise parce que bon c’est toujours pareil je j’imagine ça on voit les films on voit ceci on voit cela c’est l’idée euh + c’est le rêve américain mais bon ça se trouve ça va pas me plaire du tout + mais euh + si ça me plait j’aimerais travailler là-bas euh + une dizaine d’années + pour voir euh puis pour bon pour avoir aussi une euh + une expérience quoi parce que bon rester en France bon la France c’est bien mais bon c’est c’est un style de travail + c’est une /idé-, idée/une idéologie et c’est pas c’est différent dans tous les pays + CRFP PRI-PNE-1 

Motivations for meaning shift at the left and right periphery

51

‘but if it went well in New York in that case I’d like to work euh well/Imean not not make my whole career over there but to do at least 5 or 10 years + * hoping that I like it because well it’s always the same I I imagine that you see films you see this you see that it the idea euh + it’s the American dream but well it may turn out I may not like it at all + but euh + if I like it I’d like to work over there euh + for ten years or so + to see euh and also to well to have euh + an experience kind of as well because well staying in France well France is fine but well it’s a style of work + it’s an ide / an ideology and it not it’s different in all the countries +’

The corpora used to study bon included the Anglo-Norman corpus, the Base de Français Médiéval, FRANTEXT, the ESLO Corpus and the Corpus de Référence du Français Parlé. Key examples are given below to illustrate the main thrust of the argument. Through the Renaissance period (1,067,752 words), there are 6,326 occurrences of bon. Virtually all are adjectival, with only 5 interjective usages. One of these indicates that bon can be interpreted as ‘good up to a point’ or ‘provisional acceptance’, as shown in example (2): (2) THOMAS. Sors, l’espousée, il n’y a personne. JAQUET. Me voicy. THOMAS. Ta voix est un peu aspre. Dy en ceste sorte: me voyeecy. JAQUET. Me voyeecy. THOMAS. Bon; il faut que tu adoucisses et affiles ta langue le plus que tu pourras. JAQUET. Je feray qu’il semblera que le miel et le sucre sortent de ma bouche, que veux-tu davantage. ‘THOMAS. Come out, wife, there is no-one here. JAQUET. ‘ere I am. THOMAS. Your voice is a bit rough. Say it like this: He-w- I- am.

52

Beeching and Wang JAQUET. He-w- I- am. THOMAS. OK; you need to speak in as sweet and refined a way as you can. JAQUET. I’ll do it so it seems as if honey and sugar come out of my mouth, what more do you want?’

The fact that Thomas adds a rider to his OK, insisting that Jaquet speak in as sweet and refined a way as he can provides a bridging context to ‘provisional acceptance’. Left peripheral response-marking bon can be interpreted as ‘good up to a point’ or provisional acceptance. It is with playwrights, and particularly in the texts of Molière in the 17th century that we find the first occurrences of bon used left-peripherally as an interjection. Most indicate acceptance. But already a number suggest demurral, as in example (3). (3) ARNOLPHE : Elle . . . me vint demander, Avec une innocence à nulle autre pareille, Si les enfants qu’on fait se faisaient par l’oreille. CHYSALDE : Je me réjouis fort, seigneur Arnolphe . . . ARNOLPHE : Bon ! Me voulez-vous toujours appeler de ce nom. CHRYSALDE : Ah ! malgré que j’en aie, il me vient à la bouche, Et jamais je ne songe à Monsieur de la Souche. MOLIÈRE, L’École des femmes, 1663, p. 170, ACTE I, SCÈNE PREMIÈRE ‘ARNOLPHE: She came to ask me with unexampled innocence if children came through the ears. CHYSALDE: I greatly rejoice, Mr. Arnolphe. ARNOLPHE: What! Will you always call me by that name? CHRYSALDE: Ah! it comes to my lips in spite of me; I never remember Mr. de la Souche.’

Motivations for meaning shift at the left and right periphery

53

The context and the published prose translation demonstrate that, by the 17th century, bon cannot be considered to be a term indicating acceptance—in its interjective form it can be highly contestatory. The response to the use of bon in example (4) seems to indicate that bon is ambiguous and can mean ‘good’ acceptance or rejection: (4) CLEANTHIS: . . . Et lorsque je fus te baiser, Tu détournas le nez, et me donnas l’oreille. SOSIE : Bon ! CLEANTHIS: Comment, bon ? SOSIE : Mon Dieu ! tu ne sais pas pourquoi, Cléanthis, je tiens ce langage : J’avois mangé de l’ail, et fis en homme sage De détourner un peu mon haleine de toi. MOLIERE Amphitryon, 1668 p. 421 ACTE II, SCENE II. ‘CLEANTHIS: . . . And when I went to kiss you, you turned your nose away and offered your ear. SOSIE: ?Good !/Well! CLEANTHIS: What do you mean,?good/well? SOSIE: My God! you don’t know why, Cléanthis, I reacted that way I’d eaten garlic and did the wise thing in turning my breath away from you.’

Interjective bon is generally LP in Molière but there are also examples of concluding bon. There are in addition examples in the dramatic texts of the late 17th century of what we might call ‘quotative’ bon typically in collocations with dire. Bon evokes the beginning of an utterance and is used iconically to do so in monologual contexts.

54

DM usage: utterance medial

1

–

–

–

57 95%

–

3

–

–

–

936 89%

–

111 10.5%

5 0.5%

–

–

261 60%

–

107 25%

64 15%

–

–

390 68%

–

105 18%

77 14%

–

–

161 65%

–

58 23%

28 11%

–

246 49%

–

140 28%

115 23%

–

Concessive Bon . . mais

LP Interjection

1

Predicative adjective

42 95%

Attributive adjective

bon rate per 10,000 words

bon Tokens

Occurrences of bon in adjectival, interjectional and discourse-marking forms from the 13th–21st century

No. of words

Period Corpus

Table 3.1

Beeching and Wang

LP Medial 131,979 44 3.3 FRANTEXT Ancien Français 1200–1277 34,231 60 17.5 FRANTEXT Moyen Français 1415–1457 1,067,752 1,052 9.9 FRANTEXT Renaissance 1507–1598 Comédies 525,462 432 8.2 FRANTEXT – Molière 1660–1673 645,711 572 8.9 FRANTEXT 1701–1800 Random selection of “Théâtre” 410,820 247 6.0 FRANTEXT 1801–1900 “Théâtre” 659,533 505 7.7 FRANTEXT 1901–2000 “Théâtre”

4 0.8%

DM usage: utterance medial

LP Interjection

Predicative adjective

Attributive adjective

bon rate per 10,000 words

bon Tokens

No. of words

Period Corpus

Concessive Bon . . mais

55

Motivations for meaning shift at the left and right periphery

LP Medial ESLO 1968 “Spoken” Sample CRFP 2002 “Spoken” Sample

124,048

365

29.1

54 15%

105,537

626

59.3

67 11%

–

9 2.4%

191 52%

113 31%

1 0.3%

13 2%

53 8%

478 76%

15 3%

R = Rate of occurrences of bon per 10,000 words Bold face – Expression as percentages of total number of occurrences of bon in each sample

Table 3.1 demonstrates that rates of concessive bon . . . mais are relatively low in the corpora studied—it appears that Finell’s (dialogic) hypothesis with respect to well does not apply to bon. On the other hand, it does seem to be the case, even allowing for genre-related variation, that there are increasing occurrences of left peripheral interjectional bon over the centuries which may ease its route towards the modal usages which occur medially in contemporary spoken French. In the very early periods, an expressively fronted bon, combined with ellipsis of C’est in the predicative usage C’est bon ‘That’s good’, leads to the possibility of using bon on its own left-peripherally as an acceptance marker. In example (2), we have seen a bridging context in the 16th century, which leads to Molière being able to use it, in the seventeenth century, interjectionally, to mean ‘less than acceptance’, indeed surprise and rejection. It seems that a dialogual left-peripheral usage, reacting to a previous speaker’s words in an expressive way, either accepting, showing surprise,

56

Beeching and Wang

refuting or providing backchannel feedback (ah bon?), paves the way for monologual dialogic usages in which speakers nuance their opinions by rhetorically exploiting a dialogual style. Speakers, as it were, import a ‘second voice’, commenting on their own ongoing argument. In older speakers in the contemporary ESLO and CRFP spoken corpora this may include a bon . . . mais construction, illustrated in example (5): (5) je dirais qu’on rentre plus ou moins dans une routine bon c’est vrai que mon travail n’est pas vraiment routinier parce que + je fais énormément de choses + je je rencontre énormément de personnes + mais + non oui le week-end + je suis obligée de partir d’ici + je c’est pas que je n’aime pas Troyes + j’aime bien * notre petite ville on est tranquille + mais euh + oui au niveau euh ben des sorties euh c’est très très important parce que autrement on est là on va travailler c’est métro boulot dodo et puis + basta

‘I would say that one gets more or less into a routine well it’s true that my work is not really routine because + I do a lot of things + I I meet lots of people + but + no yes the weekend + I have to leave here + I it’s not that I don’t like Troyes + I love our little town we’re quiet + but er + yes from the point of view of er well getting away it’s very very important because otherwise you’re there you go to work it’s metro-work-sleep + and then + that’s it’

Bon in this example is arguably left peripheral in that it is positioned before a new intonation unit. However, it is janus-faced, retroactive and proactive, it provides a provisional acceptance of the previous utterance but goes on to nuance it. This is not a new observation—Roulet et al. (1985: 82) commented on this usage of bon remarking that “the speaker includes in his speech a counter-argument which has not necessarily been uttered by the addressee, or which at any rate does not appear in the co-text—and which he then rejects”. Nonetheless, it is of interest to see how this has played out historically and the ways in which the polysemies do not necessarily evolve in a linear implicational way. In contemporary spoken French, in the compound expressions such as parce que bon, mais bon and et puis bon which we saw in example (1), bon cannot be considered to be left peripheral but is rather medial—it is not the start of a concessive clause but appears to have gathered up in its meaning a type of dialogicity or concessivity which modalises the ongoing discourse. The answer to the main research question is that, for bon, dialoguality precedes dialogicity, confirming the importance of the role played by interactional factors in semantic change.

Motivations for meaning shift at the left and right periphery

57

Table 3.2 Distribution of hao in the Old Chinese data Syntactic Functions

Number

Rate per 10,000 words

Percentage

Hao as a Vt taking an Abs. N Hao as a Vt taking a Con. N Hao as a Vt taking a VP Hao as a Vi Hao as an N Hao as an attributive ADJ Hao as a predicative ADJ Total

102 14 30 16 14 3 2 181

15.96 2.19 4.69 2.5 2.19 0.47 0.31 –

56.4 7.7 16.6 8.8 7.7 1.7 1.1 100

4

Hao

Hao has a long history, developing from a verbal form (‘to like’) to an adjectival usage and thence to an agreement or acceptance marker. Its original meaning is ‘beauty’ based on its orthographic formation, the combination of the characters (radicals) nü (女), ‘female’ and zi ‘child/male’ (子). In order to investigate how these various members of hao are derived, they were examined in terms of four periods of time (Dong 1977; Wei 1994; Shi 2003): Old/Archaic Chinese (900 BC–200 BC), Middle Chinese (200 BC–900 AD), Early Mandarin/Premodern Chinese (900 AD–1500 AD), and Modern Mandarin (1500 AD–present). In each period, examples were drawn from representative works to examine the various meanings of hao. The data surveyed below come from the Academic Sinica Corpus at http://www.sinica.edu.tw/~tdbproj/handy1/ and the Chinese Text Project at http://ctext.org/zh. In the Old Chinese section of the Academic Sinica Corpus which consists of Lunyu (written in 772 BC–221 BC; 16,012 morphemes), Daxue (written in 505 BC–434 BC; 1,753 morphemes), Mengzi (written in 475 BC–221 BC; 34,685 morphemes), Laozi (written in 600 BC–470 BC; 5,000 morphemes) and Zhuangzi (written in 369 BC–286 BC; 6,466 morphemes), 181 instances of hao were found, as shown in Table 3.2. Hao is most frequently used in Early Old Chinese (11th–6th BC) as a transitive verb (56.4% in Table 3.2), as illustrated in (6):

58

Beeching and Wang

(6) 子曰：君子食無求飽，居無求安， 敏於事而慎於言，就有道而正焉， 可謂好學也已。 (《論語 ‧學而第一》) Zhi yue : Junzi shi wu qiu bao, ju wu qiu an, Confucius say Gentleman eat not seek full live   not  seek ease   min yu shi er shen yu yan,    earnest in things and careful in words    jiu   you dao er zheng yan,    approach have way and justify PRT    ke wei hao xue ye yi.    can   say   love   learning   PRT   PRT (Xue Er I, Lunyu or The Analects, 772 BC–221 BC)



‘The Master said, “He who aims to be a man of complete virtue in his food does not seek to gratify his appetite, nor in his dwelling place does he seek the appliances of ease; he is earnest in what he is doing, and careful in his speech; he frequents the company of men of principle that he may be rectified—such a person may be said indeed to love to learn.”’ (source: http://wengu.tartarie.com/wg/wengu.php?l=Lunyu&no=2)

Hao as a transitive verb often took an abstract noun as its object, as shown in (6). Gradually, its transitivity decreased and it often functioned as an attributive adjective or predicative adjective (intransitive verb or stative verb) denoting positive evaluation ‘good’ (Jing-Schmidt 2010), as shown in (7–8): (7) 所謂誠其意者，毋自欺也，如惡惡臭，如好好色， 此之謂自謙，故君子必慎其獨也！ (《禮記‧大學》) Suowei cheng qi   yi  zhe, wu zi  qi ye, So called sincerity its meaning person not self deceive PRT Ru wu    er chou, ru  hao hao   se, as hate bad smell as like beautiful colour Ci zhi wei  ziqian, This ASSOC called self modesty gu junzi    bi shen qi   du ye! so gentleman must careful his alone PRT (Da Xue, Liji or The Classic of Rites, 475 BC–221 BC) ‘What is meant by “making the thoughts sincere.” is the allowing no selfdeception, as when we hate a bad smell, and as when we love what is beautiful. This is called self-modesty. Therefore, the superior man must be watchful over himself when he is alone.’

Motivations for meaning shift at the left and right periphery

59

(8) 莊公五年，取齊女為夫人，好而無子。 (史記 衛康叔世家) Zhanggong wu nian, qu Qinü wei furen, hao er wu zi. Zhang king five year marry Qi woman as wife good but no child (House of Wei Kangshu, Shiji ‘Records of a Historian’, 109BC–91BC) ‘In the 5th year of King Zhang, (he) married a woman from Qi as his wife, who was good/beautiful but had no child.’

The distribution of hao in the Middle Chinese period (200 AD–900 AD), as reflected in examples drawn from the Shishuo Xinyu (New Account of Tales of the World written in 403 AD–444 AD; 64,989 morphemes), Yanshi Jiaxun (Family Instructions of Master Yan written in 531 AD–591 AD; 196,845 morphemes), and Wenxin Diaolong (The Mind of Literature and Carving Dragons written in 466 AD–520 AD; 1,700,011 morphemes), is summarized in Table 3.3. Table 3.3 shows that the number of occurrences of hao as a predicative adjective (14.1%) has increased, though there are still a large number of examples of hao as a transitive verb (27.5%). Table 3.3 Distribution of hao in the Middle Chinese data Syntactic Functions

Number

Rate per 10,000 words

Percentage

Hao as a Vt taking an Abs. N Hao as a Vt taking a Con. N Hao as a Vt taking a VP Hao as a Vi Hao as an N Hao as an ADV Hao as an attributive ADJ Hao as a predicative ADJ Total

39 10 28 1 24 2 18 20 142

0.2 0.05 0.14 0.005 0.12 0.01 0.09 0.1 –

27.5 7 19.7 0.7 16.9 1.4 12.7 14.1 100

In Early Mandarin/Pre-modern Chinese (900 AD–1500 AD), the adjectival function of hao replaced the transitive function to become the dominant use. The distribution of hao in Shui Hu Zhuan (also known as Outlaws of the Marsh, All Men Are Brothers, Men of the Marshes, or The Marshes of Mount Liang, written in 1290 AD–1365 AD; 920,810 morphemes), San Guo Yanyi (Romance of the

60

Beeching and Wang

Table 3.4 Distribution of hao in the Early Mandarin data Syntactic Functions

Number

Rate per 10,000 words

Percentage

Hao as a Vt taking an Abs. N Hao as a Vt taking a Con. N Hao as a Vt taking a VP [V+ hao] compound [hao + V] compound Hao as an N Hao as a DEG ADV Hao as an ADV Hao as a CONJ Hao as a resultative ADJ Hao as an attributive ADJ Hao as a predicative ADJ Hao bu hao as a tag Q Hao as a PM Total

41 10 113 55 112 68 258 451 80 121 1157 862 36 56 3420

0.21 0.05 0.58 0.28 0.58 0.35 1.33 2.33 0.41 0.62 5.97 4.45 0.19 0.29 –

1.2 0.3 3.4 1.6 3.3 2 7.5 13.2 2.3 3.5 33.8 25.2 1.1 1.6 100

Three Kingdoms, written in 1330–1400 AD; 610,453 morphemes), and Jin Ping Mei (Golden Lotus, written in 1526–1593 AD; 406,797 morphemes) is displayed in Table 3.4. Table 3.4 shows that in the Early Mandarin/Pre-Modern Chinese period, hao underwent many syntactic changes. First, it could be used as a conjunction. Both Li (2005) and Zhang and Ding (2009) treated the adverbial use of hao, occurring before a verb and meaning ‘easily’, as an auxiliary verb. Li argued that the connector use of hao, which means ‘so that’, came from its adverbial use, as seen in (9): (9) 白日放歌須縱酒，青春作伴好還鄉 (杜甫詩) Bairi  fang   ge  xu zongjiu, Day time  sing  song  must drink wine excessively qingchun zuoban hao huan xiang. youth company HAO return   home (Dufu’s poem, taken from Zhang and Ding, 2009: 33)

Motivations for meaning shift at the left and right periphery

61

‘(I) must drink wine excessively while singing songs at daytime, (my wife) will be my company for going home in spring.’

However, Jing-Schmidt (2010) considered it a modal expression. In our tabulation as summarized in Table 3.4, we treated hao occurring before a verb, meaning ‘easily’, as an adverb. When it occurred before a clause, meaning ‘so that’, we treated it as a conjunction, which suggests that hao had scope over a proposition and acquired a clause-linking function. More importantly, it emerged as a PM. Specifically, it served a textual function relating to the structure of discourse as text and/or an interpersonal function (or ‘expressive’ function in Traugott’s (1982) term) relating to the expression of speaker attitude and to the organization of the social exchange (Brinton 2006). Among the 56 tokens of hao as a PM, there are 9 co-occurring with le and 8 with ya. In L.Y. Wang’s (2004) analysis of the grammaticalisation of hao, 5 out of 151 tokens (3%) of hao were found as agreement and acceptance markers in the data coming from Zutangji (Collection from the Patriarchs’ Hall in 751 AD–834 AD). Looking at hao in Zutangji, we found that it could be used as an agreement marker. Based on this finding, we suggest that hao as an agreement and acceptance marker appeared in the Late Middle period and the number of occurrences of hao as a PM increased in the Early Mandarin/Pre-Modern Chinese period. In addition, in the data, we found that hao as an adjective was used by the speaker to make a positive evaluation, as illustrated in (10): (10) 武松笑道：“若得嫂嫂做主，最好。(金瓶梅 第二回) Wusong xiao dao: “ruo de saosao zuozhu, zui  hao. Wusong laugh say if get  elder brother’s wife determine  most good (Chap. 2, Jin Ping Mei or The Plum in the Golden Vase, 1602 AD)

‘Wusong laughs and says: “if I can have you determine it, that would be best.”’

Here, hao was used to make a subjective evaluation on the previous proposition. As this use applied to the turn-taking, it developed the function of an agreement/acceptance marker, as shown in (11): (11) “. . . 與你起個法名叫做‘孫悟空’，好麼？” 猴王笑道：“好！好！好！自今就叫做孫悟空也。” (西遊記第1回, 1590s AD)

62

Beeching and Wang “. . . yu ni  qi ge   faming  jiaozuo Sun Wukong, hao me?”  to  2SG  give CL name in religion named  Sun Wukong, okay PRT Houwang xiao dao: “Hao! Hao! Hao! monkey king laugh say  HAO HAO HAO Zi jin  jiu jiaozuo Sun  Wukong ye.” Since now ADV named Sun  Wukong PRT (Chap. 1, Xi You Ji or Journey to the West in 1590s AD) ‘So we can give you the Dharma−name Sun Wukong, which means ‘Monkey Awakened to Emptiness’. Will that do?” “Marvelous, marvelous,” said the smiling Monkey King. “From now on my name will be Sun Wukong.” (translated by Jenner (1984))

Hao appears to move from a central position as a predicative adjective to a peripheral marker of agreement on the left periphery. In line with L.Y. Wang (2004), we argue that the path of grammaticalisation of hao is from a predicative adjective to a discourse marker of agreement and acceptance. Hao in the Early Mandarin Period also co-occurred with sentence final particles to express a speaker’s positive evaluation. L.Y. Wang’s analysis of the grammaticalisation of hao did not address the collocation hao le. This combined form can be regarded as a discourse marker, as shown in (12): (12) 行者聞得此言，滿心歡喜道：「婆婆，你這裡到那解陽山有幾多 路程？」 婆婆道：「有三千里。」 行者道：「好了，好了。師父放心，待老孫取些水來你吃。」 (西遊記第 53回, 1590s AD) Xingze wen de ci yan, manxin  huaxi dao, Traveller hear get this words wholehearted happy say “Popo, ni zheli dao na Xieyangshan  you ji  duo lucheng?’ Granny 2SG here to that   Xieyang Mountain have how long distance Popo   dao, “you san qian li.” Granny say have three   thousand kilometers Xingze dao, “Hao  le,  hao   le, Traveller say HAO PRT HAO  PRT Shifu fangxin, dai lao Sun qu xie shui lai  ni chi.” Master don’t worry wait old  Sun take some water come 2SG eat (Chap. 53, Xi You Ji or Journey to the West in 1590s AD)

Motivations for meaning shift at the left and right periphery

63

‘Monkey, who was very pleased to learn all this, then asked how far it was to Mount Offspring Dissolved. “Three thousand kilometers (=Ten miles),” the old woman replied. “Fine,” said Monkey. “You can stop worrying, Master. I’ll go and get you some of the water.” (translated by Jenner (1984))

It also occurred with the negator bu ‘not’ to signal the speaker’s negative comment on a bad situation, as in (13): (13) 行者大驚道：「不好了，不好了，中了他計也！」 (西遊記第 7回, 1590s AD) Xingze da jing dao, “bu  hao le, bu hao  le, Travelor big surprise say not HAO PRT not HAO PRT Zhong le ta ji ye!” Get ASP 3SG trick PRT (Chap. 7, Xi You Ji or Journey to the West in 1590s AD)

‘“This is terrible,” said Monkey, “he’s tricked us.”’



(translated by Jenner (1984))

In early Modern Mandarin, hao as a predicative adjective plus le and its negative counterpart were frequently used to convey the speaker’s evaluation. Hao le became both a discourse marker indicating the speaker’s positive evaluation and a stage marker (marker of discourse boundary), as seen in (14): (14) 王太大道：“果然如此，好了，你到那人家說去，我等你回信。” (儒林外史第26回, 1750 AD)



Wang Tai Wang Mother ni dao na 2SG  to that wo deng ni 1SG wait  2SG



da dao, “Guoran ruci, hao le, big say really such HAO PRT renjia shuo  qu, house  say go  huixin.” reply (Chap. 26, Rulinwashi or The Scholars, in 1750 AD)

‘Mother Wang says loudly, “(It’s) really such, okay, you come to that house to tell them and I will wait for your coming to give me a reply.” ’

64

Beeching and Wang

Hao le could occur at the end of one’s message (the right periphery), as in (15): (15) 范進看了眾人，說道：“我怎麼坐在這裡？” 又道：“我這半日昏昏沉沉， 如在夢裏一般。” 眾鄰居道：“老爺，恭喜高中了！ 適才歡喜的有些引動了痰， 方才吐出幾口痰來，好了。

(儒林外史第3回, 1750 AD)

Fanjin kan le zhong ren, shuodao, “wo zeme zuo zai  zheli?” Fanjin see ASP all people say 1SG  why sit  to here You dao, “wo zhe banri hunhunchenchen, again say 1SG this half day murky ru zai meng li yiban.” like in dream LOC as such zhong linju  dao, “Laoye,  gongxi gaozhong  le!” all neighbours say Master congratulations high pass PRT shicai huanxide youxie yindong le  tan, just now happily some  cause ASP phlegm fangcai tu chu ji kou tan  lai,  hao le.” just now vomit out several CL phlegm come  H AO PRT (Chap. 3, Rulinwashi or The Scholars, in 1750 AD) ‘Fanjin sees all people and says, “why am I sitting here?’ and says “On the half day, I feel murky as in a dream.” All of the neighbours say, “Master, congratulations. You’ve got such a good grade in the national exam that there’s sputum in your lungs and now you vomit some (phlegm). It’s okay.” ’

On the other hand, according to L.Y. Wang (2004), although the agreement/ acceptance marker hao is observed as early as in Late Middle Chinese, its other discourse functions such as markers of mitigation, rebuttal, pre-closing, and topic transition, are not observed until Modern Chinese. Shao and Zhu (2005) also looked at the grammaticalisation of hao and divided its discourse functions into three types: active answering function (appraisal, promise, and affirmation), passive answering function (acceptance of indirect refusal, concession, and irony), and discourse cohesion function (discourse boundary marking, transition, and closure). They also discussed the different functions of hao de (promise given from a speaker of higher status to one of lower

Motivations for meaning shift at the left and right periphery

65

status), hao a (appraisal), hao ba (reluctant agreement) and hao le (concession or pre-closure). Shao and Zhu observed that hao as a PM appeared very frequently in the classic novel Xi You Ji (Journey to the West) during the Ming Dynasty (16th AD). They also noted that the active answering function, such as appraisal, promise, and affirmation could be found, but the other two functions, i.e. passive answering function, such as acceptance of indirect refusal, concession, and irony, and discourse cohesion function, such as discourse boundary marking, transition, and closure, could not be found, which is consistent with L.Y. Wang (2004)’s findings. Shao and Zhu found that several cases of PM hao conveying appraisal and promise also appeared in Shui Hu Zhuan (Outlaws of the Marsh) in the Ming Dynasty. However, we found that hao could be used to convey a highly positive evaluation, as in (16): (16) 這店主人吃了這一掌， 打得麻了，動撣不得， 自入屋後去躲避了。 武行者道：「好呀！你們都去了， 老爺卻吃酒肉！」把個碗 去白盆內舀那酒來， 只顧吃。 (水滸傳第32回, 1290 AD–1365 AD) Zhe dian zhuren chi le zhe yi zhang, This store host eat ASP this one slap da de ma  le, dongtan bu  de, slap DE paralyed PRT move NEG DE zi ru wu  hou qu duobi le. self enter house  LOC go hide PRT Wuxingze dao, “hao ya, nimen  dou  qu le, Wu Song say HAO  PRT  2PL all go  PRT Laoye que chi jiu rou. Ba ge wan Lord but eat wine mean take CL bowl qu baipen nei yao na jiu lai, go  whit basin LOC scoop that  wine come zhi gu  chi. just keep an eye on eat ‘The host, his face paralyzed from the slap, could scarcely walk. He went and hid himself in a rear room. “So you’re

66



Beeching and Wang all gone! Fine,” said Wu Song. “Now I can do some eating and drinking!” He scooped a bowlful of wine from the large white basin.’ (Chap. 32, Outlaws of the Marsh, 1290–1365 AD)     (source: http://www.silkpagoda.com)

Prior to (16), after the hero, Wu Song, came to a tavern for meat and wine, he wanted to order some more meat. But the tavern keeper lied and said there was none left. Wu Song slapped him and the inn keeper hid. Wu Song uses hao to express his satisfaction that, now that they had all gone, he could do some eating and drinking. Hao could also be used as a marker of concession in early Modern Mandarin, as manifested in (17): (17) 今日難得相逢，我又素昔佩服你的， 我想你應該憐惜我，同我談談； 你偏急著要走，怎麼教人不難受呢？” 老殘道：“好，好，好！我就陪你談談。”

(老殘遊記第13回, 1903 AD)

Jinri nande xiangfeng, wo you suxi peifu ni  de, Today not easy meet 1SG also usually admire 2SG DE Wo xiang ni yinggai lianxi wo, tong wo tantan; 1SG think 2SG should have pity 1SG with 1SG talk Ni pian jizhe yao zou, 2SG only hurry want leave zeme jiao ren bu nansou ne? how come make people not sad Question Particle LaoCan dao, “Hao, hao, hao! Wo jiu pei ni tantan.” LaoCan say HAO HAO HAO 1SG ADV accompany 2SG talk (Chap. 13, LaoCanYouJi or Tramp Doctor’s Travelogue, in 1903 AD, in Late Qiang Dynasty) ‘Today it’s very special that we can meet. And I usually admire you. I think you should have a pity on me to talk with me. But you want to leave in a hurry. How come I would not be sad? LaoCan says, “Okay, okay, okay! I will stay to talk with you.” ’

Motivations for meaning shift at the left and right periphery

67

In this excerpt, hao is repeated three times by LaoCan, which indicates that he is making a concession. Like French bon (Beeching 2009), it is employed as a marker of pseudo-agreement by the speaker to close down a troubling topic in conversation. Such a diminished sense of acceptance which does not involve any degree of positive evaluation might evolve from the use of hao as a marker of acceptance within the context in which it occurs (Wang and Tsai 2005). Towards a Cross-Linguistic Semantic Map of bon, hao and well

5

In Figure 3.1, we have attempted to adapt Haspelmath’s (2003) semantic map approach to capture the similarities and distinguishing features in the polysemies of hao, well and bon. All three items share the dialogual and dialogic features of (partial) agreement and concession—hao has, in addition, a verbal usage while bon has a narrative stage-marking function. Well appears to be more highly pragmaticalised as it is no longer possible to respond well and signal acceptance—it is always interpreted as signaling a dispreferred response or upcoming discourse incoherence. Compared with well, which can be used as a marker of demurral, both bon and hao alone do not have such function; however, when they co-occur with other elements, e.g. bon occurring with mais and hao with le, they can indicate demurral. It is these partially overlapping polysemies which make such expressions difficult both to translate and for non-native speakers to master. Well Demurral (ftm) Hao Concession (Pragmatic Marker) “Like”/“Love” (Transitive V)

“Good” (Predicative Adj)

Agreement (Pragmatic Marker)

Stage-marking (Pragmatic Marker) Bon

Figure 3.1 Semantic map of well, bon and hao

Dialogic

68

Beeching and Wang

Figure 3.1 shows the cross-linguistic polysemies of well, bon and hao. The degree of dialogicity (polyphony) is: well > bon > hao. Bon and hao may develop a more coded demurral function, so dotted lines in Figure 3.1 are used to indicate this tendency. 6

Terms at the Left and Right Periphery

In response to one of the LP/RP questions posed by this volume, that is, whether terms recruited to the LP are generally textual and terms recruited to the RP are more modal and interpersonal, the answer with respect to bon/ hao/well in contemporary spoken French, Chinese and English is mainly ‘no’. Left-peripheral bon, hao and well in naturally-occurring conversation are generally response markers and are arguably intersubjective in having the ability to disagree with an addressee while appearing, at least provisionally, to agree or accept what has been said. Table 3.5 shows that bon occurs in LP, RP and medially, hao in LP and RP in the construction hao le while well can occur in LP and medially but not in RP. Table 3.5 LP, Medial and RP positions of bon, hao and well    Position in  utterance

  LP     Medial

   RP

DM

Bon

√

√ (bon . . . mais; parce que bon)

Hao Well

√ √

× √ (well . . . but; because well)

√ (end marker; stage-marking) √ (hao le) ×

The semantic core of these terms is of course evaluative—suggesting ‘OK’. LP bon and hao indicate primarily acceptance (but can be used to save face by apparently accepting). LP bon can be used to ‘stop’ the previous speaker and to indicate a change of turns, thus it can be used dialogually. LP well is never used to indicate acceptance in present-day English, it indicates at least partial non-acceptance. It can be considered to be a face threat mitigator as it cushions a dispreferred response and is thus more modal than textual. In a

Motivations for meaning shift at the left and right periphery

69

medial position, bon and well introduce a concession in the ongoing discourse completed by mais/but. This is monologual dialogism and could be considered to be a way of modalising or mitigating, introducing a second viewpoint. Both bon and well can appear medially after coordinating conjunctions and serve a modalising, pause-filling or hesitatory function. Finally, well never appears on the RP. When bon appears on the RP, it has a narrative stage-marking function—in other words, speakers stop themselves, rhetorically adopting a form of monologual dialoguality. In this case, there is no evaluation, it is simply a mot de la fin. Hao can appear on the RP but only with the particle le—hao le— as both a positive evaluation and end marker. These findings are of course influenced by the core semantics of the terms themselves which are intrinsically evaluative. It seems, however, that the LP can be used to react dialogically as well as dialogually to the previous turn and that RP can be used dialogically as well as dialogually in anticipation of the following turn. Turn initial and medial positions seem to be most commonly used for the modal and interpersonal functions of ‘acceptance’ words, at least for English and French, not turn-final positions. Historically, the use of terms on the LP dialogually precedes their use monologually—in monologue, they can then be used rhetorically in both dialogual and dialogic ways. This appears to confirm hypotheses with respect to the influence of interactional factors in historical syntax and semantics and provides tentative evidence that RP elements which express modality and intersubjectivity may well have begun life as connectives or response markers on the LP. References Beeching, Kate. 2005. Politeness-induced semantic change: the case of quand même. Language Variation and Change 17/ 2: 1–27. ———. 2007. A politeness-theoretic approach to pragmatico-semantic change. Journal of Historical Pragmatics 8/1: 69–108. ———. 2009. “Sociolinguistic factors and the pragmaticalization of bon in contemporary spoken French”. In K. Beeching, N. Armstrong & F. Gadet (eds.), Sociolinguistic Variation in Contemporary French. Amsterdam/Philadelphia: Benjamins, 215–229. Brémond, Capucine. 2004. La petite marque bon, l’indice d’un accord en cours de négociations. Travaux de Linguistique 48: 7–19. Brinton, Laurel. 2006. “Pathways in the development of pragmatic markers in English”. In A. van Kemenade & B. Los (eds.), The Handbook of the History of English. Blackwell Publishing Ltd., 307–334.

70

Beeching and Wang

Dong, Tong-He. 1977. Chinese Phonology. Wenshizhe [Humanities] Press, Taipei. Defour, Tine. 2008. The speaker’s voice. A diachronic study on the use of well and now as pragmatic markers. English Text Construction 1/1: 62–82. Finell, Anne. 1989. Well now and then. Journal of Pragmatics 13: 656–6. Haspelmath, Martin. 2003. “The geometry of grammatical meaning: Semantic maps and crosslinguistic comparison”. In M. Tomasello (ed.), The New Psychology of Language, Vol. 2. Mahwah, N.J.: Lawrence Erlbaum Associates, 211–242. Jenner, William J.F. 1984. “Translator’s Afterword”. In W.J.F. Jenner (transl.), Journey to the West. Vol. 4. Seventh Edition. Beijing: Foreign Languages Press, 2341–2343. Jing-Schmidt, Zhuo. 2010. From positivity to possibility, propriety and necessary. Chinese Language and Discourse 1/1: 66–92. Jucker, Andreas. 1995. The discourse marker well: a relevance theoretical account. Journal of Pragmatics 19/5: 435–452. ———. 1997. The discourse marker well in the history of English. English Literature and Linguistics 1: 91–110. Li, Jinxia. 2005. The grammaticalisation and subjectivity of hao. World Chinese Teaching 1: 44–49. Owen, Marion. 1979. Units of natural conversation. Pragmatics Microfiche 3.5 Pomerantz, Anita. 1984. “Agreeing and disagreeing with assessments: Some features of preferred/dispreferred turn shapes.” In J. Atkinson & J. Heritage (eds.), Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press, 57–101. Roulet, Eddy & al. 1985. L’articulation du Discours en Français Contemporain. Bern: Peter Lang. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge University Press, Cambridge. Shao, Jingmin and Xiaoya Zhu. 2005. The discourse functions of hao and its evolution toward functional usage. Zhongguo Yuwn [Chinese Language] 5: 399–407. Shi, Yuzhi. 2003. Xiandai Hanyu Yufa Xitong de Jianli: Dongbu Jiegou de Chansheng ji qi Yingxiang [The Establishment of Modern Chinese Grammar: The Formation of Resultative Construction and its Effects]. Beijing: Beijing Yuyan Daxue. Traugott, Elizabeth Closs. 1982. “From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization”. In W.P. Lehmann & Y. Malkiel (eds.), Perspectives on Historical Linguistics. Amsterdam & Philadelphia: John Benjamins. ———. 2010. “Dialogic contexts as motivations for syntactic change”. In W. Kretschmar, A.-M. Hamilton-Brehm & R.A. Cloutier (eds.), Variation and Change in English Grammar and Lexicon. Berlin: Mouton de Gruyter, 11–27. Traugott, Elizabeth Closs & Richard Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press.

Motivations for meaning shift at the left and right periphery

71

Wang, Hsiao Ling. 2005. The Grammaticalization of Hao in Mandarin Chinese. M.A. Thesis. Graduate Institute of Linguistics at National Tsing Hua University, Hsinchu, Taiwan. Wang, Li Ya. 2004. The Polysemy and Grammaticalization of hao in Mandarin Chinese. Unpublished M. A. thesis. Taipei: National Taiwan Normal University. Wang, Yu-Fang and Pi-Hua Tsai. 2005. Hao in spoken Chinese discourse: relevance and coherence. Language Sciences 27: 215–243. Wang, Yu-Fang, Pi-Hua Tsai, David Goodman & Meng-Ying Lin. 2010. Agreement, acknowledgement, and alignment: The discourse-pragmatic functions of hao and dui in Taiwan Mandarin conversation. Discourse Studies 12/2: 241–267. Watts, Richard J. 1987. Relevance in conversational moves. A reappraisal of well. Studia anglica Posnaniensa 19: 37–59. Wei, Pei Chuan. 1994. Guhanyu Beidongshi De Fazhan Yu Yanbei Jizhi [The Development and Mechanisms of Passive Construction in Old Chinese]. Zhonghuo Jingnei Yuyan Yi Yuyanxue [the Dialects and Linguistics in Main China] 2: 293–319. Zhang, Ding and Haiyan Ding. 2009. The grammaticalization of the auxiliary hao and its related phenomenon of lexicalization. Language Teaching and Research 5: 31–37.

chapter 4

On the Function of the Epistemic Adverbs Surely and No Doubt at the Left and Right Peripheries of the Clause Elizabeth Closs Traugott 1 Introduction My aim in this paper is to address Beeching and Detges’ research question “What are the respective roles of LP and RP for the rise of polyphonous items?” and their hypothesis that LP and RP are asymmetric. The hypotheses most relevant to this paper may be summarized as in Table 4.1. My prime focus is on testing the last hypothesis, that items at LP are likely to be subjective, those at RP intersubjective, but the other hypotheses are addressed as well. Discussion is based on evidence from the development of two epistemic adverbs: surely and no doubt in English (see also Traugott 2012). Epistemic adverbs were chosen because epistemic modality in language use is wellknown to be at least partially speaker-oriented and subjective. Like other epistemic adverbs such as truly, surely and no doubt came to be used at LP and RP positions as what Biber and Finegan (1988) call “stance adverbials”. As such Table 4.1

Hypothesized usages of linguistic items on the left and right periphery

LP

RP

Dialogual Link to previous discourse Turn-taking/attention-getting Response-marking Subjective

Dialogic Anticipation of forthcoming discourse Turn-yielding/end-marking Response-inviting Intersubjective

* Many thanks to Sebastian Hoffmann for very helpful comments on a draft of this paper.

© koninklijke brill nv, leiden, ���4 | doi ��.��63/9789004274822_�05

On the function of the epistemic adverbs surely and no doubt

73

they comment on the discourse and therefore serve metadiscursive functions. They are particularly interesting for testing Beeching and Detges’ hypotheses because several authors have identified surely as seeking confirmation, an addressee-oriented, intersubjective use (e.g. Greenbaum 1969, Biber and Finegan 1988, Swan 1988, Downing 2001, Aijmer 2002). Furthermore, SimonVandenbergen (2007: 10) suggests that surely and no doubt share diachronic similarities and both should be accounted for in rhetorical terms rather than in logical terms of epistemic beliefs and modes of knowing. For modal adverbs LP and RP are understood as preceding or following the argument structure of the clause, i.e. they are slots at the edge and “outside of” of the propositional core. They are not necessarily utterance initial or final, since they may in English and many others languages be preceded and followed by address terms,1 and preceded by connectives such as and and but, and interjections such as Oh! (initial) as in (1):2 (1)  a. ‘But surely you hint at a very grave charge?’ (1894 Hope, Prisoner of Zenda [CL 3]) b. ‘Oh, surely there isn’t any danger of that?’ (1891 Gissing, New Grub Street [CL 3])

Schematically the LP and RP positions can be represented as in (2), where X and Y are variables for material that may optionally precede or follow: (2)

(X)

LP

Argument Structure

RP

(Y)

Their particular range of uses raises questions about the generality of the correlations proposed by Beeching and Detges. I will call surely and no doubt used at LP and RP epistemic stance adverbials (EPAs), thereby side-stepping controversies concerning how to distinguish “discourse marker” and “pragmatic marker” (see e.g. Fischer 2006, Brems, Ghesquière, and Van de Velde 2012).

1 Address terms may also appear within LP and RP, as in: (i) “Surely, sir, you would not kill him?” (1910 Brebner, The Brown Mask [CL 3]) 2 Editions used in the corpora differ with respect to using single and double quotation marks. They are cited as in the corpora.

74 2

Traugott

Theoretical Background

I approach the study from a constructionalist point of view and assume that grammar is made up of form-meaning pairs or constructions (Goldberg 2006). Surely and no doubt are constructions, the meaning of which is to be discussed below.3 Meanings are modified (“modulated”) in different schematic slots. One schematic slot that has been attested from OE4 times on is an epistemic modal slot at LP meaning approximately ‘truly’ (cf. OE soðlice), to which a large set of originally non-epistemic constructions have been recruited over time (Swan 1988, Lenker 2010). Other subtypes of adverbs that appear in this slot include evaluative adverbs ( fortunately), subject-oriented adverbs (wisely), and speech-act adverbs ( frankly). The truth emphasizers (truly, assuredly, certainly) at LP are subjective, meaning ‘certainly in my opinion, I infer’. They are used to evaluate the following discourse and function “pragmatically as speaker comments” on the upcoming discourse (Swan 1994: 249). Many of these same epistemic adverbs also occur in clause-internal, medial (Med) position, where they function as emphasizers of the truth of the proposition. Here they have scope over the clause but not the discourse. They are preferred as expressions of generally known truth (‘it is necessary that/it can’t not be that’ shading into ‘everyone knows’ and more subjective ‘I infer’). With respect to RP, in ME there are a few examples of post-clausal quotation expressions (he/she said, quod he/she in Chaucer) and of interjections such as la sentence-finally. Hoffmann’s (2006) earliest example of a tag is in a late fifteenth century play (see also Tottie and Hoffmann 2009), and Lenker (2010: 196) finds examples of post-causal adverbial connectors like however first in plays at the turn of the seventeenth to eighteenth centuries. Cumulatively, these suggest that development of a slot at RP for stance markers of various types was a relatively late development in English, beginning in the later ME period and expanding fairly rapidly, especially in conversation, during EModE. In sum, LP and Med are robustly attested as slots for ESAs throughout the history of English. However, RP as a slot for ESAs, adverbial connectors, quotation expressions and tags appears to be attested only from late ME on. 3 In Croft (2001) and Goldberg (2006) “construction” is understood to be of any size from complex schema to word to affix. The relevant criterion for construction is form-meaning pairing, not sequence in a constituent as in some other models such as Sign-based Construction Grammar (e.g. Sag 2012), therefore constructions can be monomorphemic. 4 The following abbreviations for periods of English are used: OE (Old English), c. 650–1150, ME (Middle English), c. 1150–1500, Early Modern English (EModE), c. 1500–1700, and PDE (Present Day English), c. 1970–.

On the function of the epistemic adverbs surely and no doubt

75

Much of the work on stance markers, coherence relations, and speech representation has been conducted in the context of work on information structuring, with focus on clause-initial marking (e.g. Defour 2008, Moore 2011), hence our currently rather poor knowledge of the history of clause-final markers. It has also been conducted in the context of work on grammaticalization (see e.g. Brinton 2008, Lenker 2010). The question has been raised whether such markers are truly “grammatical” because they are “pragmatic” and inferential in function, and it has been proposed that they involve “pragmaticalization” rather than grammaticalization (see Erman and Kotsinas 1993 and several researchers since then). The question assumes an architecture of grammar that is based in propositional argument structure, into which pragmatic and disjunctive structures like discourse markers cannot be easily integrated. However, discourse markers and markers of coherence relations are part of grammar understood as the linguistic structure of a language. The onset of grammaticalization has long been known to be motivated by pragmatic inferencing (e.g. Traugott 1988), and indeed has recently been construed as “frozen pragmatic anchoring” (Diewald 2011: 461). Therefore markers said to have undergone “pragmaticalization” are at the extreme end of a continuum from less to more pragmatic grammatical marking and there is no need for a distinction to be made between grammaticalization and pragmaticalization. Furthermore, construction grammar makes no fundamental distinction between elements that are primarily semantic or primarily pragmatic. From the perspective of historical construction grammar, the development of ESAs, pragmatic markers, and connectives in general is a case of grammatical constructionalization, the development of constructions with procedural function (Traugott and Trousdale 2013). As procedurals they contribute information about the speaker/author’s stance toward the proposition, and guide the hearer/reader as to how to combine a conceptual representation of the proposition with that stance (see Terkourafi 2011: 358–359). Among stances that I will discuss is dialogicity. Beeching and Detges interpret this as potentially different viewpoints that may arise when yielding a turn. In what follows I distinguish two types of dialogicity: i) that which arises in turn-taking and ii) “textual argumentative dialogicity”, a strategy where the same person may construct different points of view for rhetorical purposes (Schwenter 2000, White 2003). 3

Data and Methodology

The data were extracted by egrepping and were counted manually from three corpora: a) Corpus of Early English Correspondence (CEEC), 1403–1680, nearly

76

Traugott

2,600,000 words, b) the Early Modern English section of the Helsinki Corpus (HC), about 550,000 words, and c) the Corpus of Late Modern English Texts, Extended Version (CL, short for CLMETEV) totaling about 15 million words. Whereas the HC corpus represents a variety of texts, CL represents mainly literature, history, parliamentary debates, and philosophy. HC and CL are divided into three parts, seventy years apart. The three subperiods of HC are 1500–1570 (I), 1570–1640 (II), and 1640–1710 (III). The subperiods of CL are 1710–1780 (I), 1780–1850 (II), and 1850–1920 (III). Each subperiod of HC is about 150,000 words long, but subperiods in CL are of varied lengths: about 3,037,600 in I, 5,724,000 in II, and 6,251,600 in III. In the case of surely (also spelled suerly(e), surlie, surly), non-epistemic adverbial uses such as appear in she worked slowly but surely were not counted. In the case of no doubt (also spelled no dout(e)), nominal uses were not counted, including instances of There BE no doubt, I HAVE/MAKE no doubt, etc. In a very few cases, usually after a verb of thinking and before a that-clause where a modern interpretation might be that surely or no doubt has semantic scope over the following clause rather than over what precedes, the ESA was coded as at RP, i.e. according to the syntax. Surely and no doubt were considered to be at LP when they follow connectives (and, but, because, although), exclamatives (why), discourse markers (well), address forms (Madam), and connective (as opposed to temporal) then. They were coded as Med after dislocated clause adjuncts even though surely precedes the subject, e.g.: (3) for I will at an any time give up a good deal of regularity for a great deal of brillant (‘brilliancy’); and for the brillant surely nobody is equal to Voltaire. (1752 Chesterfield, Letters to his son [CL 1])

The raw counts of examples in the different positions in the clauses of the three corpora used are given in Tables 4.2–4.4.5 Counts of the data are raw because instances at RP are too few and unevenly distributed across text types for normalization to be meaningful. For example, 17 instances of no doubt at RP in CL 3 are from the works of Gissing (1891, 1893), and 11 from Galsworthy (1906). In most cases surely and no doubt are at LP or RP of a clause with finite verb. In some cases, however, they are at RP of a phrase, especially in elliptical responses and parentheticals, e.g.:

5 One text that appears in CEEC and HC was included only in HC. Percentages in Table 4.4 are rounded, therefore not all total 100% exactly.

On the function of the epistemic adverbs surely and no doubt

77

(4) a. ‘What kept her up? Not fear of the thunder, surely? That was over hours since.’ (1847 E. Brontë, Wuthering Heights [CL 2]) b. ‘The author,’ remarked Mivain, ‘ought to make a good thing out of this.’ ‘Will, no doubt. Ought to write at once to the papers.’ (1891 Gissing, New Grub Street [CL 3]) Table 4.2 Surely and no doubt in CEEC

Surely No doubt

LP

Med

RP

Total

36 1

15 2

0 0

51 3

LP

Med

RP

Total

13 17 5

4 6 3

0 0 0

17 23 8

0 3 1

3 3 1

0 1 0

3 7 2

Table 4.3 Surely and no doubt in HC

Surely I II III No doubt I II III

Table 4.4 Surely and no doubt in CL

Surely I II III No doubt I II III

LP

Med

RP

Total

142/44.7 335/58.0 440/59.5

170/53.5 223/38.9 252/34.0

6/1.9 20/3.5 48/6.5

318 578 740

25/18.7 60/27.6 218/40.2

105/78.4 120/55.0 233/43.0

4/3.0 38/17.4 91/16.8

134 218 542

78

Traugott

The Tables show that surely has been favored at LP, except in CL 1, and that there has been a slight increase in use at RP since 1780. As discussed below this correlates with increased (inter)subjective use. No doubt, by contrast, has been preferred at Med, although increase at LP is evidenced in CL 3. Use at RP increased in the second period of CL and has remained stable. The focus of the present volume is on the role of interactional features in semantic change. Tapping interaction in written historical texts is problematic because interaction is always represented. However, (parts of) some text types may be considered fairly close to interactional speech and talk, for example trials, plays, and dialogue in fiction (Culpeper and Kytö 2010). HC contains examples of all three text types, but as indicated above the corpus is small, and the numbers of relevant instances is minimal. By contrast, CL contains a large number of fictional works that include dialogue. It also contains a number of philosophical works and histories which, although written as monologues, and somewhat distant from norms of talk, nevertheless in many cases involve dialogic argumentation pitting one point of view against another. They are therefore valuable texts for investigating the development of ESAs. In doing so it must be remembered that many of the texts are edited. Punctuation as we now know it, particularly quotation marks distinguishing direct from indirect speech representation, developed only during the EModE period (Parkes 1992, Moore 2011). Therefore in the case of earlier texts interactional cues may have been less distinct than some editions imply. 4

The Multifunctionality of Surely and No Doubt

Surely and no doubt were borrowed from French around 1300, the first as a manner adverb, the second as NP. Surely replaced an earlier Germanic form siker(ly) that appears in texts from the thirteenth through the nineteenth centuries (OED sicker Adj and Adv). Most of the original uses of surely and no doubt have persisted until the present day. However, both now express weaker epistemic certainty, and can even express uncertainty (Aijmer 2002). Here I outline four main uses for each construction: clause-internal argument, Med, LP, and RP. Of these only the last two are fully relevant to the present paper. Since all four positions have been available in English since ME times, RP being the most recent, what is at issue here is recruitment of the two constructions to the various functions and positions, not the emergence of such functions and positions.

On the function of the epistemic adverbs surely and no doubt

79

4.1 Four Main Uses of Surely Surely as a manner adverb has a number of meanings depending on context, but in early texts most pertain to safety, carefulness, and steadfastness (5a). Doing something carefully or being steadfast may in the contexts of religion and royal institution have implied truthfulness. Very early such an implicature appears to have been semanticized and surely most often appears with the meaning ‘in a truthful manner’, especially in the context of verbs of locution, hearing, etc. (5b).6 (5) a. if it be well and surely made (1534 FitzHerbert, Book of Husbandry [HC cehand1a]) b. And I am surely enformed that the said Lordes of Scotland woll kepe their promisse to the said Duc. ‘And I am trustworthily informed that the said Lords of Scotland will keep their promise to the said Duke.’ (1524 Dacre, Letter to Wolsey [CEEC])

In both these meanings surely is used in comparatives (as/so surely as, more surely than). None of these uses were counted since they are integrated arguments of the clause. The rise of epistemic adverb use is a constructionalization: the development of a new meaning with a different syntactic distribution. In early texts it is high on the scale of certainty. It is mainly used for known certainty, especially in religious contexts (6a) but also for more subjective certainty ‘I am certain’ (6b): (6) a. for even y=e= same daye thou eatest of it, thou shalt surely dye. ‘For even the same day as you eat it (the apple) you will certainly die.’ (c1530 Tyndale, Bible [HC ceotest1]) b. The count Hollock [\is\] surely a wise, gallant gentleman. (1586 Leicester, Letter to Cecil [CEEC leyceste])

Surely is attested at LP from the time of the earliest corpora investigated. Here it has a range of uses from a relatively objective interpretation of ‘certainly, in everyone’s view’ (7a) to a subjective epistemic evaluation ‘certainly in my

6 Association with epistemic certainty may have developed in French before borrowing occurred.

80

Traugott

opinion’ (7b), to an intersubjective plea for confirmation (7c), especially in later texts (see section 5.2): (7) a. For surely to dryve the tyme as we do, . . . doing but small hurt to the Kings enemyes, it is nothing to the honor of his Highnes. ‘For surely to spend the time as we do, . . . doing only small harm to the King’s enemies, is not to the honor of his Highness.’ (1524 Dacre, Letter to Wolsey [CEEC origina1]) b. ‘Verily no, my Lorde,’ quoth I, ‘. . . I shall my self se, that I shall neither be forsworne nor swere against my conscience. Surely as to swere to the succession I see no peril.’ “‘Truly no, my Lord’, said I, ‘. . . I shall myself see that I shall neither perjure myself nor swear against my conscience. Surely as regards swearing to the succession I see no danger.’” (1534 More, Letter to M. Roper [HC cepriv1]) c. “Surely you tried to save it, Marian?” “Yes,” I said, “the housekeeper and I both did our best.” (1859–60 Collins, Woman in White [CL 3])

At RP surely is rare, as the Tables show. Two early examples, one in fictional the other in philosophical writing, are: (8) a. there was never a girl of your innocence, that set a large family in such an uproar, surely.—But let that pass. (1740 Richardson, Pamela [CL 1]) b. Usefulness is agreeable, and engages our approbation. This is a matter of fact, confirmed by daily observation. But, USEFUL? For what? For somebody’s interest, surely. Whose interest then? Not our own only: For our approbation frequently extends farther. (1751 Hume, Enquiry concerning the Principles of Morals [CL 1])

4.2 Four Main Uses of No Doubt No doubt in early texts is used primarily in formulae like There BE no doubt, I HAVE/MAKE no doubt often followed by but (9a), and sometimes as an object in less formulaic expressions (9b): (9) a. For I make no doubt but yf it please her Highnes to resolve, her Majestie’s commission . . . will sufficientlie inhable him. ‘For I am sure if it please her Highness to decide this, her Majesty’s commission . . . will sufficiently enable him.’ (1596–7 Hutton, Letter to Cecil [CEEC])

On the function of the epistemic adverbs surely and no doubt

81



b. every man to ocupie ther owne farmes, notwithstanding the premysses: put ye no doubt therin. ‘every man to occupy their own farms, regardless of prior arrangements; have no doubt about it.’ (1489 E. Plumpton, Letter to R. Plumpton [CEEC])

No doubt was used from c1400 as an adverb with subjective epistemic meaning (i.e. as ESA), usually at Med (10). This was a constructionalization as new meaning and new syntax developed. (10) [verses] which he no doubt did write upon his owne experience. (1627 Brinsley, Ludus Literarius [HC ceeduc2a])

Although Med is the preferred position in all three CL subcorpora, use at LP is attested robustly as well, especially in later periods. Here it has scope over whatever follows, from phrase to simple (11a) or complex clause (11b): (11) a. Dan. You tell of some, which haue receiued help from the hands of cunning men: And no doubt there may infinit examples be brought. (1593 Gifford, Handbook on Witches [HC cehand2a]) b. and no doubt, in some respects, it is very specious, and has more probability than the common hypothesis, by giving a plausible account of the strange mixture of good and ill which appears in life. (1779 Hume, Dialogues concerning Natural Religion [CL 1])

Use at RP is rare, but early examples appear in HC (12a). At RP no doubt may also follow a phrase rather than a clause (12b). It often occurs in ellipsis (12c) ((4b) in section 3 above, repeated for convenience): (12) a. I, quoth the king: they will be wise ones, no doubt. ‘Yes,’ said the king, ‘They will be wise ones, no doubt.’ (1608 Armin, Nest of Ninnies [HC cefict2a) b. The servant replied . . . “he did not know; but that the man looked like a sailor, and had a boy with him.” “A begging letter, no doubt,” cried Lady Clementina. (1796 Inchbald, Nature and Art [CL 2]) c. ‘The author,’ remarked Mivain, ‘ought to make a good thing out of this.’ ‘Will, no doubt. Ought to write at once to the papers.’ (1891 Gissing, New Grub Street [CL 3])

82

Traugott

At RP no doubt expresses speaker assessment of the prior discourse. There is no evidence in the corpora investigated that at right periphery no doubt “has the status of an afterthought” (Simon-Vandenbergen 2007: 14). Use as an afterthought may be a further development or may simply not be represented in writing. 5

Testing the Hypotheses with Surely and No Doubt in their ESA Uses

I now turn to the question of how evidence from the distributional uses outlined in section 4 may be used to test Beeching and Detges’ hypothesis regarding the asymmetry of (inter)subjectivity. I begin with hypotheses concerning coherence and what I call “textual argumentative dialogicity” (5.1), and then discuss (inter)subjectivity (5.2) and turn-taking (5.3). An additional hypothesis put forward in Traugott and Dasher (2002) that subjectification precedes intersubjectification historically is discussed in 5.4. 5.1 Coherence and Textual Dialogicity Given that speakers and authors generally attempt to make discourse coherent, and that if coherence is linguistically marked it is by hypothesis usually best marked at the beginning of an utterance or clause, the question arises whether surely and no doubt play a specific role in marking this coherence. Lenker (2010: 117–122) argues that in OE the epistemic adverb soþlice ‘truly’ (ME soothly) was used as an episode marker. This appears to have been a feature of soþlice and of earlier English narrative. It is not a characteristic of surely, no doubt, or of many other adverbs with sources in truth and fact that Lenker studies. However, she mentions another kind of more local coherence factor, collocations with connectives like and, but, for. Such collocations abound with surely (7a) and to a lesser extent with no doubt (11b). Lenker suggests such uses of adverbials serve “transition” functions (p. 116). She further points out that adverbs in medial “post-first-position”, which includes post-subject, also serve this kind of transition function (p. 237). This hypothesis seems intuitively correct with respect to surely and no doubt, but is a topic for further study. Surely does not appear to have a significant coherence function in terms of linking discourses to each other in earlier texts, but in later ones its use in questions triggers responses (see 5.2). In older texts it has mainly an emphasizing function, foregrounding the speaker/author’s stance toward the content of the clause which it introduces and over which it has scope. This is particularly clear in overtly dialogic contexts, e.g. those introduced by but, yet, and

On the function of the epistemic adverbs surely and no doubt

83

with content that is marked by negative polarity (not, no) (13a) or a semantically negative lexical item (13b). Emphasis may, however, be also a factor when introducing affirmative content (13c): (13) a. and yet SUERLY your frendes here did not omit any opportunite. ‘And surely your friends here did not miss any opportunity.’ (1586 Burghley, Letter to Leycester [CEEC]) b. and SUERLY I am gretly discoraged with lack of hir resolutions. ‘and surely I am greatly discouraged by the lack of her decisions.’ (1585 Burghley, Letter to Leycester [CEEC]) c. “Surely,” said the Princess, “this man is happy.” (1759 Johnson, Rasselas [CL 1])

Whereas surely may introduce opposition in the clause over which it has immediate scope, no doubt usually projects that opposition will be developed in a more distant clause. It therefore has a concessive ‘although’ function, often signaled by another clause beginning with but. Simon-Vandenbergen (2007) notes this as a frequent feature in PDE, and says of this use that the speaker/ author concedes the truth “in order to posit the counter-argument in a context of dialogic argumentation” (p. 16). She also notes that this concessive use often conveys more certainty than other contexts (p. 30). This dialogic text structuring occurs in some earlier texts and increases over time in CL. Examples include: (14) a. All that can be said against the French sublime . . . is this:—That the grandeur is MORE in the WORD, and LESS in the THING. No doubt, the ocean fills the mind with vast ideas; but Paris being so far inland, it was not likely I should run post a hundred miles out of it, to try the experiment. (1768 Sterne, Sentimental Journey [CL 1]) b. “No doubt the poor girl was horrified, but her daring did not forsake her, and she there and then made up her mind to escape at all risks”. (1885 Blind, Tarantella [CL 3])

We may conclude from these brief comments that textual argumentative dialogicity occurs extensively at LP. It may also occur at RP, however: (15) a. “Such teaching was well meant, no doubt,” said Arthur; “but it must have driven many of its victims into deserting the Church-Services altogether”. (1889 Carroll, Sylvie and Bruno [CL 3])

84

Traugott b. One thing only is clear, That Cimmerian Europe is girdling us in; that France (not actually dead, surely?) cannot march. (1837 Carlyle, French Revolution [CL 2])

From a textual point of view, do the ESAs provide evidence of links to previous discourse at LP and of anticipating upcoming discourse at RP? There is little evidence in the data that either is used in any anaphoric function, linking clauses to previous discourse. However, as stance adverbs at LP, both project that the upcoming clause is of some noteworthiness or importance (Auer 2005, Günther 2011), and furthermore that adversative content may be upcoming. As will be shown in the next two sections, RP surely additionally invites a response, whereas no doubt does not. 5.2 Subjectivity and Intersubjectivity Stance markers are by definition subjective—they express the speaker/ author’s point of view. There is a continuum of subjectivity from minimally subjective epistemic necessity (grounded in belief systems) to maximally subjective comment. In many cases it is unclear from context where on this continuum a particular use at LP is placed. In (16) Sir Thomas More reports what the Secretary (whom More considers a friend) said to him by way of warning in an interview with the Chancellor (an adversary). More interjects his own personal judgment about the Secretary’s son with the epistemic stance adverb of trouth; the following surely is part of what the Secretary is reported to have said. It could be interpreted as the Secretary’s personal opinion, or as something obvious to all; given the adversarial context of the meeting, it may well have been intended as the latter: (16) Vpon this Maister Secretary . . . saide and sware a gret oth, that he had leuer that his owne only sonne (which is of trouth a goodly yonge gentilman . . .) had lost his hedde, than that I shoulde thus haue refused the oth. For surely the Kynges Highnes wolde now conceiue a great suspicion against me. ‘Upon this Master Secretary . . . said and swore a great oath, that he would rather that his own only son (who is truly a handsome young gentleman . . .) had lost his head than that I should have refused to say the oath. For surely the King’s Highness would now have great suspicion against me.’ (1534 More, Letter to M. Roper [HC cepriv1])

Particularly striking about the difference between surely and no doubt is their orientation, no doubt inward to the speaker/author’s private conclusion, surely

On the function of the epistemic adverbs surely and no doubt

85

outward to the world in general and to the addressee/reader in particular (Traugott 2012). This can be seen most clearly in two passages in the data where no doubt occurs with surely: (17) a. ‘Did I move them, Gashford?’ said Lord George. ‘Move them, my lord! Move them! They cried to be led on against the Papists, . . . they roared like men possessed—’ ‘But not by devils,’ said his lord. ‘By devils! my lord! By angels.’ ‘Yes—oh surely—by angels, no doubt,’ said Lord George, thrusting his hands into his pockets, . . . and looking uncomfortably at the fire. ‘Of course by angels—eh Gashford?’ (1841 Dickens, Barnaby Rudge [CL 2]) b. Whereupon, he wished me a good morning and withdrew, disconcerted and offended, no doubt; but surely it was not my fault. (1848 A. Brontë, Tenant of Wildfell Hall [CL 2])

In (17a) Lord George starts by agreeing without much thought to what his secretary Gashford has said, then thinks about it, and internalizes Gashford’s positive rather than his own negative view of the event. In (17b) the narrator gives her conclusion about Mr. Boarham’s thoughts, and then reorients her protest it was not my fault to whoever will pay attention, in this case the imagined reader. No doubt is inherently inward-oriented and normally subjective at both LP and RP. Surely is inherently outward-oriented and likewise normally subjective at both LP and RP. It can also be used intersubjectively in both positions, especially with second person subjects. In (18a) surely may absorb intersubjectivity from the interjection why, the address man, and the second person subject, which all imply orientation to the addressee, but (18a) also suggests that the “fighting words” use that Downing (2001: 165) identifies in PDE uses of surely with second person is quite old. In (18b) Chesterfield clearly wants his son to assent (compare no doubt in place of surely—no doubt would mean ‘I conclude X’ rather than ‘I am certain and ask your confirmation’). (18) a. Why, surely, man, thou forgettest whom thou talkest to. (1740 Richardson, Pamela [CL I]) b. Are you pleased with, and proud of the reputation which you have already acquired? Surely you are, for I am sure I am. Will you do anything to lessen or forfeit it? Surely you will not. And will you not do all you can to extend and increase it? Surely you will. (1749 Chesterfield, Letters to his Son [CL 1])

86

Traugott

Perhaps the best evidence for the difference in orientation can be found in the presence of surely but absence of no doubt in questions. In the data questions are always in non-inverted statement form. They are outward-oriented and answered, even in dialogic monologues (where they are rhetorical questions), or the author specifically remarks that the response was silence. There are two examples of surely in questions in CL1 both with negative polarity, 33 in CL 2, of which 28 have negative polarity, and 40 in CL 3, of which only 17 have negative polarity (see (4a) above and (19) below with negative polarity, (7c) above with positive polarity). The shift to positive polarity suggests expansion of the intersubjective use of surely. Use in the different positions is summarized in Table 4.5 (percentage is of total occurrences in the subperiod): Table 4.5 Use of surely in rhetorical questions in CL CLMETEV

LP

Med

RP

Total

%

I II III

1 17 34

1 9 3

0 7 3

2 33 40

0.6 5.6 5.3

Especially instructive for present purposes is the fact that surely appears at LP in most questions. In conclusion, where surely and no doubt are concerned, subjectivity or intersubjectivity appear to be correlated more closely with their inherent meaning and immediate context than with use at LP or RP. 5.3 Turn-taking Here I turn to the question whether surely and no doubt support the hypotheses that attention-getting and response-marking occur at LP, end-marking and response-inviting at RP. I limit my examples to those that occur in represented conversation, although they are also used in monologues involving dialogic debate, where the authorial self is imagined as divided into two interlocutors with different opinions, and in letters where an answer may be expected, but not immediately. When they simply follow the end of a turn, they may be preceded by Oh! When surely is used at RP in a question and is thus end-marking and turninviting, it is used to seek corroboration, as in:

On the function of the epistemic adverbs surely and no doubt

87

(19) Poor, poor girl, is she the accused? But it is wrongfully; every one knows that; no one believes it, surely, Ernest?” “No one did at first; but several circumstances came out . . .” (1818 Shelly, Frankenstein [CL 2])

No doubt appears to disfavor up-take, being limited in the data to statements, sometimes sarcastic. In (20) Shelton responds to the question You don’t get dhrunk?, not to the comment with no doubt: (20) “You don’t get dhrunk, I suppose?” he said suddenly—“too much of ’n Englishman, no doubt.” “Very seldom,” said Shelton.

While both surely and no doubt can be used at LP in answer to questions or as follow-up of an interlocutor’s comment, this does not appear to be an important function. Nor do they appear to be used as attention-getters. For example, there are no instances of ‘Surely/No doubt’, she interrupted. On the contrary, there are a few examples of represented interruption, breaking off utterances starting with surely and no doubt, e.g.: (21) a. “Where is it?” said Glory. “Gorn, my dear.” “Surely you don’t mean—” “No, not dead, but I ‘ad to put it out, pore thing!” (1897 Caine, The Christian [CL 3]) b. “. . . but I know what he DID.” “And what did he?” “Nothing at all for the poor.” “If any of them applied to him, no doubt—” “Oh! they knew better than all that comes to; for if they asked for anything, he was sure to have them sent to Bridewell, or the workhouse.” (1796 Inchbald, Nature and Art [CL 2])

The ESAs are preferred within a larger discourse as in (21b) rather than in turntaking positions. 5.4 The Historical Order of Semantic Changes One of the research questions Beeching and Detges raise is whether subjective meaning precedes intersubjective as suggested in Traugott and Dasher (2002: 281):

88

Traugott

(22) non-subjective > subjective > interpersonal

This semantic cline is confirmed by surely, which derives from the manner adverbial ‘carefully, steadfastly’, was subjectified to an epistemic adverbial that scales the proposition with respect to the truth, and intersubjectified in the meaning ‘I am certain and I want you to agree’. No doubt meaning ‘no fear’ (a non-subjective meaning), appears to have been routinized with epistemic necessity meaning in There BE no doubt/I HAVE/MAKE no doubt. It is also used from ME on in the weaker sense of ‘there is no question’. It therefore supports the first part of the cline. However, speakers have not recruited no doubt to cue intersubjective meaning, so its development does not support the second part. In addition to the semantic cline in (22), Lenker (2010: 117) proposes another more structural one for expressions with the source domain TRUTH/FACT:7 (23) clause-internal adverbial > sentence adverbial > discourse particle

The cline in (23) models origin as a construction integrated in the argument structure of the clause as in (5a), change to an adverbial with scope over the clause (“sentence adverbial”) as in (6), and finally to use as a particle with scope that can extend over “discourse” (complex clause structure), as in (7a,b). The history of surely follows this trajectory, but that of no doubt is less direct since it originated in an NP, not an adverbial. While it was clause-internal in the sense that it is either a predicate nominal (There BE no doubt) or an object (I MAKE/ HAVE no doubt), the formula itself was the first in a complex clause construction. The EPA no doubt can function as a sentence adverbial, in other words, a chunk that in construction grammar terms is a fixed expression equivalent to a mono-word like surely, as in (11a), but as a pragmatic marker it can have scope over the following discourse (11b). As has often been pointed out (e.g. Bybee and Pagliuca 1987), earlier meanings tend to shape later use. In the case of no doubt, it appears that the original meaning ‘no fear’, an internal experience, and presumably continued association with doubt, a cognitive state, led to the restriction of no doubt to primarily inward-turning stance. The absence of collocations with negative polarity or semantic negativity may be attributed to the presence of no. If so, no has not been entirely bleached in the context of the construction. By contrast, the origins of surely in meanings related to security and steadfastness may 7 Lenker refers to (23) as a semantic cline, however.

On the function of the epistemic adverbs surely and no doubt

89

be reflected in the preponderance of outward-directed stance—one secures things against or is steadfast against something external. 6 Conclusion The main finding of this study of surely and no doubt is that they do not support a deterministic or even strong correlation between use at LP and RP and any of the factors Beeching and Detges propose. It remains to be investigated to what extent this is a characteristic of English, and of the late development of RP as a slot where interactional markers may occur. That it is not unique to English is suggested by the identification of intersubjective use of pragmatic markers of various types at LP in French (Degand and Fagard 2011) and Japanese (Onodera and Suzuki 2007). The findings suggest that while Beeching and Detges’ hypotheses doubtless pertain to effects of communicative interaction, we should not expect all constructions used at LP and RP to cue the hypothesized effects, at least in written representations of interaction. A distinction needs to be made between turn-taking as interlocutors’ practice, whether or not it is marked linguistically, and explicit marking of turn-taking (see also Moore 2011 on making vs. marking of speech turns).

Data Sources

CEEC Corpus of Early English Correspondence. 1998. Compiled by Terttu Nevalainen, Helena Raumolin-Brunberg, Jukka Keränen, Minna Nevala, Arja Nurmi, and Minna Palander-Collin. Department of English, University of Helsinki. http://www.helsinki.fi/varieng/CoRD/ corpora/CEEC/index.html. CL The Corpus of Late Modern English Texts (Extended Version) (CLMETEV). 2006. Compiled by Hendrik De Smet. Department of Linguistics, University of Leuven. http://www.helsinki.fi/varieng/ CoRD/corpora/CLMETEV/. HC Helsinki Corpus of English Texts. 1991. Compiled by Matti Rissanen (Project leader), Merja Kytö (Project secretary); Leena Kahlas-Tarkka, Matti Kilpiö (Old English); Saara Nevanlinna, Irma Taavitsainen (Middle English); Terttu Nevalainen, Helena Raumolin-Brunberg (Early Modern English). Department of English, University of Helsinki. http:// www.helsinki.fi/varieng/CoRD/corpora/HelsinkiCorpus/index.html. OED Oxford English Dictionary. http://www.oed.com/.

90

Traugott

References Aijmer, Karin. 2002. “Modal adverbs of certainty and uncertainty in an English-Swedish perspective”. In H. Hasselgård, S. Johansson, B. Behrens & C. Fabricius-Hansen (eds.), Information Structure in a Cross-Linguistic Perspective. Amsterdam: Rodopi, 97–112. Auer, Peter. 2005. Projection in interaction and projection grammar. Text 25: 7–36. Beeching, Kate & Ulrich Detges. This volume. “Introduction”. Biber, Douglas & Edward Finegan. 1988. Adverbial stance types in English. Discourse Processes 11: 1–34. Brems, Lieselotte, Lobke Ghesquière & Freek Van de Velde (eds.). 2012. Intersections of intersubjectivity. Special issue of English Text Construction 5. Brinton, Laurel J. 2008. The Comment Clause in English. Cambridge: Cambridge University Press. Bybee, Joan L. & William Pagliuca. 1987. “The evolution of future meaning”. In A. Giacalone Ramat, O. Carruba & G. Bernini (eds.), Papers from the 7th International Conference on Historical Linguistics. Amsterdam: John Benjamins, 109–122. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Culpeper, Jonathan & Merja Kytö. 2010. Speech in Writing: Explorations in Early Modern English Dialogues. Cambridge: Cambridge University Press. Defour, Tine. 2008. The speaker’s voice: A diachronic study on the use of well and now as pragmatic markers. English Text Construction 1: 62–68. Degand, Liesbeth & Benjamin Fagard. 2011. Alors between discourse and grammar: The role of syntactic position. Functions of Language 18: 29–56. Diewald, Gabriele. 2011. “Grammaticalization and pragmaticalization”. In H. Narrog & B. Heine (eds.), The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press, 450–461. Downing, Angela. 2001. ‘Surely you knew!’ Surely as a marker of evidentiality and stance. Functions of Language 8: 253–285. Erman, Britt & Ulla-Britt Kotsinas. 1993. Pragmaticalization: The case of ba’ and you know. Studier i Modernspråkvetenskap 10: 76–93. Stockholm: Almqvist and Wiksell. Fischer, Kerstin (ed.). 2006. Approaches to Discourse Particles. Amsterdam: Elsevier. Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press. Greenbaum, Sidney. 1969. Studies in English Adverbial Usage. London: Arnold. Günther, Susanne. 2011. “Between emergence and sedimentation: Projecting constructions in German interactions”. In P. Auer & S. Pfänder (eds.), Constructions: Emerging and Emergent. Berlin: De Gruyter, 156–185. Hoffmann, Sebastian. 2006. Tag questions in Early and Late Modern English: Historical description and theoretical implications. Anglistik 17: 35–55.

On the function of the epistemic adverbs surely and no doubt

91

Lenker, Ursula. 2010. Argument and Rhetoric. Adverbial Connectors in the History of English. Berlin: De Gruyter Mouton. Moore, Colette. 2011. Quoting Speech in Early English. Cambridge: Cambridge University Press. Onodera, Noriko O. & Ryoko Suzuki (eds.). 2007. Historical Changes in Japanese: Subjectivity and Intersubjectivity. Special issue of Journal of Historical Pragmatics 8 (2). Parkes, Malcolm B. 1992. Pause and Effect: An Introduction to the History of Punctuation in the West. Aldershot: Scolar Press. Sag, Ivan A. 2012. “Sign-based construction grammar: An informal synopsis”. In H.C. Boas & I.A. Sag (eds.), Sign-based Construction Grammar. Stanford, CA: CSLI Publications, 69–202. Schwenter, Scott A. 2000. “Viewpoints and polysemy: Linking adversative and causal meanings of discourse markers”. In E. Couper-Kuhlen & B. Kortmann (eds.), Cause—Condition—Concession—Contrast: Cognitive and Discourse Perspectives. Berlin: Mouton de Gruyter, 257–281. Simon-Vandenbergen, Anne-Marie. 2007. “No doubt and related expressions”. In M. Hannay & G.J. Steen (eds.), Structural-Functional Studies in English Grammar. Amsterdam: Benjamins, 9–34. Swan, Toril. 1988. Sentence Adverbials in English. Oslo: Novus. ———. 1994. “A note on Old English and Old Norse initial adverbials and word order with special reference to sentence adverbials”. In T. Swan, E. Mørck & O.J. Westvik (eds.), Language Change and Language Structure. Berlin: Mouton de Gruyter, 233–270. Terkourafi, Marina. 2011. The pragmatic variable: Toward a procedural interpretation. Language in Society 40: 343–372. Tottie, Gunell & Sebastian Hoffmann. 2009. Tag questions in British and American English. Journal of English Linguistics 34: 283–311. Traugott, Elizabeth Closs. 1988. “Pragmatic strengthening and grammaticalization”. In S. Axmaker, A. Jaisser & H. Singmaster (eds.), Berkeley Linguistics Society 14: General Session and Parasession on Grammaticalization, 405–416. Berkeley, CA: Berkeley Linguistics Society. ———. 2012. “Intersubjectification and clause periphery”. In Brems, Ghesquière & Van de Velde (eds.), 7–28. Traugott, Elizabeth Closs & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs & Graeme Trousdale. 2013. Constructionalization and Constructional Changes. Oxford: Oxford University Press. White, P.R.R. 2003. Beyond modality and hedging: A dialogic view of the language of intersubjective stance. Text 23: 259–284.

chapter 5

Setting Up a Mental Space: A Function of Discourse Markers at the Left Periphery (LP) and Some Observations about LP and RP in Japanese1 Noriko O. Onodera 1 Introduction The concept of ‘periphery’ was first introduced in the context of historical pragmatics several years ago (cf. Traugott, 2005), and has since been broadly examined cross-linguistically. What forms and functions appear at LP and RP and whether there are cross-linguistic differences in such a form-function-position mapping have been among the intriguing questions researched. Nonetheless, many of the questions surrounding “periphery” still remain unsolved or even unexamined. The aim of this paper is to shed light on at least two points surrounding the notion of periphery: (1) to illustrate one particular function of LP elements, “setting up a mental space for ‘what is upcoming’ in discourse”, and (2) to sketch the association between position (left and right periphery) and function (e.g. ‘subjective’ and ‘intersubjective’). The mental space that LP elements set up for what is upcoming can be seen as the frame/base of understanding through which the coming part (e.g. main clause) is defined. The concept of “mental space”2 has been discussed by many researchers including Fauconnier (1985). However, in this study, it is the positioning of the clause that is focused upon rather than the theoretical pursuit of ‘mental space’. The concepts of subjective and intersubjective functions are understood following both Lyons (1982) and Traugott and Dasher 1 The earlier versions of sections 1–7 in this paper were presented at the Historical Pragmatics Workshop in March, 2011 (Gakushuin University, Tokyo), at the workshop on periphery, the 12th International Pragmatic Conference in July, 2011 (University of Manchester, UK), and at the Aoyama Linguistics and Literature meeting in November, 2011 (Aoyama Gakuin University, Tokyo). The author would like to thank the organizer, participants and audience of these conferences for their valuable comments. Part of this study was supported by a grant from the Research Institute for Humanities, Gakushuin University. 2 I appreciate the audience of the 12th IPC (July, 2011) for their comments and information on this concept.

© koninklijke brill nv, leiden, ���4 | doi ��.��63/9789004274822_�06

Setting up a Mental Space: Discourse Markers in Japanese

93

(2002). Lyons’ (1982: 102) definition of ‘subjectivity’, which has inspired many subsequent works on this theme, refers to “the way in which natural languages, in their structure and their normal manner of operation, provide for the locutionary agent’s expression of himself and of his own attitudes and beliefs”. Traugott and Dasher (2002: 21–22) further narrow down the workings of “subjectivity” in discourse as “. . . explicitly encod[ing] SP[eaker]/W[riter]’s point of view, for example in deixis, modality, and marking of discourse strategies”. “Intersubjectivity” defined by Traugott (2003: 128) is: the explicit expression of the SP[eaker]/W[riter]’s attention to the ‘self’ of the addressee/reader in both an epistemic sense (paying attention to their presumed attitudes to the content of what is said), and in a more social sense (paying attention to their ‘face’ or ‘image needs’ associated with social stance and identity). These definitions are the background for this study. In the use of naturally-occurring speech in different individual languages (e.g., Japanese, English, or French), we see both subjective and intersubjective functions at both left and right peripheries. To answer the research questions posed by the editors of this volume, further reports on periphery phenomena from diverse varieties of language are called for in order to make overarching theoretical generalizations about the role of the periphery in semantic change. 2

Key Concepts and Problem Statement

2.1 Periphery What we are tasked with doing is fine-tuning currently somewhat ill-defined ideas about periphery and setting them into a more theoretically well-defined framework. The notion of periphery refers to both edges of a unit of talk, or utterance. The left periphery signifies utterance-initial position (e.g. well as in “Well, I didn’t mean that.”) and the right periphery utterance-final position (e.g. y’know as in “I didn’t mean it, y’know.”). 2.2 Periphery and Discourse Markers I suggest that, from the outset, the concepts of periphery and discourse markers (DMs, henceforth), theoretically at least, partially overlap. The operational definition of DMs (Schiffrin, 1987: 31) is that “[DMs are] sequentially dependent elements that bracket units of talk”. “Bracket” here

94

Onodera

derives from Goffman’s (1974: 251–269) concept, which “provide[s] frames of understanding through which social life is defined” (Schiffrin, ibid.: 36). When these brackets are applied to talk, the bracketed activity is not social life but the units of talk such as utterances. Hence, the opening bracket of an utterance is the utterance-initial DM at the LP, while the closing bracket of an utterance is the utterance-final DM at the RP. Focusing on the symbolism and predominance of opening brackets over closing counterparts, Goffman further suggests that the beginning bracket will not only establish an episode but will establish a slot for signals informing and defining what kind of transformation is to be made of the materials within the episode (1974: 255). I also note that DMs are a similar category to Blakemore’s (1987) “procedurals”. Some DMs, such as na-elements in Japanese, appear both at the LP and RP. They emerge as utterance-initial interjections at LP (e.g. nee (summons use) as in nee, ojoosan (‘hey, miss’)), and as utterance-final particles (often called sentence-final particles (shuu-joshi) in Japanese linguistics) at RP (e.g. ne (confirmation use) as in wakarimasita ne? (‘did you understand ne?’)). However, other DMs seem to appear exclusively at the LP or the RP. One Japanese DM, dakara, is used exclusively at the LP, and at the RP, its structural counterpart predicate + kara is used. See (1) and (2). (1)3 Dakara watashi wa    ikanakatta. So I TP  go-NEG-PST ‘So, I didn’t go’. (2) Watashi wa ikanakatta. Okaasan ga soo  itta kara. I TP  go-NEG-PST mother SB so say-PST  because ‘I didn’t go. ‘Cause my mother told me not to.’

In (2), at the RP (of the second utterance), the same form as the LP DM, dakara, cannot occur, but its RP counterpart, V-kara, occurs. RP V-kara seems to have appeared earlier than LP dakara in the course of historical development of LP dakara (Onodera 2004: 212).

3 In this chapter, the following abbreviations are used in the glosses: COP (copula); DO (direct object); EMP (emphatic); FP (sentence-final particle); GER (gerundive form); IP (sentenceinternal particle); NEG (negative); NOM (nominalizer); PASS (passive); PST (past/perfect tense); Q (assigned to a FP ka only); QT (quotative marker); SB (subject marker); TAG (tagquestion-like expressions such as auxiliary verb forms (e.g. desho, ja-nai)); TP (topic marker).

Setting up a Mental Space: Discourse Markers in Japanese

95

2.3 Main Question As periphery research is still in its infancy as a field of study, there is some merit, in the first instance, in focusing on empirical data with a view to discerning which elements appear at the LP and RP in different languages and what such elements do on their respective peripheries. The current chapter reviews and synthesizes existing literature on elements which appear at the LP and RP in Japanese and examples are provided to illustrate the functions of LP and RP elements in conversational Japanese and to throw light on their historical evolution. On the basis of such studies across different languages, it may be possible to arrive at some cross-linguistic generalizations and a contribution may be made to theory in this area. 3

Suggestions

From my observations of Japanese conversational data (both modern conversations and some historical data) in my research, I would like to make a few general points concerning the issue of periphery. In Japanese, at both the LP and RP, a number of elements with abundant pragmatic functions are used, and they signal speakers’ subjective or intersubjective meanings. Some are clearly discourse markers. Japanese is a language well-known to have a great number of so-called sentence-final particles. That is, Japanese speakers frequently use one or a combination of sentence-final particle(s) at the end of an utterance, in other words at the RP. As a consequence, many Japanese utterances terminate with one or more sentence-final particle(s) that add diverse and subtle subjective/intersubjective expressiveness to the propositional meaning. LP discourse markers in Japanese have been less well covered in the literature and the first point I would like to suggest is that LP discourse markers ‘create the ground’ or ‘set up a mental space’ for the upcoming focal part in discourse. In this chapter, I will discuss this in sections 4–6. Related to the above general observation, the second point I will suggest is that at least for Japanese the association between LP-subjectivity and RP-intersubjectivity does not hold. In Japanese, both LP and RP elements can mark both subjective and intersubjective meanings. I will return to this point in section 8. 4

Setting Up a Mental Space—A Function of Lp Discourse Markers

I begin the discussion by drawing on Auer’s (2000) observations concerning left-peripheral elements, which sparked my interest in the function of LP

96

Onodera

elements for ‘setting up a mental space’. Looking at both pre-positioned and post-positioned wenn-clauses, Auer (2000: 173) suggests the preference of prepositioned wenn-clauses in spoken German. (3) exemplifies such a case. (3) ((job interview)) (Auer 2000: 181) wenn ich (–) grad WEIter ausführen darf; (0.5) Sie wissen ja in de: in der AUtoinduschdrie .h herrschen SEHR groβe k’ konkurRENZ, markt ‘if I may continue elaborating on that; (0.5) you know that in the car industry there is a lot of competition ((etc.))’

Auer explains that “[t]his is the case for “speech-act related”4 wenn-clauses which do not conjoin two propositions on the content level”, but “they are [often] used in order to mitigate subsequent face-threatening acts” (181). In (3), during a job interview, the speaker draws on a negative politeness strategy (a request for permission) before making what could be perceived as a facethreatening act, “interruption” (Auer 2000: 181). Auer (2000) further suggests that the reason why speakers choose the pre-positioned wenn-clause is due to “some kind of cognitive ‘naturalness’ in the way in which conditionals [one of which is a wenn-clause] create the ground—or, in more recent [. . .] parlance, set up a ‘mental space’ (Fauconnier, 1985). For cognitive reasons, it is the grounding which (iconically) precedes the focal proposition, not the other way [a]round” (184). In (3), the pre-positioned wenn-clause sets up a mental space for the following main clause. With the use of this wenn-clause, the speaker can mitigate the potential face-threatening of the main clause. Ford (1993) puts forward a similar idea on the function of the English ifconditional clause: “The prevalence of initially placed if-clauses may reflect the general tendency to signal ((. . .)) that the interpretation of the coming clause will be, [. . .] limited by the contents of the if-clause” (15). Here, Schiffrin’s operational definition of “discourse markers” is brought to mind. Remember that this definition contains the concept of “bracket” (Goffman, 1974), as shown in 2.2. Discourse markers are brackets applied to talk, and these brackets provide frames of understanding through which the containing activity/talk is defined (cf. Goffman, 1974: 251–659; Schiffrin, 1987: 36).

4 Auer (2000: 201) mentions that this term is used in a broader sense than in Sweetser (1990). See note 11 in Auer (2000: 201).

Setting up a Mental Space: Discourse Markers in Japanese

97

Auer, Ford, and Goffman all arrive at the same conclusion: the prepositioned clause sets up a mental space, and it limits/frames/grounds the upcoming focal clause. 5

Cognitive Naturalness

Why do the speakers use wenn-clauses utterance-initially? Auer (2000: 184) ascribes the reason for the frequency of pre-positioned wenn-clauses in contemporary German conversation to “cognitive naturalness” as in (3) above. As Auer (ibid.) writes, for cognitive reasons, the grounding part (iconically) precedes the focal part, and not the other way around. This suggested cognitive natural flow and framing function are illustrated in (4). (4) LP element + subsequent part grounding part focal part

frame/ground

This kind of ‘cognitive natural flow’ can indeed be observed in several linguistic sequences we know. Examples are:

• (if/wenn) conditionals [if/wenn clause + main clause] • topic-comment structure [topic + comment] • cause-effect sequence [cause + effect] • adjacency pairs [first pair part + second pair part] • LP discourse markers [LP discourse marker + subsequent part] Another instance may also reside in “academic presentations”: i.e. the ‘introduction—body—conclusion’ sequence is usually observed. For all these sequences, the grounding segment occurs initially, followed by the second element, not the other way around. 6

Examples of LP Discourse Markers

From the list of sequences that carry the cognitive natural flow, two LP discourse markers will be highlighted: German obwohl in 6.1., and Japanese dakedo in 6.2. These LP DMs set up a mental space for the upcoming part of discourse.

98

Onodera

6.1 LP Discourse Marker obwohl The first example is German obwohl (although). According to Günthner (2000), the obwohl construction has been used as a “subordinate” concessive in traditional German grammar. But at the present time, people increasingly use this construction with main-clause syntax. This recent use of obwohl has led to heavy criticism among “caretakers” of the German language who wish to save the German subordinate-clause order (ibid.: 439). A causal subordinator, weil (because) in traditional grammar, which is also recently increasing in use in main-clause syntax, is another target for such criticism. Observing such “ungrammatical” uses in everyday colloquial German interaction, Günthner (2000) writes that her study “aims at uncovering how grammatical resources are exploited by interactants to fulfill communicative functions”. The development of obwohl introduced here is one that moves from a concessive subordinator to an independent LP discourse marker. The following shows an example of this. (5)

SUMMER HEAT　(Günthner, 2000: 451) 1 Klaus: das is echt s’BESTE BIER. (–) 2 ich mein von den alkoholfreien. 3 (–) 4 Hans: hhm. obwohl es gibt schon BESSERE. 5 zum Beispiel BECKS is bei weitem TRINKBARER.



‘1 Klaus: this is really the best beer (–) 2 I mean among the alcohol-free ones 3 (–) 4 Hans: hhm. although there are better ones 5 for example Becks is far more drinkable.’

In lines (1–2), Klaus makes an assessment about the high quality of the beer he and Hans are drinking. In line (3), a short pause already marks upcoming dissent. Then in line (4), Hans shows hesitation (hhm), utters obwohl (although), and makes his disagreeing assessment. Here obwohl (in translation ‘although’) in line (4) is used as an initial discourse marker, so it rather reads like ‘however’. The motivation for obwohl adopting a DM function in utterance-initial position which emerges from its concessive subordinator usage in traditional grammar lies in the fact that the speaker adopts a negative-face saving strategy before providing his disagreement (about the quality of beer). Without this DM, the following disagreement (‘there are better ones’) would sound abrupt or too challenging.

Setting up a Mental Space: Discourse Markers in Japanese

99

Günthner (2000: 452) reports that this ‘discourse marker corrective use’ is pervasive in today’s German conversations. She suggests that this new use of obwohl now links two utterances. 6.2

LP Discourse Marker dakedo

The second example of an LP discourse marker is Japanese dakedo. Like obwohl in 6.1., the English translation for this discourse marker is also ‘although’. There seem to be several parallels between German obwohl and Japanese dakedo. The first is observable in their diachrony. While obwohl is a concessive subordinator in traditional German, in Japanese the clause-linking da (copula) + kedo (connective particle) (meaning ‘although’) functions as a concessive subordinator (for more on this view, see Morioka, 1973: 15; Tokieda, 1990: 148–149; Kyogoku and Matsui, 1973: 118–119). If utterance-initial dakedo indeed originated from the clause-final da + kedo, obwohl and dakedo share the similar developmental process, i.e. from a concessive subordinator to an independent initial DM. An example of LP dakedo is shown in (6). (6) Yuujoo (1920) Nakata: a. Maa, sonna koto    o Well, like.that  N OM DO

b. iu mono ja nai yo. say NOM COP.IP NEG FP

Hayakawa: c. Dakedo, boku wa ujimushi  But, I TP maggot

d. atsukai sarete damatte wa　 treat PASS keep.silent-GER EMP



e. irare nai   yo.    can NEG FP



‘Nakata: a–b. Well, you shouldn’t say something like that. Hayakawa: c–e. But, I cannot keep silent, being treated like a maggot.’

In the conversation from which excerpt (6) is extracted, Hayakawa and another participant, Nojima, were having a heated argument. Nakata tries to arbitrate

100

Onodera

between the two, addressing Hayakawa in (a). Hayakawa then says dakedo, and rebuts by uttering (c–e). Dakedo in (c) marks the contrastive action, ‘claiming the turn’, as well as the referential contrast, ‘Hayakawa shouldn’t say something like that’ (a–b) vs. ‘Hayakawa (I) cannot keep silent’ (c–d). Here dakedo in (c) indicates the upcoming disagreement with Nakata, and it reads like “[as] against what you just said” (Günthner, 2000: 452), the same reading as that of obwohl in (5). Without dakedo, Hayakawa’s disagreement would sound too abrupt or direct. In addition to the developmental process from a concessive subordinator to an utterance-initial discourse marker, obwohl and dakedo share a synchronic feature, the pragmatic function of framing the upcoming disagreement. The cross-linguistic examination in (6.1) and (6.2) not only reveals a similar development in German and Japanese LP elements but also hints at possible avenues of research in the functions of LP elements in other languages. 7

Why are d-Connectives used at LP?—Two Types of Mechanisms that Explain d’s Function

Above, I showed two instances of LP discourse markers, German obwohl and Japanese dakedo. Section 7 addresses the following questions. Why are dconnectives so conspicuous in terms of the number of members in this class and their frequency in Japanese conversations? In spite of the fact that the Japanese language carries a relatively small number of DMs (compared to languages such as English), why is it that most of the d-connectives seem to be DMs and are exclusively applied as LP DMs? 7.1 D-connectives Overall, Japanese does not appear to have a large number of items which would qualify as discourse markers. Of the items which could be judged as DMs, d-connectives form a very major part. D-connectives appear only at the LP, and are fused or univerbised forms, made up of two elements. The d-element refers back anaphorically to previous text, while the second element may have a variety of functions in relation to the ongoing discourse. Connectives begin with da (copula), de (gerundive form of da), or na (pre-nominal form of da). D-connectives include the forms dakara, datte, dakedo, demo, datoshitemo, dattara, daga, dakedomo, dewa, datoshitara, de, and nanoni (Maynard, 1989; Mori, 1996; Matsumoto, 1988; Kyogoku and Matsui, 1973; Mio, 1995: 209; Onodera, 2004: 211–214 and elsewhere). In previous discourse studies, only a few items out of potential members, dakara (Maynard, 1989), datte (Mori, 1996) dakedo and demo (Onodera 2004) have been judged to be DMs.

Setting up a Mental Space: Discourse Markers in Japanese

101

The reason why d-connectives appear in utterance-initial positions (at the LP) appears to be that the element d prefacing d-connectives (dakedo, demo, dakara, datte, datoshitemo, dattara, etc.) carries one particular function (pro-predicate) and, owing to this function, d-connectives can set up a frame for what is upcoming. At the LP, d, at the beginning of a d-connective, can easily take in context residing in prior discourse, and with this appropriate context for the hearer’s interpretation, a d-connective can set up a frame for the upcoming focal part in discourse. Thus, the exploration into the mechanisms for the function of d indeed answers the onomasiological question—“why did this set of items (d-connectives) move into one given domain (i.e. LP conjunctions)” (cf. Traugott 1985: 298). Kyogoku and Matsui (1973: 118–119) put forward the historical development through which “auxiliary or verb [da or de] + connecting particle [kedo, keredo, kara, wa, etc.]” shifted into d-connectives. Referring to expressions including daga, they further state that “‘auxiliary + connecting particle’ separated from [the end of] the preceding sentence and became independent” (115). They state that among d-connectives, daga, dakara, datte, and dewa emerged during the Edo era (1603–1867). Their suggested historical development of d-connectives is illustrated in Table 5.1: Table 5.1

verb da da da da da de de

Word formation of d-connectives

+

connecting particle

→

d-connective

+ + + + + + +

ga kara tte (quoting form) keredo kedo wa mo

→ → → → → → →

daga (but) dakara (so) datte (‘cause) dakeredo (but) dakedo (but) dewa (then) demo (but)

The word formation of d-connectives, as can be seen in this figure, seems quite ‘productive’, as Huddleston (1984: 27) defines the term. Da/de and the subsequent connecting particles are related due to agglutination, one of the typological features of the Japanese language. There can be more utterance-initial elements (conjunctions) in Japanese, such as denakereba (de + nakere + ba) and denaito (de + nai + to) , that go through a similar

102

Onodera

formation process. As stated in Huddleston (1984: 27), “a morphological process is fully productive if it can apply to all members of a large and independently definable set of stems”. Da/de exhibits a fairly high degree of productivity with connecting particles, as seen above. In the sections which follow, the pragmatic function of d will be explained in the light of two mechanisms suggested by Okutsu (1978) (7.2) and Kuno (1978) (7.3), its ‘replacement’ function and its ‘recovery’ function (the da strategy), respectively. 7.2 D’s Replacement Function (Okutsu, 1978) According to Okutsu’s well-known Boku wa Unagi Da no Bunpoo (I am the Eel Grammar 1978: 10–13, and elsewhere), da (d), widely-recognized as a copula in Japanese, does not always serve the linking function as a copula, but rather its fundamental function is “to replace a predicate in a prior discourse”. The following examples (i), (ii), and (iii) illustrate that the copula da (d) can replace different types of predicates: in (ii), da (d) replaces an adjective, and in (iii) a passive predicate. (i) seems to be the basic replacement case where da replaces “a case marker (such as o, ni, e, de) + a verb” (Okutsu, 1978: 18). (i) Q: (Kimi wa nani o chuumon suru?) (at a fish restaurant) ‘What are you going to order?’ A: (a1) (a2)

Boku wa unagi da. I TP eel  C OP ‘I am the eel (literally).’ Boku wa unagi o  chuumon suru. I TP eel  DO order ‘I’m going to order eel.’

In (i), for example, when asked the question at a fish restaurant, “what are you going to order ?”, if you intend to have eel, you may say something like “I’m going to order eel”. In such a case, instead of “I’m going to order eel (Boku wa unagi o chuumon suru)” (a2), in Japanese, you can also say “I am (the one who orders) the eel (Boku wa unagi da)” (a1). Da in (a1) replaces the predicate “order ～” (o chuumon suru). So, (a1) and (a2) convey exactly the same meaning. With this replacement function of d, there is no need for the speaker to repeat the same predicate, if there is felicitous context. In the case of (i), the context is provided in the question Q. Due to this context, the content of d (what d replaces), “order ～”, is easily recovered by the hearer. (ii) shows how da (d) replaces a predicate which consists of an adjective.

Setting up a Mental Space: Discourse Markers in Japanese

103

(ii) Q: (Dare ga chiisai ka?) ‘Who is small?’ A: (a1) Taro da. COP ‘Taro is.’ (a2) Taro ga chiisai. SB small ‘Taro is.’

In (ii), the question given is “Who is small?” To answer this, we can say either (a1) Taro da or (a2) Taro ga chiisai (‘Taro is small’)”. Since da replaces the predicate (subject marker ga + adjective), which is recoverable from the preceding question, (a1) and (a2) both convey ‘Taro is small’. The third case showing the replacement of d is when the predicate is in the passive voice, as in (iii). (iii) Q: (Kimi wa dono  kuruma  ni You TP which car  by ‘By which car were you hit?’ A: (a1) Boku wa  ano kuruma I TP that car   ‘I was hit by that car.’ (a2)

hanerareta ka?) hit-PASS-PST Q

(ni)  d a. by COP

Boku wa ano  kuruma ni hanerareta. I TP   that car by  hit-PASS-PST ‘I was hit by that car.’

Now let us consider an answer to the question “By which car were you hit?” To answer this, we can utter either (a1) Boku wa ano kuruma (ni) da or (a2) Boku wa ano kuruma ni hanerareta. In example (iii), the copula da in (a1) replaces the predicate ni hanerareta in the passive form which already resides in the question Q. If we hear the answer with the d-replacement (a1), we can easily calculate and recover the content of d (being hit by ～) because there is already a shared assumption “the answerer was hit by some car” derived from the question. As shown in (i)–(iii) above, in Japanese conversations, the d-replacement strategy is used frequently and efficiently. D can indeed replace various types of predicates; in addition to transitive verbs, adjectives, and passives, the range

104

Onodera

includes progressive aspect, causatives, and negation, among others (Okutsu, 1978: 44–52). This strategy helps avoid redundancy and engenders economy and brevity in Japanese conversations. 7.3

Kuno’s (1978) da Strategy—Another Explanation for the Function of d5

As mentioned above, the frequent emergence and use of d-connectives as LP DMs is very noticeable in Japanese. Researchers other than Okutsu (1978) have also paid attention to the behavior of d-connectives. Kuno (1978) is one such example. Kuno (1978: 4–124) devotes a whole chapter to the explication of “ellipsis” in Japanese discourse. In this chapter, Kuno investigates the mechanism of d’s function with “ellipsis”, which approaches this function from a slightly different angle than Okutsu’s (1978) “replacement”. Kuno (1978) illustrates several rules (principles, strategies, and constraints) which are at work in Japanese discourse. The following rules (I), (II), and (III) are those related to d’s function. Rule (I) is the fundamental rule for ellipsis (omission) in Japanese discourse. Rule (I): Fundamental Principle of Ellipsis The elements to be omitted must be ones recoverable from the verbal or nonverbal context (Kuno, 1978: 8).6

Then, the two principal strategies for omission in Japanese discourse are illustrated as follows: Rule (II): Repetition of Main-Verb Strategy Omit the recoverable elements. However, the main verb should remain (Kuno, 1978: 8).

5 I would like to thank Professor Kazuo Nakazawa and Professor Shigeo Tonoike for their comments on the function of d and the validity of Kuno’s (1978) work. 6 Kuno (1978) is written in Japanese. So, the citations from this volume are my translations. Glosses and free translation for examples (7)–(11) are also mine.

Setting up a Mental Space: Discourse Markers in Japanese

105

Rule (III): Da Strategy (Apply this strategy only when the main verb is recoverable.) Omit the recoverable elements. For the remaining elements to qualify as a sentence, embed them in the da structure (Kuno, 1978: 8).

Rule (III) is more directly relevant to d’s function, the issue of this paper. However, Rule (II) is offered here, because it is helpful to comprehend the omission system in Japanese. The following (7), (8), and (9) are examples of omission by Rule (II) (Kuno, 1978: 8–9), and (10) and (11) are those of omission by Rule (III). In the examples, A and B stand for speakers A and B, while [φ] marks where the omission occurs. In the ‘free translation’, [ ] marks the omitted and recovered elements. (7) A: Kimi wa,   kinoo,  sanji koro, Hanako to Kanda o aruite ita  ka? You TP yesterday  3 o’clock around  with walk-GER       Q ‘Were you walking in Kanda (area) with Hanako around 3 o’clock yesterday?’ B: Un, [φ] aruite ita. Yes  walk-GER ‘Yes, [I was] walking [in Kanda (area) with Hanako around 3 o’clock yesterday].’ (8) A: Kimi wa, amerika ni itta  koto ga aru ka? You TP America to go-PST NOM SB   experience Q ‘Have you ever been to America?’ B: Un, [φ] aru. Yes experience ‘Yes,  [I] have [been to America].’ (9) A: Kimi  wa, itsu, Hanako to Kanda o  aruite ita  ka? You TP when with walk-GER Q ‘When were you walking in Kanda (area) with Hanako?’ B: [φ] Kinoo sanji     koro, [φ] aruite ita. yesterday 3 o’clock around  walk-GER ‘[I was] walking [in Kanda (area) with Hanako] around 3 o’clock yesterday.’

The examples (7), (8), and (9) all show situations where Speaker B answers Speaker A’s question. In these situations, a considerable amount of context is given in the question and is therefore shared by both A and B. So, according to Rule (II), all the recoverable elements are omitted, but the main verb remains. Kuno (1978: 81) ascribes the reason for retaining the main verb to ensure that B’s answer qualifies as a “sentence”.

106

Onodera

Consequently, in (7) and (8) where the given question is a yes-no type, the answerer B utters un ‘yes’ and omits most of the elements that are shared information except the main verb aruite ita ‘walking’ in (7) and aru ‘experience’ in (8). B’s short answers in (7) and (8) thus sound perfectly grammatical as Japanese utterances (sentences), owing to the system of omission. In (9), A uses a wh-question to request a piece of information concerning the time when B was walking in the Kanda area with Hanako. In his/her response, B fills in the requested information with ‘around 3 o’clock yesterday’, and repeats the main verb aruite ita ‘walking’, so that his/her whole answer qualifies as a sentence. As seen in the above (7)–(9), due to the omission (ellipsis) Rule (II), recoverable elements can be all omitted, thus redundancy is avoided in Japanese discourse. Now, we can further see that in Japanese the “da strategy” allows of the most economical omission pattern. The following examples (10) and (11) show the da ellipsis pattern (Kuno, 1978: 80–82). (10) A: Doko de umareta  no? Where in be-PST born  FP ‘Where were you born?’ Ba: Tokyo de umareta. Tokyo in   be-PST born ‘[I] was born in Tokyo’. Bb: Tokyo da. Tokyo COP ‘[I was born in] Tokyo’. (11) A: Kono hanashi o dare kara kiita  ka?7 This story   D O whom from hear-PST    Q ‘From whom did you hear this story?’ Ba: Yamada kun kara kiita. Yamada Mr. from  hear-PST ‘[I] heard [this story] from Mr. Yamada’. Bb: Yamada kun (kara) da. Yamada Mr. (from) COP ‘[I heard this story (from)] Mr. Yamada’.

7 In Kuno (1978: 81), the original example of (11) shows a polite situation with honorific forms. Here, I show a simpler case with plain forms.

Setting up a Mental Space: Discourse Markers in Japanese

107

In Ba of (10) and (11), the main verbs umareta ‘be-PST born’ and kiita ‘hear-PST’ remain, although they are recoverable from the context. Without umareta and kiita here, Ba (10) and Ba (11) are not grammatically well-formed sentences. This follows Rule (II), as previously seen in (7)–(9). Notice that Bb cases make possible answers even shorter than Ba due to Rule (III): the da strategy. In (10) Bb, all the elements (subject ‘I’ and main verb ‘be-PST born’) are recoverable from the context (from A’s question), so they are omitted. But, the new information ‘Tokyo’ (a noun) alone cannot be a sentence, so it is embedded in a da-structure. As a consequence, Ba and Bb convey exactly the same meaning ‘I was born in Tokyo’. Likewise in (11) Bb, all the elements (subject ‘I’, main verb ‘heard’, direct object ‘this story’, preposition ‘from’) are omitted. (See in the free translation, the bracketed [I heard this story (from)] which indicates the omitted and recovered elements.) For Bb, the preposition kara is bracketed. This means that the short answer with or without kara is possible: ‘Yamada kun da’ or ‘Yamada kun kara da’. Only the new information ‘Mr. Yamada’ remains and is embedded in a da structure in the former. In (11) Bb, ‘Yamada kun da’ consists of ‘Yamada kun’, the non-omittable new information, and ‘da’ which enables B’s whole answer to be in ‘sentence’ form. 7.4 The Answer to the Question—Why are d-Connectives Used at the LP? In 7.1–7.3, I described d-connectives and illustrated the function of d in two ways based on the mechanisms proposed by Okutsu (1978) and Kuno (1978), respectively. I now offer an answer to the original question—Why are dconnectives used at the LP? By Okutsu’s (1978) argument, d’s function is revealed by ‘replacement’, whereas by Kuno, it is by ‘ellipsis’. Here, the example (6) of dakedo is given again (as (6’)) below and accounted for in order to show how the two mechanisms work in Japanese conversational management. (6’) Yuujoo (1920) 　　 Nakata: a. Maa, sonna Well, like.that

 koto  o NOM DO

b. iu mono ja  nai   yo. say NOM COP.IP NEG FP

Hayakawa: c. Dakedo, boku wa ujimushi But,  I  TP maggot

108

Onodera

d. atsukai sarete damatte   wa　 treat  PASS keep.silent-GER EMP e. irare nai      yo. can   N EG FP

‘Nakata: a–b. Well, you shouldn’t say something like that. Hayakawa: c–e. But, I cannot keep silent, being treated like a maggot.’

The speaker, Hayakawa, says dakedo in (c). The meaning (content) of da in dakedo is ambiguous since there are theoretically “10 or 20 [possible] patterns of content” (Okutsu, 1978: 151) that reside in the precedent textual context. However, in the case of (6’), d’s content/what d replaces seems to be the preceding message by Nakata ‘Hayakawa shouldn’t say something like that’ in (a–b). Using Okutsu’s idea of replacement, da in dakedo replaces this message, and through Kuno’s theory of ellipsis, this message portion is omitted in Hayakawa’s utterance in (c–e), and is recovered by the da-strategy in dakedo. In either interpretation, d’s content is successfully recovered and conveyed to the hearer and participants. Morioka (1973: 20) suggests one further characteristic of the LP d-connectives, that dakedo, demo, daga, danoni, or datte take in content from somewhere in the preceding discourse and serve as a ‘subordinate clause’. So, for example, one short word, dakedo, serving as a mini subordinate clause, grounds the following main clause (in the case of (6’), ‘I cannot keep silent . . .’). According to Morioka (ibid., 19–20), d-type conjunctions “are regarded as an ‘omitted form’ or ‘replacement’ of a full subordinate clause”. This view perfectly accords with Kuno and Okutsu’s expositions. Hence, the answer to the question of why d-connectives are used at the LP is that d-connectives have the function of setting up a mental space for the upcoming focal part in discourse, owing to d’s pragmatic function of replacement (Okutsu, 1978) or of recovering what is omitted (da-strategy) (Kuno, 1978). Above, it has been shown that in either Okutsu or Kuno’s expositions, d ’s function enables d-connectives to be applied at the LP and work as powerful utterance-initial connecting devices. This device serves a discourse-linking textual function. 8

Pragmatic Elaboration at Both LP and RP

Section 8 addresses the second aim of this paper mentioned at the outset: to explore the association between the peripheral slot (LP/RP) and function

Setting up a Mental Space: Discourse Markers in Japanese

109

(subjective/intersubjective). In the past two decades, many discourse studies have paid attention to the fact that both edges of an utterance are typical sites for pragmatic elaboration in Japanese (cf. Suzuki 1998, Ohori 1998, Onodera 2004 among others). Table 5.2, below, provides a brief sketch of the pragmatic functions (subjective/intersubjective) that appear at the LP and RP in Japanese. Table 5.2 lists a few expressions from each category. As seen in this Table, in Japanese, for the LP, at least two categories (traditional word classes), interjections and conjunctions, are used. The interjection aa typically expresses a human subjective exclamation by giving vent to one’s innermost feelings, whereas there are many Japanese interjections that are intersubjective. For conjunctions, a single expression seems to be able to convey either a subjective or intersubjective meaning, depending on the situation. However, another possibility is that all the LP discourse markers function intersubjectively when they convey subjective meanings, because they transmit subjective meanings to other people. Table 5.2 Pragmatic functions (subjective/intersubjective) at the LP and RP in Japanese8 LP/RP category



expression sb./intsb. example (function)

nee intsb. “Nee, ojoosan.” (summons) hey Miss ‘Hey, Miss.’ Interjections noo intsb. “Noo ikani Tokashi dono.” (summons) hey what Tokashi Mr. ‘Hey, Mr. Tokashi.’ LP yo intsb. “Yo, neesan.” (reply) yep sis ‘Yep, sis.’ aa sb. “Aa, samui!” (exclamation) oh cold ‘Oh, it’s cold!’ 8 In Table 5.2, the function column provides the specific function/meaning of the underlined expression in each example. A portion of the examples are from Nihon Kokugo Dai Jiten 2nd edition Editorial Committee and Shoogakkan Kokugo Jiten Editorial Board (eds.) (2001) and Uchio and Okamura (1973).

110 LP/RP category

Onodera expression

sb./intsb. example (function)

demo intsb. “Demo, kono natsu dokoe ikoo?” (open conversation) but this summer where to go ‘By the way, where are we goin’ this summer?’ Conjunctions demo sb. “Demo, kooiu kangae mo arimasu.” (disagreement) but this like idea also exist ‘But, I rather think this way.” dakara intsb. “Dakara, itta desho!” (blame)   so told TAG ‘I told you!’ dakara sb. “Dakara kinoo kesseki sita.” (cause-effect) so yesterday absence did ‘So, I was absent yesterday.’ na sb. “Samui na.” (exclamation) cold FP ‘It’s cold!’ ne sb. “Watashi wa tadasii to  omou ne.” (emphasis) TP right    QT think FP ‘I think I’m right.’ ne intsb. “Kayoobi, gakkoo iku ne.” (confirmation) Tuesday   school go FP RP FPs ‘I’ll go to school on Tuesday, OK?’ yo intsb. “Suiyoobi ni iku yo.” (insistence) Wednesday go FP        ‘I’ll go on Wednesday.’ wa sb./intsb.? “Asu samui daroo to omou wa.” (judgment) tomorrow cold  will.be QT think FP ‘It’ll be cold, tomorrow.’

Setting up a Mental Space: Discourse Markers in Japanese LP/RP category

expression

111

sb./intsb. example (function)

mono sb./intsb. koto sb. NOMs koto intsb. no sb. no intsb.

“Watashi nakimusi nan desu mono.” (insistence/rebuttal) I crybaby COP.NOM COP NOM ‘I’m a crybaby.’ “Yoku ganbatta koto! (exclamation) well work.hard NOM ‘Good job!’ “Moo shinai koto!” (order, request) again do.NEG NOM ‘Don’t do it again!’ “Samui no.” (exclamation) cold NOM ‘I’m cold.’ “Konakute ii no!” (order, request) come.NEG.GER all.right NOM ‘Don’t come!’

On the other hand, for RP, two categories, sentence-final particles (FPs) and nominalizers (NOMs) are used. The nominalizers listed here can be labeled as pragmatic particles after their pragmatic development, so FPs and NOMs here work in a similar way in discourse. It is important to remember that in Japanese discourse, both subjective and intersubjective meanings are expressed at both the LP and RP. Sometimes it is difficult to judge whether a specific meaning in an example is subjective or intersubjective (e.g. a FP wa). What is even more complicated, but typical in Japanese, is that at the RP, at the end of an utterance, combinations of the above pragmatic particles are perfectly acceptable: e.g. . . . nai no yo ne (. . . negation sb. intsb. intsb.) (Ohori 1998: 194), . . . itta wa yo ne (. . . say-PST sb. intsb. intsb.). This pragmatic elaboration (especially at the RP) has been examined in many works in traditional Japanese grammar/linguistics. Minute expressiveness of the speaker’s stance/angle/attitude or feeling is marked at Japanese RP by modals and the subsequent pragmatic particles. In short, the end of a Japanese predicate seems to consist of ‘main verb, (negation), modal element, and final pragmatic particle(s)’. The best way to show this structure is with the following Figure.

112

Onodera

D (dentatsu ‘communication’ = intersubjectivity) C (handan ‘judgment’ = subjectivity) B ( jitai ‘events’)

Nee Doomo

Yukiko ga

A (doosa ‘acts’) henji wo dasa-

IP

SBJ

reply OBJ send-

somehow

nakat-ta

yooda

NEG-PST seem =Belief

Yo I tell you =Assert

‘It seems that Yukiko didn’t send a reply, (I tell you)’ Figure 5.19 Layered structure models (Shinzato, 2007: 177): cited from Minami (1974), Takubo (1987), Noda (1997)

sentence initial

predication judgment sentiments interaction sentence final

a sentence Figure 5.2 Hayashi’s layered model of a Japanese sentence

“This layered model elegantly depicts the core to peripheral extensions as well as the correlation between the sentence-initial and sentence-final elements” (Shinzato, 2007: ibid.). This layered structure has also been scrutinized by other Japanese traditional linguists (e.g., Hayashi 1983, Watanabe 1971, Okutsu 1978). I will provide one additional Figure (5.2) to support their view.

9 In Figure 5.1, Shinzato (2007) uses the abbreviations: IP (interjectional particle); NEG (negative); OBJ (object); PST (past); SBJ (subject).

Setting up a Mental Space: Discourse Markers in Japanese

113

As we can see in Figures 5.1 and 5.2, in a Japanese utterance, the intersubjective meaning appears in the outermost layer, i.e., in the utterance-initial and final positions. The subjective meaning appears one layer inside of this layer. If this view on the layered structure of a Japanese utterance is accepted, the somewhat mysterious occurrence of the combination of multiple pragmatic particles at the RP can be resolved. For the typical RP string . . . nai no yo ne (. . . negation sb. intsb. intsb.) mentioned above, each element emerges in a specific layer; negation nai in Layer B, subjective particle no in Layer C, and intersubjective markers yo and ne (working like a tag) in Layer D, as exemplified in Figure 5.1. A Japanese utterance seems to consist of Layers A and B that convey the core information, with the optional use of elements in Layers C and D. The layered structure model neatly explains the optionality in the use of subjective/intersubjective particles. That is, for the case where an utterance ends with a subjective particle (Layer C) only (e.g. Samui no. (sb.)), we can say that the speaker optionally did not use an intersubjective particle. Thus, the pragmatic elaboration at the RP in Japanese is captured very accurately in the layered structure model shown above. In Japanese at least, the core and peripheries of an utterance seem to be structured in this way. The view visualized here suggests that both subjective and intersubjective functions appear on both the LP and RP according to the speaker’s subjective judgment and that these pragmatic functions do their work in this order. 9 Conclusion Though Japanese DMs can have subjective or intersubjective interpretations at the LP or RP, it seems that d-connectives are constrained to appear at the LP, perhaps unsurprisingly given their anaphoric role in recapturing previous discourse (through ellipsis or replacement). The ‘d’ connective expressions thus bracket units of talk as discourse connectives and set up the mental frame for the interpretation of the ensuing discourse. Interestingly, German obwohl appears to be developing a DM usage at the LP of main clauses (derived from a subordinating conjunction) but cannot yet appear at the RP, and the same is true of English although. English though, on the other hand, can appear as a conjunction and as a hedging DM on LP or RP, respectively, as can but (at least in some varieties of Australian and Scottish English). As we have seen, Japanese LP dakara/dakedo derive from RP V-kara/V-kedo. It seems that, whereas in some European languages there is a shift from LP to RP, in Japanese there is rather a reverse shift, i.e. from RP to LP.

114

Onodera

Acknowledgments I am deeply indebted to Kate Beeching and Ulrich Detges for their insightful comments on the earlier versions of this chapter. I am also grateful for all the comments from Rumiko Shinzato, Yuko Higashiizumi, and an anonymous reviewer. Thanks are also due to Maggie Camp for proofreading and correcting the manuscript. References Auer, Peter. 2000. “Pre- and post-positioning of wenn-clauses in spoken and written German”. In Couper-Kuhlen, E. & B. Kortmann (eds.), Cause-Condition-ConcessionContrast: Cognitive and Discourse Perspectives. Berlin & New York: Mouton de Gruyter, 173–204. Blakemore, Diane. 1987. Semantic Constraints on Relevance. Oxford: Blackwell. Fauconnier, Giles.1985. Mental Spaces. Cambridge: Cambridge University Press. Ford, Cecilia E. 1993. Grammar in Interaction: Adverbial Clauses in American English Conversations. Cambridge: Cambridge University Press. Goffman, Erving. 1974. Frame Analysis. Cambridge, MA: Harvard University Press. Günthner, Susanne. 2000. “From concessive connector to discourse marker: The use of obwohl in everyday German interaction”. In E. Couper-Kuhlen & B. Kortmann (eds.), Cause-Condition-Concession-Contrast: Cognitive and Discourse Perspectives. Berlin & New York: Mouton de Gruyter, 439–468. Hayashi, Shiro. 1983. “Nihongo no bun no katachi to shisei (The structure and modality of a Japanese sentence)”. In Danwa no Kenkyuu to Kyooiku 1. [Nihongo kyooiku shidoo sankoosho 11.]. Tokyo: National Language Research Institute, 43–62. Huddleston, Rodney. 1984. Introduction to the Grammar of English. Cambridge: Cambridge University Press. Kuno, Susumu. 1978. Danwa no Bunpoo (Discourse Grammar). Tokyo: Taishuukan,. Kyoogoku, Okikazu & Matsui, Eiichi. 1973. “Setsuzokushi no hensen (The change of conjunctions)”. In K. Suzuki & O. Hayashi (eds.), Setsuzokushi∙Kandooshi (Conjunctions∙Interjections) [Hinshibetsu Nihonbunpoo Kooza 6]. Tokyo: Meiji Shoin, 89–136. Lyons, John. 1982. “Deixis and subjectivity: loquor, ergo sum?” In R.J. Jarvella & W. Klein (eds.), Speech, Place, and Action: Studies in Deixis and Related Topics. New York: Wiley, 101–124.

Setting up a Mental Space: Discourse Markers in Japanese

115

Matsumoto, Yo. 1988. From bound grammatical markers to free discourse markers: History of some Japanese connectives. BLS 14: 340–51. Maynard, Senko K. 1989. Functions of the discourse marker dakara in Japanese conversation. Text 9/4: 389–414. Minami, Fujio. 1974. Gendai Nihongo no Koozoo. Tokyo: Taishukan Shoten. Mio, Isago. 1995 [1942]. Hanashi Kotoba no Bunpoo: Kotobazukai Hen (Grammar of Spoken Japanese: Language Use). Tokyo: Kurosio. Mori, Junko. 1996. “Historical change of the Japanese connective datte: its form and functions”. In N. Akatsuka, S. Iwasaki & S. Strauss (eds.), Japanese/Korean Linguistics 5. Stanford: CSLI Publications, 201–18. Morioka, Kenji. 1973. “Bunshoo tenkai to setsuzokushi/kandooshi (Discourse cohesion and conjunctions/interjections)”. In K. Suzuki & O. Hayashi (eds.), Setsuzokushi∙ Kandooshi (Conjunctions∙ Interjections) [Hinshibetsu Nihonbunpoo Kooza 6]. Tokyo: Meiji Shoin, 7–44. Nihon Kokugo Dai Jiten 2nd edition Editorial Committee and Shoogakkan Kokugo Jiten Editorial Board (eds.). 2001. Nihon Kokugo Dai Jiten 2nd edition (Japan Japanese Linguistics Grand Dictionary). Tokyo: Shoogakkan. Noda, Harumi. 1997. No (da) no Kinoo. Tokyo: Kurosio. Ohori, Toshio. 1998. “Close to the edge”. In T. Ohori (ed.), Studies in Japanese Grammaticalization. Tokyo: Kurosio, 193–197. Okutsu, Keiichiro. 1978. Boku wa Unagi Da no Bunpoo (‘I Am the Eel Grammar’). Tokyo: Kurosio Publishers. Onodera, Noriko. O. 2004. Japanese Discourse Markers: Synchronic and Diachronic Discourse Analysis. Amsterdam: John Benjamins. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Shinzato, Rumiko. 2007. “(Inter)subjectification, Japanese syntax and syntactic scope increase”. In N.O. Onodera & R. Suzuki (eds.), Historical Changes in Japanese: Subjectivity and Intersubjectivity. Special Issue of Journal of Historical Pragmatics 8 (2), 171–206. Suzuki, Ryoko 1998. “From a lexical noun to an utterance-final pragmatic particle: wake”. In T. Ohori (ed.), Studies in Japanese Grammaticalization. Tokyo: Kurosio, 67–92. Sweetser, Eve E. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Takubo, Yukinori. 1987. Toogo koozoo to bunmyaku joohoo (Syntactic structure and contextual information). Nihongogaku 6/5: 37–48. Tokieda, Motoki. 1990. Nihon Bunpoo Koogohen (Japanese Grammar for Spoken Language). Tokyo: Iwanami.

116

Onodera

Traugott, Elizabeth C. 1985. “Conditional markers”. In J. Haiman (ed.), Iconicity in Syntax. Amsterdam: John Benjamins, 289–307. ———. 2003. “From subjectification to intersubjectification”. In R. Hickey (ed.), Motives for Language Change. Cambridge: Cambridge University Press, 124–139. ———. 2005. Comments at the workshop on “Historical Changes in Japanese: With Special Focus on Subjectivity and Intersubjectivity”. The 9th International Pragmatics Conference, in Riva del Garda, Italy. Traugott, Elizabeth C. & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Uchio, Kumi & Okamura, Kazue. 1973. “Joshi sooran (List of particles)”. In K. Suzuki & O. Hayashi, (eds.), Joshi (Particles) [Hinshibetsu Nihonbunpoo Kooza 9]. Tokyo: Meiji Shoin, 249–88. Watanabe, Minoru. 1971. Kokugo Koobunron (A Study on Japanese Sentence Structure). Tokyo: Hanawa Shoboo.

Chapter 6

Italian guarda, prego, dai. Pragmatic Markers and the Left and Right Periphery Chiara Ghezzi and Piera Molinelli 1 Introduction1 In Italian, verbs like guarda “look.imp.2sg”, prego “pray.prs.1sg”, dai “give. imp.2sg” are frequently used with pragmatic functions, particularly in the spoken language (ex. 1–3).2 (1) no guarda allora guarda se io pot ess i       te lo regal er ei neg look.imp.2sg so look.imp.2sg if I  can.subj.past.1sg you it  give. cond.1sg volentieri (lipfb14) gladly ‘No, look, so, look if I could I would gladly give it to you’ (2) A: ti ringrazio o you thank.1sg ‘Thank you’ B: prego (cr-ifamdl06) prey.1sg ‘You are welcome’ 1 This paper is a product of the PRIN project Contact and Change in the History of Mediterranean Languages coordinated by Marco Mancini. More specifically, the authors have worked within the Research Unit at University of Bergamo, whose coordinator was Piera Molinelli (PRIN 2008, prot. EHLWYE, sponsored by Italian Ministry of Education and Research, MIUR). Although the present contribution is conceived together. Sections 2–3 have been written by Piera Molinelli and Sections 0, 3 and 4 by Chiara Ghezzi. 2 Data analysis is based on two corpora of contemporary spoken Italian. The two corpora are Lessico dell’Italiano Parlato (De Mauro, Mancini, Voghera & Vedovelli 1993) and the Italian Section of the parallel corpus C-Oral-Rom (Cresti & Moneglia 2005). Examples in this contribution are taken from the first (LIP) or from the second corpus (CR) together with their respective transcription conventions. Some of the illustrative examples are also taken from everyday interactions. © koninklijke brill nv, leiden, ���4 | doi 10.1163/9789004274822_007

118

ghezzi and molinelli

(3) dai / sbrig a ti (cr-ifamcv24) give.imp.2sg hurry.imp.2sg.refl ‘Come on, hurry up’

These verbs have been variously classified as discourse markers (Dostie 2004, Bazzanella 2006, Waltereit 2002, Fagard 2010), pragmatic markers (Brinton 2008), courtesy markers (Akimoto 2000) or interjections (Poggi 2009, Norrick 2009). Their uses allow us to define them as pragmatic markers (pms), i.e. a class of elements which includes “a number of different kinds of expressions of stance to text and addresses” (Traugott 2010: 109). Elements often generically classified as pms are characterized by different functions and orientations to a) discourse coherence and textual cohesion (e.g. topic shifters), b) identity, personal coherence relating to the expression of the speaker’s point of view (e.g. interjections), c) social cohesion (e.g. courtesy markers). Several markers are prototypically polyfunctional as they can perform different functions in different contexts, but also within the same context. This polyfunctionality is often described and analysed on the basis of data that necessarily rely on contextual interpretations and, in some cases, on intuitions. Given this functional versatility of pms, a purely functional approach to their description is therefore to complement with reflections on their formal properties.3 In particular, like other pms, verb-based pragmatic markers (vpms) in focus here

• are functionally extra-sentential and frequently appear peripherally in the utterance, • have a variable position, • have specific prosodic contours, • have variable scope, • have procedural, rather than propositional meaning (Sperber & Wilson 1995). Wilson (2011) explains how procedural expressions guide the inferential process by imposing constraints on the construction of the context for message interpretation. What procedural expressions have in common is not the cognitive function they trigger, but rather their triggering role. Therefore pms can 3 Although functional approaches to the analysis of pms are more common, a number of studies have also taken into consideration their formal properties as Schiffrin (1987: 31, 323, 328), Bazzanella (1995), Brinton (1996), Waltereit and Detges (2007: 63).

italian guarda, prego, dai and the left and right periphery

119

have procedural dialogual functions in guiding the interlocutor to focus on the process of enunciation (enunciation level) and to act accordingly, or on the content of the utterance (utterance level) and to to seek for its cognitive effects. Both functions are exemplified through the use of guarda in (4) and (5), respectively. (4) Guarda, se le chiedi scusa, tutto si sistema. vpm if her ask.2sg pardon everything refl fix ‘Look, if you apologize, everything will be all right’ (5) Guarda che non è mica difficile! vpm that neg be.3sg neg difficult ‘Look, this is not difficult!’

Moreover, Sperber, Clèment, Heintz, Mascaro, Mercier, Origgi and Wilson (2010) consider the role of some procedural elements (e.g. modal particles or evidential markers) as relating to “epistemic vigilance”, described as what forces the speaker to present the information by relating it to the hearer’s background assumptions. On this account, the role of pms is dialogic, since their function is not to guide the comprehension process in one direction or another, but to display the speaker’s stance towards the enunciation process (ex. 6) or the content of the utterance (ex. 7). (6) È meglio che tu faccia i compiti, guarda. is better that you do.subj.1sg the homework vpm ‘You should do your homework, look’ (7) Il problema è davvero complesso, guarda. the problem be.3sg really complex vpm ‘The problem is really complicated, look’

Other formal properties, instead, are verb-specific, as the fact that vpms don’t admit complementation or negation, only sporadically appear with the subject, or with a vocative coreferential with it (Martín Zorraquino & Portolés Lázaro 1999, Dostie 2004). The property of these elements to appear peripherally, in the leftmost (lp) or rightmost (rp) position of a turn, utterance or phrase seems particularly significant, because it correlates with other formal properties and is relevant in the development of peculiar pragmatic functions.

120

ghezzi and molinelli

Distributional properties together with morphosyntactic and semantic characteristics of the verb source, play an important role in the process of (inter)-subjectification,4 which has semantic as well as syntactic correlates; pragmatic-semantic change occurs sequentially as new meanings appear in syntactic most favouring contexts, advance to less favourable ones and finally innovative meanings become more abstract and polysemous, less dependent on the surrounding syntactic and semantic context (cf. Company Company 2006). The process giving rise to (inter-) subjectification has correlates in the function-form relationship and can be observed through constructional (formmeaning) variability in synchronic distributions. For these reasons this paper intends to examine, through a semasiological approach, the interaction between the pragmatic functions of three classes of vpms in spoken Italian and the position they occupy in different discourse units. Its aim is to describe this interaction also considering the type of discourse unit in which vpms appear and they have scope over, the dialogual vs dialogic context of occurrence, and corresponding variations in types of functions performed. The organization of the paper is as follows: in Section 2 we will present a brief account of how the relationship between vpms and syntactic periphery has been treated in literature; in Section 3 we will describe the three classes of Italian vpms and the methodological basis for their analysis; in Section 4 is examined the empirical evidence for correlations between pragmatic functions of three vpms, their syntactic position and the type of discourse unit in which they appear; finally, in Section 5, we will try to summarize results and discuss them. 2

vpms: Formal Properties and Syntactic Periphery

Although rarely studied systematically, the process giving origin to vpms is regular and widespread since it systematically exploits linguistic strategies (Dostie 2004); vpms are widespread interlinguistically,5 but also intralinguistically, as can be seen from the purely illustrative list below of vpms in contemporary Italian. 4 Cf. Traugott & Dasher (2002). 5 Consider for instance pragmatic values acquired in different languages by verbs of perception (Fagard 2010, Brinton 2008, Pons Bordería 1998, Fedriani, Ghezzi & van Olmen 2012), verbs of exchange and movement (Fedriani & Ghezzi 2013), and by performatives (Ghezzi & Molinelli 2012, Ghezzi forth.).

italian guarda, prego, dai and the left and right periphery

121

andare “go” (e.g. va’, va bene, vabbe’, andiamo bene) ascoltare “listen” (e.g. ascolta, ascolti, ascolta un attimo) capire “understand” (e.g. capisco, capisci, capite, ho capito, capirai) chiedere “ask”(e.g. chiedo) consentire “consent” (e.g. se mi consente, se mi è consentito) credere “believe” (e.g. credo, credi? non credi?, credimi, credo bene, crederei) dare “give” (e.g. dai, maddai) dire “say/tell” (e.g. ti dico, (tu) dici, dice X, diciamo, direi, come si dice, come posso dire, ti dico, non dico, non mi dire, dica, di’, per così dire, come dici tu, dicono, come dicono tutti, dico bene/male) fare “do/make” (e.g. fai, faccia, fa, mi fa) guardare “look” (e.g. guarda, guardi, guarda un po’, guarda che) immaginare “imagine” (e.g. immagino) mettere “put” (e.g. mettiamo, metti) parere “seem” (e.g. mi pare) pensare “think” (e.g. penso, non penso, pensi, pensa un po’, ma pensa, non pensi?) permettere “permit” (e.g. se mi permetti/permette) pregare “pray” (e.g. prego) sapere “know” (e.g. lo sai, sapete, saprai bene, che io sappia, non so, come sai, come tutti sanno) sbagliare “mistake” (e.g. se non sbaglio) scusare “excuse” (e.g. scusa, scusami, scusi un attimo, ma scusa) seguire “follow” (e.g. mi segui, ti seguo) sembrare “seem” (e.g. mi sembra) sentire “listen” (e.g. senti, senti un po’) tenere “take” (e.g. tieni, tie’) trovare “find” (e.g. trovo, trovi/trovate, non trovo) vedere “see” (e.g. vedi (un po’), vedete, veda Lei, vediamo) venire “come” (e.g. ma vieni) volere “want” (e.g. se vuoi/vuole) Be it deverbal or not, a pm can be identified formally on the basis of different criteria (Company Company 2006: 100, Bazzanella 1995): 1) 2) 3)

the form appears between pauses, the expression constitutes a whole predication in and of itself, the expression can be deleted from the proposition without altering its meaning, 4) the form does not take the usual modification, complementation and subcategorization.

122

ghezzi and molinelli

vpms,6 in particular, are characterized by recurrent morphosyntactic properties pertaining to: 1)

(2) (3)

(4)

(5)

the order of elements which includes the use of a) the verb in isolation (e.g. ascolta “listen”), b) the verb and a pre-or post-posed co-referential vocative (Dai Alberto! “come on, Alberto,” Alberto dai! “Alberto, come on”), c) the verb and a post-posed pronoun or noun (ascoltami “listen to me”, ascolta un attimo “listen a second”), d) the verb and a pre-posed pronoun (mi ascolti? “are you listening to me?”, non mi dire! “don’t tell me”). the logical type, more commonly positive, although interrogative and negative forms can be exploited pragmatically;7 tense and mood are more frequently the present indicative or the imperative, the present subjunctive or conditional; other tenses are more rarely attested (e.g. future capirai! lit. “you will understand” or past ho capito lit. “I have understood”); person is typically the first or second, more rarely the third (especially with verbs of saying); grammatical person has specific correlates in terms of pragmatic functions performed, especially in relation to the development of subjective (first person, prego “pray.prs.1sg”) or intersubjective uses (second person, guarda “look.imp.2sg”, dai “give.imp.2sg”); number is more frequently singular, but instances of second plural and first plural are attested with specific pragmatic functions (e.g. dico “say.1sg” vs diciamo “say.1pl”, dici “say.2sg” vs dite “say.2pl”).8

Morphological properties are interrelated with the syntactic structure of the clause/utterance. For instance tense and mood correlate with syntactic clause type: negative sentences are frequently in imperative (e.g. non mi dire! “don’t tell me!” with a mirative function), while interrogative structures are typically at the second person (certo che ho fatto una festa in giardino, cosa credi? “of course I held a party in the garden, what do you think?”). 6 Cf. Dostie (2004) on verb-based markers in French. 7 Consider for example the case of back channels like credi? “do you think?” vs non credi? “don’t you think?”, where the first is used as hedging device to express disagreement, and the second is used as agreement marker, to imply the interlocutor’s agreement. 8 The use of the first person plural is exploited pragmatically to include the interlocutor’s point of view (ma noi, diciamo, siamo giovani di spirito “but we, let’s say, are young at heart”), while the first singular can easily encode modal and subjective values (credevo di prendere di avere un voto non dico buono ma discreto “I thought I could have got a mark, I do not say good, but decent”).

italian guarda, prego, dai and the left and right periphery Table 6.1

123

Degrees of proceduralization

Pragmatic-semantic tendencies

non-subjective content scope within proposition truth-conditional

> subjective > content/procedural > scope over-proposition >

> > > >

intersubjective procedural scope over discourse non-truth conditional

From a syntactic point of view, the degree of freedom of vpms and, consequently, their peripheral occurrence, ties in inherently with their degree of proceduralization,9 (inter) subjectivity, and variability of scope (cf. Table 6.1, Traugott & Dasher 2002: 40). In the following sequences the vpm is dialogically motivated, as its use implies a perspectivisation. In ex. (8) guarda encodes the speaker’s belief, and mitigates a partial disagreement, is used in the rp and has scope on the whole sequence of turn exchange (i.e. on discourse), while in (9) it is used as an adversative marker in the lp and has scope only on the following speech act. (8) A: Mi   sembra che la zia stia meglio. to-me seem.3sg that the aunt be.subj.3sg better ‘I think auntie feels better’ B: Non lo so, guarda. neg it know.1sg vpm ‘Well, I don’t know’ (9) Guarda che Marco viene alle cinque, non alle sei. vpm that Marco come.3sg at-the five   neg at-the six ‘Mind that Marco will come at 5, not at 6’

9 The debate on the nature and properties of this development has been intense in recent years, especially in relation to the terminological labels that better capture its characteristic. More frequently labels as grammaticalization or pragmaticalization are used. Considering that the focus of this contribution is on vpms, we find that the term ‘pragmaticalization’ better captures the nature of their developments. Cf. Ghezzi (2014) for an overview on different theoretical approaches and consequent terminological choices, but also Erman & Kotsinas (1993), and Dostie (2004) for similar terminological choices.

124

ghezzi and molinelli

Syntactic properties also correlate with specific functional features which are dialogually motivated, as in the case of turn-managing in (10) or of discourse coherence in (11). (10) do la parola al segretario Gori, prego give.1sg the word to-the secretary Gori vpm ‘I hand it over to the secretary Gori, please’ (11) A: Piove governo ladro! rain.3sg government thief ‘It rains, thievish government!’ B: prego? vpm ‘Pardon me?’

From a functional-pragmatic perspective, the relationship between lp or rp and specific pragmatic functions has rarely been the focus of studies.10 These studies have underlined that lp is relevant for interactional ends, since elements tend to be “syntactically loose and thus interactionally and textually attractive” (Aijmer 1997: 29). From an interactional point of view lp “projects something else to come, but does not oblige the speaker to subscribe to one particular syntactic project at a time when s/he may still be in the phase of planning” (Auer 1996: 313). Moreover lp is textually ‘appealing’ since it contains the sentence theme, which is associated with different functions as topic/ subtopic introduction, or relates what is said to the preceding context. From a discourse-analysis perspective lp and rp do not behave in a specular way: rp is in principle expandable and, therefore, flexible and underdetermined (Schegloff 1996), while lp is syntactically ambiguous. Therefore early constituents can be progressively redefined as belonging to different discourse units. Aijmer (1997) considers the exploitation of each of these positions as a technical solution to interactional problems: rp expandability makes the termination of a discourse unit negotiable, thus providing an important repair space which minimizes gaps and secures understanding; while lp syntactic ambiguity helps the speaker to take over the turn without the need to plan fully the subsequent utterance. 10

In the generative paradigm, on the contrary, the interrelation between structural position and pragmatic functions has been extensively studied. In such a perspective left periphery, by virtue of its precedence to predication and prosodic prominence, has been associated with marking topic continuity, contrastive topic and focus narrowing, while right periphery with marking focus narrowing or widening (cf. Rizzi 1997, Adger, de Cat & Tsoulas 2004).

italian guarda, prego, dai and the left and right periphery

125

Moving from Aijmer’s (1997) conceptualization of lp and rp as technical solutions to interactional problems, we will analyse how their respective properties interact with different classes of vpms in spoken Italian. 3

Classes of vpms and Periphery

Having spoken Italian as basis of research, we focus on three functional classes of vpms. The classes include verbs which 1) share the properties of conveying the speaker’s point of view in relation to the content of the utterance, the speech event or the interlocutor, 2) can be identified on the basis of their formal properties, 3) show different degrees of pragmaticalization. The first class of vpms includes a group of elements derived from perception verbs, here exemplified through guarda “look”, which are used pragmatically as attention getters (ex. 12). (12) Guarda, io mi sono informato oggi (lip-mb30) vpm i refl aux inform.ptcp today ‘Look, I got some information today’

In Italian vpms can formally be used at the first person plural, the second person singular and plural, or at the third person singular11 of the imperative (guard-a “look-2sg”, guard-i “look-3sg”, guard-iamo “look-1pl” guard-ate “look2pl”); these vpms have a perlocutive value as they are used to get the interlocutor’s attention through the exploitation of the intrinsic deictic feature derived from the semantics of verb (cf. Waltereit 2002). Although they are used frequently as vpms, formally these verbs still have a recognizable paradigm, can be used with modulation devices as nouns or adverbs (cf. guarda un po’ lit. “look a bit”, guarda bene lit. “look well”) and they do not reach the status of frozen forms. A second group of vpms is derived from performative verbs (as prego “I pray” and chiedo “I ask”) which are used as courtesy markers (cms) (ex. 13). (13) Prego, mi segua vpm me follow.imp.3sg ‘Please, follow me’

11

In Italian the third person singular corresponds to the first person of the present subjunctive and it is typically used as courtesy form.

126

ghezzi and molinelli

These vpms can be employed pragmatically only at the first person singular of the present indicative. Their use as cms and their propositional uses as ordinary verb constructions exhibit some degree of ‘divergence’ (Thompson & Mulac 1991). While the verb can be modulated by an adverb, the cm cannot (ex. 14). In such contexts the degree of syntactic integration of the verb is relevant in determining the possibility of modulation. (14) a. *Prego vivamente, mi segua vpm warmly me follow.imp.3sg ‘*Please warmly, follow me’ b. La prego vivamente di seguirmi acc.3sg pray.prs.1sg warmly to follow.inf-me ‘I warmly pray you to follow me’

For these vpms it is possible to distinguish verbs maintaining a syntactic integration and semantic propositional content, which can also be used as pms (chiedo, ex. 15), and verbs, as those in focus here, which are ‘frozen’ as cms (prego).12 (15) presidente / chiedo scusa // non so se ci sia qualche teste president ask.1sg apology neg know if there be.subj.3sg some witness in aula (cr-inatla03) in  courtroom ‘President, I apologize, I do not know if there is any witness in the courtroom’

A third group of vpms derive from verbs of exchange (dai “give”, to’, tie’ “take”) or movement (va’ “go”) which are pragmatically employed as injunctive, phatic or expressive markers (ex. 16). (16) dai raccontaci (lipfa2) vpm tell.imp.2sg-us ‘Come on, tell us!’

Such vpms, analysed here through the case of dai, appear only at the second person singular of the imperative and are generally labelled as “interjections” in reference grammars and dictionaries (Poggi 2009). This treatment highlights

12

Cf. Akimoto (2000), Traugott & Dasher (2002), Ghezzi & Molinelli (2012).

italian guarda, prego, dai and the left and right periphery

127

Table 6.2 Formal properties of the three classes of vpms Lexical source

Perception verbs

Example

guarda, guardi, guardiamo, guardate

Performative verbs prego Exchange or movement verbs

dai

Mood and tense

Person and number Pragmatic funtion

Attention getters I person plural II person singular/ plural III person singular (polite form) Present indicative I person singular Courtesy markers Imperative II person singular Phatic markers Imperative

the fact that their form is totally frozen, their original semantic meaning is bleached, they are incompatible with modulation, and their use is ritualized and characteristic of informal registers (cf. Fedriani & Ghezzi 2013). At the synchronic level, formal properties of the three classes of vpms are summarized in Table 6.2. The formal properties summarily described here highlight that vpms are characterized by different and increasing degrees of pragmaticalization moving from attention getters, to courtesy markers and phatic markers. The class of attention getters is not fully pragmaticalized, as deictic uses still co-exist with pragmatic functions, and modulation is still possible for all verbs of the class; the same happens for some, but not all, verbs belonging to the class of cms (it holds for chiedo, but not for prego), while verbs belonging to the third class show a completed pragmaticalization pattern. If morphological properties of the verbs at the synchronic level are indicators of the degree of functional pragmaticalization, their syntactic position contributes to the definition of their functional roles. It is possible to identify correlations between the position a marker occupies and the discourse units it affects. Particularly relevant are: (a) the correlations between the classes of markers and their positions within different discourse units ((lp), (rp) or intermediate position (ip)); (b) the correlations between the functions performed by each marker and the position it occupies in these units; (c) variations in the distribution of such correlations in the three classes.

128

ghezzi and molinelli

Table 6.3 Discourse units in the Val.Es.Co. model Levels

Dimensions

Dialogic Monologic

Structural

Social

Dialogue exchange Intervention, Act

Turn-taking Turn

Informative

Subtract

Relevant methodological issues relate to a) what type of discourse unit the vpm affects (a single constituent or phrase, a speech act, a whole intervention), b) in what dimension of language (structural, social or informative) the vpm functions. Basing our analysis on the Val.Es.Co. model (Briz & Val.Es.Co, 2003), we have taken into consideration the system of discourse units represented in Table 6.3, which has the advantage of highlighting how the structural, social and informative dimensions of language interact in defining the structure of conversation. Relevant elements for data analysis are: 1) 2)

3)

4

the social dimension, which is characterized by the system of turns and identified by the change of speaker, the structural dimension, which is characterized by structures of conversation, i.e. structural units as interventions and acts. Interventions, structural units resulting from the purely structural fact that a speaker has uttered a chain of sounds, are delimited by a change of speaker and have immediate identifiable, isolable and autonomous constituents (i.e. acts), the informative dimension, characterised by structures which are clearly distinguishable, but not autonomous (i.e. subacts). Corpus Analysis

vpms belonging to the three classes under analysis here appear in various positions in different discourse units, showing different trajectories of development and patterns of distribution, and peculiar functions in correlation with specific positions.

italian guarda, prego, dai and the left and right periphery

129

4.1 Attention getters: Guarda The pragmaticalization of the attention getter guarda has been widely documented for Italian as well as for its correspondents in other languages.13 Although positions can vary consistently in this regard,14 from the literal meaning of the imperative as an invitation to direct the sight toward something in the deictic context, the marker progressively acquires a primary phatic value which is dialogually motivated; in these contexts the vpm has a performative nuance which invites the listener to direct her attention towards the process of enunciation (paraphrase listen to me).15 The primary phatic value develops a secondary diologic function through which the speaker expresses her stance toward the previous/following utterance by inviting the listener to pay attention to its content (paraphrase consider, pay attention to what I am saying); in these contexts the vpm functions as marker of relevance and can be interpreted as an instruction to the listener to process previous or subsequent utterances as relevant information (Pons Bordería 1998).16 direct the sight toward something in the deictic context > direct the attention toward the process of enunciation > direct the attention toward the content of utterance

Pragmatic functions, therefore, are based on both the original deictic meaning of the verb and subsequent developments of the extended meaning “consider”. These developments have morphological as well as syntactic correlates. The first imply a partial loss of alternation in flexional morphology and a tendency to fixation of the form in the second person singular of the imperative; nevertheless, as Figure 6.1 shows, pragmatic uses of the marker in spoken corpora still admit variation in person (guard-ate “look-2pl” vs guard-iamo

13

14 15 16

Consider for example Brinton (2001) on English, Pons Bordería (1998) on Spanish, Fagard (2010) on Romance languages, Iliescu (2009) on Romanian and French, Waltereit (2002) and Ghezzi (2012) on Italian perception verbs. Cf. also Fedriani, Ghezzi & van Olmen (2012) for a typological study of pragmatic markers derived from verbs of perception. Cf. for example the diverging explanations given in Waltereit (2002) and Ocampo (2009). Martín Zorraquino and Portolés Lázaro (1999) categorize this use of the vpm as “alterity focalizing”. Sweetser (1993) shows how this is a common trend in meaning change which consists in a metaphorical shift from the sociophysical, content world, to mental world of reason, to the world of speaking. Both Sweetser (1993) and Ocampo (2009) suggest that the development of extended meaning “consider” is secondary.

130

ghezzi and molinelli 8 0 4 guarda guardi guardiamo guardate

Figure 6.1 Guarda: Fixation of forms and loss of alternation in flexional morphology

“look-1pl”), number (guard-a “look-2sg” vs guard-ate “look-2pl”), in degree of formality (guard-a “look-2sg” vs guard-i “look-3sg”), although one form (second person singular guarda) is by far more frequent than others. From a syntactic point of view, the units at left or right of the verb cannot be interpreted as constituents of the verbal phrase, but the verb maintains regular concordance in all cases. In addition, the verb can appear with a vocative, more frequently in post-verbal position (ex. 17). (17) guarda / maria / ma tanto / io nella vita / io ho imparato / che perché non vpm Mary but anyway I in-the life     I aux learn.ptcp that why not posso   vivere / quest’attimo? (cr-ifamdl20) can.1sg live.inf this-moment ‘Look, Mary, but anyway, in life, I learned that why can’t I live this moment?’

Pragmatic functions acquired by the marker can be correlated with its syntactic position. As expected, guarda occurs more frequently in lp, where it operates in structural, social and informative dimensions. The centrality of the phatic function, derived from the semantic of the verb and from the imperative mood, implies calling the addressee’s attention to the process of enunciation. This phatic function, focused on the interlocutor, is at the origin of social and intersubjective uses of the marker as a turn-taking device, especially in competitive turns (ex. 18). (18) A: [ allora vengo nel pomeriggio va bene? then come.1sg in-the afternoon go.3sg well ‘Then I’ll come in the afternoon, all right?’

italian guarda, prego, dai and the left and right periphery

137

M: prego prego (lipma13) vpm vpm ‘please, please’

As in the case of guarda, the use of prego originates from a context in which the speaker asks the interlocutor to perform an action. The request shifts to enunciation level, where prego can introduce a request for repetition or reformulation, typically with rising intonation (ex. 33). (33) A: Sinistra Democratica // video connection left democratic video connection ‘Democratic Left, video connection’ B: prego? (cr-imedts03) vpm ‘Pardon me?’

Similar uses of the marker, operating on the level of enunciation, are frequently found in multi-party debates where turn managing is relevant. In such contexts, prego is used as a turn-giving device, in the lp or in rp (ex. 34, 35). Its different positions typically correlate with its degree of syntactic freedom and lp has a lower degree of freedom compared to rp. (34) ecco / prego Gianluca di [/] di approfondire subito (cr-ipubmn01) here pray.1sg Gianluca to to elaborate.inf now ‘OK, I pray Gianluca to elaborate on this now’ (35) la parola al presidente Marcucci prego (cr-IPUBMN01) the word to-the president Marcucci vpm ‘the word to president Marcucci, please’

Similarly, prego is used as a turn-managing device typically in competitive turns, where the marker appears in the lp (ex. 36) or as an independent speech act. (36) Prego, prego, finisca pure. vpm vpm finish.im.2sg also ‘go ahead, you can finish’

138

ghezzi and molinelli

Finally, the vpm can be found in the last turn exchanges in interventions where it functions as a closing or pre-closing device; in these contexts the marker occurs more frequently in the lp (ex. 37) or as independent speech act in minimal pairs as Prego, arrivederci “you are welcome, goodbye”. (37) [ maddai). (38) Dai, venite con noi! vpm come.imp.2pl with us ‘Come on, come with us!’

Its functions derive from the semantics of its source, the exchange verb dare “to give” and are deeply embedded in the basic canonical schema of interactional verbal exchange.19 The vpm has a conative and intersubjective meaning which implies the speaker’s appeal for hearer cooperation in an action or a discourse task (Schiffrin, 1987: 63). More prototypical functions are focused on the deictic contexts or on the act of enunciation. In both uses the marker is dialogually motivated and appears with an injunctive meaning in directive acts. The vpm can be used in both the lp and rp as action initiator in orders, requests and suggestions; it has a perlocutionary value and, depending on

19

Cf. Fedriani & Ghezzi (2013) on the pattern of development of vpms from verbs of exchange in Italian, Latin and Greek.

italian guarda, prego, dai and the left and right periphery

141

prosody and position, functions as a politeness-motivated booster or hedge (ex. 39 and 40, respectively). (39) dai / sbrigati // (cr-ifamcv24) vpm hurry.imp.2sg.refl ‘come on, hurry up’ (40) [ ‘yesterday came rel people’. This head-final relativization strategy is similar to that found in verb-final languages such as Japanese and Korean, and is a mirror image of the head-initial relativization strategy in verb-medial languages such as English (cf. the people, the ones who came yesterday > the people who came yesterday). Similar to verb-medial languages such as English, complement-taking predicates in Chinese (e.g. Cantonese jan4dei6 waa6 ‘People say’ and Mandarin wo juede ‘I feel/think’) can also develop into evidential and epistemic markers,

© koninklijke brill nv, leiden, ���4 | doi 10.1163/9789004274822_009

180

yap, yang and wong

first at the left periphery preceding the insubordinated complement clause (e.g. People say evidential markers / I think epistemic marker[she will dump him] new main ), as in (1a) and (2a), and subsequently at the right periphery after the clause insubordinated complement clause, via right-dislocation (e.g. [She will dump him] main clause, people say evidential marker + speaker’s disclaimer / I think epistemic marker + ), as in (1b) and (2b). speaker’s disclaimer (1)

Cantonese hearsay evidential jan4dei6 waa6 ‘people say’ a. 人哋話佢會飛起佢個男朋友 jan4dei6 waa6 keoi5 wui5 fei1hei2 keoi5 go3 naam4pang4jau5 people say 3sg will dump 3sg cl boyfriend ‘People say she will dump her boyfriend.’

b. 佢會飛起佢個男朋友，人哋話 keoi5 wui5 fei1hei2 keoi5 go3 naam4pang4jau5, jan4dei6 waa6 3sg will dump 3sg cl boyfriend people say ‘She will dump her boyfriend, people say.’ (2) Mandarin epistemic parenthetical wo juede ‘I think’ a. 我覺得她會拋棄她的男朋友 wo juede ta hui paoqi ta de nanpengyou 1sg think 3sg will dump 3sg gen boyfriend ‘I think she will dump her boyfriend.’ b. 她會拋棄她的男朋友，我覺得 ta hui paoqi ta de nanpengyou, wo juede 3sg will dump 3sg gen boyfriend 1sg think ‘She will dump her boyfriend, I think.’

On the other hand, similar to verb-final languages such as Japanese and Korean, the Chinese language deploys numerous other strategies whereby pragmatic markers emerge in sentence final position without recourse to right dislocation from the left periphery to the right. One highly productive strategy in Mandarin involves the reanalysis of complementation structures (essentially a form of relativized or nominalized clauses) as stand-alone finite structures. For example, Mandarin shi . . . de focus constructions, which comprise of the copula focus particle shi and a complementation clause marked by nominalizer de, can have its focus particle elided to yield a shi-less assertion that is amenable to a wide range of pragmatic interpretations, often still with strong assertive force that gives rise to its use as a marker of speaker’s reassurance to

on the development of sentence final particles in chinese

181

the addressee (e.g. ta shi xihuan ni de ‘It’s (true) that he likes you’ > ta xihuan ni de ‘He likes you (I assure you)’. An intriguing follow-up question is whether new strategies not found in either predominantly verb-medial (vo) languages such as English or predominantly verb-final (ov) languages such as Japanese and Korean have emerged in Chinese, which show mixed characteristics involving head-initial (e.g. vo) and head-final (e.g. complement clause + head noun) structures. The quick answer is, yes we do find some such structures. Recall the example of the Cantonese hearsay evidential marker jan4dei6 waa6 ‘people say’ from (1a) and (1b) above. Note, in particular, that subject omission is common in Chinese. This allows right-dislocated hearsay evidential markers to develop into subject-less evidential markers, often accompanied by other sentence final particles. In Cantonese, as seen in (3) below, the subject-less hearsay evidential marker waa6 can combine with the emphatic sentence final particle o3 to form a hearsay evidential sentence final particle wo3 with pragmatic nuances such as counterexpectation marking (mirative reading) (see Leung 2006). In other words, we see the following development: hearsay evidential utterance tag jan4dei6 waa6 + emphatic particle o3 > subjectless hearsay evidential marker waa6 + emphatic particle o3 > hearsay evidential sentence final particle wo3 (often with mirative meaning). (3) Cantonese hearsay evidential jan4dei6 waa6 ‘people say’ 佢會飛起佢個男朋友喎 keoi5 wui5 fei1hei2 keoi5 go3 naam4pang4jau5 wo3 3sg will dump 3sg cl boyfriend evid mirative reading possible ‘She will dump her boyfriend, they say(!)’

What is interesting is that not all utterance tags readily merge with the preceding clause that they modulate. There is asymmetry in frequency not only across languages (e.g. unlike Cantonese, English tends to resist reanalyzing their utterance tags—such as hearsay evidential people say or epistemic marker it seems—into sentence final evidential particles), but also there is asymmetry across pragmatic markers within the same language (e.g. in Cantonese, speakers prefer to retain the first person subject ngo5 ‘I’ in the epistemic utterance tag ngo5 gok3dak1 ‘I think’, which then impedes the emergence of gok3dak1 as a subjectless epistemic sentence final particle). Clearly, various strategies are used in the development of pragmatic markers across languages, and various factors contribute to variation not only across but also within languages. In this paper, we will identify six major restructuring processes that contribute to the rise of sentence final particles—i.e. pragmatic

182

yap, yang and wong

markers at the right periphery—in Chinese. More specifically, we will examine the following six processes: verb serialization that gives rise to syntactic relabeling (§2), clausal integration (§3), right-dislocation (§4), a combination of right-dislocation and clausal integration (§5), main-clause ellipsis (§6), and insubordination in the form of ‘stand-alone’ nominalization (§7). We will then conclude with some observations on the similarities and differences in the strategies used in Chinese, a somewhat atypical verb-medial language, and some verb-final languages such as Japanese and Korean (§8). In this way, we hope our analyses of the diachronic development of various types of Chinese pragmatic markers at the right periphery will contribute to a better understanding of functional overlaps at the left and right peripheries, which nevertheless often differ in meaningful ways, either in terms of frequency or in terms of subtle pragmatic nuances, or both. 2

Verb Serialization and Syntactic Relabeling: Semantic Scope Expansion, Syntactic Restructuring, and the Rise of Sentence Final le

Semantic extension is a ubiquitous phenomenon that is endemic to human cognition, and extensions that result in scope expansion will then trigger syntactic restructuring, sometimes very subtle and covert, within the language system. In this section, we will briefly discuss the development of Mandarin le to illustrate how such semantic scope expansion and syntactic restructuring works in Chinese as a result of verb serialization. The exact etymology of le remains somewhat controversial. One hypothesis (e.g. Cao 1995) is that le is derived from a lexical verb liao meaning ‘finish’, attested in both Early as well as Late Middle Chinese as seen in (4a) and (4b) respectively.1 In serial verb (i.e. V 1 V2) constructions such as shai liao (‘dry up’, lit. ‘dry finish’), liao in V2 position was reinterpreted as a completive aspect marker, as in (4c).2 As a completive aspect marker arising from a serial verb construction, liao in Middle Chinese was still interchangeable with other aspect markers with completive marking functions such as jing 竟, bi 畢, qi 訖, and yi 已. In Late Middle Chinese and increasingly from Early Modern

1 Another hypothesis is that le as a perfect marker is derived from lai ‘come’ (Chao 1968; see also Sun 1996). 2 Note that the second liao, which forms the reverend’s reply, is a lexical verb usage of liao in the sense of ‘finish’, similar to (4a). Hence the reverend’s reply liao ye simply means ‘It’s finished’ / ‘It’s done.’

on the development of sentence final particles in chinese

183

Chinese, within a sequential-type biclausal construction, i.e. (np) vp-liao, (np) vp, where the completion of the first event is crucial to the realization of the second event, liao (or its reduced phonological variant le) came to be reinterpreted as an anterior or perfect tense-aspect marker, as in (4d). As an anterior or perfect marker, liao ~ le marks not only completion of an event but also a change of state with current relevance to the speech situation (crs).3 Ellipsis of the subsequent clause still allowed for the retention of the anterior or perfect marker interpretation of liao ~ le, as in (4e), (4f) and (4g). With phonological reduction (liao > le), particularly in Early Modern Chinese, postverbal completive aspect marker le came to be reanalysed as a perfective suffix (i.e. vo le > V-le O). By the 15th century, the V-le (O) construction has become the dominant perfective structure (Sun 1996: 85). In Modern Chinese, the phonologically reduced tense-aspect marker le can thus either be suffixed to the preceding verb as a perfective aspect marker, with scope over the predicate, as in (4h), or it can be used as a clause-final perfect marker with sentential scope and pragmatic import, as in (4i) and (4j).4 3 See Comrie (1976, 1985) for helpful discussions on the distinctions between the different types of tense-aspect markers within the perfect(ive) domain, and see also Bybee, Perkins and Pagliuca (1994: 51−105) for a diachronic development perspective. 4 Roughly speaking, previous studies identified two types of le: le1 (which functions as a perfective aspect suffix, i.e. a verbal aspect marker) and le2 (which functions as a perfect marker, and often as a sentence final particle (Chao 1968; Li & Thompson 1981; Sun 1996). Crucially for our present study, in terms of pragmatic function, this sentence final particle le2 can convey an implicit message that the speaker is done with his/her turn and now invites the addressee to take up the conversational floor (Lu & Su, 2009). In their quantitative analysis of Taiwanese Mandarin conversational data, 71.56% of sentence final le2 tokens are followed by a floor change from speaker to hearer, which suggests that the addressee responds to le2 as an indicator that the speaker has completed his turn. This turn-completion marking function of le2 in sentence final position is illustrated below, Note that the interviewer (ir) completes her speech turn with sentence final le, and this is followed by an enthusiastic uptake by the hearer, as seen in his effusive agreement with the reduplicated affirmative expression dui dui ‘yes, yes’. This type of solidarity-marking uptake following le2 occurs with sufficient and significant frequency, as reported in Lu & Su (2009), to be identified as a turn-completion and turn-transition marker in Mandarin conversations. IR: 就它現在，反正現在我們就另一個眼光看它了。 jiu ta xianzai, fanzheng xianzai women jiu yong ling yi ge just 3sg now anyway now 1pl just use another one cl yanguang kan ta le. viewpoint see 3sg crs ‘Now it’s . . . we just look at it from another perspective anyway.’

184

yap, yang and wong

(4) a. 官事未易了也 guan-shi wei yi liao ye official-matter neg easy complete prt ‘The government business is not easy to finish.’ ( Jinshu Fuxian zhuan; Tang period, 648 ad) b. 事了矣 shi liao yi thing finish asp ‘The thing is done.’ (Da Tang Xin Yu, Vol. 4, Tang period, 807 ad) c. 問僧：曬了也未？ wen seng, shai liao ye wei? ask monk dry finish prt neg (Someone) asked the monk, ‘Have you finished drying (it) in the sun?’ 僧云：了也。 seng yun, liao ye. monk say finish prt
 The monk replied, ‘It’s done.’
 (Yunmen Kuangzhenshanshi Guanglu, Tang period, 9th century) d. 理會得這箇了，他日若有材料， lihui-de zhe-ge liao,     tari ruo you cailiao, understand-pot dem-cl complete future if have material 却依此起將去，只此一箇道理 que yi ci qi jiang qu, zhi ci yi ge daoli. then according.to this build prt go only this one cl principle

IE: 對對，也，也是，也是記錄那個，那段兒嘛。 dui dui, ye, ye shi, ye shi jilu nage, na duanr ma. yes yes also also cop also cop record dem dem cl sfp ‘Yeah yeah, also, (it) also, also represents that, that period of history.’ (Data from the Mandarin corpus of the project “Stance Marking in Asian Languages: Linguistic and Cultural Perspectives” ).

on the development of sentence final particles in chinese

185

‘Once (one) has understood this, one can erect (a house) on this (foundation) when materials are ready. It’s just this very (simple) principle.’ (Zhuzi Yulei 14/250; Southern Song period, 1270 ad) e. 喫飯了也 chi fan liao ye eat food complete crs ‘(One) had finished eating the meal.’ (Zutangji 1/166/7; Five Dynasties period, 952 ad; cited in Liu 1985: 132, and Sun 1996: 88)

f. 作此語了 zuo ci yu liao make this words complete ‘(He) finished making this statement.’ (Dunhuang Bianwen, Tang period, 7th–10th century; cited in Sun 1996: 88)5

g. 皆變壞了 jie bian huai liao all turn bad crs ‘All (of them) have turned bad.’ (Zhuzi Yulei, Southern Song period, 1270 ad) h. 喫了酒也 chi liao jiu ye eat pfv wine crs ‘(One) has drunk (some) wine.’ (Laoqida A/58/7, Ming period, 15th century; cited in Sun 1996: 99) i. 季子之罪，不在放走了慶父， jizi zhi zui, bu zai fangzou liao qingfu, pn attr crime neg at release pfv pn 先已自有罪過了。 xian yi zi you zuiguo liao. before already self have error crs

5 Dunhuang Bianwen is a collection of manuscripts from the Tang period that were discovered in the Mogao Caves of Dunhuang in 1899; the exact dates of these texts are unknown.

186

yap, yang and wong ‘Jizi’s crime is not that he has released Qingfu, but that he has already commit ted an error before.’ (Zhuzi Yulei, Southern Song period, 1270 ad) j. 我早做完了 wo zao zuo wan le 1sg early do finish crs ‘I’ve finished (it) a long time ago.’ (with the implication ‘so now I can or should be able to do X’) (Contemporary Chinese)

Liu (1985) noted that prior to the 10th century, le was not used in sentence final position; there had to be either another clause following the le-clause, or le had to be followed by a sentence final perfect marker such as ye or yi. This suggests that prior to the 10th century, le was still lexical (hence pronounced as liao), as in (4a), or le was a completive aspect marker (still pronounced as liao) and could be accompanied by a perfect marker, as in (4c), not unlike modern Mandarin zuo wan le (lit. ‘do finish already’ > ‘have finished doing’), where wan and le are the contemporary completive and perfect markers respectively. In terms of semantic scope expansion, we see an extended use of a lexical verb liao ‘finish’ being reanalyzed as a tense-aspect marker with perfective or perfect (i.e. anterior aspect) meaning, and subsequently further used as a pragmatic marker to signal the completion of the speaker’s turn and an invitation to solicit the addressee’s involvement (Lu & Su 2009).6 In functional terms, this is an extension from the propositional domain to grammatical and interpersonal domains (à la Traugott 1989; 1995; 2003).7 In syntactic terms, we see a series of restructuring and relabeling arising from verb serialization as in Figure 8.1 below. Crucially, we see scope expansion of liao ~ le from the vp (verbal) domain to the tp (tense-aspect) domain and ultimately to the cp (speaker stance) domain, where sentence final particles typically reside. Such expansions, which Roberts and Roussou (2003) discuss in terms of V>v>(T)>C movements, are widespread across the languages of the world.8 6 In earlier work, van den Berg & Wu (2006) have shown that le is also used as a ‘common ground coordinator’ between discourse participants. 7 Recent work on sentence final particle liao in Singapore Mandarin reveals a wider range of pragmatic uses than seen for Standard Mandarin le. Among the functions are the speaker’s expression of seriousness, surprise, hopelessness and frustration (Lee & Cheong 1999). 8 The term ‘V>v>(T)>C movement’ refers to semantic and syntactic scope expansions involving structural reanalyses whereby verbal elements (V) are sometimes reinterpreted as

on the development of sentence final particles in chinese Stage 1 Stage 2 Stage 3a Stage 3b Stage 4a Stage 4b

187

(main verb) [TP NP [VP liao ]] (sequential verb) [TP NP1 [VP V (NP2)]], [VP [VP liao]] (completive aspect)9 [TP NP1 [AspP [VP V ] liao (NP2)]] (perfect aspect)10 [TP NP1 [AspP [VP V (NP2)] liao ~ le]] (perfective aspect) [TP NP1 [AspP [VP V ] liao ~ le (NP2)]] (sentence final particle) [CP [TP NP1 [AspP [VP V (NP2)]] le]

figure 8.1 Stages in the syntactic restructuring and relabeling of Mandarin interactional particle le

Semantic scope expansion and syntactic restructuring are part and parcel of the grammaticalization of all sentence final particles, not only in Chinese but also in other languages. In verb-final languages such as Japanese and Korean, semantic scope expansion and syntactic restructuring readily take place at the right periphery, as illustrated with Japanese adversity marker -te shimatta (and its phonologically reduced variant chatta), with examples shown in (5a–f).11 As seen in (5a), shimau is a lexical verb meaning ‘to put away’. It can occur in a converbal V1-te V2 construction, e.g. yatte shimaimashoo ‘let’s finish doing (this)’ in (5b). In realis contexts, the past form -te shimatta can be used as a completive and perfective aspect marker, as in (5c) and (5d). As seen in (5e) and (5f), V-te shimatta constructions are often phonologically reduced to V-chatta, and



9

10

11

aspectual markers (v), with the possibility of further being reinterpreted as tense markers (T) and complementizers (C). Extensions to T are optional in languages such as Chinese, which are often referred to as ‘tenseless’ languages, given the lack of obligatory inflectional tense marking in the language, unlike tensed languages such as English. Completive aspect marker le can be suffixed to the verb to form perfective le as shown in (i) to (ii) below. (i) [TP NP1 [AspP [VP V] liao/le (NP2) ]] (ii) [TP NP1 [AspP V-le (NP2) ]] Note that the perfect tense-aspect marker is widely recognized as a ‘relative tense marker’ (see Comrie 1985). That is, its deictic reference point need not be the moment of speaking, but can be a designated point in the past, present or future within a narrative or other type of discourse. See Strauss (2003) for a fuller discussion of the near-parallel developments of Japanese –te shimau and Korean –a/e pelita as subjectivity markers (see also Ono & Suzuki 1992). Strauss and Sohn (1998) and Yoshida (1994, 1995) have also identified Japanese chau as a social dialect and group identity marker. Note that –te shimau and chau are the non-past forms of –te shimatta and chatta.

188

yap, yang and wong

both forms are often used to mark non-volitional, unexpected or inadvertent outcomes. (5) a. kodomo ga omocha o shimatteiru. child nom toy acc put.away.prog ‘The child is clearing his/her toys away.’

b. A: nokori wa ashita shimashoo ka. rest top tomorrow do.adhort q ‘Shall we do the rest of (the work) tomorrow?’ B: demo, kore dake yatte shimaimashoo. but this only do.conv finish.adhort ‘But, let’s just finish doing this one.’

c. moo, kaite shimatta no!? already write.conv finish.pfv sfp ‘You have already finished writing (it), haven’t you?’ d. watashi wa moo ronbun o kaite shimaimashita. 1sg top already thesis acc write.conv finish.pfv ‘I have already written the thesis.’ e. sukkari ookikunatte, michigaete shimatta/michigae chatta wa quite grow.up look.different.conv.non-vol.pfv sfp ‘You have grown up and look so different (that I hardly recognize you).’ f. okiniiri no kappu ga warete shimatta/ware-chatta. favorite gen cup top break.conv.advers ‘(My) favorite cup has been broken.’ (expressing sadness/regret)

Note that the verb concatenation process that we earlier discussed in terms of verb serialization, which is a crucial intermediate stage in the development of Mandarin sentence final particle le, is more commonly referred to as converblinking in verb-final languages such as Japanese and Korean (see the use of –te linkage for Japanese –te shimau and –a/e linkage for Korean –a/e pelita sentence enders). The grammaticalization of –te shimau as an adversity marker is summarized in Figure 8.2 below.

on the development of sentence final particles in chinese Stage 1 (main verb) Stage 2 (converbal construction) Stage 3 (completive aspect) Stage 4 (perfect(ive) aspect) Stage 5 (non-volitional marker) Stage 6 (adversity marker)

189

[TP (NP1) [VP NP2 shimau]] [TP (NP1) [VP (NP2) V-teLNK shimau]] [TP (NP1) [VP (NP2) V-te shimatta/chatta]] [TP (NP1) [VP (NP2) V-te shimatta/chatta]] [CP [TP (NP1) [VP (NP2) V]-te shimatta/chatta]] [CP [TOP (NP2)] [TP [VP V]]-te shimatta/chatta]

figure 8.2 Stages in the syntactic restructuring and relabeling of Japanese lexical verb shimau to adversity marker –te shimau/chatta

With the exception of a converbal use of –te shimau in Japanese where Chinese uses a serial verb construction for liao ~ le, and the reanalysis of –te shimau/-te shimatta as a sentence final pragmatic particle (often reduced to chau/chatta) with non-volitional and adversity readings, the development from verbal to aspectual to (inter)subjective use (i.e. V > v > C trajectory) is remarkably similar for the two languages. This development, is also attested in neighbouring languages such as Korean, and indeed is crosslinguistically robust and attested in other language families as well (e.g. Greek and Italian; see Roberts & Roussou 2003). In the case of Chinese, as seen in Figure 8.1 above, there is sometimes a postverbal object np that may intervene between V1 and V2 in a multi-clausal construction, which may reduce the amount and rate of V2 being reanalyzed as a grammatical or pragmatic marker at the right periphery (rp). Nevertheless, there is still ample syntactic relabeling activity involving V > v > (T) > C movements in Chinese, particularly in the southern Sinitic varieties, that makes Chinese well-known as a language rich in sentence final particles. 3

Clausal Integration: The Rise of Mitigative and Adhortative Particles er yi yi, bale and haole

Another fairly productive syntactic restructuring process that contributes to the rise of sentence final particles in Chinese is clausal integration. This strategy was attested in Old Chinese and continues to be used in Modern Chinese. It is also crosslinguistically robust, and is more commonly referred to in the literature as ‘clause-combining’ (e.g. Laury 2008; Givón 1985, 2001; Haiman & Thompson 1988). Here we use the term ‘clausal integration’ because of our special interest in the development of sentence final particles, which deals

190

yap, yang and wong

with a late phase in the grammaticalization process, and which involves the merging of clauses in which highly subjective (i.e. evaluative or emotive) constituents in the second clause have themselves already undergone substantial semantic and syntactic reanalyses such that they can readily be reinterpreted as pragmatic markers that scope over the preceding clause. We illustrate this type of development here with mitigative and adhortative particles from Classical and Modern Chinese. In Classical Chinese, a propositional clause is sometimes followed by an evaluative clause such as er yi yi, which means ‘and that’s all’, with er conveying a connective or anaphoric meaning and yi yi conveying a double perfective aspectual reading, which contributes to the reanalysis of er yi yi as a complex sentence final particle (Yap, Wang & Lam 2010), often with a mitigative reading, as in (6) and (7). (6) 學問之道無他，求其放心而已矣。 xuewen zhi dao  wu  ta, qiu qi fang xin er yi yi. learn.ask gen way not.have others seek 3sg missing heart sfp ‘The great end of learning is nothing else but to seek for the lost mind.’ (Mencius 11/11, late Warring States period, 2nd−3rd century bc; translated by James Legge 1960) (7) 我竭力耕田，共為子職而已矣。 wo jie li geng tian , gong wei zi zhi er yi yi. 1sg exhaust strength cultivate field all for son duty sfp ‘I toil in the fields and all this simply as my duty as a son.’ (The speaker intends to make the point that he has no ulterior motive.) (Mencius 9/1, Warring States period; cited in Yap, Wang & Lam 2010: 69)

Figure 8.3 captures the syntactic restructuring which leads to clausal integration (see also Yap, Wang & Lam 2010). Note that er yi yi originated as an evaluative terminal clause in a multi-clausal construction, and was then reanalyzed as a mitigative marker that combined with the preceding clause to form a complex sentence final particle. In other words, clausal integration with er yi yi resulted in the restructuring of a biclausal construction into a monoclausal one, as seen in the transition from Stage 1 to Stage 2 in Figure 8.3 below. Crucially, this merger results in semantic scope expansion where a propo­ sitional clause is now expanded to also encode the speaker’s subjective evaluation.

on the development of sentence final particles in chinese Stage 1 (multi-clausal): Stage 2 (monoclausal):

191

Clause-1, Clause-2. .................., [er yi yi ]. .................., [CONN finish ASP] .................., [and that’s it!] Clause-1 + SFPmitigative marker

figure 8.3 Clausal integration involving mitigative sentence final particle er yi yi

A similar development involving clausal integration can be seen in the rise of mitigative marker bale in Early Modern Chinese, as seen in (8). Originating in a verb meaning ‘stop’, ba followed by the perfect(ive) aspect marker le likewise developed into a mitigative ‘and that’s it’ reading (see Yap, Wang & Lam 2010). High frequency usage triggered phonological compression that gave rise to bale, which then merged with the preceding clause to become its mitigative sentence final particle. (8) 這不過是個田單火牛之計罷了 zhe buguo shi ge tiandan huo niu zhi ji bale dem just cop cl pn fire cattle gen strategy sfp ‘It’s just Tian Dan’s Fire Cattle Columns tactic (which has been used two thousand years ago). (San Bao Tai Jian Xi Yang Ji, Ming period, 16th century)

Clausal integration can also be seen in the grammaticalization of haole as an adhortative (i.e. urging, nudging or encouraging) marker in Contemporary Chinese, as seen in (9). Note that clausal reduction often precedes clausal integration. In this particular case, ( jiu) hao le ‘(then) it’d be good’ is reduced simply to haole ‘just’. The former, as part of a biclausal conditional construction as seen in (9a), functions as a weak adhortative expression, while the latter (i.e. haole), now merged as a sentence final particle in a monoclausal construction as in (9b), has a more direct and insistent adhortative quality. (9) a. 你明天走就好了 ni mingtian zou jiu hao le. 2sg tomorrow walk then good sfp Lit. ‘If you leave tomorrow, then it’d be good.’ ‘It would be good if you leave tomorrow.’

192

yap, yang and wong b. 明天走好了 mingtian zou haole tomorrow walk sfp ‘Just leave tomorrow (then).’ (often uttered in a suggestive manner)

In terms of semantic scope expansion and syntactic restructuring, clausal integration facilitates the rise of haole from the predicational domain (AspP) to the pragmatic domain (cp).12 Here we see an extension from the subjective (evaluative predicate hao le) domain to the intersubjective (adhortative and interpersonal utterance) domain that involves a larger constituent than single morphemes such as the completive verb le discussed in section 2 earlier. As seen in Figure 8.4, we see a clausal integration process whereby the evaluative haole clause is integrated with (rather than simply to) the preceding clause. That is, Propositional clause 1 + Evaluative haole clause 2 > Subjective clause with haole as a mitigative marker. In structural terms, we see a deve­ lopment within the second clause where evaluative adjective hao ‘good’ merges with sentence final particle le to form an evaluative utterance tag haole (Stage 2). This utterance tag in turn interacts with the sentence final evaluative prosody of the preceding clause to form an adhortative sentence final particle haole that scopes over the entire preceding clause (Stage 3). In this way, sentence final adhortative particle haole overtly manifests the speaker’s stance within a monoclausal structure. Stage 1 (evaluative Adj + sfp) [CP1 [TP np [VP mingtian qu]]](,) [CP2  jiu [AdjP hao] le] Stage 2 (evaluative utterance tag) [CP1 [TP np [VP mingtian qu]]](,) [CP2  jiu haole] ⇒ [CP1 [TP np [VP mingtian qu]]](,) [CP2 haole] Stage 3 (adhortative particle) [CP2 [TP np [VP mingtian qu]] haole] Figure 8.4 Stages in the clausal integration of Mandarin adhortative haole

The same developmental stages can be seen for a number of other sentence final adhortative particles in Mandarin Chinese such as dele and suanle. While all three adhortative markers (haole, dele and suanle) can convey impatience,

12

We indicate the AspP predicational domain rather than just the AdjP predicational domain because haole comprises not only of adjective hao but also perfect aspect marker le.

on the development of sentence final particles in chinese

193

haole is more often mildly suggestive, as seen in (9b) above, while the use of dele tends to convey impatience often accompanied by a subtle sense of imposition, as seen in (10) below, and the use of suanle often conveys not only a sense of impatience but also a sense of resignation, as seen in (11). (10) a. 你明天走就得了 ni mingtian zou jiu de le 2sg tomorrow walk then be.fine (< be.able) sfp Lit. ‘If you leave tomorrow, then that’s fine.’ ‘It’s fine if you leave tomorrow.’ b. 明天走得了 mingtian zou dele tomorrow walk sfp ‘Just leave tomorrow (then).’ (often uttered with imposition and impatience) (11) a. 他不去就算了 ta bu qu jiu suan le 3sg neg go then be.settled (< count) sfp ‘(If) he doesn’t go, then (we) just forget it.’ b. 不去算了 bu qu suanle neg go sfp ‘Forget it if he doesn’t go.’ (uttered with resignation and impatience)

Clausal integration is attested in other neighbouring languages as well. In Modern Japanese, for example, sureba ‘if’ conditional clauses are typically followed by their consequent clauses, as in (12a). However, the conditional clause is sometimes followed by the deontic-evaluative predicate ii, meaning ‘(it should be) good’, as in (12b). High frequency usage of this evaluative mainclause predicate has resulted in its integration with the preceding clause, and the concomitant reanalysis of sureba ii as a deontic-evaluative sentence final particle, as in (12c). (12) a. benkyoo sureba, tesuto ni gookaku suru yo study do.cond test acc pass do sfp ‘(If) you study (hard), (then you will) pass the test.’

194

yap, yang and wong

b. benkyoo sureba, ii study do.cond be.good ‘(If) you study (hard), (then it should) be good.’ c. benkyoo sureba ii study sfp ‘You should study (hard).’

Contemporary Japanese also has a related expression benkyoo sure-ba? as shown in (13) below, where sentence final particle sure-ba functions as a deontic-adhortative marker. This construction, however, emerged via another strategy, namely ‘main-clause ellipsis’, which we will examine later in section 6. (13) benkyoo sureba? study sfp ‘Why don’t you study (hard)?’

4

Right-Dislocation: Emergence of Epistemic Utterance Tag kongpa

A more common process that gives rise to utterance tags and sentence final particles in Chinese is right-dislocation (see Cheung 2009; Lin 2008). This process is not necessary for verb-final languages such as Japanese and Korean, whose evaluative, expressive and attitudinal verbal complexes at the right periphery can be more directly recruited to form sentence-final pragmatic markers. However, right-dislocation is often necessary for verb-medial languages such as English and Chinese, giving rise to utterance tags in postpredicate position. Well-known examples in English include the epistemic marker I think (Thompson & Mulac 1991; Kärkkäinen 2003), with parallel studies in other languages (see for example Lim 2011 and Endo 2013 on Mandarin wo juede ‘I think’ as epistemic and pragmatic markers). In this section we will focus on the process of right-dislocation in Chinese. We illustrate with Mandarin epistemic marker kongpa. Diachronically, kongpa emerged from a combination of two ‘fear’ verbs kong and pa in Late Middle Chinese, during the Tang and Song periods, to express the speaker’s anxiety. In Early Modern Chinese, kongpa developed into an epistemic marker meaning ‘probably’ that is typically used in clause-initial position and sometimes used parenthetically in clause-medial position (see Yap, Chor & Wang 2012; see also Endo 2006 for the development of Mandarin ‘fear’ verb pa on its own as a pragmatic marker). Epistemic marker kongpa is also

on the development of sentence final particles in chinese

195

sometimes used as an utterance tag with a pragmatic hedging function in clause-final position, where it helps to soften an epistemic claim, or to attend to the face-needs of the interlocutors (either the speaker or the addressee, or both) when the speaker may be certain or confident about the veracity of his/ her claim yet for pragmatic reasons considers it best to downgrade the strength of their epistemic claim. Consider the uses of kongpa in (14) below (see also Yap, Chor & Wang 2012). In (14a), we see wo kongpa yielding an ambiguous interpretation: it can either express anxiety or epistemic probability, both anchored in the speaker’s subjective stance. The epistemic interpretation emerges via semantic extensions whereby the speaker’s anxiety comes to reflect his/her concern about the likelihood of impending negative outcomes, which facilitates the reanalysis of kongpa as an epistemic marker meaning ‘possibly’ or ‘probably’. This development contributes to the insubordination of the complement clause ta bu xihuan wo le ‘he doesn’t like me’ as an independent ‘main clause’ construction as in (14b), with kongpa as its epistemic stance marker at the left-periphery. As an epistemic adverbial, kongpa can now also occur parenthetically in clause-medial position, as in (14c). While epistemic adverbial kongpa favors the clause-initial and clause-medial position, it can also occur as an utterance tag in clause-final position, as in (14d). (14) a. 我恐怕他不喜歡我了 wo kongpa ta bu xihuan wo le 1sg fear 3sg neg like 1sg sfp ‘I’m afraid/Probably he doesn’t like me anymore.’ b. 恐怕他不喜歡我了 kongpa ta bu xihuan wo le fear 3sg neg like 1sg sfp ‘Probably he doesn’t like me anymore.’

c. 他恐怕不喜歡我了 ta kongpa bu xihuan wo le 3sg fear neg like 1sg sfp ‘He probably doesn’t like me anymore.’



d. 他不喜歡我了，恐怕。 ta bu xihuan wo le, kongpa. 3sg neg like 1sg sfp fear ‘He doesn’t like me anymore, probably (< I’m afraid).’

196

yap, yang and wong

Figure 8.5 below highlights the stages in the development of kongpa from a lexical verb to an epistemic marker and a right-dislocated utterance tag with a pragmatic hedging function. Stage 1 (lexical verb) [CP2 [IP2 (wo) [VP kongpa [CP1 [IP1 ta bu xihuan wo] le]]]] Stage 2 (epistemic marker) [CP2 (wo) kongpa [CP1 [IP1 ta bu xihuan wo] le]] Stage 3 (right-dislocated epistemic marker ⇒ utterance tag pragmatic hedger) [CP2 [CP1 [IP1 ta bu xihuan wo] le], kongpa] figure 8.5 Stages in the right-dislocation of Mandarin pragmatic hedger kongpa

Right-dislocation is not uncommon in other languages either, and in some varieties of Malay (e.g. Perak Malay and Kedah Malay, which are spoken in the northwestern part of peninsular Malaysia), the epistemic marker kot, which is derived from the ‘fear’ verb takut, has developed into a sentence final parti­ cle (Yap, Chor & Wang 2012). While Mandarin kongpa has not grammaticalized beyond the utterance tag stage to become an epistemic sentence final particle, arguably because of its relatively low usage frequency, there are examples of Chinese utterance tags at the right periphery that have developed into sentence-final particles. We explore this in the next section, using sentence final evidential particles in Cantonese as examples. 5

Right-Dislocation and Clausal Integration: Emergence of wo-Type Sentence Final Evidentials in Cantonese

Mandarin Chinese does not make productive use of grammaticalized evidential markers at the right periphery, but as noted earlier in the introduction (see §1), a number of Chinese varieties spoken in the south have developed some highly grammaticalized evidential sentence final particles via a combination of right-dislocation and clausal integration. We will here examine this more elaborate process with examples involving the development of Cantonese hearsay evidential wo3 and wo5. The Cantonese lexical verb waa6 ‘say’ is known to have developed evidential markers at both the left and right periphery, with the latter type further developing into a wide range of pragmatic sentence final particles (Matthews 1998; Leung 2006, 2010; Yeung 2006; Yap & Ahn 2012). At the left periphery, hearsay

on the development of sentence final particles in chinese

197

evidential uses of the matrix clause jan4 waa6 ‘people say’ were attested in early 17th century opera lyrics, with the plural form jan4dei6 waa6 ‘people say’ attested in early 19th century song lyrics (Jiu 1828). More elaborate forms such as ngo5 teng1gin3 jan4 waa6 ‘I hear people say’ and ngo5 teng1gin3 waa6 ‘I hear say’ were both attested in the late 19th century (The Forty Exercises, Anonymous 1877). A Comprehensive Dictionary of Chinese Topolects (Xu & Miyata 1999) includes teng1waa6 ‘hearsay’, undated but probably from a mid-20th century text. Contemporary Cantonese now favours teng1gong2 (also ‘hearsay’) as its leftperiphery evidential marker. Movies from the 1960’s provide us with examples of both teng1gong2 waa6 (lit. ‘hear.say say’) and teng1gong2 (lit. ‘hear.say’). Examples of hearsay evidential jan4dei6 waa6 ‘people say’ and teng1waa6 ‘hearsay’ (> ‘It’s said’) at the left periphery are shown in (15) and (16) below (see Yap & Ahn 2012 for more detailed diachronic discussion). (15) 人話天孫今夜會牛郎 jan4 waa6 tin1syun1 gam1je6 wui6 ngau4long4 people say pn tonight meet pn Lit. ‘(People say) Tin-Syun (goddess of fertility) will meet Ngau-Long (the cowherd).’ ‘For to night, it is said, heaven’s bride and bridegroom unite.’ (From the opera lyric Faa1 Zin1 Gei3 “The Flower Scroll”, early 17th century, translated by Peter Perring Thoms 1824: 1) (16) 聽話好熱鬧， teng1waa6 hou2 jit6naau6, hear.say int exciting 點解咁高興你都唔去呀？ dim2gaai2 gam3 gou1hing3 nei5 dou1 m4 heoi3 aa3 ? why ( wo3). This would involve phonological changes that include segment reduction and syllable fusion. Note that this process also involves tone-rising changes, more specifically from low-tone waa6 to mid-tone wo3. The shift to a higher tone favours the expression of tentativeness and lower epistemic commitment, and is more compatible with evidential uses. Hearsay evidential wo4 was attested as a phonological variant in the late 19th century (Ball 1888), and hearsay evidential wo5 was attested slightly later in the early 20th century (Ball 1912), as shown in (20), and continues to be used to this day, as seen in (21). It is worth noting that wo5 (with the rising tone) is the one that is favoured to carry on as the evidentiality marker in Contemporary Cantonese.

13 14

Tone 2 is a high-rising tone and tone tone 5 is a low-rising tone, while tone 3 is a mid-tone. All these three tones are higher than tone 6, which is the original tone used for waa6 ‘say’. Note that we are using the Jyutping Romanization system for the Cantonese examples.

on the development of sentence final particles in chinese

199

(19) 佢聽日至黎和 keoi5 ting1jat6 zi3 lai4 wo3 3sg tomorrow only.then come evid ‘He’ll come tomorrow.’ (Morrison 1828: Part III, Section X, “Friendship”) (20) A: 佢話有乜野錯呢 keoi5 waa6 jau5 mi1je5 co3 ni1 3sg say exist what.thing wrong sfp Lit. ‘He said have what thing wrong, eh?’ ‘What fault did he find with it?’ B: 乜野都錯啝 mi1je5 dou1 co3 wo5 what.thing also wrong evid Lit. ‘Everything even wrong, so-he-says.’ ‘Everything is wrong about it.’ (Ball 1912: 66, 67) (21) (聽講) 佢唔嚟喎 (teng1gong2) keoi5 m4 lai4 wo5 hear.say 3sg neg come evid ‘He’s not coming, I hear (< ‘people say’).’ (Contemporary Cantonese)

Leung (2006: 66) noted that an example with utterance tag waa1 in Ball (1888), as shown in (22) below, was later reproduced with sentence final particle wo5 in his later 1924 edition. This suggests a strong link between waa-type and wotype evidentials, which we suggest is derivational in nature (see Figure 8.6 for further discussion). (22) 佢打我口話 (later replaced by 啝 in Ball’s 1924 edition) keoi5 daa2 ngo5 waa1 (> wo5) 3SG hit 1sg evid evid ‘He said he would strike me.’ (Ball 1888: 90)

Interestingly, while rising-tone wo5 has developed into the dominant hearsay evidential in Contemporary Cantonese, mid-tone hearsay evidential wo3 developed instead into a counter-expectation marker (Yap, Chor & Wang 2012), as illustrated in (23). There are also phonological variants, depending on sentence final prosody, with Cantonese movies from the 1960’s providing examples of low-tone counter-expectation marker wo4. Previous scholars have observed that, in contrast to the high tones, which convey a sense that “the speaker is not fully committed to what s/he just said” (Sybesma & Li 2007: 1768), the low tones function instead as pragmatic strengtheners (ibid; see also Law 1990;

200

yap, yang and wong

Matthews & Yip 1994). It is therefore not surprising then that mirative and counter-expectation readings come to be associated with low-tone wo4. (23) (係喎！) 你唔講我都唔知喎！ (hai6 wo3!) nei5 m4 gong2 ngo5 dou1 m4 zi1 wo3! yes ce 2sg neg say 1sg also neg know ce ‘Oh yes/right!? If you didn’t tell me, I wouldn’t have known!’ (Contemporary Cantonese)

Essentially, two major types of phonological changes were involved in the grammaticalization of waa6 ‘say’ evidentials at the right periphery. One type involved segment reduction and syllable fusion when waa-type evidentials combined with sentence final particles such as o3 to form wo-type evidentials (e.g. waa6 + o3 > wo3). Another type involved tone changes as both waa-type and wo-type evidentials combined with higher tones to form a wider range of evidentials. This development points to an important role for speaker’s sentence final prosody in the grammaticalization of ‘say’ evidential markers at the right periphery.15 These observations are consistent with the findings of previous studies, in particular Sybesma and Li (2007: 1768; citing Law 1990), which suggest that “non-segmental tonal sfps are highly localized intonation.” From a structural perspective, as highlighted in Figure 8.6 below, using the utterance jan4dei6 waa6 keoi5 sat1-zo2-zung1 ‘people say he’s gone missing’ as a point of departure, we see a process of insubordination in which the subject and predicate-taking verb of the matrix clause (namely, jan4dei6 waa6 ‘people say’) is first grammaticalized into a left-periphery evidential marker, triggering reanalysis of the complement clause as an insubordinate ‘main clause’. That is, the complement clause keoi5 sat1-zo2-zung1 ‘he’s gone missing’ now becomes an insubordinate (i.e. independent) construction. Right-dislocation, on the other hand, gives rise to utterance tags and sentence final particles at the right periphery, both for the waa-type and wo-type evidentials. Combinations of waa-type and wo-type evidentials with other sentence final particles (e.g. emphatic particle o3) and with the speaker’s sentence final prosody give rise to phonological variants with subtle shades of attitudinal and other pragmatic 15

It is also interesting to note the use of different Chinese characters over time, from 話 waa6 ‘say’ to 話 waa-type utterance tags and 啝 or 喎 wo-type sentence final particles. As discussed elsewhere (e.g. Yap & Ahn 2012), it is worth further noting that the Chinese characters for the utterance-final variants of waa6 (話) are written with an additional semantic radical meaning ‘mouth’ (口) to indicate its colloquial usage as well as its sentence final particle (sfp) status.

on the development of sentence final particles in chinese

201

functions such as confirmation-seeking, reminding and counterexpectation marking. Phonological reduction of these waa-type and wo-type ‘say’ sentence final particles facilitates clausal integration, whereby the evidential utterance tag at the right periphery of the insubordinate complement clause is reanalyzed as its sentence final particle. In this way, the hearsay evidential particle comes to explicitly mark the insubordinate (i.e. independent and finite) status of the erstwhile complement clause. Stage 1 (lexical verb) [IP2 jan4dei6 [VP waa6 [CP1 [IP1 keoi5 sat1-zo2 zung1]]]] Stage 2 (hearsay evid at lp) [CP2-EVID jan4dei6 waa6 [IP1 keoi5 sat1-zo2-zung1]] Stage 3 (hearsay evid at rp) [CP2-EVID [IP1 keoi5 sat1-zo2-zung1], [EVID ( jan4dei6) waa6]] Stage 4 (attitudinal evid ‘so it’s said’) [CP2-EVID [IP1 keoi5 sat1-zo2-zung1], [CP2 ( jan4dei6) gam2 waa6 o3]] Stage 5 (hearsay evid at rp) [CP2-EVID [IP1 keoi5 sat1-zo2-zung1] [EVID  waa6 + o3]] 3 Stage 6 (sentence-final evid wo ) [CP2-EVID [IP1 keoi5 sat1-zo2-zung1] wo3] Stage 7a (sentence-final evid wo5) ⇒[CP2-EVID [IP1 keoi5 sat1-zo2-zung1] wo5] (+ tone change) Stage 7b (sentence-final ce wo3) [CP2-EVID [IP1 keoi5 sat1-zo2-zung1] wo3] (+ mirativity) figure 8.6 Stages in the right-dislocation and clausal integration of Cantonese hearsay evidential sentence final particle wo3 and wo5

As seen in Figure 8.6 above, the later stages in the development of the waatype and wo-type ‘say’ constructions include a strong pragmatic interpretation (e.g. confirmation-seeking, reminding, counterexpectation marking). The development of these pragmatically-nuanced ‘say’ constructions from waatype utterance tags to wo-type sentence final particles is consistent with crosslinguistic tendencies which, in the words of Hunston and Thompson (2000: 143; paraphrasing Martin 2000), show that “the expression of attitude is not, as is often claimed, simply a personal matter—the speaker ‘commenting’ on the world—but a truly interpersonal matter in that the basic reason for advancing an opinion is to elicit a response of solidarity from the addressee” (see also Kärkkäinen 2003: 185). From a crosslinguistic perspective, it is worth noting that numerous other languages—particularly verb-final languages—have also developed sentence-

202

yap, yang and wong

final evidentials from ‘say’ verbs (see Aikhenvald 2004; Aikhenvald & LaPolla 2007; Grunow-Hårsta 2007; inter alia). An interesting difference is that the syntactic mechanism of right-dislocation is not necessary for verb-final languages; they simply rely on verb-serialization or converbal construction to achieve similar semantic and syntactic scope expansions. Despite the apparent greater ease with which verb-final languages form their ‘say’ evidential constructions at the right periphery, it is also interesting to note that the development of ‘say’ constructions into sentence-final evidential and pragmatic particles in verb-final languages such as Japanese and Korean also undergo many phonological reduction processes similar to the ones we have identified for Cantonese, including segment reduction and syllable fusion (e.g. Cantonese waa6 + o3 > wo3; Japanese to itteiru > tte; Korean tako hanta > tanta) (see S. Suzuki 1998 and R. Suzuki 2007 on the development of Japanese tte, and Ahn & Yap 2012 on the development of Korean tanta and other ‘say’ evidentials such as tako, tamye, tamyense and tanun). 6

Main-Clause Ellipsis and Right-Dislocation: Emergence of Concessive Utterance Tag buguo

Recent work on Japanese and Korean sentence final particles have revealed extensive use of ‘main-clause ellipsis’ in the formation of pragmatically-laden sentence final particles, often derived from elliptical processes that reanalyze connectives at the right periphery of subordinate clauses as sentence final particles of ‘stand-alone’ insubordinated clauses (e.g. Ohori 1998; Higashiizumi 2006, 2012; Shibasaki 2007, in press; Shinzato 2007, 2011; Rhee 2012). Such elliptical processes are more restricted in Chinese, partly because many connectives in Modern Chinese are typically found at the left (rather than right) periphery of subordinate clauses.16 The main clauses of concessive constructions tend to be highly amenable to ellipsis, particularly in conversational discourse contexts where the concessive connective is being right-dislocated for pragmatic reasons and thus appears at the right-periphery of the subordinate 16

Clause-final (i.e. right periphery) connectives were not uncommon in Old Chinese, with clause-final conditional zhe (‘when/if’) subordinate clauses attested in Classical Chinese texts (see Yap & Wang 2011). This clause-final conditional subordinator structure is still retained in Modern Chinese (e.g. Mandarin dehua (‘if’) conditional clauses). Note that these clause-final connectives are derived from clause-final nominalizers zhe and di (>de) in Old Chinese and Late Middle Chinese respectively).

on the development of sentence final particles in chinese

203

clause to attenuate the epistemic claims of the prior speaker, as seen in the Mandarin Chinese example in (24) and the Cantonese Chinese example in (25). (24) A: 明天問問他。 mingtian wen-wen ta. tomorrow ask-ask 3sg ‘(We’ll) just ask him tomorrow.’ B: 他不會來不過。 ta bu hui lai buguo. 3sg neg fut come however ‘He won’t be coming here though.’ (25) A: 佢次次都遲到架啦。 keoi5 ci3ci3 dou1 ci4dou3 ga3laa1. 3sg always also late sfp ‘As we all know, he’s always late.’ B: 係呀。但係佢今次早到 (喎) hai6 aa3. daan6hai6 keoi5 gam1ci3 zou2 dou3 (wo3) yes sfp however 3sg this.time early arrive ce 不過。 bat1gwo3. however ‘Yes, but this time he’s early though.’

As seen in (24) and (25) above, the interactive nature of conversational talk lends itself to the formation of elliptical constructions, as dyadic talk often makes repetition of the prior speaker’s claims unnecessary. As such, ellipsis can serve as a politeness strategy in that it can signal to the prior speaker that the current speaker is tacitly acknowledging what has been said, and crucially ellipsis then allows the current speaker to focus on making his/her point and deftly adding a pragmatic touch, by conceding that what the prior speaker has proposed may be a possible solution, as in (24) above, or may be true, as in (25), but at the same time ellipsis of the main clause allows the current speaker to draw attention to the particular semantics of the connective and the assertion in the subordinate clause. In the Mandarin example in (24), through the ellipsis of the main clause, Speaker B tacitly agrees with Speaker A that it would be a good idea to confirm the facts with a certain ta ‘he’ who is in a position to

204

yap, yang and wong

provide the answer to their question. By subsequently adding concessive connective buguo ‘however’ as a right-dislocated utterance tag to his reply, Speaker B is able to also point out a potential problem, namely that the person who is in a position to provide the relevant information unfortunately will not be available for questioning. In the Cantonese example in (25), by eliding the main clause, Speaker B tacitly accepts Speaker A’s assessment that a certain person they have been talking about is often late, and by subsequently adding concessive markers daan6hai6 ‘however’ in clause-initial position and bat1gwo3 in clause-final (right-dislocated) position, both of which happen to be optional, Speaker B is at the same time able to refute the relevance of the prior information provided by Speaker A. The combined use of main-clause ellipsis and the right-dislocated concessive utterance tag buguo/bat1gwo3 thus allows the current speaker (B) to engage in dispreferred moves, such as disagreeing with the prior speaker, in a more subtle and less face-threatening manner. Figure 8.7 highlights the stages in the development of Mandarin sentence final concessive buguo. Stage 1 (bi-clausal concessive): Stage 2 (main-clause ellipsis): Stage 3 (right-dislocation of concessive connective ⇒ insubordinated clause with utterance tag):

Main clause + buguo subordinate clause Buguo subordinate clause

Insubordinate clause buguo

figure 8.7 Stages in main-clause ellipsis of Mandarin concessive buguo

As is often the case in Japanese and Korean (as well as other languages), mainclause ellipsis is thus used in Chinese as a face-saving device to avoid undue attention to a point of contention.17 At the same time, grammatical devices such as the concessive connectives buguo in Mandarin and bat1gwo3 in Cantonese are often right-dislocated to serve as sentence final pragmatic markers that allow the speaker to pragmatically hedge his/her counter-claims or alternative proposals. There has not been as much work done in Chinese conversational discourse on these connective-type sentence-final particles that emerge as a result of main-clause ellipsis and right-dislocation, and more studies are clearly welcome. From a typological perspective, main-clause ellipsis is known to be far more productive in verb-final languages with abundant 17

For studies of ‘main clause omission’ and the rise of negative attitudinals in Korean, see Rhee (2012).

on the development of sentence final particles in chinese

205

clause-final connectives, both in terms of usage frequency and structural types, as seen in numerous studies on Japanese and Korean. This is largely because right-dislocation of clause-final connectives is not even an option for these verb-final languages, which means that Stage 3 in Figure 8.7 above is altogether unnecessary, thereby reducing (for these languages) the number of steps required for the development of sentence final particles derived from clausal connectives. 7

Insubordination of Nominalization Constructions: Reanalysis of Nominalizers as Tense-Aspect-Mood and Attitude Markers

In this section we will examine how versatile nominalizers are reinterpreted as sentence final particles. This phenomenon is robust crosslinguistically among verb-final languages (for Tibeto-Burman languages, see Matisoff 1972, Noonan 1997, 2008 & 2011, Simpson 2008, Grunow-Hårsta 2011, inter alia; for Japanese, see Horie 2011; for Korean, see Ahn & Yap 2013, Rhee 2008, 2011; for a crosslinguistic perspective across Asian languages, see Yap & Grunow-Hårsta 2010; inter alia). Interestingly, as briefly noted in §1 earlier, although Modern Chinese is essentially a verb-medial language (svo), some of its nominalization constructions are head-final, which structurally enables some of its nominalizers to also develop into sentence final particles. A case in point is Middle Chinese nominalizer di, as seen in (26a), which has undergone phonological reduction and character substitution and now is realized as de, as seen in (26b). Nominalizer di or de often occurs at the right periphery of an utterance or sentence, which over time facilitates its reanalysis into a sentence final particle, in large part mediated by shi . . . di~de cleft constructions, as seen in (27) and (28) (see Yap, Choi & Cheung 2010).   

(26) a. 人人盡有底 ren ren jin you di people people all have nmlz ‘that which everyone has’18 (Zutangji, Five Dynasties period, 10th century)

18

This refers to one’s character, which in this discussion between two monks is referred to as one’s skin, and metaphorically symbolized by one’s garment.

206

yap, yang and wong

b. 對於這個問題， duiyu zhe ge wenti, as.for this cl question 我知道的很有限 wo zhidao de  hen youxian 1sg know nmlz very limited ‘As for this question, what I know is very limited.’ (Academia Sinica Balanced Corpus of Modern Chinese) (27) a. 上下兩輪月，若個是真底？ shang xia liang lun yue, ruo ge shi zhen di? up down two cl moon which cl cop real nmlz/sfp ‘The moon up (in the sky) and the moon down (in the water), which one is the real one / which one is real?’ (Xia Ye Wan Yue, Southern Song period, 12th–13th century)

b. 此寺是則天皇后蓋造的， ci si shi zetian huanghou gaizao di, this temple foc pn queen build sfp 後來崩損，又是崔相國重修的。 houlai bengsun , you shi cui xiangguo chongxiu di . later collapse int foc pn prime.minister rebuild sfp ‘This temple was built by Queen Zetian. Later it collapsed. It was Prime Minister Cui who rebuilt it.’ (Xixiangji, Yuan period, 13th–14th century)

(28) a. 是我先找到這個答案的 shi wo xian zhaodao zhe-ge daan de foc 1sg first find.out this-cl answer sfp (< nmlz) ‘It was I who found the answer first.’ (Modern Chinese) b. 這個答案（，）是我先找到的 zhege daan (,) shi wo xian zhaodao de this.cl answer foc 1sg first find.out sfp ‘It was I who found the answer first.’ (Modern Chinese) c. 這個答案（，）我先找到的 zhege daan (,) wo xian zhaodao de this.cl answer 1sg first find.out sfp ‘I found the answer first.’ (Modern Chinese)

on the development of sentence final particles in chinese

207

At the right periphery, as seen in (27) and (28) above, de is in an ideal position to serve as the landing site for the speaker’s illocutionary force as conveyed through the sentence final prosody. The interpretation typically associated with sentence final particle de is one of assertion. This is partly because the emergence of sentence final particle de is mediated by the use of de nominalization constructions in cleft constructions, as highlighted in (27b). However, given that nominalization constructions can also serve as complement constructions in interrogative, mirative and other contexts, sentence final de can also host other types of sentence final prosodic features, ranging from dubitative to sceptical, hesitant or even playful, and not just assertive ones (see Yap, Choi & Cheung 2010). Sentence final de can also combine with other sentence final particles such as ba, ne and ma to form complex sentence final particles such as deba, dene and dema, as shown in (29) below. (29) a. 他們挺像的吧 tamen ting xiang deba 3PL int be.like sfp ‘They are quite alike.’ (conveying some hesitation) b. 他們挺像的呢 tamen ting xiang dene 3pl int be.like sfp ‘They are quite alike.’ (conveying some assurance) c. 他們挺像的嗎？ tamen ting xiang dema? 3pl int be.like sfp ‘They are quite alike?’ (conveying some query, doubt, or scepticism)

Figure 8.8 highlights the various stages in the development of di~de from nominalizer to sentence final particle. The cleft construction, involving focus particle shi and a complement clause headed by nominalizer di~de, facilitated the emergence of sentence final particle di~de (Stage 3), particularly since nominalizer di~de is conveniently positioned at the right periphery within the shi . . .  di~de cleft construction, which makes it ideally situated to host the speaker’s sentence final prosody. In much the same way that sentence final di~de is able to combine with other sentence final particles to convey a wide range of speaker’s mood, evaluation and attitude, sentence-final di~de is also able to merge with various types of sentence final prosody cues to serve a wide range of pragmatic functions.

208

yap, yang and wong

Stage 1 (nominalizer di) Stage 2 (nominalizer di in cleft) ⇒ (sentence final di in cleft) Stage 3 (sentence final di) or (sentence final di + topicalization)

[CP1 [IP1 wo zhidao] di] [CP2 shi [CP1 [IP1 wo xian zhaodao zhege daan] di]] [CP2 shi [CP1 [IP1 wo xian zhaodao zhege daan]] di] [CP2 [CP1 [IP1 wo xian zhaodao zhege daan]] di] [TOP zhege daan], [CP2 [CP1 [IP1 wo zhidao]] di]

figure 8.8 Stages in the reanalysis of Mandarin di~de from nominalizer to sentence final particle

8

Some Crosslinguistic Observations: Strategies in the Development of Sentence Final Particles (and Utterance Tags) in Chinese, Japanese and Korean

We have thus far identified six pathways in the development of sentence final particles (and utterance tags) in Chinese. These pathways are by no means exhaustive, nor are the phonological and morphosyntactic processes involved unique to the Chinese language. We have noted that the use of sentence final particles is particularly prominent among verb-final languages, such as Japanese and Korean for example. Regardless of their canonical word order (vo vs. ov), verb concatenation (e.g. verb serialization and converb-linking) and clausal integration are common strategies in the formation of grammatical and pragmatic markers, some of which develop into sentence final particles such as Mandarin le, Japanese –te shimau Korean –a/e pelita. What is also noteworthy is that syntactic restructuring mechanisms such as right-dislocation are essentially superfluous and thus absent in verb-final languages. As such, whereas verb-medial languages such as Chinese (and English) sometimes resort to right-dislocation to convey speaker moods such as epistemic uncertainty, counter-expectation marking, speaker detachment, solidarity marking, or a combination of these, as in the case of Mandarin kongpa and Cantonese wo, verb-final languages can dispense with right-dislocation and simply rely on converb-linking (along with the necessary phonological changes such as segment reduction and syllable fusion) to give rise to sentence final particles. As noted earlier, this simpler grammaticalization process has given rise to numerous evidential markers, including Japanese –tte and Korean tako, tamye, tamyense, tanun and tanta. In this respect, the development of some types of sentence final particles is structurally more costly for verb-medial languages such as Chinese. This also explains why, depending on the degree of grammaticalization of a right-dislocated constituent, we see utterance tags that have not (yet) gone the full length of clausal integration to become

on the development of sentence final particles in chinese

209

sentence final particles in verb-medial languages such as English and Chinese. As discussed in section 7 above, some utterance tags are phonologically and morphosyntactically further integrated with the preceding clause to form sentence final particles, as attested in southern Chinese dialects that are rich in sentence final particles (e.g. Cantonese wo-type evidentials), but the reanalysis of utterance tags as sentence final particles is rare in languages such as English that are impoverished in sentence final particles. This suggests the possibility that an environment rich in sentence final particles, with a strong predilection for clausal integration at the right periphery, facilitates the reanalysis of utterance tags into sentence final particles, when supported by high usage frequency. Elliptical constructions that involve the elision of main clauses and the subsequent insubordination of subordinate clauses, and the concomitant reanalysis of their connectives as sentence final particles, are highly productive in verb-final languages (see for example discussions of ‘suspended clauses’ in Ohori (1995), as well as Japanese kara and node utterance-final particles in Higashiizumi (2006, 2012) and various Korean pragmatic sentence enders such as tanikka in Rhee (2012)). The use of ‘main-clause ellipsis’ in this fashion that gives rise to connective-based sentence final particles is much more restricted in verb-medial languages such as English (but see Mulder, Thompson & Williams (2009) for a discussion of recent developments of sentence-final but in Australian, New Zealand and Falkland Islands English), and it is also quite restricted in Chinese, with concessive sentence final particle buguo and negator-turned-interrogative sentence final particle bu among the few examples. In this paper, as we examined various pathways by which sentence final particles emerge in Chinese, we frequently saw a progression in which constructions evolve over time from propositional to subjective and intersubjective uses, consistent with Traugott’s observations (e.g. Traugott 1989, 1995, 2003, 2010).19 One important observation that keeps converging across the various pathways we have examined is the formation of ‘finite’ structures, either in the form of monoclausal subjective constructions derived from biclausal or multiclausal constructions (e.g. Mandarin le, bale, haole, dele), or insubordinate 19

(Inter)subjectivity refers to the process by which speakers express their personal and interpersonal feelings, views, evaluations and attitudes in the course of human interaction and communication, while (inter)subjectification focuses on the language change processes that give rise to markers of speakers’ (inter)subjective stances. Research within Traugott’s framework has provided us with a diachronic and typological perspective to (inter)subjectivity phenomena, while Langacker (1989) provides us with a cognitive linguistic model to conceptualize these (inter)subjective phenomena.

210

yap, yang and wong

‘suspended’ constructions from subordinate complements (e.g. Mandarin sentence final buguo and kongpa, and Cantonese wo-type evidentials), or insubordinate ‘main-clause predicates’ from stand-alone nominalized complement clauses (e.g. Mandarin sentence-final de constructions). In all these cases, we obtain ‘finite’ structures, where the term ‘finite’ is to be understood in a broad sense to include various strategies by which clauses become semantically, phonologically and morphosyntactically independent structures (see Nikolaeva 2007 and papers therein, in particular Evans 2007; see also Nikolaeva 2010). Within the cognitive linguistic tradition, such ‘finiteness’ is construable in terms of ‘clausal grounding’ or ‘clausal anchoring’ (Taylor 2003). What we have shown thus far, then, is that Chinese sentence final evidentials and other pragmatic markers are ‘finiteness markers’ or ‘clausal grounders’ in that they contribute to the formation of independent clauses. This helps to explain how ‘tenseless’ languages such as Chinese form ‘finite’ (or independent) clauses. Whereas English relies heavily on inflectional tense markers and modals, in addition to remnants of a once-robust case and agreement system, Chinese deploys a variety of strategies that include not only the use of sentence-final aspect markers, modal auxiliaries, intensifiers or degree adverbials (e.g. hen ‘very’) and comparative markers (e.g. bijiao ‘in comparison’), among others, it also uses a wide array of sentence final particles. The ‘finite’ sentence final particles that we have identified in the preceding sections include: (i) Mandarin perfect and conversation turn-transition marker le; (ii) Classical Chinese mitigative marker er yi yi, and Mandarin mitigative and adhortative particles bale, haole, dele and suanle; (iii) Mandarin concessive marker buguo in ‘main-clause ellipsis’ constructions, (iv) Cantonese evidential and counterexpectation marker wo3, and (v) Mandarin nominalizer and default assertive marker de. These ‘finite’ particles are grounded not only in terms of temporal deictic information but also in terms of the speaker’s illocutionary force. As discussed in section 4, Mandarin epistemic marker and pragmatic hedger kongpa has developed into an utterance tag but not (yet) into a sentence final marker; nevertheless, it can be said to indirectly contribute to the finiteness of its host predicate, which previously was its erstwhile complement clause that is now insubordinated via a right-dislocation process that has back-shifted (or postpositioned) kongpa from the matrix verb position to the adjunct-like utterance tag position. Crucially, utterance tag kongpa retains its subjective epistemic ‘probably, I’m afraid to say’ reading. This provides a means whereby the insubordinated complement clause can be grounded in the discourse, not so much via temporal deixis by means of tense marking, but more by means of anchoring onto the speaker’s subjective epistemic mood—in other words, by relying on pragmatic deixis. Not surprisingly,

on the development of sentence final particles in chinese

211

this broader definition of finiteness, which can rely on a wide range of sentence final particles to serve as temporal and pragmatic indexicals (i.e. deictic elements) that can ground a clause as independent structures within discourse, is attested in other languages too, including Japanese and Korean, many of whose sentence final particles are marked for tense while simultaneously engaging in pragmatic functions. 9 Conclusion In this paper, we have examined how sentence final particles emerged in Chinese. More specifically, we identified a number of morphosyntactic and phonological processes that contribute to the rise of pragmatic markers at the right periphery, many of which are also found in other languages. One type of morphosyntactic process is verb concatenation in the form of verb serialization, which is then followed by a form of clause combining. In Chinese, this process gives rise to a wide range of temporal aspect markers at the right periphery, which go on to develop into pragmatic markers with (inter)subjective functions, as seen in the use of Mandarin le to signal the completion of the speaker’s turn and by implicature invite the addressee to take up the next conversational turn (Lu & Su 2009). Clausal combining can involve constructions larger than serial verb constructions. In Chinese, evaluative expressions—whether phrasal or clausal— often occur at the end of a series of prior clauses (thus operating at the level of discourse rather than syntax), and these expressions often undergo phonological reduction and subsequent clausal integration to form evaluative sentence final particles, such as Classical Chinese er yi yi and Mandarin bale, haole, dele and suanle. Evaluative and other (inter)subjective speaker stances are often expressed in matrix predicates, as in Mandarin epistemic (wo) kongpa expressions, which are roughly equivalent to English I think epistemic phrases. These (inter)subjective expressions often grammaticalize into epistemic adverbials that occur in clause-initial and clause-medial (parenthetical) positions, but they are also sometimes found in clause-final position. In the case of Mandarin kongpa and English I think, these epistemic expressions occur as utterance tags that have been right-dislocated (or ‘backshifted’) to serve pragmatic functions as well, often as hedges that serve to protect the face-needs of the speaker or the addressee, or both. In Chinese, utterance tags derived via right-dislocation often further merge with the preceding clause, via a process of clausal integration, to form

212

yap, yang and wong

sentence final particles. This is facilitated by the extensive use of subject np and object np omission in Chinese in cases where these referents are easily retrievable from context. Examples of this type of development can be seen in the semantic-pragmatic extension of Cantonese waa-type evidential utterance tags which undergo phonological changes and structural reanalysis to form wo-type sentence final particles that also serve (inter)subjective pragmatic functions such as counter-expectation marking. Sentence final particles also often emerge when subordinate or embedded clauses undergo insubordination and are reanalyzed as finite and independent ‘main-clause predicates’. In this paper, we have discussed two types of insubordination processes for Chinese. One involves a ‘main-clause ellipsis’ phenomenon that gives rise to the reanalysis of concessive connective buguo as a concessive sentence final particle. Another one involves the reinterpretation of de-nominalization constructions as ‘stand-alone’ finite structures; this type of insubordination process has been documented in numerous other languages (e.g. Noonan 1997, DeLancey 2011; Rhee 2008, 2011; Simpson 2008; Horie 2011; Yap, Grunow-Hårsta & Wrona 2011; inter alia). Crucially, this process involves the reanalysis of erstwhile nominalizers as sentence final particles. Typically, these particles at the right periphery convey assertive force. However, because they are highly bleached semantically, they can also serve as the landing site for a relatively wide range of sentence final prosodic cues, and can also combine with other sentence final particles, to yield numerous shades of speaker moods, evaluations and attitudes, including surprise, disbelief, doubt, hesitation and ridicule. In addition to identifying the different processes that give rise to sentence final particles in Chinese, and comparing these processes with those found in other languages such as Japanese and Korean, we have also probed the question of whether sentence final particles are ‘finiteness markers’. We adopt a broad definition of finiteness, a la Nikolaeva (2007, 2010), and in this paper we have argued that sentence final particles either develop within the context of pre-existing finite structures, or contribute to the formation of finite structures. In this regard, we show that tense-aspect-mood (tam) markers at the right periphery not only serve as temporal markers but also as pragmatic markers; at the same time, we also show that non-tam markers such as evidential markers and nominalizers can also develop extended pragmatic uses with ‘finitizing’ or ‘clausal grounding’ functions, since these pragmatic markers also contribute to the formation of independent clauses that can stand alone semantically, prosodically and morphosyntactically.

on the development of sentence final particles in chinese



Appendix A: List of Abbreviations

1sg 1pl 2sg 3sg 3pl acc adhort advers asp attr ce cl compl cond conv cop crs dem evid foc fut gen int neg nmlz nom non-viol pfv pn pot prog prt q sfp top

first person singular first person plural second person singular third person singular third person plural accusative adhortative adversity marker aspect marker attributive counterexpectation marker classifier completive conditional marker converb linker copula current relevant state; also anterior/perfect marker demonstrative evidential marker focus particle future marker genitive intensifier negator nominalizer nominative non-volition marker perfective person’s name potential progressive particle question marker sentence final particle topic marker

213

214

yap, yang and wong

Acknowledgements We wish to gratefully acknowledge generous research funding support from the Research Grants Council of Hong Kong (General Research Fund, PolyU 5513/10H) to the first author for the project “Stance Marking in Asian Languages: Linguistic and Cultural Perspectives”. We also wish to thank the Faculty of Humanities Dean’s Reserve for their generous support through the research project “Clause-medial Particles in Chinese” (hkpu 1-zv8a). We further wish to thank Professor Samuel Hung-nin Cheung for access to his Early Canton Dialect Database and to Dr Andy Chi-on Chin for access to A Linguistic Corpus of Mid-20th Century Hong Kong Cantonese Movies. We would also like to express our thanks to the editors and two anonymous reviewers, and to Mikyung Ahn, Ariel Chan, Weirong Chen, Lawrence Cheung, Winnie Oi-wan Chor, Lily He, Yuko Higashiizumi, Kathy Lin, Stephen Matthews, Mizuho Tamaji, Jiao Wang, and Brian Wai for their valuable comments and suggestions. References Ahn, Mikyung & Foong Ha Yap. 2013. Negotiating common ground in discourse: A diachronic and discourse analysis of maliya in Korean. Language Sciences 37: 36–51. ———. “On the extended uses of evidential markers in Korean: Diachronic and discourse perspectives”. Paper presented at the Conference on the Nature of Evidentiality, Leiden University, June 13–14. Aikhenvald, Alexandra. 2004. Evidentiality. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. & Randy J. LaPolla. 2007. New perspectives on evidentials: A view from Tibeto-Burman. Linguistics of the Tibeto-Burman Area. 30/2, Special issue on Evidentiality. Anonymous. 1996 [1824]. Chinese Courtship: in Verse, translated by Peter Perring Thoms. Macao: The Honorable East. ———. 1877. The Forty Exercises. Unknown publishing details. Ball, J. Dyer. 1883. Cantonese Made Easy: a Book of Simple Sentences in the Cantonese Dialect with Free and Literal Translations, and Directions for the Rendering of English Grammatical Forms in Chinese. Hong Kong: China Mail Office. ———. 1888. Cantonese Made Easy: A Book on Simple Sentences in the Canton Dialect. 2nd ed. Hong Kong: China Mail Office. ———. 1912. How to Speak Cantonese: Fifty Conversations in Cantonese Colloquial: with the Chinese character, free and literal English translations, and romanised spelling with tonic and diacritical marks, etc. 4th ed. Hong Kong: Kelly & Walsh.

on the development of sentence final particles in chinese

215

———. 1924. Cantonese Made Easy: A book on Simple Sentences in the Canton Dialect, with free literal translations, and directions for the rendering of English grammatical forms in Chinese. Revised and enlarged. 4th ed. Hong Kong: Kelly & Walsh. Bybee, Joan L., Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: The University of Chicago Press. Cao, Guangshun. 1995. Jindai Hanyu Zhuci. Beijing: Yuwen Chubanshe. Chao, Yuen Ren. 1947. Cantonese Primer. New York: Greenwood. ———. 1968. A Grammar of Spoken Chinese. Berkeley/Los Angeles: University of California Press. Cheung, Lawrence. 2009. Dislocation focus construction in Chinese. Journal of East Asian Linguistics 18/3: 197–232. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University. ———. 1985. Tense. Cambridge: Cambridge University. DeLancey, Scott. 2011. “Finite structures from clausal nominalization in TibetoBurman”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 343–360. Endo, Tomoko. 2006. “Experiencer defocusing and the grammaticalization of emotion predicates: The case of love and fear in Mandarin Chinese”. In Proceedings of the Sixth Annual Meeting of the Japanese Cognitive Linguistics Association, 75–84. ———. 2013. “Epistemic stance in Mandarin conversation: The positions and functions of wo juede (I feel/think)”. In Y. Pan & D. Kádár (eds.), Chinese Discourse and Interaction: Theory and Practice. London: Equinox, 12–34. Evans, Nicholas. 2007. “Insubordination and its uses”. In I. Nikolaeva (ed.), Finiteness: Theoretical and Empirical Foundations. Oxford: Oxford University, 366–431. Givón, Talmy. 1985. “Iconicity, isomorphism and non-arbitrary coding in syntax”. In J. Haiman (ed.), Iconicity in Syntax. Amsterdam: John Benjamins, 187–220. ———. 2001. Syntax, Volumes 1 & 2. Amsterdam: John Benjamins. Grunow-Hårsta, Karen. 2007. Mirativity and evidentiality in Magar. Linguistics of the Tibeto-Burman Area 30/2: 151–194. ———. 2011. “Innovation in nominalization in Magar: A Tibeto-Burman language of Nepal”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 215–254. Haiman, John & Sandra A. Thompson (eds.). 1988. Clause Combining in Grammar and Discourse. Amsterdam: John Benjamins. Higashiizumi, Yuko. 2006. From a Subordinate Clause to an Independent Clause: A History of English Because-clause and Japanese Kara-clause. Tokyo: Hituzi Syobo.

216

yap, yang and wong

———. 2012. Ellipsis and stance-taking in Japanese. Paper presented at the Workshop on Epistemicity, Evidentiality and Attitude in Asian Languages: Typological, Diachronic and Discourse Perspectives, Hong Kong Polytechnic University, September 3–5. Horie, Kaoru. 2011. “Versatility of nominalizations: Where Japanese and Korean contrast”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.). Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 473–496. Hunston, Susan & Geoff Thompson (eds.). 2000. Evaluation in Text: Authorial Stance and the Construction of Discourse. Oxford: Oxford University. Jiu, Ji-yung. 1828. Yue-ou. Canton: Dengyungge. ———. 1992 [1828]. Cantonese Love Songs: An English Translation of Jiu Ji-yung’s Cantonese Songs of the Early 19th Century, translated by Peter T. Morris. Hong Kong: Hong Kong University. Kärkkäinen, Elise. 2003. Epistemic Stance in English Conversation. Amsterdam: John Benjamins. Laury, Ritva (ed.). 2008. Crosslinguistic Studies of Clause Combining: The Multi­ functionality of Conjunctions. Amsterdam: John Benjamins. Law, Sam-po. 1990. The Syntax and Phonology of Cantonese Sentence Final Particles. PhD dissertation, Boston University. Lee, Cher Leng & Zheng-yin Agnes Cheong. 1999. Pragmatics of liao in Singapore Mandarin. Paper presented at the 17th Annual Meeting of the International Association of Chinese Linguistics, crlao-ehess, Paris. Legge, James. 1960. The Chinese Classics: With a Translation, Critical and Exegetical Notes, Prolegomena, and Copious Indexes. Hong Kong: Hong University Press. Leung, Wai-mun. 2006. On the Synchrony and Diachrony of Sentence Final Particles: The Case of wo in Cantonese. PhD dissertation, University of Hong Kong. ———. 2010. On the Synchrony and Diachrony of Sentence-Final Particles: The Case of wo in Cantonese. Saarbrücken: Lambert Academic Publishing. Li, Charles N. & Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Lim, Ni Eng. 2011. “From subjectivity to intersubjectivity: Epistemic marker wo juede in Chinese”. In Y. Xiao, .L. Tao & H.L. Soh (eds.), Current Issues in Chinese Linguistics. Cambridge: Cambridge Scholars Press, 265–300. Lin, Tzong-Hong Jonah. 2008. Complement-to-specifier movement in Mandarin Chinese. Manuscript, National Tsing Hua University, Hsinchu, Taiwan. Liu, Xunning. 1985. Xiandai hanyu juwei le de laiyuan [The origin of the sentence final le in Modern Mandarin]. Fangyan 2: 128–33.

on the development of sentence final particles in chinese

217

Lu, Louis Wei-lun & Lily I-wen Su. 2009. Speech in interaction: Mandarin particle le as a marker of intersubjectivity. Zeitschrift für interkulturellen Fremdsprachenunterricht 14/1: 155–168. Martin, James Robert. 2000. “Beyond exchange: Appraisal systems in English”. In S. Hunston & G. Thompson (eds.), Evaluation in Text: Authorial Stance and the Construction of Discourse. Oxford: Oxford University, 142–175. Matisoff, James. 1972. “Lahu nominalization, relativization, and genitivization”. In John Kimball (ed.), Syntax and Semantics, Volume 1. Studies in Language Series. New York: Seminar, 237–57. Matthews, Stephen. 1998. Evidentiality and mirativity in Cantonese: wo3, wo4, wo5. Manuscript, University of Hong Kong. Matthews, Stephen & Virginia Yip. 1994. Cantonese: A Comprehensive Grammar. London: Routledge. Morrison, Robert D. D. 1828 [2001]. Vocabulary of the Canton Dialect. London: Ganesha. [First ed. published in Macao: East India Company.] Mulder, Jean, Sandra A. Thompson & Cara Penry Williams. 2009. “Final but in Australian English conversation”. In P. Peters, P. Collins & A. Smith (eds.), Comparative Studies in Australian and New Zealand English: Grammar and Beyond. Amsterdam: John Benjamins, 337–358. Nikolaeva, Irina (ed.). 2007. Finiteness: Theoretical and Empirical Foundations. Oxford: Oxford University Press. ———. 2010. Typology of finiteness. Language and Linguistics Compass 4/12: 1176–1189. Noonan, Michael. 1997. “Versatile nominalizations”. In T. Givón, J. Bybee, J. Haiman & S.A. Thompson (eds.), Essays on Language Function and Language Type. Amsterdam: John Benjamins, 373–394. ———. 2008. “Nominalization in Bodic languages”. In M.J. Lopez-Couso & E. Seoane (eds.), in collaboration with T. Fanego, Rethinking Grammaticalization: New Perspectives. Amsterdam: John Benjamins. ———. 2011. “Aspects of the Historical Development of Nominalizers in the Tamangic Languages”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 195–214. Ohori, Toshio. 1995. “Remarks on suspended clauses: A contribution to Japanese phraseology”. In M. Shibatani & S.A. Thompson (eds.), Essays in Semantics and Pragmatics: In Honor of Charles J. Fillmore. Amsterdam: John Benjamins, 201–218. ——— (ed.). 1998. Studies in Japanese Grammaticaliztion: Cognitive and Discourse Perspectives. Tokyo: Kurosio.

218

yap, yang and wong

Ono, Tsuyoshi & Ryoko Suzuki. 1992. “The development of a marker of speaker’s attitude: The pragmatic use of the Japanese grammaticized verb shimau in conversation”. In Proceedings of the Eighteenth Annual Meeting of the Berkeley Linguistics Society: General Session and Parasession on The Place of Morphology in a Grammar (bls 18), 204–213. Rhee, Seongha. 2008. “On the rise and fall of Korean nominalizers”. In M.J. LopezCouso & E. Seoane (eds.), in collaboration with T. Fanego, Rethinking Gramma­ ticalization: New Perspectives. Amsterdam: John Benjamins, 239–264. ———. 2011. “Nominalization and stance marking in Korean”. In F.H. Yap, K. GrunowHårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 393–422. ———. 2012. Context-induced reinterpretation and (inter)subjectification: The case of grammaticalization of sentence-final particles. Language Sciences 34/3: 284–300. Roberts, Ian & Anna Roussou. 2003. Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University. Shibasaki, Reijirou. 2007. Ellipsis and discourse-syntactic structures in Japanese interview discourse: The emergence of the evidential marker to. Language and Linguistics 8/4: 939–966. ———. (In press). “From relativization to nominalization and more: Evidence from the history of Okinawan”. In Proceedings of the 36th Annual Meeting of the Beerkeley Linguistic Society (bls 36). Shinzato, Rumiko. 2007. (Inter)subjectification, Japanese syntax and syntactic scope increase. Journal of Historical Pragmatics 8/2: 171–206. ———. 2011. “Nominalization in Okinawan: From a diachronic and comparative perspective”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Amsterdam: John Benjamins, 445–472. Simpson, Andrew. 2008. “The grammaticalization of nominalizers in Burmese”. In M.J. Lopez-Couso & E. Seoane (eds.), in collaboration with T. Fanego, Rethinking Grammaticalization: New Perspectives. Amsterdam: John Benjamins, 265–288. Strauss, Susan. 2003. Completive aspect, emotion, and the dynamic eventive: Korean V-a/e pelita, Japanese V-te shimau, and Spanish se. Linguistics 41/4: 653–679. Strauss, Susan & Sung-Ock Sohn. 1998. “Grammaticalization, aspect, and emotion: The case of Japanese -te shimau and Korean -a/e pelita”. In D.J. Silva (ed.), Japanese/ Korean Linguistics, volume 8. Stanford: csli. Sun, Chaofen. 1996. Word Order Change and Grammaticalization in the History of Chinese. Stanford: Stanford University.

on the development of sentence final particles in chinese

219

Suzuki, Ryoko. 2007. (Inter)subjectification in the quotative tte in Japanese conversation: Local change, utterance-ness and verb-ness. Journal of Historical Linguistics 8/2: 207–237. Suzuki, Satoko. 1998. Tte and nante: Markers of psychological distance in Japanese conversation. Journal of Pragmatics 29: 429–462. Sybesma, Rint & Boya Li. 2007. The dissection and structural mapping of Cantonese sentence final particles. Lingua 117: 1739–1783. Taylor, John R. 2003. Cognitive Grammar. Oxford: Oxford University Press. Thompson, Sandra A. & Anthony Mulac. 1991. “A quantitative perspective on the grammaticalization of epistemic parentheticals in English”. In E.C. Traugott & B. Heine (eds.), Approaches to Grammaticalization, Volume 2: Focus on Types of Grammatical Markers. Amsterdam: John Benjamins, 313–327. Thoms, Peter Perring (translated). 1996 [1824]. Chinese Courtship: In Verse. Macao: The Honorable East. Traugott, Elizabeth C. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65/1: 33–65. ———. 1995. “Subjectification in grammaticalization”. In D. Stein & S. Wright (eds.), Subjectivity and Subjectivisation. Cambridge: Cambridge University Press, 37–54. ———. 2003. “From subjectification to intersubjectification”. In R. Hickey (ed.), Motives for Language Change. Cambridge: Cambridge University, 124–139. ———. 2010. “Revisiting subjectification and intersubjectification”. In K. Davidse, L. Vandelanotte & H. Cuyckens (eds.), Subjectification, Intersubjectifica­tion and Gramamticalization. Berlin: De Gruyter Mouton, 29–70. Van den Berg, Marinus & Guo Wu. 2006. The Chinese Particle le: Discourse Construction and Pragmatic Marking in Chinese. New York: Routledge. Watters, David E. 2008. Nominalization in the Kiranti and central Himalayish languages of Nepal. Linguistics of the Tibeto-Burman Area 31/2: 1–44. Whitman, John. 2001. “Relabeling”. In S. Pintzuk, G. Tsoulas & A. Warner (eds.), Diachronic Syntax: Models and Mechanisms. Oxford: Oxford University Press, 220–240. Xu, Baohua & Ichiro Miyata (eds.). 1999. Hanyu Fangyan Da Cidian [A Comprehensive Dictionary of Chinese Topolects]. Beijing: Zhonghua Shuzhu. Yap, Foong Ha & Mikyung Ahn. 2012. “On the development of evidentiality in Cantonese and Korean”. Revised manuscript of paper presented at the Workshop on Eviden­ tiality and Mood, National Seoul University, October 20, 2011. Yap, Foong Ha, Pik-ling Choi & Kam Siu Cheung. 2010. “De-lexicalizing di3: How a Chinese locative noun has evolved into an attitudinal nominalizer”. In A. Van Linden, J.-C. Verstraete & K. Davidse (eds.), in collaboration with H. Cuyckens.

220

yap, yang and wong

Formal Evidence in Grammaticalization Research. Typological Studies in Language 94. Amsterdam: John Benjamins, 220–240. Yap, Foong Ha, Winnie Chor & Jiao Wang. 2012. “On the development of epistemic ‘fear’ markers: An analysis of Mandarin kongpa and Cantonese taipaa”. In W. Abraham & E. Leiss (eds.), Covert Patterns of Modality. Cambridge, uk: Cambridge Scholars. Yap, Foong Ha & Karen Grunow-Hårsta. 2010. Non-referential uses of nominalization constructions: Asian perspectives. Language and Linguistics Compass 3/1: 1–21. Yap, Foong Ha, Karen Grunow-Hårsta & Janick Wrona. 2011. “Nominalization strategies in Asian languages”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Typological Studies in Language 96. Amsterdam: John Benjamins, 1–57. Yap, Foong Ha & Jiao Wang. 2011. “From light noun to nominalizer and more: The grammaticalization of zhe and suo in Old and Middle Chinese”. In F.H. Yap, K. Grunow-Hårsta & J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives. Typological Studies in Language 96. Amsterdam: John Benjamins, 61–107. Yap, Foong Ha, Jiao Wang & Charles Tsz-Kwan Lam. 2010. Clausal integration and the emergence of mitigative and adhoratative sentence-final particles in Chinese. Special Issue on Stance Phenomena in Chinese: Diachronic, Discourse and Processing Perspectives. Taiwan Journal of Linguistics 8/2: 63–84. Yeung, Ka-Wai. 2006. On the status of the complementizer waa6 in Cantonese. Taiwan Journal of Linguistics 4/1: 1–48. Yoshida, Eri. 1994. “Expectation motivating the use of the auxiliary shimau in Japanese spoken narratives”. In S. Iwasaki, T. Ono, H. Tao & H.S. Lee (eds.), University of Santa Barbara Working Papers in Linguistics, Volume 5: East Asian Linguistics. Santa Barbara: University of California, Santa Barbara, Department of Linguistics, 245–263. ———. 1995. “Speaker’s subjectivity and the use of shimau in Japanese spoken narratives”. In Noriko Akatsuka (ed.), Japanese/Korean Linguistics, volume 4. Stanford, ca: clsi Publications, 183–199.

Chapter 9

The Interplay of Discourse and Prosody at the Left and Right Periphery in Korean: An Analysis of kuntey ‘but’ Sung-Ock S. Sohn and Stephanie Hyeri Kim 1 Introduction1 In recent years, there has been a growing body of research on the grammatical or discourse functions of left and right peripheries of a sentence or an utterance (Shinzato 2007; Waltereit and Detges 2007; Mulder and Thompson 2008; Mulder, Thompson and Williams 2009; Traugott 2010). Cross-linguistic data indicate that there are distinctive communicative functions which are specific to the left or right peripheries. For instance, studies show that the left periphery (lp) is mainly concerned with discourse organization (e.g. topic shift), whereas the right periphery (rp) indicates a modal function (e.g. speaker’s stance towards the message or the addressee) (Beeching and Detges, this volume). Previous studies have provided functional linguists with invaluable insights into the functional differences between the two peripheries. Nonetheless, there are some pending issues to be explored. First of all, while periphery typically refers to both left and right edges of an utterance or a sentence, it remains unclear how we define and identify the scope of a periphery (i.e., periphery of what?). Most previous studies defined the periphery on the basis of grammatical units such as sentence, clause, phrase, or argument structure and very few studies have examined the peripheral expressions using a turn unit in natural discourse (for example, see Mulder and Thompson 2008; Kim and Sohn 2011, forthcoming; Haselow 2011). Second, while previous studies demonstrated functional differences between left and right peripheral expressions in terms of pragmatic meanings, it remains unexplored how speakers use prosodic features to achieve such 1 The authors would like to thank Kate Beeching and two anonymous reviewers for their invaluable comments on an earlier version of this paper. Also, the authors gratefully ack­ nowledge Seunggon Jeong’s insightful comments and help on the prosodic features of the Korean data.

© koninklijke brill nv, leiden, ���4 | doi 10.1163/9789004274822_010

222

sohn and kim

different communicative functions at the two peripheries. Communicative functions cannot be achieved through syntax and a lexical choice alone. Prosodic features play a crucial role in identifying the scope of periphery by providing empirical evidence for any prosodic juncture. For example, prosodic features can offer evidence whether discourse markers are intonationally disjunct from the rest of the utterance at lp or rp. While most discourse markers in English occur at lp and are disjunct intonationally, in many Asian languages they often occur at rp and are not always disjunct from the preceding element (e.g. Kim and Sohn 2011, forthcoming; Rhee 2013; Traugott 2013; Sohn, forthcoming). Since discourse markers can occur at lp and/or rp syntactically, when a speaker takes more than one utterance to complete his/her turn and includes a discourse marker at the intersection of the two utterances, it may not be clear whether the discourse marker is at the end of the prior utterance (i.e. rp), or at the beginning of the subsequent utterance (i.e. lp), if prosodic features are not taken into consideration. For instance, consider an example drawn from Korean conversation. (1)

A: paper-lul ppalli mos nay-l kes kath-ayo kuntey paper-lul ppalli paper-acc soon not submit-pro nml seem-pol kuntey paper-acc soon nay-myen-un hankwuk tuleka-ki swip-ci anh-canh-ayo2 not submit-if-top Korea enter-nml easy-nml not-modal-pol ‘It appears that I may not be able to submit a paper soon kuntey if I don’t submit the paper soon, you know, it would be difficult to go to Korea.’



B: kuleh-ki-n ha-ta ku-ci e be so-nml-top do-dec right-comm yeah ‘Yeah, it appears so, right?’

In the above, the discourse marker kuntey expresses the meaning of ‘but, however, by the way’, and is permissible at lp and/or at rp. In (1) the question arises as to whether the discourse marker is produced at the end of a current utterance or at the beginning of the next utterance. Here, prosodic features can serve as the key factor in determining which of the two cases is valid.3 Consider another example where kuntey is syntactically ambiguous between lp and rp. In (2) kuntey appears between two utterances produced by one speaker. 2 The modality suffix -canh- expresses the meaning of ‘you know’ (Sohn 2010). 3 A pitch analysis for this study shows that the kuntey in (1) appears at the beginning of the following utterance, that is, it occurs at lp.

Discourse and prosody in Korean: the case of kuntey ‘but’

223

(2) B: .hhhh ku chonsulep-ta-nun key cincca cey:il ccille-ss-ta .hhhh that countrified-dec-mod thing really most stab-pst-dec kuntey hye-ka ki-ney-yo4 kulen kes-pota kuntey   tongue-nom long-ind-pol such thing-than ‘That (expression) tacky is really most insulting kuntey more than any other expressions such as ‘you have a long tongue’.

A: hehehhh ‘Ha ha . . .’

The syntactic ambiguity of kuntey can be disambiguated by differences in prosodic boundaries. A pitch analysis for this study indicates that kuntey in (2) is prosodically attached to the preceding predicate, that is, it occurs at rp. (This will be illustrated in Figure 9.9 in section 5.2.) Previous studies have mainly focused on the use of kuntey at lp since kuntey typically occurs at the beginning of an utterance or a turn (e.g. Kim and Suh 1996; Park 1999; Choi 2007). The use of kuntey at rp is a recently observed phenomenon; Kim and Sohn’s study (2011, forthcoming) is the first in which the usage of kuntey at rp is investigated. They demonstrate that the discourse marker kuntey in Korean, whose default position is known to be the beginning of an utterance (i.e. lp), can also appear at the end of an utterance (i.e. rp). Consider the following example for the use of kuntey at rp. (3) A: The problem is serious kuntey. B: Really?

Kuntey at rp typically occurs at the end of a turn unit. As defined in the field of conversation analysis, a single turn-at-talk may be built out of several turn constructional units (tcus), in which a tcu, simply speaking, is defined as the most fundamental segment of speech that is grammatically, intonationally, and actionally complete (Sacks, Schegloff and Jefferson 1974). Thus, after a tcu that ends with kuntey the speaker may extend his/her turn by adding another tcu as shown in (2), or there could be a turn transfer to another person as shown in (3). Such final kuntey at rp, which is prosodically attached to the sentence-ending suffix of the preceding predicate, exhibits unique functions

4 The suffix -ney is a familiar-level declarative ender, which expresses indicative mood (Sohn 1999).

224

sohn and kim

and prosodic features. While, at lp, kuntey is mainly used for discourse organization, at rp it is placed at the unit end to avoid showing the speaker’s disaffiliative stance from the outset (Kim and Sohn 2011, forthcoming). The present chapter investigates the relationship between the discoursepragmatic functions of periphery elements and prosodic features by analyzing the intonation patterns of kuntey at lp and rp in Korean conversation. It raises the following research questions: 1. 2.

Are there any differences in the intonation patterns of kuntey at lp and rp? How is the difference, if any, relevant to the grammatical and/or communicative functions of kuntey at lp and rp?

The Korean language offers a unique opportunity to examine the relationship between the peripheral expressions and the use of prosodic features. Typologically, Korean is an intonational language in which prosodic prominence is based on phrasal tones, not on lexical tones or stress. Phrasal tones mark prosodic groupings and the tonal pattern of the prosodic phrase is highly predictable (Jun 1996, 1998, 2011; Park 2003). In addition, Korean is an agglutinative verb-final language. As such, Korean displays a wide range of sentence-final suffixes which express the speaker’s modality and speech style (e.g. formal, informal polite, intimate, and plain styles) (Sohn 1999). A boundary tone which is marked at the end of the largest prosodic unit is realized on the final syllable of sentence-final suffixes. The organization of this chapter is as follows. We begin with the description of the data and methods for this study in section 2. We discuss the origin of kuntey in section 3 and introduce the prosodic structures of Korean in section 4. In section 5, we examine the prosodic features of kuntey at lp and rp (section 5.1 and section 5.2, respectively), and in section 5.3 we examine the prosodic features of kuntey at both lp and rp within a single turn by the same speaker. In section 6 we demonstrate how the usages of kuntey at rp are interactionally different from the usages of kunety at lp. 2

Data and Method

The data for this study is drawn from the Korean Linguistics Data Consortium (ldc) spoken corpus, which is a collection of naturally occurring telephone conversations (approximately 44 hours of audio) made in the early 2000s, involving more than 100 speakers. It contains a total of 464,581 words in Korean

Discourse and prosody in Korean: the case of kuntey ‘but’

225

and a total of 1,805 tokens of kuntey. All conversations are between native speakers of Korean. Participants for the project were given a free calling card (30 min.) to call friends and/or family. Both callers and recipients were informed that their conversations would be recorded, but were given no guidelines concerning topics or speech styles. The speakers in the corpus range in age from early-twenties to 70’s (with both genders) and include students, housewives, social workers, etc., and the corpus consists of 100 telephone conversations among friends and family members. The ldc corpus is one of very few spoken corpora in Korean drawn from spontaneous conversation and accompanied by digitized sound files and transcripts. The corpus is publically available at http://www.ldc.upenn.edu/.5 In order to identify kuntey at lp and rp, we analyzed the intonation contour of kuntey at peripheries using the Praat program available at http://www.fon .hum.uva.nl/praat/. The analysis of intonation patterns provides important insights into whether kuntey is used at lp or rp. Out of a total of 1,805 kuntey tokens in the ldc corpus, this chapter will present the findings from a pilot study based on the prosodic analyses of 40 tokens (20 tokens each at lp and rp). The analysis of prosodic units in Korean and the notation of tonal configurations in this study are based on an intonational approach (Pierrehumbert and Beckman 1988; Jun 1998, 2000, 2007, 2011). The notion of periphery in this study is defined in relation to a turn constructional unit. lp represents the beginning of a turn constructional unit, and rp represents the final position of a turn constructional unit. Drawing from the studies in interactional linguistics (Ochs, Schegloff, and Thompson 1996; Selting and Couper-Kuhlen 2001; Ford, Fox, and Thompson 2002), we analyzed how speakers of kuntey use prosodic features at peripheries to achieve respective communicative goals. 3

The Origin of kuntey

The discourse marker kuntey originates from the deictic demonstrative predicate kule- ‘that is so; to be such’ and the contrastive connective nuntey ‘but’ (Suh 1996; Choi 2007). Nuntey has been primarily characterized as a clause connective which expresses a contrastive relation or provides background information. The sequence of the demonstrative predicate kule- and the clause connective nuntey has become grammaticalized into the discourse connective kulentey, which has been sequentially developed into the reduced form kuntey 5 The University of Pennsylvania is the ldc’s host institution.

226

sohn and kim

in spoken discourse. The grammaticalization path of kuntey is illustrated below (An 2000; Choi 2007; Kim and Sohn 2011, forthcoming). (4) kule- ‘that is so’ + nuntey/ntey ‘but’ > kulentey ‘but’ > kuntey ‘but’

Being derived from the deictic predicate and the contrastive connective, both the full form kulentey and the short form kuntey are frequently used at the beginning of an utterance or a sentence (i.e. lp) as a discourse connective. While the two expressions were traditionally considered to be free variants of a single form, Choi (2007) reveals that there are visible functional differences between the two forms in spoken discourse. Using a corpus-based frequency analysis, Choi demonstrates that the full form kulentey mainly serves as a referential connective, whereas kuntey is chiefly used to initiate a topic. Another major difference between the two forms is their position at a turn unit; while both forms occur at lp, only the reduced form kuntey is observed at rp (Kim and Sohn 2011, forthcoming). The difference in position has to do with the lexical meaning of the source form which originates in the sequence of the deictic predicate (kule- ‘that is so’) and the clause connective nuntey/ntey ‘but’. That is, the full form is mainly used as a referential connective at lp since kulentey retains the erstwhile lexical meaning. As the full form is frequently used at the beginning of a turn or tcu, it gains a new pragmatic meaning as discourse marker and becomes reduced (kulentey > kuntey). Once the lexical word has become a discourse connective, it further develops into a final particle at rp. The fact that only the reduced form is used at rp indicates that the kuntey at rp has lost a referential meaning and is on the path to grammaticalization as a final particle, as illustrated in Kim and Sohn (forthcoming). The grammaticalization of kuntey is further supported by tonal configurations. In the present study, the prosodic features of kuntey at lp and rp will be further discussed in the following sections. 4

Prosodic Structure of Korean

Before discussing the prosodic features of kuntey at lp and rp, we will briefly describe the prosodic structure of Korean. The prosodic structure of Korean for this study is based on the intonational approach which assumes a hierarchical prosodic structure (Pierrehumbert and Beckman 1988; Jun 1998, 2000; Park 2003). This approach defines the prosodic units larger than a word based on the surface phonetic form of an utterance by looking at suprasegmental features (e.g. intonation and final lengthening).

Discourse and prosody in Korean: the case of kuntey ‘but’

227

According to Jun’s study (2011), the intonational structure of the standard dialect of Korean has three intonationally defined prosodic units above a word: accentual phrase (ap), intermediate phrase (ip), and intonation phrase (ip).6 An ip is marked by a boundary tone (%) which is realized on the final syllable of an intonation phrase. An intermediate phrase is larger than an ap and smaller than an ip. The criterion is defined by pitch reset, either by higher apinitial tone (+H) or by a high intermediate phrase boundary tone (H-), and minor phrase-final lengthening when marked by H-. The hierarchical organization of the prosodic units in Korean is schematically represented in Figure 9.1 (adopted from Jun 2011). ip ip

(ip)

(ap)

ap w s

(w)

s ... s

T +H

s

L+ Ha

(H-)

T%

Figure 9.1 Intonation model of Seoul Korean ip: Intonation Phrase; ip: Intermediate phrase; ap: Accentual Phrase w: phonological word; s: syllable T = H, when the ap-initial segment is aspirated or tense consonant or /h, s/; Otherwise, T = L +H: H tone realized on the second syllable of ap L+: L tone realized on the penultimate syllable of ap Ha: an ap-final boundary tone. (H-): an optional ip-final boundary tone T%: an ip-final boundary tone (e.g., H%, L%, lh%, hl%, lhl%, hlh%) (when boundary tones of different prosodic units are realized on the same syllable, the tone of a higher prosodic unit overrides that of a lower prosodic unit.)

6 Standard Korean is based on the Seoul dialect.

228

sohn and kim

As illustrated in Figure 9.1, an ap is smaller than an intermediate phrase (ip) and larger than a phonological word (w), which is a lexical item plus a case marker or postposition. An ap may contain one or more words and is typically marked by phrase initial Low-High (lh) or High-High (hh). The underlying default tonal pattern of an ap in Korean is Low-High-Low-High (lhlh) or High-High-Low-High (hhlh) (Jun 2007). When the ap-initial segment is aspirated or tense, it is marked with H, otherwise it is marked with L. Regardless of the tonal difference on the first syllable of an accentual phrase, the second syllable of an accentual phrase is H when the ap has more than 3 syllables. The end of the largest prosodic phrase (i.e. intonation phrase) is marked by a boundary tone (%) such as a low boundary (L%), a high boundary (H%), etc.7 The boundary tone is realized in the final syllable of an intonation phrase. Korean boundary tones which mostly coincide with sentence-enders deliver discourse-pragmatic meaning as well as information about the sentence type such as declarative, interrogative, etc. (Park 2003). For instance, sentenceenders marked with a high boundary tone tend to indicate seeking information or confirmation, whereas those marked with a low boundary tone tend to make a statement. The tonal configuration and boundary of prosodic units play an important role in determining if the discourse marker kuntey occurs at lp or rp. In section 5, we will compare the prosodic features of kuntey at two peripheries. All tokens of kuntey in this study are drawn from the ldc spoken corpus. We will begin with the discussion of kuntey at lp. 5

Prosodic Features of kuntey at lp and rp

5.1 Prosodic Features of kuntey at lp As discussed in previous studies, the discourse marker kuntey that is most frequently used for discourse organization functions typically occurs at the beginning of a turn (e.g. Park 1997, Kim and Suh 1996). The prosodic analysis for this study shows that the kuntey at lp typically forms a separate prosodic unit such as an accentual phrase, an intermediate phrase, or an intonation phrase. This is illustrated in Figure 9.2 in which kuntey at lp constitutes an intermediate phrase. (See Appendix B for the notations of intonation contours.) 7 According to Jun (2007), there are at least nine boundary tones (L%, H%, lh%, hl%, lhl%, hlh%, hlhl%, lhlh%, lhlhl%) in Korean. The nine boundary tones can be classified in three main tonal groups based on their meaning: monotonal H% and L%, bitonal lh% and hl%, and multitonal hlh%, lhl%, lhlh%, hlhl%, lhlhl% (cf. Park 2003).

Discourse and prosody in Korean: the case of kuntey ‘but’

229

Figure 9.2 Kuntey it has already been six months.

The segment in Figure 9.2 (a total length of approximately 1.41 second) is excerpted from a telephone conversation between friends in the ldc corpus. Note that the discourse marker kuntey at lp is marked by a High intermediate phrase boundary tone (H-). As discussed earlier, an intermediate phrase is higher than an accentual phrase and smaller than an intonation phrase (Jun 2007). The largest prosodic unit (ip) in Figure 9.2 is marked by final lengthening and a low boundary tone (L%) at the end of an utterance. The tonal configuration confirms that kuntey at lp constitutes a separate prosodic unit; it has a clear juncture with the rest of the utterance. A salient juncture marked by kuntey at lp is also observed in Figure 9.3. (The discourse context of kuntey in Figure 9.3 will be discussed in example (6) in section 6.1.) Note in Figure 9.3 that kuntey forms one prosodic unit with the preceding interjection ‘oh’. A pitch reset triggers a bigger prosodic juncture than an accentual phrase boundary. Kuntey at lp is marked by an intermediate phrase juncture.8 The boundary juncture between the kuntey and the following phrase cincca ‘really’ is larger than an accentual phrase boundary.

Figure 9.3 Oh kuntey I am really hungry.

8 This juncture (i.e. an intermediate phrase) is not large enough to be categorized as an intonation phrase boundary because the final syllable of kuntey is not lengthened.

230

sohn and kim

Consider another example where kuntey at lp is marked by an intermediate phrase boundary and a pitch reset. (The segment in Figure 9.4 is drawn from example (6) in section 6.1.)

Figure 9.4 Kuntey it turns out that (Sangok sister can smile with the eyes.).

The prosodic juncture marked by kuntey at lp in Figure 9.4 is a High intermediate phrase boundary. (See Figure 9.1 for the tonal description.) It is not large enough to be categorized as an intonation phrase boundary because the final syllable of kuntey is not lengthened. However, a pitch reset in the following accentual phrase (i.e. al-ko po-nikka) triggers a bigger prosodic juncture than an accentual phrase boundary. Kuntey at lp marks an explicit break for the rest of utterance. (See example (6) for the complete utterance of Figure 9.4.) The salient prosodic juncture marked by kuntey (i.e. an intermediate phrase boundary) appears to signal a topic shift in the upcoming utterance. This will be discussed in more detail in conjunction with the discourse function of kuntey in example (6) in section 6.1. Kuntey at lp can be marked by a bigger juncture than an intermediate phrase juncture. This is illustrated in Figure 9.5. (The segment in Figure 9.5 appears after the last turn in example (6) in section 6.1.)

Figure 9.5 Kuntey in the photo you were only smiling softly with your eyes.

Note in Figure 9.5 that kuntey at lp is marked by the largest prosodic unit, i.e., an intonation boundary tone (H%). A pitch reset is realized by a higher

Discourse and prosody in Korean: the case of kuntey ‘but’

231

initial H tone of the accentual phrase after kuntey. The explicit break and juncture after kuntey confirms the discourse function of kuntey. As discussed earlier, kuntey at lp is most frequently used for discourse organization at the beginning of a turn. Marked with a salient juncture, kuntey in Figure 9.5 signals a shift to another topic (i.e. ‘In the photo you were only smiling softly with your eyes’). Similarly, kuntey in Figure 9.6 is marked by an intonation phrase boundary tone and a pitch reset. The segment in Figure 9.6 is excerpted from example (5) where kuntey is used as a topic shifter at the beginning of a turn. (See example (5) in section 6.1 for the discourse context of kuntey in Figure 9.6.)

Figure 9.6 Kuntey what photos are (they)?

As shown in Figure 9.6, kuntey is marked by an intonation phrase boundary (bitonal hl%), which is the largest prosodic unit in Korean. Also, note phrase final lengthening at the second syllable of kuntey and a pitch reset. Marked by the boundary tone and final lengthening as well as the pitch reset, kuntey in Figure 9.6 forms a salient prosodic juncture.9 So far, we have discussed the prosodic features of kuntey at lp. What is common among the aforementioned examples is that kuntey at lp forms a distinct prosodic unit such as an intermediate phrase and an intonation phrase. A salient disjuncture is marked by the tonal patterns of kuntey at lp. While kuntey at lp forms a separate intonation unit, kuntey at rp shows quite a different pattern. In section 5.2 we will discuss the prosodic features of kuntey at rp. 5.2 Prosodic Features of kuntey at rp Unlike kuntey at lp, kuntey at rp is prosodically attached to the preceding element. This is illustrated in Figure 9.7. (See example (8) in section 6.2 for the discourse context of the final kuntey.) 9 See Jun (2007, 2011) for the criterion of prosodic units in Korean.

232

sohn and kim

Figure 9.7 The problem is serious kuntey.

The analysis of intonation contours shows that the kuntey at rp is prosodically integrated into the preceding predicate and forms one intonation unit.10 Marked by the boundary tone (L%) and final lengthening, the final kuntey is appended to the predicate. This stands in sharp contrast with the prosodic feature of kuntey at lp which is intonationally disjunct. Similarly, kuntey at rp in Figure 9.8 is integrated into the preceding predicate.

Figure 9.8 It’s a general education section at the biology department kuntey.

Attached to the preceding predicate which ends in a declarative suffix (-ya in an intimate speech style), kuntey at rp in Figure 9.8 is prosodically integrated into the predicate. The prosodic features of kuntey can provide empirical evidence for disambiguating the syntactic obscurity of kuntey between lp and rp. In the following segment drawn from an earlier example (2), it is unclear whether kuntey occurs at lp or rp. (2) B: .hhhh ku chonsulep-ta-nun key cincca cey:il .hhhh that countrified-dec-mod thing really most kuntey kuntey

10

ccille-ss-ta stab-pst-dec

The predicate is marked with the sentence-ending suffix -ay which indicates an intimate speech style.

Discourse and prosody in Korean: the case of kuntey ‘but’

233

hye-ka ki-ney-yo kulen kes-pota tongue-nom long-ind-pol such thing-than ‘That (expression) tacky is really most insulting kuntey than any other expressions such as ‘you have a long tongue’.

The prosody analysis in Figure 9.9 shows that the final kuntey occurs at rp. Attached to the preceding predicate which ends with a declarative suffix (-ta), the final kuntey forms one prosodic unit (i.e., an intonation phrase boundary) with the preceding predicate.

Figure 9.9 That (expression) tacky is really most insulting kuntey

As discussed earlier, as an agglutinative language, Korean displays a wide range of sentence-final suffixes. The final kuntey which is attached to the preceding predicate tends to occur with diverse sentence-final suffixes such as interrogative, imperative as well as declarative. For instance, the final kuntey in Figure 9.10 is attached to a predicate which is marked with an imperative suffix. The imperative predicate (ponay ‘send’) forms one intonation unit with the final kuntey.

Figure 9.10 Hey, send me a photo soon kuntey.

Likewise, the final kuntey constitutes one prosodic unit when it occurs with an interrogative predicate. In Figure 9.11 and Figure 9.12, kuntey is attached to the preceding predicate marked with a final interrogative suffix -nya (in a plain speech style).

234

sohn and kim

Figure 9.11 How can you find a lodging (with that money) kuntey?

In both Figure 9.11 and Figure 9.12, the final kuntey is attached to the interrogative suffix (-nya), and is prosodically integrated into the prosodic unit of the predicate. Also note in Figure 9.12 that the voiceless sound /s/ in moksoli ‘voice’ is marked with a High tone since it is the second syllable of an accentual phrase. As discussed in Jun (2007), the second syllable of an ap is always H when the ap has more than 3 syllables.

Figure 9.12 What’s with your voice kuntey?

Thus far, we have discussed the prosodic features of the final kuntey. While kuntey at lp forms an independent prosodic unit, kuntey at rp is attached to the preceding predicate. Attached to the preceding predicate, the final kuntey forms one intonation unit together with the predicate. The difference in prosodic features of kuntey at lp versus rp is further illustrated in examples where kuntey is produced at both lp and rp in the same turn by one speaker. In the next section, we will discuss an example in which kuntey appears at both lp and rp. 5.3 Prosodic Features of kuntey at Both lp and rp In Figure 9.13, kuntey at lp forms an intonation phrase marked with a high boundary tone (H%). A salient juncture after kuntey at lp confirms that the initial kuntey is phrased separately. In contrast, kuntey at rp is attached to the

Discourse and prosody in Korean: the case of kuntey ‘but’

235

preceding predicate with no pause and is integrated into the preceding predicate. The final lengthening which is realized at the end of the low boundary tone (L%) provides further evidence that the final kuntey forms one prosodic unit with the preceding predicate.

Figure 9.13 Kuntey I cannot say for sure kuntey

The contrast in tonal configurations between lp and rp is also observed when kuntey occurs with another discourse marker. As a typical agglutinative language, it is quite common in Korean to have a sequence of pragmatic markers (Sohn, forthcoming). This is illustrated in Figure 9.14 where kuntey occurs at both lp and rp.

Figure 9.14 Kuntey by the way how’s living kuntey?

Note in Figure 9.14 that kuntey at lp is followed by the discourse particle mwe which is a contracted form of the question word mwues ‘what’.11 Originating from an interrogative noun, mwe is frequently used as a discourse particle to mitigate the speaker’s stance (cf. Cha 2010). The low boundary tone (L%) marked in mwe confirms the function of mitigation. Preceded by kuntey without any pause, mwe is prosodically as well as morphologically attached to 11

According to Cha’s corpus-based study (2010), 85% (678 out of 790 tokens) of mwe in conversation is used as a discourse marker.

236

sohn and kim

the preceding element. That is, the sequence of ‘kuntey mwe’ is marked by an intonation phrase (with a low boundary tone), thus forming one intonation unit.12 In contrast, kuntey at rp is attached to the preceding predicate which ends with an interrogative ending suffix (-nya) and is integrated into the prosodic unit of the predicate. Thus far, we have compared and contrasted the prosodic features of kuntey at lp and rp. The analysis of tonal configurations confirms that kuntey at lp is disjunct prosodically, whereas kuntey at rp is integrated into the preceding predicate. In the following section, we will explore whether and how the differences in prosody are relevant to the discourse functions of kuntey at lp and rp. 6

Discourse Functions of kuntey at lp and rp

6.1 Usages of kuntey at lp Previous studies have shown that kuntey at lp is used primarily for discourse organizing functions such as topic resumption and shift. For example, Park (1997) points out that kuntey at lp (or, at turn-initial position in Park’s terms) is frequently used in resuming the speaker’s own prior talk or interrupting the immediately prior sequence. Similarly, Kim and Suh (1996) introduce kuntey as a counterpart of the English “by the way” and demonstrate kuntey’s function as a topic shifter. Specifically, they show that the lp kuntey is used when the speaker proffers a new topic in order to deal with a problematic situation and to avoid the problem that the prior turns have presented to the speaker. The lp kuntey’s function as a topic shifter is illustrated in the following example from a telephone conversation between two girlfriends living in different cities. The conversation took place in the early 2000s when film cameras were still popular. A and B went on a trip together months ago and used B’s camera to take pictures; however, B has yet to send the photos to A. As will be shown in the excerpt, A does not remember which photos are being discussed.

12

It appears that mwe is in the process of grammaticalization from a discourse marker into a particle. The fact that there is no pause between mwe and the preceding element supports the grammaticalization process of mwe. Furthermore, that ‘kuntey mwe’ forms one prosodic unit (i.e., an intonation phrase) demonstrates that mwe has lost its lexical status and become like a particle.

Discourse and prosody in Korean: the case of kuntey ‘but’

237

(5) [ko_6808 (00:17:19)] 01

B: ai sacin-un cengmal : : , dm photo-top really ‘Well, the photos, really’

02 A: ung := ‘yes’ 03 B: =cey-ka toytolok-i-myen i:hon-nom possibly-be-if 04 05 -> A:

ponay tuli-lkkey-yo : : ceng[mal. send     send:hon-will-pol really ‘I will try to send ((them)), really’ [kuntey mwusun sacin-i-ya? kuntey  what photo-be-Q KUNTEY   what photos are ((they))?

06 sayngkak-i cal an (h)na hhhhh thought-nom well neg remember : ie ‘Can’t remember well’ 07 B: enni ↑phulinsuthen-eyse ccik-un ke-yo : : sister Princeton-loc take-mod thing-pol ‘Sister, the ones ((we)) took in Princeton’

In response to B’s promise to mail the photos (lines 1, 3–4), A asks a kunteyinitiated question that proffers a topic associated with, but different from, the prior topic. The new topic still addresses the photos taken during the trip; however, the question initiates a new sequence (Schegloff 1968, 2007) and takes the conversation in a clearly different direction that departs from the prior turn. Here, kuntey is used to signal the turn as being related to the prior turn in some way, but at the same time as being shifted to a new topic. Topic shift also occurs between two turns produced by the same speaker. This is shown in the following example from the same telephone conversation. The speakers are still discussing the photos taken during the trip. Sangok, who was also on the trip, is in the photos, too. In the immediately prior talk,

238

sohn and kim

B has stated that Sangok always complains about her eyes looking swollen in photos. (6) [ko_6808 (00:18:15)] 01

B: Sangoki enni-nun : : wenlay sangoki enni nwun-i pwu-un name sister-top originally name sister eye-nom swollen-mod

02 ke-l cal molu-nun kes kath-te [la-kwu-yo : : thing-acc well don’t : know-mod nml seem-retro-qt-pol ‘It seems that Sangok doesn’t realize her eyes are actually swollen.’

03 A:

[HAHEHAHAhhhhhhhhhhhh

04 [°ettek-ha-myen coh-a. how-do-if like-ie ‘Oh dear’ 05 B: [wenlay com pwu-ess-cahn-ayo: nwun-i? originally little swollen-pst-you:know-dec:pol eye-nom ‘((Her)) eyes are a little swollen, you know: ((Her)) eyes are?’ 06 A: hhhhhhh 07 (.) 08 -> B: kuntey al-ko po-nikka Sangoki enni-to kuntey know-and see-when name sister-also 09 nwun-wusum-i iss-tela eye-smile-nom exist-retro ‘KUNTEY it turns out that Sangok sister can smile with the eyes.’ 10

A: al-ko po-nikka? know-and see-after ‘it turns out?’

Discourse and prosody in Korean: the case of kuntey ‘but’ 11

B: ce-n eps-nun cwul al-ass-ketun-yo? i : hon-top don’t : exist-mod dn know-pst-correl-pol ‘because ((I)) never knew ((she)) had ((it))’

12

A: a : ‘oh’

239

After stating that Sangok often complains about her swollen eyes in photos, B dismisses Sangok’s complaint (in a joking manner) by saying Sangok does not realize her eyes are actually swollen. A’s big laughter (line 3) shows that she finds B’s frankness humorous. However, the laughter displays neither agreement nor alignment (Glenn 2010). Thus, to seek A’s agreement, B further asks ‘((her eyes)) are already a little swollen, you know’ (line 5). In designing this turn, she uses the -cahn-a-‘you know’ construction, which supposes the recipient as a party who shares the same knowledge, as well as com ‘a little’ that mitigates her previous stance. Then, B even further extends this turn by adding ‘((her)) eyes are?’ in a question intonation that solicits a response. However, A only laughs in response (line 6) again, and a hearable micro pause ensues (line 7). Facing trouble in securing an agreement from A, B attempts to remedy the trouble by shifting to a related but new topic: Sangok’s nwun-wusum ‘smiley eyes’ (line 8). This turn is initiated with kuntey. The turn is topically associated with the speaker’s prior turn in that it concerns Sangok’s eyes, but it is also different in that it concerns Sangok’s smiley eyes and praises this new aspect of her eyes (rather than focusing on the unfavorable aspect: swollenness). After using kuntey at lp to index the turn as having shifted to a new topic, B also uses the retrospective suffix -tela to ground her new finding as an objective one based on her observation of Sangok’s eyes in the photos (Kim 2005). The prosodic juncture marked by kuntey at lp supports the discourse function of kuntey in (6). As discussed in section 5, kuntey at lp is marked by a salient juncture (an intermediate phrase boundary). The explicit break and disjuncture signal a topic shift in the upcoming utterance. As illustrated in Figure 9.15, kuntey at lp in line 8 is produced with a bigger prosodic boundary than an accentual phrase. This suggests that Korean speakers tend to produce a prosodic boundary between kuntey at lp and the subsequent phrase to signal a topic shift.

240

sohn and kim

Figure 9.15 Kuntey it turns out that Sangok sister can smile with the eyes.

Kuntey at lp not only indexes topic shift, but also marks the resumption of a prior sequence that has been mentioned earlier in the conversation. B’s hungriness has been mentioned twice earlier in the call (about 10 and 15 minutes prior to the segment), and is being returned to in line 2 below. Note that A is the caller and has volunteered to have a 30-minute phone conversation for a small payment in return. (7) [ko_6808 (00:19:40)] 01 (4.0) 02 -> B: e kuntey cincca paykophu-ta. Oh kuntey really hungry-dec ‘Oh KUNTEY I am really hungry’ 03

A: al-ass-e al-ass-e cenhwa kkunh-ulkkey.13 know-pst-ie know-pst-ie phone hang:up-promissive ‘okay okay ((I))’ll hang up’

04 mwe-lato sa-mek-ko wa. something-even:if buy-eat-and come:ie ‘go out and get something to eat’

Immediately prior to this excerpt, a previous sequence has come to a close, and a (4.0) pause, which is not only noticeable but can present awkwardness, has ensued. Earlier in the call, there were extensive discussions about B’s hungriness and the debate whether B should go out to grab a late dinner. The call has 13

The sentence ending suffix -(u)lkkey expresses the speaker’s promise or assurance (Sohn 1999).

Discourse and prosody in Korean: the case of kuntey ‘but’

241

now continued for approximately 19 minutes, and when the opportunity (i.e. silence) comes for B to return to the topic (and possibly, to end the call), she takes it and initiates the turn with kuntey to signal that she is “only” returning to the earlier topic. A, who is the caller, understands this to be a complaint for the lengthy conversation, and responds, ‘okay okay I’ll hang up’ (line 3) and suggests B get something to eat (line 4). As the examples demonstrate, the lp kuntey regularly indexes discourseorganizing functions such as topic shift and resumption. Although more research is necessary, such functions appear to correlate with the prosodic features discussed earlier in 5.1, where kuntey at lp is described as typically forming a separate prosodic unit and as showing a clear juncture from the impending turn. We hypothesize that these prosodic features characterizing the lp kuntey are motivated by, and contribute to, the discourse functions demonstrated in this section. Whether kuntey is indicating topic shift or resumption, it indexes that speakers are departing from the immediately prior turn in some way. This departure is marked in the prosody, which, at the same time, furnishes Korean speakers with a tool to communicate the discourse functions. 6.2 Usages of kuntey at rp As discussed earlier, compared to the left periphery phenomena, the right periphery phenomena has received scant attention until recently (Beeching and Detges, this volume). In Korean, previous studies on the right peripheral expressions have primarily focused on the development of clause-connectives, which are morphologically attached to the predicate stem of the first clause in medial position, into utterance-final particles. For example, Park (1997, 1999) examined the use of the clause-connective nuntey ‘but’ in turn-final position in Korean and demonstrated that the connective is used for marking the speakers’ contrastive stance or dispreferred response to the other speaker’s prior utterance. Being derived from the sequence of the deictic predicate kule- ‘that is so’ and the connective nuntey, the use of kuntey inherits some of the erstwhile meanings. However, the right peripheral kuntey is distinct from the left peri­pheral kuntey in terms of discourse functions and tonal configurations, as discussed earlier. First of all, the use of kuntey at rp appears to be a recent development in Korean. Kim and Sohn (2011, forthcoming) note that they only found 21 instances of kuntey at rp during approximately 40 hours of spoken discourse, in contrast to 107 cases of kuntey at lp in only 200 minutes of conversational data (Park 1997). Second, while the lp kuntey is most frequently used for discourse organization (e.g. topic shift or resumption), Kim and Sohn (2011, forthcoming) show that the rp kuntey most frequently occurs in dispreferred responses. Responses

242

sohn and kim

that disalign with the previous action’s preference are referred to as dispreferred responses in conversation analytic literature. For example, an invitation prefers an acceptance from the recipient, and thus a rejection is a dispreferred response. When the recipient of an invitation rejects the invitation, it is likely to be delayed with hedges, turn-initial gaps, and in-breaths (Pomerantz 1984; Schegloff 2007: 63–78), or accompanied by accounts or elaborations, while a preferred response, acceptance, tends to be unelaborated and immediate. Kim and Sohn (forthcoming) show that the most common context for the rp kuntey in their data is refusing to accept (or disagreeing with) a suggestion. That the rp kuntey is often used in a dispreferred response is illustrated below, drawn from a telephone conversation between two female friends. Speaker A advises B not to take truancy too seriously (lines 1–4). (8) [School life] 01

A: ha : kky[o : : : ‘school’

02

B:

[e     ‘Yeah’

03 A: kkok nemwu cwungyohakey sayngkakha-ci mal-ko surely very seriously think-comm stop-and ‘Don’t take school too seriously’ 04

mwe     swuep- ppattuli-myen cwuk-nun-ta sayngkakha-ci mal-ko?= what class miss-if die-ind-dec think-comm     not-and ‘Missing class is not the end of the world.’

05 B: =kulehky sayngkak an ha-ko iss-ess-e that way think not do-prog-pst-dec : ie ‘((I)) wasn’t thinking like that.’ 06

→ mwuncey-ka simkakh-ay kuntey problem-nom serious-dec:ie kuntey ‘The problem is serious kuntey.’

07 A: um? ‘Really?’

Discourse and prosody in Korean: the case of kuntey ‘but’ 08

243

B: nemwu- nemwu- (.) nemwu very very very ‘Very- very- very-’

09 A: ung ‘yeah’ 10

B: nemwu kulehkey sayngkaha-ki ttaymwuney very like that think-nml because

11 mwuncey-ka simkakha-ta-ko, nay-ka. problem-nom serious-dec-qt i-nom ‘((I)) really took it like that (easy), so the problem became serious.’

Speaker B begins aligning with A’s suggestion, but uses the past progressive form (-ko iss-ess-) that foreshadows a disagreement (line 5). In line 6, B shows her disagreement by stating that ‘the problem is serious’ in the present tense to contrast it with the previous utterance. This dispreferred response is designed with kuntey at rp. Such construction of the turn allows B not to dismiss A’s advice, yet still enables B to express something that may contradict A’s remark, similar to the English though (Pomerantz 1984). The next related example comes from a telephone conversation between two girlfriends, analyzed in Kim and Sohn (forthcoming). A is the oldest in her cohort at the university language program. In lines 4–5, when A reports that she usually spends time by herself apart from her younger peers, B proposes that A have fun (line 8). (9) [KO_6808_07:44] (Kim and Sohn, forthcoming) 01

A: elinay-tul: (0.1) keuy tongki-ka eps-ci? kid-pl almost same:age-nom not exist-Q ‘Many kids: (0.1) There isn’t anyone the same age, you know.’

02 (0.1) 03 B: um : : : [ : uhm

244 04

sohn and kim A:

[kuliko ayey elinay-tul-un icey citul-kkili nol-ko? ahh   and just    kid-pl-top now they-among play-and ahh ‘And the younger kids just hang out with their own peers? Ahh’

05 na-nun honca nol-ko? i-top alone play-and ‘I hang out by myself?’ 06 B: a~ : ~ : : ~ : : : ((elongated upward and downward intonation)) ‘Ah~~~~~~’ 07 A: Jihuy-lang nol-canha? kakkum? ahahh name-with play-you know sometimes ahahh ‘((I)) hang with Jihuy? Sometimes? Ahahaa’ 08 B: kulemyen an toy-ci enni, sinnakey nol-aya-cyeng : : : if so not okay-comm old sister excitingly play-have to-comm ‘((It)) shouldn’t be like that, sister, {you} should have fun.’ 09 (0.6) 10 A:-> kulssey ku no-nun kes-to swiwun key  ani-ya kun(h)tey.hehehh well that play-rl nml-also  easy nml be not-dec:ie kuntey ‘Well having fun isn’t easy either kuntey. Hehehee’ 11

B: .hhhha hakin kukes-to nunglyek-i cham kathi: pwuth(h)-ecweya right that-too ability-nom truly together put-only only if

12 nol swu iss-nun ke(h)ki-n haciman: : play:rl possible be-rl nml-nom-top but

‘.hhhha Right, that’s true, it also requires some ability’

In response to B’s playful suggestion to join the peers and have fun (line 8), A provides a disagreeing response by introducing a physical aspect: it is not so ‘easy to have fun’ because of her age. It is clear that A refuses B’s suggestion, but she constructs the turn with the resources such as a pause, kulssey ‘well’, and laughter, to soften the force of the disagreement. Added to these coordinating resources is kuntey at the end that marks B’s contrastive stance.

Discourse and prosody in Korean: the case of kuntey ‘but’

245

In both examples, the rp kuntey allows the speaker to delay the marking of her contrastive stance from the outset (Kim and Sohn forthcoming). Furthermore, the prosodic feature of the rp kuntey, which is downplayed with final lengthening (see section 5.2), contributes to communicating the speaker’s attempt to mitigate his/her disaffiliative stance. 7

Discussion and Conclusion

The findings demonstrate the effect of prosody in grammar and discourse. Moreover, the analysis of boundary tones shows the importance of examining spoken corpora drawn from natural discourse in studying lp and rp elements. In particular, the study shows how Korean speakers achieve their interactional goals with lp and rp elements, and provides insights into how the prosodic features may be tightly connected with the interactional functions of kuntey at lp and rp. That is, while the lp kuntey is produced with a salient juncture to index a departure from the immediately prior turn, the rp kuntey is prosodically appended to the preceding turn and its downplayed prosody contributes to mitigating the speaker’s contrastive stance. The study also has implications for the grammaticalization of discourse connectives at the two peripheries. The prosodic integration of kuntey at rp indicates an ongoing grammaticalization into a final particle (Kim and Sohn, forthcoming). Having originated from the sequence of the deictic predicate kule- ‘that is so’ and the connective nuntey ‘but; however’, kuntey loses its erstwhile meaning and is further developing into an utterance-final particle. Given that Korean has an agglutinative morphology, the grammaticalization process of kuntey has some implications on other expressions at lp and rp. There is a series of discourse connectives in Korean which originate from the deictic predicate (kule- ‘that is so’). While these discourse connectives typically appear at the beginning of an utterance or a sentence (i.e. lp), some connectives can occur at a final position (i.e. rp) as well as at lp. More research is needed to explore how prosody interacts with these discourse connectives at two peripheries and how prosody affects the grammaticalization process of final particles. Finally, the present study is limited in that the finding is based on a pilot study. It is necessary to examine more data drawn from a spoken corpus to fully investigate the effect of prosody in lp and rp.

246

sohn and kim

Appendix A: Abbreviations Used in the Korean Gloss

acc Accusative comm Committal correl Correlation dec Declarative dm Discourse marker dn Defective noun hon Honorific ie Intimate ending ind Indicative loc Locative mod Modifier modal Modality neg Negation nml Nominalizer nom Nominative pl Plural pol Polite speech level pro Prospective prog Progressive pst Past tense q Question marker qt Quotative particle retro Retrospective marker rl Relativizer top Topic marker

Appendix B: Notations for Intonation Contours

ap Accentual Phrase ip Intonation Phrase +H High tone realized on the second syllable of an accentual phrase H- An optional intermediate phrase-final boundary Ha An accentual phrase-final boundary tone L+ Low tone realized on the penultimate syllable of and accentual phrase % Intonation phrase boundary tone

Discourse and prosody in Korean: the case of kuntey ‘but’



247

Appendix C: Transcription Conventions

[ The point where overlapping talk starts (0.5) Length of silence in tenths of a second (.) Micro-pause; hearably a silence but not readily measurable . Falling, or final intonation; not necessarily the end of a sentence ? Rising intonation; not necessarily a question , Continuing intonation : Sound stretch - Cut-off or self-interruption word Underlining indicates some form of stress or emphasis WOrd Upper case indicates especially loud talk ° ° Portions quieter than the surrounding talk ­¯ Sharper rise or down in pitch hhh Laughter, or hearable exhalation or aspiration (outbreaths); .hhh Hearable inhalation or inbreath (( )) Transcriber’s remarks References An, Cwu-Ho. 2000. ‘Kule’ kyeyel cepsoksa-ui hyengseng kwaceng-kwa mwunpephwa (The formation process and grammaticalization of ‘kule’ connectives). Kukehak (Korean Language Studies) 35: 113–141. Cha, Ji-Hyeon. 2010. Interaction of prosody and discourse function: Korean discourse marker mwe in the sentence-final position. Language Facts and Perspectives 25: 227–256. Choi, Jane. 2007. A Corpus-based Discourse Analysis of Korean Discourse Markers: An Analysis of Spoken and Written Use. (PhD dissertation, University of California, Los Angeles). Ford, Cecilia E., Barbara A Fox and Sandy A. Thompson (eds.). 2002. The Language of Turn and Sequence. Oxford: Oxford University Press. Glenn, Phillip. 2010. Laughter in Interaction. Cambridge: Cambridge University Press. Haselow, Alexander. 2011. Discourse marker and modal particle: The functions of utterance-final then in spoken English. Journal of Pragmatics 43: 3603–3623. Jun, Sun-Ah. 1996. Influence of microprosody on macroprosody: A case of phrase initial strengthening. ucla Working Papers in Phonetics 92: 97–116. ———. 1998. The accentual phrase in the Korean prosodic hierarchy. Phonology 15(2): 189–226.

248

sohn and kim

———. 2000. K-ToBI (Korean ToBI) labelling conventions. Manuscript. University of California, Los Angeles. ———. 2007. “The intermediate phrase in Korean intonation: Evidence from sentence processing.” In C. Gussenhoven and T. Riad (eds.), Tones and Tunes: Studies in Word and Sentence Prosody, Mouton de Gruyter, 143–167. ———. 2011. “Prosodic markings of complex np focus, syntax, and the pre-/post-focus string,” In M. B. Washburn et al (eds.), Proceedings of the 28th West Coast Conference on Formal Linguistics, Somerville, ma: Cascadilla Proceedings Project, 214–230. www.lingref.com, document #2454. Kim, Stephanie Hyeri and Sung-Ock Sohn. 2011. “The turn-final kuntey ‘but’ in Korean conversation” (paper presented at the ipra-12, Manchester, July 2011). ———. forthcoming. Grammar as an emergent response to interactional needs: A study of final kuntey ‘but’ in Korean conversation, Journal of Pragmatics. Kim, Kyu-Hyun and Kyung-Hee Suh. 1996. “Dealing with prior talk: Discourse connectives in Korean conversation,” In N. Akatsuka, S. Iwasaki, and S. Strauss (eds.), Japanese/Korean Linguistics 5. Stanford: csli, 83–99. Kim, Mary Shin. 2005. Evidentiality in achieving entitlement, objectivity, and detachment in Korean conversation. Discourse Studies 7(1): 87–108. Mulder, Jean and Sandy Thompson. 2008. “The grammaticalization of but as a final particle in conversation,” In L. Ritva (ed), Crosslinguistic Studies of Clause Combining: The Multifunctionality of Conjunctions, John Benjamins, Amsterdam, 179–204. Mulder, Jean, Sandy Thompson and Penry C. Williams. 2009. “Final ‘but’ in Australian English conversation,” In P. Peters., P. Collins and A. Smith (eds.), Comparative Studies in Australian and New Zealand English: Grammar and Beyond Varieties of English Around the World, v. G39. Amsterdam: John Benjamins, 339–359. Ochs, Elinor, Emanuel A. Schegloff and Sandy A. Thompson (eds.). 1996. Interaction and Grammar. Cambridge: Cambridge University Press. Park, Mee-jeong. 2003. The Meaning of Korean Prosodic Boundary Tones (PhD diss., University of California, Los Angeles). Park, Yong-Yae. 1997. A Cross-Linguistic Study of the Use of Contrastive Connectives in English, Korean and Japanese Conversation (PhD diss., University of California, Los Angeles). ———. 1999. Korean connective ‘nuntey’ in conversational discourse. Journal of Pragmatics 31: 191–218. Pierrehumbert, Janet B. and Mary E. Beckman. 1988. Japanese Tone Structure. Cambridge, Mass.: mit Press. Pomerantz, Anita. 1984. “Agreeing and disagreeing with assessments: some features of preferred/dispreferred turn shapes,” In J.M. Atkinson and J. Heritage (eds.), Structures of Social Action. Cambridge: Cambridge University Press, 57–101.

Discourse and prosody in Korean: the case of kuntey ‘but’

249

Rhee, Seongha. 2013. “lp and rp in grammaticalization of rhetorical interrogative forms in Korean,” (paper presented at the IPra-13, New Delhi, September 2013). Sacks, Harvey, Emanuel A. Schegloff and Gail Jefferson. 1974. A simplest systematics for the organization of turn-taking for conversation, Language 50(4): 696–735. Schegloff, Emanuel. 1968. Sequencing in conversational openings, American Anthropologist 70: 1075–1095. ———. 2007. Sequence Organization. Cambridge: Cambridge University Press. Selting, Margret and Elizabeth Couper-Kuhlen (eds.). 2001. Studies in Interactional Linguistics. Amsterdam: John Benjamins Publishing Company. Shinzato, Rumiko. 2007. “(Inter)subjectification, Japanese syntax and syntactic scope increase,” In N.O. Onodera and R. Suzuki (eds.), Historical Changes in Japanese: Subjectivity and Intersubjectivity, Special Issue of Journal of Historical Pragmatics 8 (2): 171–206. Sohn, Homin. 1999. The Korean Language. Cambridge: Cambridge University Press. Sohn, Sung-Ock. 2010. “The role of frequency and prosody in the grammaticalization of Korean -canh-,” In A.V. Linden, J.-C. Verstraete and K. Davidse (eds.), Formal Evidence in Grammaticalization, Amsterdam: John Benjamins, 245–273. ———. forthcoming. “Grammaticalization,” In L. Brown and J. Yoon (eds.), The Handbook of Korean Linguistics, Wiley-Blackwell. Suh, Jung-Swu. 1996. Kuke Mwunpep [Korean Grammar]. Seoul, Korea: Hanyang University Press. Traugott, Elizabeth C. 2010. “Dialogic contexts as motivations for syntactic change,” In W. Kretschmar, A.-M. Hamilton-Brehm and R.A. Cloutier (eds.), Variation and Change in English Grammar and Lexicon, Berlin: Mouton de Gruyter, 11–27. ———. 2013. “Remark on the panel of form-function periphery structure: Focus on exchange structure and action structure” (paper presented at the ipra-13, New Delhi, September 2013). Waltereit, Richard and Ulrich Detges. 2007. “Different functions, different histories: Modal particles and discourse markers from a diachronic point of view,” In M.J. Cuenca (ed.), Contrastive Perspectives on Discourse Markers, Special issue of Journal of Catalan Linguistics 6: 61–82.

Author Index Adger, David 124n Aguilar, Marta 173 Ahn, Mikyung 196, 197, 200ff, 202, 205 Aijmer, Karin 3, 6, 73, 78, 124, 152 Aikhenvald, Alexandra Y. 202 Akimoto, Minoji 117, 126n An, Cwu-Ho 226 Andersen, Gisle 5ff Apothéloz, Denis 38 Auer, Peter 84, 124 Bach, Carme 162 Ball, J. Dyer 198ff Barnes, Betsy 26 Barth-Weingarten, Dagmar 152, 158 Bazzanella, Carla 5, 117, 118n, 121 Beckman, Mary E. 225ff Beeching, Kate 5, 6, 14, 17, 34ff, 39n, 44, 48, 50, 72, 73, 75, 82, 87, 89, 221, 241 Benveniste, Emile 38 Berrendonner, Alain 154 Biber, Douglas 72, 73 Bilger, Mireille 154 Birner, Betty 158 Blanche-Benveniste, Claire 154 Blommaert, Jan 5 Bolly, Catherine 153, 159, 160, 161 Bossong, Georg 27, 36 Bouacha, Abdelmadjid 161 Bras, Myriam 153 Breban, Tine 158 Brémond, Capucine 47, 49 Brems, Lieselotte 73 Brinton, Laurel J. 4, 5, 6, 61, 75, 117, 118n, 120n, 129n, 151, 152, 158 Briz Gómez, Antonio 128 Bybee, Joan L. 88, 183ff Cao, Guangshun 182 Carlier, Anne 158, 159 Caron, Jean 5 Caron-Prague, Josiane 5 Carroll, Susanne 26 Cat, Cécile de see De Cat Cechetto, Carlo 2

Cha, Ji-Hyeon 236n Chao, Yuen Ren 182ff, 183ff, 198 Cheong, Zheng-yin Agnes 186ff Cheung, Kam Siu 205, 207 Cheung, Lawrence 194, 205, 207 Choi, Jane 223, 225, 226 Choi, Pik-ling 205, 207 Chor, Winnie 194ff, 199, 210 Ciabarri, Federica 162 Cinque, Guglielmo 2 Clément, Fabrice 119 Clift, Rebecca 158 Company Company, Conceptión 119, 121 Comrie, Bernard 183ff, 187ff Couper-Kuhlen, Elisabeth 152, 158, 225 Cresti, Emanuela 117n Croft, William 1, 74n Crompton, Peter 158 Cuenca, Maria Josep 162 Culpeper, Jonathan 78 Cuyckens, Hubert 8, 159 Danon-Boileau, Laurent 8, 24 Dasher, Richard B. 1, 4, 82, 87, 123, 126n, 135n Davidse, Kristin 8, 159 Davidson, Brad 33 De Cat, Cécile 2, 26, 124n, 158 De Mauro, Tullio 117n De Mulder, Walter 158, 159 Defour, Tine 50, 75 Degand, Liesbeth 5, 6, 9ff, 15, 18, 89, 151, 152, 153, 154, 155, 158, 159, 160, 161, 173 DeLancey, Scott 212 Detges, Ulrich 4, 8ff, 13, 17 25, 31, 33, 34, 35, 44, 72, 73, 75, 82, 87, 89, 118n, 221, 241 Diessel, Holger 158 Diewald, Gabriele 75 Dister, Anne 153 Doetjes, Jenny 26 Dong, Tong-He 57 Dostie, Gaétane 117, 119, 120, 121n, 123n Downing, Angela 73 Dowty, David 12 Ducrot, Oswald 4, 37 Durham, Mercedes 3

252

Author Index

Edmondson, Willis 5 Endo, Tomoko 194 Erman, Britt 5, 6, 75, 123n Evans, Nicholas 210 Evers-Vermeul, Jacqueline 160

Hogan, John T. 158 Horie, Kaoru 205, 212 Hunston, Susan 201 Hybertie, Charlotte 153 Hyland, Ken 5

Fagard, Benjamin 89, 117, 120n, 129n, 153, 154, 158, 159, 160, 173 Fairbanks, Brendan 151 Fedriani, Chiara 120n, 126, 129n, 140n Ferrari, Anna 154 Fetzer, Anita 27, 28, 37, 38 Fillmore, Charles 5 Finegan, Edward 72, 73 Finell, Anne 48, 49, 50, 55 Fischer, Kerstin 5, 73, 151 Foolen, Ad 152 Ford, Cecilia E. 155, 225 Fox, Barbara A 225 Francard, Michel 153 Franckel, Jean-Jacques 153 Frascarelli, Mara 2 Fraser, Bruce 3, 5, 152 Fries, Peter H. 158

Iliescu, Maria 129n

Gerecht, Marie-Jeanne 153 Ghesquière, Lobke 73 Ghezzi, Chiara 6, 15, 18, 120n, 123n, 126, 126n, 129n, 135, 135n, 140n Givón, Talmy 189 Glenn, Phillip 239 Goldberg, Adele E. 74, 74n Greenbaum, Sidney 73 Grunow-Hårsta, Karen 202, 205, 212 Grupo Val.Es.Co. 128 Gülich, Elisabeth 162 Günther, Susanne 84 Haiman, John 189 Hambye, Philippe 153 Hansen, Maj-Britt M. 5, 31, 151, 153, 154, 158 Haselow, Alexander 7, 152, 221 Haspelmath, Martin 48, 67 Heine, Bernd 151, 157 Heinz, Christophe 119 Higashiizumi, Yuko 202, 209 Ho-Dac, Lydia-Mai 158 Hoffmann, Sebastian 74

Jahnke, Nathan 152 Jayez, Jacques 153 Jefferson, Gail 223 Jenner, William J.F. 62,63 Jespersen, Otto 12 Jing-Schmidt, Zhuo 58, 61 Jiu, Ji-yung 197 Johansson, Marjut 27, 28, 37, 38 Jucker, Andreas 48,49,50 Jun, Sun-Ah 224, 225, 226, 227, 229, 231n, 234 Kärkkäinen, Elise 194, 201 Keller, Eric 152 Kerbrat-Orecchioni, Cathérine 5 Kim, Stephanie Hyeri 17, 221, 222, 223, 224, 226, 242, 243, 245 Kim, Kyu-Hyun 223, 228, 236 Kim, Mary Shin 239 Kim, Min-Joo 152 Koch, Peter 3 Kotschi, Thomas 162 Kotsinas, Ulla-Britt 75, 123n Kytö, Merja 78 Lam, Charles Tsz-Kwan 190, 191 Lambrecht, Knud 2, 26, 33, 158 Lamiroy, Béatrice 5 LaPolla, Randy J. 202 Laury, Ritva 189 Law, Sam-po 198–200 Le Draoulec, Anne 153 Lee, Cher Leng 186ff Legge, James 190 Lenk, Uta 152, 158 Lenker, Ursula 74, 75, 82, 88 Leung, Wai-mun 181, 196, 199 Li, Boya 198ff Li, Charles N. 183ff Li, Jinxia 60 Lim, Ni Eng 194

253

Author Index Lin, Tzong-Hong Jonah 194 Liu, Xunning 185 Lu, Louis Wei-lun 183ff, 186, 211 Mancini, Federico 117n Martín Zorraquino, María A. 119, 129n Martin, James R. 201 Mascaro, Olivier 119 Matisoff, James 205 Matthews, Stephen 196, 198, 200 Mauranen, Anna 173 Mayol, Laia 29 Mercier, Hugo 119 Mertens, Piet 24, 36 Miyata, Ichiro 197 Molinelli, Piera 6, 15, 18, 120n, 126n, 135, 135n Moneglia, Massimo 117n Moore, Colette 75, 89 Morel, Mary-Annick 8, 12, 24, 27, 30, 35, 36, 37, 43 Morrison, Robert D. D. 198, 199 Mortier, Liesbeth 160 Mulac, Anthony 194, 125, 135 Mulder, Jean 152, 158, 209, 221 Nikolaeva, Irina 210, 212 Nikula, Tarja 5 Nølke, Henning 4, 27, 35 Noonan, Michael 205, 212 Norrick, Neal R. 117 Ocampo, Francisco 129n Ochs, Elinor 225 Oesterreicher, Wulf 3 Ohori, Toshio 202, 209 Olmen, Daniël van 120n, 129n Ono, Tsuyoshi 187ff Onodera, Noriko O. 6, 10–11, 15, 17, 89 Oppermann Marceaux, Evelyne 141n Origgi, Gloria 119 Östman, Jan-Ola 5 Owen, Marion 47 Pagliuca, William 88, 183ff Pander Maat, Henk 154, 160 Park, Mee-jeong 224, 227 Park, Yong-Yae 223, 228, 236, 241, 242 Parkes, Malcolm B. 78

Perkins, Revere 183ff Pierrehumbert, Janet B. 225, 226 Poggi, Isabella 117, 126 Pomerantz, Anita 47, 49, 242, 243 Pons Bordería, Salvador 12, 120n, 129n Portolés Lázaro, José 119, 129n Prideaux, Gary D. 158 Redeker, Gisela 5 Rhee, Seongha 202, 204ff, 205, 209, 212, 222 Rizzi, Luigi 2, 124n, 151 Roberts, Ian 186, 189 Rossari, Corinne 154 Rouget, Claire 154 Roulet, Eddy 4, 56 Roussou, Anna 186, 189 Sacks, Harvey 142, 223 Sag, Ivan A. 74 Sanders, Ted 160 Sankoff, David 5 Saussure, Ferdinand de 1 Schegloff, Emanuel A. 124, 142, 223, 225, 238, 242 Schiffrin, Deborah 5, 6, 48, 118n, 140, 142,152, 161 Schmale, Günter 157 Schourup, Laurence 5, 151 Schwenter, Scott A. 4, 75 Selting, Margret 155, 225 Shao, Jingmin 65 Shi, Yuzhi 57 Shibasaki, Reijirou 202 Shinzato, Rumiko 10n, 202, 221 Simon, Anne Catherine 153, 155, 160 Simone, Raffaele 36 Simon-Vandenbergen, Anne-M. 6, 38, 41, 73, 152 Simpson, Andrew 205, 212 Sohn, Homin 223n, 224, 240n Sohn, Sung-Ock S. 6, 17, 18, 187ff, 221, 222, 226, 227, 235, 242, 243, 245 Sperber, Dan 118, 119 Stark, Elisabeth 30, 31 Strauss, Susan 152, 158, 187ff Su, Lily I-wen 183ff, 186, 211 Suh, Jung-Swu 225 Suh, Kyung-Hee 223, 228, 236

254 Sun, Chaofen 182ff, 183, 183ff, 185 Suzuki, Ryoko 89, 187ff, 202 Suzuki, Satoko 202 Swan, Toril 73, 74 Sweetser, Eve 129n Sybesma, Rint 198ff Tanguy, Noalig 155 Taylor, John R. 210 Tchizmarova, Ivelina 140, 142 Terkourafi, Marina 75 Thompson, Geoff 201 Thompson, Sandra A. 125, 135, 152, 155, 158, 189, 183ff, 194, 209, 221, 225 Thoms, Peter Perring 197 Tottie, Gunell 74 Traugott, Elisabeth C. 1, 2, 4, 6, 14, 17, 47, 49, 61, 72, 75, 82, 85, 87, 117, 123, 126n, 135n, 153, 158, 186, 209, 209ff Trousdale, Graeme 75 Tsoulas, George 124n Unger, Christopher 5 Van Damme, Thomas 155 Van de Velde, Freek 73 Van den Berg, Marinus 186ff Van Den Eynde, Karel 154 Van der Wouden, Ton 152 Vandelanotte, Lieven 8, 159 Vedovelli, Massimo 117n Verschueren, Jef 5 Verstraete, Jean-Christophe 158

author Index Vincent, Diane 5, 6, 151, 157 Virtanen, Tuija 158 Vlemings, Jeroen 154 Voghera, Miriam 117n Waltereit, Richard 4, 8ff, 12, 13, 17, 31, 117, 118n, 125, 129n, 221 Wang, Hsiao-Ling 48 Wang, Jiao 190, 191, 194ff, 199, 202ff Wang, Li-Ya 48, 62, 64, 65 Wang, Yu-Fang 6, 14, 17, 47 Ward, Gregory 158 Watts, Richard J. 5, 47, 152 Wei, Pei-Chuan 57 White, P.R.R. 75 Williams, Cara P. 197, 198, 209, 221 Wilson, Deirdre 118, 119,153 Wong, Tak-Sum 6, 16, 18 Wrona, Janick 212 Wu, Guo 186ff Xiang, Xuehua 152, 158 Xu, Baohua 197 Yang, Ying 6, 16, 18 Yap, Foong Ha 6, 16, 18, 190, 191, 194ff, 199, 200ff, 202, 202ff, 205, 207, 212 Yeung, Ka-Wai 196 Yip, Virginia 198, 200 Yoshida, Eri 187ff Zénone, Anna 154 Zhang, Ding 60

Subject Index Abtönungspartikel 5, 151n addressee (-oriented) 48, 50, 56, 68, 73, 84f, 85, 130f, 157ff, 160, 161, 162f, 170ff, 180f, 183n, 186, 195, 201, 211 adverb 74, 76 epistemic 72ff manner 78, 88 stance 72, 73, 84 alors 153, 159, 163, 167, 171 anti-topic 35, 43, 44 argumentative relation 153, 158, 160 attention getter 125, 129ff bon 50ff clausal integration 182, 189, 196f, 208, 211 coded meaning 153, 159, 161, 170 cognitive naturalness 97 coherence 3, 5, 6, 13, 17, 27, 29, 30, 31, 32, 33, 34, 35, 39, 42, 43, 44, 75, 82, 118, 124, 146, 151 concessive 15, 49, 54, 55, 56, 83, 98, 99, 100, 202, 204, 209, 210, 212 conjunction 5, 6, 60f, 69, 101, 108ff, 113, 151, 153 connective 4, 5, 6, 16, 38, 47, 69, 73, 75, 76, 82, 99, 100ff, 113, 132, 151, 153, 158, 159, 166, 190, 202ff, 202n, 209, 212, 220ff, 241, 245 construction 13, 26, 31, 33, 35, 36, 37, 43, 44, 56, 68, 74f, 74n, 78, 88, 89, 98, 118, 120, 126, 157, 158, 179, 180, 182f, 187, 189, 190, 191, 194, 195, 200, 202ff, 205ff, 209ff, 211f, 223, 239, 243 constructionalization 75, 79, 81 context 7, 8, 14, 39, 53, 67, 75, 79, 86, 88, 101, 102, 107, 108, 118, 120, 124, 126, 129, 131ff, 135, 137, 139f, 141f, 142f, 145ff, 152, 187, 212, 229, 231, 232 bridging 52, 55, 157 dialogic 82, 83, 84, 86, 88 continuum 75, 84 contrast 29f, 34, 35, 38, 39, 43, 44, 100, 124n, 225ff, 241, 244f corpus 25f, 31n, 48, 51, 57, 75f, 89, 117n, 128ff, 153, 153n, 184n, 224, 225 courtesy marker 125f, 127, 135ff

dakedo 99f, 113 dai 126, 127, 140ff d-connectives 100ff, 113 da strategy 104ff, 113 motivation 100ff replacement 102ff, 107, 113 word formation 101 deictic 125, 127, 129, 135f, 138, 140, 143, 145ff, 187n, 210f, 225, 226 dialogic 4, 11, 13, 15, 47f, 82, 83, 84, 86, 88, 72, 75, 78, 82, 83, 86, 119, 123, 129, 133, 142, 143, 145, 146f dialogual 4, 11, 14, 15, 47f., 118f, 123f, 129, 133, 135, 136f, 140, 143, 145f text 72 discourse 221, 224, 226, 227 context 229, 231, 232 function 230, 231, 236, 241 global level 152, 158, 161, 164ff local level 152, 158,162, 165ff marker 3, 4, 5ff, 16, 17, 93ff, 113, 151, 152, 222, 225, 226, 228, 235 organization 224, 228, 231, 242 particle 236 dislocation left 2, 26 right 2, 3, 12, 26, 27, 180, 182, 194ff, 208, 211 donc 153, 156, 159, 160, 165, 169, 172 epistemic 7, 14, 16 adverb 72ff marker 179, 194ff, 210 evidential marker 181, 196, 197, 200, 208, 212 hearsay 180f, 196ff, 201 exchange verbs 126 final position 152ff grammaticalization 11, 16, 226, 236n, 245 guarda 125, 127, 129ff hao 57ff hedging 195, 196 historical pragmatics 92

256 initial position 151ff insubordination 182, 195, 200ff, 212 interaction 78, 89 interjections 94, 109 interpersonal 27, 37, 42 intersubjectification 119f, 122f intersubjective 3, 7f, 10, 11, 13, 14, 15, 16, 17, 18, 19, 43, 92, 95, 109ff, 189, 192, 209ff, 211, 212 intonation 224, 227, 232, 233 phrase 227, 228, 230, 235 layered structure 111ff lengthening 227, 231, 232, 235, 245 main-clause ellipsis 182, 194, 202, 209, 212 medial position 74, 82 mental space 92, 93, 95ff metadiscursive 153, 159, 161, 162 mitigative 189, 190ff, 210 modal 6, 12, 27, 33, 34, 37, 42, 43, 44 particle 5, 12 modalising function 1, 3, 4, 11, 12, 13, 17, 19 modality 34, 39, 44, 72ff multifunctionality 78 nominalizers 111 objectivity 79 obwohl 98f, 113 orientation 84ff parenthetical 28, 37ff, 43, 180, 194, 195, 211 particle 226, 236, 245 Abtönung 5, 151n modal 5, 12 sentence final 94, 95, 110, 111, 179, 191, 208 perception verbs 125 perfect aspect 187, 192 performative verbs 125f peripheral meaning 153, 170 periphery 1ff, 7, 8ff, 10ff, 24ff, 72, 92, 93, 153ff left 24, 26ff, 29ff, 38ff, 41, 42, 43ff, 95ff, 124, 130ff, 134, 135, 137, 138, 139, 143f, 145ff

Subject Index right 24, 27, 28, 34ff, 38, 40ff, 43f, 124, 132, 134, 135, 139, 143f, 145ff, 179, 187, 194, 196, 202, 205, 209, 211 shift 101, 113 syntactic 124, 130, 137, 138, 140, 145 phatic markers 126, 127, 140f pitch 227, 229, 230, 231 post-rheme 24, 27, 35, 36, 37, 43, 44 Praat 225 pragmatic marker 73, 75, 88, 89, 117f polyfunctionality 118, 133, 134 properties 118, 121 pragmaticalization 123n, 127, 129, 135, 140, 145, 147 preamble 24, 25, 30, 41, 42, 44 prego 125f, 127, 135ff pro-predicate 101 prosodic 221 boundaries 223 feature 222 juncture 222, 230 units 225, 227, 229, 231 prosody 24, 26, 36, 118, 124n, 139f, 157, 192, 199, 200, 207, 212, 233, 236, 241, 245 representation 75, 78, 89 responses 76, 82 dispreferred 241, 242, 243 rheme 24, 25, 34, 35, 36, 37, 41, 42, 43 scope 74, 76, 81, 82, 83, 88 semantic map 67 sentence-ending suffix 223, 224 sentence final particle 94, 95, 110, 111, 179, 191, 208 slot 73, 74, 89 speaker-oriented 158, 162f, 171, 173 stance 7, 10, 16, 18, 19, 27, 44, 186, 192, 195, 209ff, 211, 214, 236, 239, 241, 245 adverb 72, 73, 84 disaffiliative 224, 245 subjectification 119f, 122f subjective 7f, 10, 11, 13, 14, 15, 16, 17, 18, 19, 27, 92, 95, 109ff subjectivity 7f, 11, 13, 15, 17, 82, 84ff, 93, 112, 158f syntactic relabeling 182, 189

257

Subject Index tone 224, 228 tone boundary 227, 229, 231 high 227, 228, 230, 234 low 227, 228, 229, 235, 236 topic 24, 26, 30, 31, 32, 33, 34, 43, 44, 236, 238, 239, 241 resumption 221, 236, 241 shift 221, 230, 231, 236ff, 240, 242 turn 221, 223, 226, 236, 238, 239 constructional unit 223, 225 taking 1, 3f, 8, 9, 10, 11, 13, 15, 16, 17, 31, 32, 34, 43, 44, 72, 75, 82, 86, 89

transition 157, 159, 161, 164, 168, 171 unit 221, 223, 225, 226 yielding 72, 75 utterance tag 181, 192, 194ff, 204, 208ff Val.Es.Co. model 128, 145 verb serialization 182, 186, 188, 202, 208, 211 verb-based markers 117f well 49f