Discourse Particles in Asian Languages Volume II: Southeast Asia [1 ed.] 9781138482449, 9781032532615, 9781351057752

Citation preview

Discourse Particles in Asian Languages Volume II

This volume is the second in a two-part collection of research on discourse particles focusing exclusively on the languages of Asia from the perspective of formal as well as non-formal semantics and pragmatics. Despite increasing interest in discourse particles, most research in the area (particularly within formal semantics and pragmatics) focuses on a restricted set of languages, and there has been little consensus on the proper formal treatment of particles. The term “discourse particles” has been used to cover a broad range of phenomena, including such things as “sentence-final particles,” “discourse adverbs,” and other related phenomena. In recent years, there has been extensive development of the formal approach to discourse particles, which often treats these words as devices for marking information updates. It is vital, however, to extend this data to non-Western languages, like Malay, Thai, or Vietnamese. These two volumes are the first to give an exclusive focus on particles in non-European languages (in this case, Asian languages), from the perspective of formal and non-formal semantics and pragmatics. This second volume includes chapters on Tagalog, Kimaragang Dusun (Malaysia), Malay, Singlish (Colloquial Singapore English), Thai, and Vietnamese. The chapters are informed by recent theoretical work in formal and non-formal semantics and pragmatics relating to the meaning of particles. The collection contributes to our theoretical understanding of the meaning of discourse particles and to empirical knowledge of discourse particles in the languages of Asia. It will be of interest to postgraduate students and scholars of semantics and pragmatics. Hiroki Nomoto is Associate Professor of Malay Language and Linguistics at Tokyo University of Foreign Studies, Japan. Elin McCready is Professor in the Department of English Language and Literature at Aoyama Gakuin University, Japan.

Routledge Studies in Linguistics

34 Metonymies and Metaphors for Death Around the World Wojciech Wachowski and Karen Sullivan 35 Cross-cultural Genre Analysis Investigating Chinese, Italian and English CSR reports Danni Yu 36 Significance in Language A Theory of Semantics Jim Feist 37 Two Dimensions of Meaning Similarity and Contiguity in Metaphor and Metonymy, Language, Culture and Ecology Andrew Goatly 38 Researching Metaphors Towards a Comprehensive Account Edited by Michele Prandi and Micaela Rossi 39 The Referential Mechanism of Proper Names Cross-cultural Investigations into Referential Intuitions Jincai Li 40 Discourse Particles in Asian Languages Volume I East Asia Edited by Elin McCready and Hiroki Nomoto 41 Discourse Particles in Asian Languages Volume II Southeast Asia Edited by Hiroki Nomoto and Elin McCready For more information about this series, please visit: www.routledge.com/ Routledge-Studies-in-Linguistics/book-series/SE0719

Discourse Particles in Asian Languages Volume II Southeast Asia

Edited by Hiroki Nomoto and Elin McCready

First published 2024 by Routledge 4 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 605 Third Avenue, New York, NY 10158 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2024 selection and editorial matter, Hiroki Nomoto and Elin McCready; individual chapters, the contributors The right of Hiroki Nomoto and Elin McCready to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-PublicationData A catalogue record for this book is available from the British Library ISBN: 978-1-138-48244-9 (hbk) ISBN: 978-1-032-53261-5 (pbk) ISBN: 978-1-351-05775-2 (ebk) DOI: 10.4324/9781351057752 Typeset in Galliard by Apex CoVantage, LLC

Contents

List of figures List of tables List of contributors

Introduction 1 Tagalog pala: an unsurprising case of mirativity

vii viii ix

1 9

SCOTT ANDERBOIS

2 Discourse particles in Tagalog: the case of e

37

NAONORI NAGAYA

3 A Kimaragang status particle: accessible information

67

PAUL R. KROEGER

4 A syntactic universal in a contact language: the story of Singlish already

91

MICHAEL YOSHITAKA ERLEWINE

5 On the discourse marker dah in Colloquial Malay (and sudah in Sabah Malay)

121

HOOI LING SOH

6 On the apparently non-additive use of Malay additive pun

147

HIROKI NOMOTO

7 A unified analysis of (some) discourse particles in Thai UPSORN TAWILAPAKUL AND ELIN MCCREADY

173

vi

Contents

8 Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai conversation

191

KIYOKO TAKAHASHI

9 A scalar semantics for the Vietnamese sentence-final particle cơ

214

ANNE NGUYEN

10 Syntax-information structure interface in Vietnamese

237

THUAN TRAN

Index

260

Figures

2.1 Prosody of ha in (14). 2.2 Prosody of e in (23). 6.1 Classification of various uses of pun. 7.1 The structure of super-QUD and sub-QUDs of (4).

45 50 148 177

Tables

2.1 2.2 3.1 8.1 8.2 8.3 8.4 8.5

Conversational corpus for the present study Frequency of SFPs e, a, ha, and o in the conversational corpus Template for 2P clitic ordering Plausible extension of the usage of /kɔ̂ɔ/ Main types of the pragmatic particle /kɔ̂ɔ/ Usage types of the turn-initial /kɔ̂ɔ/ Usage frequencies of each type of the turn-initial /kɔ̂ɔ/ Subtypes of “reaction” speech act pertinent to uses of the turn-initial /kɔ̂ɔ/ 8.6 Syntactic subtypes of the turn-initial /kɔ̂ɔ/ 8.7 Syntactic subtypes of the turn-internal /kɔ̂ɔ/ 8.8 All the attested speech-act types pertaining to uses of /kɔ̂ɔ/

40 42 70 193 195 196 199 204 205 206 208

Contributors

Scott AnderBois is an associate professor of cognitive, linguistic, and psychological sciences at Brown University. His research uses primary fieldwork to explore questions in semantics and pragmatics, including work on Yucatec Maya, Tagalog, English, and A’ingae. He also is co-director of a community-engaged language documentation project working with speakers of A’ingae. Michael Yoshitaka Erlewine is Presidential Young Professor in linguistics at the National University of Singapore. His work investigates the syntax and semantics of languages of East and Southeast Asia through original fieldwork. Paul R. Kroeger is a professor of applied linguistics at the Dallas International University in Dallas, Texas, and a senior linguistic consultant for SIL International. His research has focused primarily on the syntax and semantics of Western Malayo-Polynesian languages, with a particular interest in languages of Borneo. Elin McCready is a professor in the Department of English Language and Literature at Aoyama Gakuin University. Her research is on semantics and pragmatics, with focus on expressive content, social meaning, and evidentiality. She is the author of Reliability in Pragmatics (Oxford University Press, 2015) and The Semantics and Pragmatics of Honorification: Register and Social Meaning (Oxford University Press, 2019), as well as several edited volumes in linguistics and philosophy. Naonori Nagaya received his PhD in linguistics from Rice University and is currently an associate professor at the University of Tokyo in Tokyo, Japan. He specializes in the Austronesian languages of the Philippines and Indonesia. His research interests are in the areas of linguistic typology, cognitivefunctional linguistics, and interactional linguistics. Anne Nguyen is a postdoctoral fellow in linguistics at National University of Singapore. Her research interests are formal semantics, pragmatics, syntax,

x

Contributors and their interface. She is working on discourse and sentence-final particles and non-canonical questions.

Hiroki Nomoto is an associate professor of Malay language and linguistics at Tokyo University of Foreign Studies. His linguistic research has focused on syntax and semantics of Malay and related Austronesian languages. Besides linguistics, he has also been active in developing language resources, such as corpora, and lexical resources of Malay/Indonesian and other languages in East and Southeast Asia. Hooi Ling Soh is a professor of linguistics at the University of Minnesota. Her research interests include topics in syntax and semantics and how they interact, with focus on languages spoken in Southeast Asia. Her current research focuses on the syntax and semantics of discourse particles. She has worked on discourse particles in Mandarin Chinese, English, Malay, and Hmong. Kiyoko Takahashi is a professor at Kanda University of International Studies, Japan. She specializes in Thai linguistics. Her research interests are in functional analyses of Thai grammar. Upsorn Tawilapakul is a lecturer at Thammasat University’s Language Institute in Bangkok, Thailand. Her areas of interest include information structure, the operations of discourse particles, and temporality in the Thai language based on the formal semantic and formal pragmatic approaches. Thuan Tran is an independent researcher. His research interests are semantics and pragmatics, with a special focus on Asian languages.

Introduction Hiroki Nomoto and Elin McCready

In linguistics, especially in the subfield of discourse analysis, there has been substantial interest in “discourse markers” since the introduction of the concept by such scholars as Schifrin (1987) and Schourup (1985). For them, “discourse markers” were restricted to interjections and conjunctives. The term “discourse particles” has been used to cover a much broader range of phenomena, including such things as “sentence-final particles,” “discourse adverbs,” and other particles in addition to “discourse markers” as defined earlier. The meanings conveyed by discourse particles are extremely heterogeneous, leading to them also being called “pragmatic particles” or “modal particles.” They can indicate the information status of a sentence or part of a sentence, provide question meanings or mark politeness, or even indicate exclamation or surprise. Consequently, the best way to characterize their meanings remains controversial both from formal and informal perspectives. Within the formal semantics and pragmatics literature, various approaches have been proposed to the meanings of particles, but no consensus is currently available. Existing resources for the semantics and pragmatics of discourse particles are collections of papers on the topic covering a somewhat-arbitrary selection of approaches and phenomena (Abraham 1991; Fischer 2006; Bayer & Struckmeier 2017). These books typically consider the meanings of particles in European languages, particularly German and English. The present volumes are the first to give an exclusive focus on discourse particles in non-European languages, more specifically, East and Southeast Asian languages, which are well-known for their richness of discourse particles. Volume I discusses four East Asian languages: Japanese, Mandarin Chinese, Cantonese, and Korean. Volume II discusses five Southeast Asian languages: Tagalog, Kimaragang Dusun, Malay, Thai, and Vietnamese. The chapters in these volumes ofer descriptions and specific analyses of discourse particles in these languages. Considering data from these languages will deepen our empirical and theoretical understanding of discourse particles. First, it allows a characterization of the ingredients required for the general analysis of discourse particles. Second, one can deploy the resulting tools in the analysis of discourse particles from a range of diverse languages. DOI: 10.4324/9781351057752-1

2

Hiroki Nomoto and Elin McCready

The present volumes consist of 16 chapters discussing nine languages in East and Southeast Asia. The chapters are ordered according to the areas in which the languages discussed are used. Some chapters are theoretical and are informed by recent theoretical work in formal semantics and pragmatics relating to the meaning of discourse particles. Others are more descriptive and present empirical facts that are quite unique in the literature and should be of great interest to researchers on both discourse particles and Asian languages in general. The chapters in Volume I consider several issues relating to particles in the languages of East Asia, specifically, Japanese, Korean, and Chinese. The first two chapters discuss Japanese. In the first chapter in Volume I, David Y. Oshima aims to provide an adequate semantic description of one of the major uses of the no (da) construction. The relevant use has been labeled as the “discovery,” “grasping,” or “comprehension” use by Japanese grammarians and is exemplified by a sentence such as A, koko ni ita n da “Oh, there you are.” In light of the recent discussion of the notion of mirativity (“linguistic marking of an utterance as conveying information which is new or unexpected to the speaker,” DeLancey 2001) in the typological literature, Oshima argues that no (da) in its “discovery” use can sensibly be regarded as a mirative marker. He further discusses some subtle aspects of its meaning that do not automatically follow from its being a mirative marker. The second chapter in Volume I, by Lukas Rieser, discusses the evidential particle no in Japanese, which has received considerable attention in the formal and descriptive literature. There is no consensus on how to distinguish it from and connect it to the homophonous complementizer typically occurring in the no da construction discussed in Oshima’s chapter. Based on observations about various uses, Rieser argues that the particle no can be sharply distinguished from the complementizer no in the no da construction. He analyzes it as an utterance modifier introducing an evidence condition parallel to that on assertion but requiring mutually accessible evidence. On the background of comparison with no da construction counterparts in other languages, he proposes that evidential no has developed from the complementizer in bridging contexts of evidence-based inference. The next two chapters are concerned with Mandarin Chinese. In the third chapter in the volume, Hooi Ling Soh presents new observations regarding discourse restrictions and interpretative efects of Mandarin Chinese sentencefinal de in a bare de sentence and proposes an analysis of de as a discourse marker, marking “private evidence.” She then considers distributional restrictions of de in yes/no questions and shows that they follow from the analysis, coupled with a specific proposal about the syntax of de, and certain standard assumptions about the syntax of yes/no questions and modal auxiliaries. Soh’s analysis has implications for the syntax of modal auxiliaries, the relation between bare de sentences and shi . . . de sentences, and the syntax of discourse particles. It connects de with particles that mark the speaker’s belief about whether the (evidence for the) asserted proposition is shared knowledge between the speaker and the hearer and whether it is “verifiable on the spot.”

Introduction

3

Satomi Ito, in Volume I’s fourth chapter, discusses the particle ne in Mandarin Chinese in light of partition semantics, according to which the meaning of an interrogative is represented as a partition on the set of possible worlds (Hamblin 1973; Karttunen 1977). Ito shows that the particle ne marks the contrast between the cells of the partition induced by an interrogative and that the contrastive use of the particle can be applied to declaratives and rhetorical questions. She further argues that the particle ne following nominals marks topic. The next two chapters discuss another Chinese language, namely, Cantonese. The fifth chapter, by Yurie Hara, analyzes four kinds of polar questions, that is, ho2, me1, aa4, and A-NOT-A questions, in the framework of radical inquisitive semantics (Groenendijk & Roelofsen 2010; Aher 2012; Sano 2015). Ho2, me1, and A-NOT-A questions exhibit a multi-dimensional semantics. In addition to their primary speech act of questioning, ho2 and me1 interrogatives encode secondary assertive acts of positive and negative expectations, respectively, while A-NOT-A interrogatives conventionally encode lack of expectation, hence the neutrality requirement. In contrast, aa4 interrogatives are semantically simplex question acts and so can be used in both biased and neutral contexts. The analysis is further supported by one forced-choice experiment and one naturalness-rating experiment. Grégoire Winterstein, Regine Lai, and Zoe Pei-sui Luk’s chapter, the sixth in Volume I, examines the linguistic correlates of the notions of “assertiveness” and “softness” in Cantonese. Specifically, it focuses on a set of sentence-final particles often described as conveying these two values: the particles aa3, ge3, and gaa3. They argue that these elements do not directly encode assertiveness and softness, but rather, that these come as side efects of the conversational and dialogical nature of these particles. On one hand, they argue that aa3 encodes an explicit call on the addressee to mirror the content of the conversational move of the speaker, while on the other hand, ge3 indicates that the utterance of the speaker addresses (and is presented as solving) a salient decision problem in a way comparable to the Japanese particle yo. The particle gaa3 combines the efects of ge3 and aa3. It is then the competition between ge3 and gaa3 that explains the assertiveness attached to the former. The last chapter in Volume I is a study on the Korean particle yo by Soo-Hwan Lee. The repeated use of expressives in general gives rise to a strengthening efect referred to as repeatability (Potts 2007). The notion of repeatability, however, requires further scrutiny. Unlike other formality-denoting expressives, medial yos in Korean show an uncommon pattern of behavior: the more yos you have, the less sense of formality you get. Adopting many of the concepts introduced in Potts (2005, 2007) and McCready (2019), Lee ofers a formal analysis of how a repeated use of an expressive can weaken the expressive content. Further, medial yos can only be realized in the presence of another formality marker, such as (su)pni. His findings suggest that there are predictable restrictions on expressives which have largely been understudied in the literature. Specifically, Lee argues that medial yos are underspecified in terms of their expressive content and that they are dependent on a fully specified expressive elsewhere in the derivation.

4

Hiroki Nomoto and Elin McCready

This volume, the second in the pair, discusses particles in several languages primarily spoken in Southeast Asia. The first two chapters in this volume are concerned with Tagalog, an Austronesian language spoken in the Philippines. Scott AnderBois investigates the second-position particle pala. Similar to many descriptions of miratives cross-linguistically, Schachter and Otanes’s (1972) classic descriptive grammar of Tagalog describes it as “expressing mild surprise at new information, or an unexpected event or situation.” Drawing on recent work on mirativity in other languages, however, AnderBois shows that this characterization needs to be refined in two ways. First, while pala can be used in cases of surprise, pala itself merely encodes the speaker’s sudden revelation with the counterexpectational nature of surprise arising pragmatically or from other aspects of the sentence, such as other particles and focus. Second, he presents data from imperatives and interrogatives, arguing that this revelation need not concern “information” per se but rather the illocutionary update the sentence encodes. Finally, he explores the interactions between pala and other elements which express mirativity in some way and/or interact with the mirativity pala expresses. Naonori Nagaya’s chapter, the second chapter in this volume, provides an overview of discourse particles in Tagalog by observing naturalistic data produced by Tagalog speakers. In addition, Nagaya ofers a case study of the sentence-final particle e, which is the most frequent sentence-final particle in his corpus. This particle has three diferent uses: as a marker of reason, a stance-justifying marker, and a marker of negative evaluation. It is suggested that a polysemy approach rather than a list or general-meaning approach can best account for the behavior of such a multifunctional particle. Paul R. Kroeger, in this volume’s third chapter, describes the particle gima in Kimaragang Dusun, an Austronesian language spoken in the northern part of Sabah, Malaysia. Its meaning and functions seem to be similar (but not identical) to those of German unstressed ja. The core descriptive meaning associated with gima seems to be that the propositional content of the utterance is uncontroversial and accessible to both speaker and addressee. In addition, this particle may convey expressive content, particularly in exclamatory utterances, conveying the speaker’s surprise, annoyance, disapproval, etc. In the fourth chapter, Michael Yoshitaka Erlewine investigates the syntax and semantics of Singlish (Colloquial Singapore English) sentence-final already and its development in the contact ecology of Singapore. Bao (2005) identifies the Mandarin Chinese perfective sufx -le and sentence-final particle le as two substrate sources for Singlish already. Erlewine argues instead that sentence-final le – together with its cognate liao/laa in other Chinese languages of Singapore – is the sole substrate source for the syntax and semantics of Singlish already. The various interpretational efects of already can be derived from a uniform semantics shared with its substrate cognate particles le/liao/laa, together with consideration of the telicity of the predicate. He furthermore shows that the syntax of already difers subtly from that of its

Introduction

5

substrate cognates. Singlish sentence-final already unambiguously scopes over the entire clause, whereas its substrate cognate particles le/liau/laa are in a clause-medial position. It is proposed that this diference can be explained through a process of reanalysis forced by the interaction of a syntactic universal, the Final-over-Final Constraint, together with independent diferences between Singlish and its substrate Chinese languages. The next chapter, by Hooi Ling Soh, is also about “already,” but in Malay: dah in Colloquial Malay spoken in Peninsular Malaysia and sudah in Standard Formal Malay and Sabah Malay. Presenting new empirical facts, Soh argues that two uses of dah must be distinguished: one as an aspectual marker and one as a discourse marker, whose use indicates that the speaker holds a certain belief about the common ground. The two uses difer in their syntactic positioning. Soh also compares the use of dah with sudah in Sabah Malay. While sudah has been noted to exhibit a preference to appear post-verbally in Sabah Malay (Hoogervorst 2011), she shows that post-verbal sudah patterns like sentencefinal sudah and unlike pre-verbal sudah. The results have implications on the cross-linguistic properties distinguishing “already” from perfect aspect (Vander Klok & Matthewson 2015). Furthermore, Soh connects Colloquial Malay dah and Sabah Malay sudah with Mandarin Chinese particle le, which is known to have an aspectual or a discourse function, depending on its syntactic positioning, whether post-verbally or sentence-finally (Li & Thompson 1981). As the aspectual and discourse le have been analyzed as sharing a core meaning, with diferent semantic efects arising from their distinct syntactic positions (e.g., Huang & Davis 1989; Soh 2008), Soh’s chapter adds two varieties of Malay to the group of languages that are fruitful to consider when determining how the syntactic positioning of a grammatical item may afect its meaning. Hiroki Nomoto’s chapter is also on Malay. Nomoto discusses the particle pun in Standard Malay, which is polyfunctional. Besides the simple additive (≈ too, also), scalar additive (≈ even), and final particle uses, it has another use that has been described as expressing a temporal meaning in the literature. He ofers formal analyses of these uses, with a primary focus on the last kind, namely, the so-called “event-sequence” pun. He shows that this use is in fact more general and is not restricted to temporal relations between events. He argues that it is essentially a type of additive: “propositional additive.” Propositional additive difers from the ordinary “individual additive” in that the alternative propositions share a discourse-structural property (= properties of propositions) rather than properties concerning individuals. Specifically, they answer the same immediate Question Under Discussion (QUD). The temporal meaning is not encoded by the pun itself but arises pragmatically when the QUD is “What happened to X?” Upsorn Tawilapakul and Elin McCready also utilize the QUD model and analyze discourse particles in Thai, in the seventh chapter in this volume. Taking the status of information in the common ground and information management by the speaker into account, they propose an approach which incorporates the mechanism of QUD. They connect QUD-based approaches with other

6

Hiroki Nomoto and Elin McCready

non-ofcially QUD-based particle analyses and exemplify its application with new analyses of the particles in Central Thai. The relation between particles and intonation is briefly discussed through the analysis of the particle ná. Kiyoko Takahashi presents a data-driven analysis of spoken uses of the Thai pragmatic particle /kɔ̂ɔ/. Drawing on an examination of instances in dialogic discourse, Takahashi’s chapter identifies several speech-act types related to uses of /kɔ̂ɔ/ and reveals that /kɔ̂ɔ/ interpersonally functions as a reaction marker when the speaker uses it turn-initially and takes into consideration the interlocutor’s previous utterance as a communicatively given presupposition. A fine classification of the instances enables us to see the plausibility of /kɔ̂ɔ/ having broadened past its original, text-procedural function to a logical and modal one, and further to an interpersonal one. The last two chapters deal with Vietnamese. Anne Nguyen documents the usage and distribution of the Vietnamese sentence-final particle cơ in the ninth chapter in this volume. The particle is found in diferent sentence types and seemingly used for various purposes. Nguyen argues that those uses attributed to the particle can be unified and proposes that cơ is a scalar particle, reflecting a conflict between speaker belief and expectations that arise from the preceding conversation. The proposition marked by cơ is required to be stronger than its contrasting alternatives on a given scale. Thuan Tran’s chapter is concerned with the interface between information structure and syntax in Vietnamese; this is the final chapter in this volume. Tran argues that left dislocation in Vietnamese is triggered by contrast, an information structural category as autonomous as topic and focus. However, contrast difers from topic and focus in a fundamental way. While the former is semantic in nature and is syntactically operative, the latter are pragmatic and syntactically inert. We believe that with these chapters collected in one place, the present volumes will contribute to our theoretical understanding of the meaning of discourse particles and to empirical knowledge of discourse particles in the languages of Asia and beyond. We would like to express our gratitude to Christopher Davis, Henrison Hsieh, Wataru Okubo, Jozina Vander Klok, and Mengxi Yuan, who read some of the chapters and ofered insightful comments to improve them. We would also like to thank Yukinori Takubo, who was originally part of the editorial team. It was Yuki who first proposed putting together the research on discourse particles on Asian languages into a volume. Without his initial initiative and encouragements, this volume would not have been possible. Lastly, we would like to note that many of the contributors of this volume were members of the joint project “Semantics of Discourse Particles in East and Southeast Asian Languages” at the Research Institute for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies. This project played an important role of gathering researchers of various Asian languages to discuss the same topic and learn from each other, which finally resulted in this volume.

Introduction

7

References Abraham, Werner (ed.). 1991. Discourse Particles: Descriptive and Theoretical Investigations on the Logical, Syntactic and Pragmatic Properties of Discourse Particles in German. Amsterdam: John Benjamins. http://doi.org/10.1075/pbns.12. Aher, Martin. 2012. Free choice in deontic inquisitive semantics. In Maria Aloni, Vadim Kimmelman, Floris Roelofsen, Galit W. Sassoon, Katrin Schulz & Matthijs Westera (eds.), Logic, Language and Meaning, 18th Amsterdam Colloquium, Amsterdam, 22–31. Berlin, Heidelberg: Springer. http://doi.org/10.1007/978-3-64231482-7_3. Lecture Notes in Computer Science. Bao, Zhiming. 2005. The aspectual system of Singapore English and the systemic substratist explanation. Journal of Linguistics 41. 237–267. http://doi.org/10.1017/ S0022226705003269. Bayer, Josef & Volker Struckmeier (eds.). 2017. Discourse Particles: Formal Approaches to Their Syntax and Semantics. Berlin: Walter de Gruyter. http://doi. org/10.1515/9783110497151. DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33. 369– 382. http://doi.org/10.1016/S0378-2166(01)80001-1. Fischer, Kerstin (ed.). 2006. Approaches to Discourse Particles. Leiden: Brill. http:// doi.org/10.1163/9780080461588. Groenendijk, Jeroen & Floris Roelofsen. 2010. Radical Inquisitive Semantics. https:// projects.illc.uva.nl/inquisitivesemantics/assets/files/papers/GroenendijkRoelofsen2010_CLC.pdf. Hamblin, Charles L. 1973. Questions in Montague English. Foundations of Language 10(1). 41–53. www.jstor.org/stable/25000703. Hoogervorst, Tom G. 2011. Some introductory notes on the development and characteristics of Sabah Malay. Wacana 13. 50–77. http://doi.org/10.17510/wjhi. v13i1.9. Huang, Lillian Meei Jin & Philip W. Davis. 1989. An aspectual system in Mandarin Chinese. Journal of Chinese Linguistics 17. 128–166. www.jstor.org/stable/23757128. Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1(1). 3–44. http://doi.org/10.1007/BF00351935. Li, Charles N. & Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley, CA: University of California Press. McCready, Elin. 2019. The Semantics and Pragmatics of Honorification: Register and Social Meaning. Oxford: Oxford University Press. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford: Oxford University Press. Potts, Christopher. 2007. The expressive dimension. Theoretical Linguistics 33(2). 165–197. http://doi.org/10.1515/TL.2007.011. Sano, Katsuhiko. 2015. Avoiding impossibility theorems in radical inquisitive semantics. In Shier Ju, Hu Liu & Hiroakira Ono (eds.), Modality, Semantics and Interpretations. Berlin, Heidelberg: Springer. http://doi.org/10.1007/978-3-662-47197-5_6. Schachter, Paul & Fe T. Otanes. 1972. Tagalog Reference Grammar. Berkeley, CA: University of California Press. Schifrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schourup, Lawrence C. 1985. Common Discourse Particles in English Conversation. London: Routledge.

8

Hiroki Nomoto and Elin McCready

Soh, Hooi Ling. 2008. The syntax and semantics of change/transition: Evidence from Mandarin Chinese. In Susan Rothstein (ed.), Theoretical and Cross-Linguistic Approaches to the Semantics of Aspect, 387–419. Philadelphia: John Benjamins. Vander Klok, Jozina & Lisa Matthewson. 2015. Distinguishing already from perfect aspect: A case study of Javanese wis. Oceanic Linguistics 54(1). 172–205. http:// doi.org/10.1353/ol.2015.0007.

1

Tagalog pala An unsurprising case of mirativity Scott AnderBois

1

Introduction1

Like many languages of the Philippines, Tagalog has a prominent set of discourse particles which express a variety of diferent evidential, attitudinal, illocutionary, and discourse-related meanings. Morphosyntactically, these particles have long been known to be second-position clitics, with a number of authors having explored fine-grained details of their distribution, relative order, and the interaction of this with diferent types of sentences (e.g., Schachter & Otanes 1972; Billings & Konopasky 2003; Anderson 2005; Billings 2005; Kaufman 2010). With a few recent exceptions, however, comparatively little has been said about the semantics/pragmatics of these diferent elements beyond Schachter and Otanes’s (1972) pioneering work (which is quite detailed, given the broad scope of their work). In this chapter, we explore in detail the contribution of one second-position particle, pala, as illustrated in (1). Schachter and Otanes (1972) – henceforth S&O – describe pala as being “used in expressing mild surprise at new information, or an unexpected event or situation, .  .  . or in expressing an afterthought.” While the category of mirativity as such did not exist at that time and pala has not been discussed in the literature on miratives, even this brief description makes clear that pala is a mirative morpheme. (1) Ikaw pala ang kapatid ni Pedro. you mir dir sibling indir Pedro “So you’re Pedro’s sister.” (Schachter & Otanes 1972: 462) Since first gaining broad currency as a category in DeLancey (1997), literature on mirativity has considered a number of diferent conceptions of mirativity. Aikhenvald (2012) summarizes by listing the following conceptions of mirativity: (i) “new information”; (ii) “sudden discovery, revelation or realisation”; (iii) “surprise”; (iv) “counterexpectation”; and (v) “unprepared mind.” While often regarded as theories or analyses of the concept of mirativity, recent works, such as Salanova and Carol (2016) and AnderBois (2016a, 2018), instead take the diference between these to be an empirical matter, DOI: 10.4324/9781351057752-2

10

Scott AnderBois

with individual mirative morphemes expressing one or another of these related, but distinct, meanings. Since S&O’s earlier description makes reference to several of these notions, the question we ask here is what the relationship is between pala and these various notions. Here, we break this question down into two subparts. Q1: Which of these various meanings does pala conventionally encode? Q2: How do the more specific uses of pala (e.g., as identified by S&O) arise from the co-occurrence of pala with other discourse particles, focus, and context? Looking ahead, we make the following claims. A1: Similar to AnderBois’s (2016a, 2018) account of Yucatec Maya bakáan, I argue that pala semantically encodes a “sudden revelation” about the illocutionary update being performed. New or surprising information is a likely source of such a revelation, but these more specific meanings are not a necessary feature for the felicitous use of pala. A2: In addition to pala, Tagalog has a number of other elements which play key roles in the expression of the various kinds of mirativity in combination with pala. Some of these, such as the sentence-final particle a in (2), are also miratives, conventionally encoding a mirative meaning of some sort. Others, such as the use of contrastive focus in (3), do not conventionally encode a mirative meaning per se but can be used (together with pala) to express one under certain conditions. We term the latter mirativity strategies by analogy to Aikhenvald’s (2004) well-known distinction between evidentials proper and evidentiality strategies.2 Finally, we find elements like the particle nga in (4) which do not themselves express mirativity in any of these senses but which express other meanings which are only compatible with certain kinds of mirative uses, thereby indirectly constraining the kind of mirativity expressed. (2) Ma-ganda pala ito, a! A dj-beauty mir dem part “Oh, but this is pretty!”

(Schachter & Otanes 1972: 462)

(3) Context: Dogs are supposed to eat dog food, but you suddenly realized that your dog Rufe is eating cake. [Yung keyk]Foc pala yung kinain ni Rufe. dem.lnk cake mir dem.lnk eat.AV.ipfv indir Rufe “Oh, Rufe is eating CAKE!” (4) Context: I knew that it was raining, but it slipped my mind. I suddenly remember and say: Umuulan nga pala. rain.ipfv part mir “Oh yeah it’s raining (of course).” The road map for the rest of the paper is as follows: §2 provides background on Tagalog; §3 discusses previous empirical and theoretical properties of

Tagalog pala 11 miratives and mirativity strategies across languages; §4 examines declaratives with pala with an eye toward distinguishing distinct mirative notions; §5 extends the investigation to non-declarative types; §6 examines the interactions between pala and other aspects of Tagalog grammar; and §7 concludes. 2

Background on Tagalog

Tagalog is an Austronesian language of the Philippines, spoken as a first language by tens of millions in the Philippines and diasporic communities around the world. It is also spoken by at least as many people as a lingua franca in its standardized form, known as Filipino. The data here constitute a mix of naturalistic data and elicited data from four college-aged speakers of the Manila dialect that hews close to this standard. We are not aware of dialect variation in the core data presented in this chapter. Tagalog is a predicate-initial language with various types of non-verbal predicates possible with no copula. For verbal predicates, this most typically results in a VOS word order in basic clauses. However, as is common for languages of the Philippines, verbal predicates have a rich system of “voice” morphology which may produce diferent word orders, such as VSO. Beyond word order, the voice system determines the morphological case which occurs with each of the various arguments. The proper analysis of this system has been a central focus in Philippine linguistics (see Chen 2017 for a recent survey). Since the details of how best to analyze this system are not relevant here, we will label these three cases, exemplified in (5), as follows: ang Dir(ect), ng Indir(ect), and sa Obl(ique). (5)

Sumusulat ng liham sa titser ang estudyante write.av.ipfv indir letter obl teacher dir student “The student is writing a letter to the teacher.”

S&O, p. 73

While Tagalog patterns typologically as a predicate-initial language, it also has robust pre-verbal topic and focus positions characterized in detail by Kaufman (2005) (and explored more recently by Latrouite and Riester 2018). First, there are topics such as sa Bulakan in (6a) which appear preverbally, are followed by ay (often omitted in spoken forms), and do not act as host for pronominal clitics such as kami. Second, there are foci such as sa Bulakan in (6b) which also appear pre-verbally (but following topics), have no specific morphological marking, and do act as hosts for pronominal clitics like kami. (6)

a. Context: Responding to the question “What did you do in Bulacan?” Sa Bulakan ay nagpiknik kami. obl Bulacan top picnic.av.ipfv dir.1pl.excl  “In Bulacan, we picnicked.” Kaufman (2005)

12

Scott AnderBois b. Context: Responding to the question “Where did you picnic?” Sa Bulakan kami nagpiknik. obl Bulacan dir.1pl.excl picnic.av.ipfv  “It was in Bulacan that we picnicked.” Kaufman (2005)

In addition to the aforementioned pronominal clitics, Tagalog also possesses a set of 18 second-position “enclitic particles,” including our focus here, the mirative pala. 3

Mirativity cross-linguistically

From the inception of “mirativity” as a term in DeLancey (1997), the nature of mirativity and even its very existence have been hotly debated. While certain aspects of mirativity have been largely agreed upon, the essential question of what kind of mental attitude miratives express has received diferent answers from diferent authors, often in service of the analysis of diferent languages. Often unclear in these discussions is whether these disputes are primarily theoretical – What is the correct conception of mirativity? – or empirical – Which possible mirative notion is appropriate for the analysis of a given mirative in a given language? Following Salanova and Carol (2016) and AnderBois (2016a, 2018), we take this debate to be an empirical one. This, of course, is not to say that works that engage these issues on more theoretical or conceptual terms are not useful. On the contrary, if formulated with sufcient precision, they give us a sense of what data are likely to prove useful in the empirical task of establishing the properties of diferent mirative morphemes. AnderBois (2018) breaks down the semantics of miratives into two questions which, in principle, are orthogonal to one another. The question which most literature has grappled with is what sort of mental attitude a speaker who uses a mirative is expressing. Equally pressing, however, is a question much less discussed in most literature: What is the object of this attitude? Most literature implicitly assumes that this is the proposition that the rest of the sentence denotes. Considering the parallel with evidentials, however, there are other plausible answers, such as an illocutionary or speech-act update, an event, etc. We return to these later in §5 when we consider non-declarative sentences. As far as the mirative attitude, Aikhenvald (2012) identifies five distinct conceptions of mirativity from previous literature. (7) Five conceptions of the mirative attitude 1. 2. 3. 4. 5.

New information Sudden discovery, revelation, or realization Surprise Counterexpectation Unprepared mind

Tagalog pala 13 As has been discussed by previous authors, these various notions are, of course, closely related to one another. For example, it would seem that one cannot experience counterexpectation or be surprised by information which is not new. Similarly, psychologists (see, for example, Huron (2006) for a survey) regard surprise as a basic emotion triggered from experiencing particular kinds of counterexpectation (as opposed to more general notions like surprisal), and Peterson (2016) argues specifically that sudden revelation is a necessary component of surprise. I set aside the fifth notion, “unprepared mind,” in what follows. This term is used most extensively by DeLancey (1997), but it remains unclear what distinguishes it from the other notions. Despite the close connections between these diferent conceptions of mirativity, the relationships between them would seem to be asymmetrical ones. One may be surprised by new information, but new information itself need not be surprising. While new information is one potential cause of a sudden revelation or realization, one may also suddenly remember something or suddenly piece together a realization based on old information they already had. Something may go against one’s expectations (at least to a slight extent) without one experiencing the emotional sensation of surprise. In an information theoretic sense, any such expectation violation could be said to result in surprisal, but we would be hard-pressed to say that this constitutes surprise either intuitively or in the sense discussed in psychological literature. In addition to the notional connections between these diferent conceptions of mirativity, there are additionally some shared background assumptions about what mirativity is like. Though not always explicitly addressed, mirativity, for all authors, it seems, is anchored in the deictic center of the conversation. Whichever mirative attitude it encodes, it expresses that the speaker (or occasionally the addressee, for some authors) holds it at the moment of the utterance, or near to it. An expression of the speaker’s having held such an attitude a week ago would not be considered a case of mirativity, nor would an expression of a third party’s holding such an attitude at the time of the utterance. To use Faller’s (2002) term, miratives are mperformative, directly reflecting the speaker’s subjective mental estimation at the moment of utterance. One caveat to the ideas here, discussed some in AnderBois (2018), is that – like any speaker-oriented expression – there may be cases of direct quotation, free-indirect discourse, and other related forms of perspective shift, especially in narratives, which facilitate uses which appear to violate the preceding assumptions. For example, in (8), we see that the narrator makes the story more vivid for the hearer by using pala to refer to the sudden realization they had at that point in the narrative rather than at the time of the story’s telling. Such uses, however, are clearly distinct from non-mirative lexical expressions, such as the English verb “surprise,” which encode a similar attitude but allow the speaker to freely attribute this attitude to any arbitrary agent with no such tie to the deictic center of the narrative.

14

Scott AnderBois

(8) Context: Speaker is telling a story about the other day about trying to get in touch with a mutual friend. Kahapon, nasira pala ang kanya-ng telepono, kaya yesterday break.pfv mir dir obl.3sg-lnk phone thus di ko siya tinawagan. neg indir.1sg dir.3sg call.pfv “So, yesterday, (I was like) oh, his phone’s broke, so I stopped calling him.”

One potential assumption from some previous literature which we do not adopt here is the presumption that mirativity and evidentiality have any necessary relationship to one another. While they have in common the m-performativity just discussed in the preceding text, they express notionally distinct kinds of attitudes (much like evidentiality and epistemic modality). This is not to deny that there are connections between mirativity and evidentiality or that there are no reasons that indirect evidentials in many languages can be used as mirative strategies (or perhaps miratives proper). However, they are, in principle, distinct, and we take it as a matter of empirical investigation to determine what further properties, if any, they may have in common. Since Tagalog pala only expresses mirativity and never indirect evidentiality, as illustrated in (9), we do not take up this relationship here further. (9) Context: I have not been outside all day and go outside and see mud on the ground. Some hours later, I report to you that it was raining earlier. #Umuulan pala. rain.ipfv mir Intended: “Apparently, it was raining.” In this section, we have explored various notions of mirativity from prior literature and their interrelationships. We turn in the next section to examine data from context-relative felicity judgments aimed at disentangling these notions empirically. 4

Mirative pala in declaratives

As with all miratives, Tagalog pala can, of course, be used in scenarios where the speaker is experiencing surprise, as in (10)–(11): (10) Context: The speaker is surprised to suddenly learn that their friend’s boyfriend lives far away and is therefore sad. Kaya pala malungkot ang kaibigan ko. thus mir sad dir friend indir.1sg “Oh, that’s why my friend is sad!” (11) Context: The speaker sees a light on in their child’s room and expects that their child is playing video games, but then suddenly realizes that they are reading the Bible instead and says:

Tagalog pala 15 Nagbabasa pala ng biblia. read.av.ipfv mir indir bible “Oh, they’re reading the Bible!” Such uses are, of course, consistent with the idea that, of the various mirative notions introduced earlier, pala encodes surprise. However, they are equally consistent with the other three notions – new information, sudden revelation, and counterexpectation – since the speaker’s surprise in these examples is triggered by suddenly realizing a new piece of information which violates their expectations. To distinguish between these diferent notions, then, we need felicity judgments of examples in contexts which are consistent with some but not all the candidate mirative meanings. Following AnderBois (2018), we take there to be two clear-cut cases of this sort: (i) cases where a speaker’s prior expectation is suddenly met, and (ii) cases where a speaker suddenly remembers a piece of information. 4.1

Prior expectation suddenly met

The first situation which clearly distinguishes between diferent mirative notions are cases where a speaker has a prior belief or expectation and then receives some new piece of information which confirms this prior expectation. Such a scenario is clearly consistent with new information, a, or revelation, b. For miratives which encode surprise, c, or counterexpectation, d, however, this sort of scenario is incompatible. These predictions are summarized in (12). (12) Felicity in scenarios where speaker’s prior expectation is suddenly met Type

Felicitous?

a b c d

   

New information Revelation Surprise Counterexpectation

As seen in (13)–(14), Tagalog pala is felicitous in such scenarios, supporting the claim that it encodes either new information, a, or sudden revelation, b. (13) Context: I am supposed to meet my friend Juan, who is very punctual, at the library. I’m checking my phone for the time and suddenly look up and see that he is there, on time as always, and say: Ah, nandito na pala si Juan. ok here now mir dir Juan “Oh, Juan’s here.”

16

Scott AnderBois

(14) Context: I am talking to my friend about my stomachache and trying to figure out what to do about it. Eating always helps with this kind of stomachache, so I say: (Ah,) kailangan ko pala-ng kumain ok need indir.1sg mir-lnk eat.av ‘Oh, I need to eat!’ 4.2

Speaker suddenly remembers

The second sort of scenario AnderBois (2018) discusses for Yucatec Maya bakáan are two diferent sorts of cases where a speaker has a sudden revelation or realization with no new outside information. Within this category, there are two diferent subcases, one more airtight than the other. First, we have situations where the speaker knows a particular fact but it slips their mind temporarily and then pops back into their head, what we will call pure remembering. Second, we have cases where a speaker knows a set of premises and has them more or less in mind but suddenly realizes that they can draw a given inference from them that they had not previously drawn, what we can call sudden inference. Of course, these two situations are not mutually exclusive either, since one may suddenly remember a fact p that they had forgotten and then draw a new inference based on p together with other information. As seen in (15)–(17), Tagalog pala is felicitous in both sorts of scenarios, though the line between them is not always entirely clear. (15) Context: You are asking me what I did yesterday, and I am listing things I did before I momentarily draw a blank and then remember suddenly and say: Tapos, nagbeach (nga) pala ako. after go.to.beach.pfv part mir dir.1sg “Oh yeah, then I went to the beach.” (16) Context: The singer recounts being nauseated and confused, looking for their own place to live and wondering what has happened in the wake of a breakup. Suddenly remembering, the singer sings: Oo nga pala, hindi nga pala tayo yes part mir neg part mir dir.1pl.incl “Oh yeah, there’s no ‘us’” (i.e., we are no longer a couple) AnderBois (2018: 28) (17) Context: We are talking about someone I haven’t been in touch with for a while. I suddenly realize why she’s out of town on vacation and has been busy of late and say: Ay, kasal (nga) pala si Maribel. interj marry.pfv part mir dir Maribel “Oh, Maribel got married!”

Tagalog pala 17 For pure remembering cases like (15), it seems quite clear that such a context is compatible only with sudden revelation, b, and therefore that Tagalog pala must encode that notion. For cases of sudden inference like (17), whether such an inference counts as “new information” or not depends on how we conceive of information. Under a fairly standard semantics which captures information in terms of propositions modeled as sets of possible worlds, one makes the simplifying idealization that belief is closed under entailment. Under such a theory of belief, then, sudden inference cases seemingly must be regarded as old information, since the speaker’s belief worlds already include only worlds where the proposition denoted by the mirative’s scope holds. This is, in some sense, an intuitive result, since a speaker in such a scenario does not receive any new information from the outside world. On the other hand, we, of course, know that ultimately beliefs of human agents are not always closed under entailment, even if this is an ideal state which we do, indeed, tend toward. Depending on how our theory addresses this issue, we may find that sudden inference cases could indeed be regarded as new information, despite arising entirely from propositions which are old information. We see this state of afairs summarized in (18). (18) Felicity in pure remembering and sudden inference scenarios Type

Pure remembering

Sudden inference

a b c d

   

??   

New information Revelation Surprise Counterexpectation

We set aside the issue of how to interpret sudden inference scenarios here, since the felicity of pala in pure remembering cases renders it moot and, as noted earlier, distinguishing the two is potentially quite fraught. Moreover, as we will see in §5 – and as AnderBois (2018) argues for Yucatec Maya bakáan – the use of pala across sentence types suggests that information is not the relevant notion in the first place but rather illocutionary updates. I elaborate this point further in what follows, but note here merely that the distinction between sudden inference and pure remembering cases is ultimately irrelevant in this case. In sum, I have argued that the Tagalog particle pala encodes sudden revelation or realization, rather than surprise, counterexpectation, or new information (though it is consistent with the speaker experiencing any of these). This chapter, then, joins a growing body of work arguing that many mirative markers do not, in fact, encode surprise but rather one of these other notions: AnderBois (2016a, 2018) on Yucatec Maya bakáan, Salanova and Carol (2016) on Guaraní ra’e, Simeonova (2015) on Turkish and Bulgarian, and possibly

18

Scott AnderBois

Tawilapakul (2013) on Thai lεεw45. We turn now to address the interactions between pala and diferent sentence types and other operators. 5

Mirative pala across sentence types

Thus far, we have examined the use of the second-position clitic pala in declarative sentences. Given the obvious connections and similarities between miratives and evidentials, however, there is obvious reason to consider other sentence types, as these have often been key sources of evidence for or against diferent analyses in diferent languages. For evidentials, there are four main sorts of interactions which are wellestablished in previous literature. First, there are evidentials which are simply incompatible with some or all types of non-declarative sentences. In some cases, this appears to be for independent syntactic reasons, while in others, no such independent reason is found (cf. Aikhenvald (2004: §8.1)). Second, evidentials in interrogatives in some languages undergo what has been dubbed “interrogative flip,” where instead of encoding some evidential attitude on the part of the speaker, they express that the speaker expects that the addressee will hold such an attitude for the answer they give (cf. Murray 2017; Faller 2002), as illustrated in (19a). A third possibility attested for evidentials is to form biased questions that simultaneously express the speaker’s evidential attitude toward a proposition and ask the addressee to confirm (cf. Bhadra 2017), (19b). This option has been less discussed cross-linguistically but, to my knowledge, is only attested with indirect evidentials.3 Finally, we find evidentials, as in (19c), that express that the speaker holds a given evidential attitude about the illocutionary update the sentence itself encodes (cf. Faller 2002; AnderBois 2017). This “outside the speech act” case is attested in both interrogatives and imperatives, but only for reportative evidentials, presumably since other information sources do not apply readily to illocutionary content. (19) a. Interrogative flip Context: I know your friend went to a concert and you didn’t, but that you probably talked to your friend about it. I’m asking you about the concert, expecting your answer will be based on the roommate’s report. Ma-galing ba daw yung kumanta? adj-good polq rep dem.lnk singer “Was the singer good (according to what you heard)?” Tagalog b. Biased question  Context: Ram and Sita live together, and as they are going out for a movie, Ram asks Sita to leave their house keys with the neighbor and NOT the caretaker. When they return, Ram sees Sita dialing the caretaker’s number and asks:

Tagalog pala 19 cabi-ta kyaartekar-er kache rekh-e esh-e-chish naki? key-cl caretaker-loc close keep-ipfv come-ipfv-perf.2p infer “(Iinfer) you left the keys with the caretaker, (is that true)?” Bangla (Bhadra 2017: 95) c. Outside the speech act Context: Our mother has told me to make sure that my younger sibling eats their bread. I tell my sibling: Kainin mo daw ang tinapay mo eat.pv indir.2sg rep dir bread your “Eat your bread (she orders)” Tagalog (AnderBois 2017: 467) For miratives, the paucity of discussion of miratives in non-declarative sentences in prior literature presumably reflects the fact that in at least some of these cases, the first situation obtains and the combination is not possible. While not logically impossible, an interrogative flip interpretation for miratives seems to be of limited utility. The flipped mirative would have the speaker expressing their expectation that the addressee will answer their question with an answer the addressee finds surprising/suddenly realized/etc., and yet they are still asking the addressee the question and so presumably do not themselves have access to this new information. For the biased question case, it is hard to see how a sensible mirative analogue would be possible, since unlike indirect evidentials, for which the addressee may be expected to have better evidence, a speaker who holds any of the mirative attitudes has no such epistemic weakness (except insofar as the mirative also encoded indirectness or uncertainty of some sort or co-occurs with other expressions that do). In the remainder of this section, we show that pala in interrogatives and imperatives is possible and has an “outside the speech act” interpretation. Moreover, we argue in §5.3 that even in declaratives, there is evidence that the speaker’s sudden revelation concerns the sentence’s illocutionary update rather than the proposition itself. That is to say that all uses of pala are “outside the speech act” uses as opposed to non-declaratives having some sort of special interaction. 5.1

Imperatives with pala

For evidentials, imperatives more or less only ever co-occur with reportatives and have an “outside the speech act” use, as illustrated with Tagalog daw in (19c). Imperatives with Tagalog pala are readily possible and, as illustrated in (20), have clearly illocutionary uses. In particular, such sentences convey that the speaker has a sudden revelation that motivates the imperative speech act they perform with the rest of the sentence. This is most typically in cases where the speaker had previously intended to issue the command/advice/ ofer/wish/etc. and forgot or had otherwise been unable to do so and then suddenly remembers.

20

Scott AnderBois

(20) a. Context: While working in the library, we are trying to remember the Tagalog word for “blue.” I suddenly realize that, being in the library, there are dictionaries nearby, and say:  Gamitin mo pala yung diksyonario. use.pv.imper indir.2sg mir dem.lnk dictionary  “Oh yeah, use the dictionary!” b. Context: I am going to buy us concert tickets. You already bought them, though, and suddenly remember you never told me not to as you had intended, and say: Huwag na pala don’t now mir “Oh, don’t (buy them).” While pala occurs readily in imperatives, there are restrictions on the contexts in which this is possible. For very similar data in Yucatec Maya, AnderBois (2018) argues that these restrictions are related to the illocutionary update that imperative mood conventionally encodes. Here, we will not delve into the details of the analysis but will summarize a bit the main empirical restriction and refer the reader to that work for further discussion. Intuitively, imperatives, in their most prototypical uses, are about the speaker’s desire for the addressee to perform some action. While this is clearly true in command uses, for example, it’s not clear that the speaker’s desire is always necessary for an imperative to be used, which has led to vigorous debate over whether or not imperative mood cross-linguistically encodes the speaker’s desire as (part of) its meaning (see AnderBois (2018) for recent discussion and references). However, as AnderBois (2018) claims for Yucatec Maya bakáan MIR, we find instead that pala is infelicitous in situations where the speaker’s sudden realization concerns their own individual desires, as in (21b) and (22b). In contrast, the same imperatives with pala are completely acceptable in minimally diferent situations where the speaker’s sudden revelation concerns some other aspect of the speech act besides a shift in their own wishes. (21) a. Context: A mother is in the kitchen, cooking, and remembers that there are no beans in the house because she forgot to tell her son to go buy some, and says:  Bumili ka (nga) pala ng monggo  buy.av.imper dir.2sg part mir indir beans  “Oh yeah, go buy beans.” b. Context: A mother is in the kitchen, cooking, and realizes that she wants her son to go buy beans, and says: #Bumili ka (nga) pala ng monggo buy.av.imper dir.2sg part mir indir beans  Intended: “Oh yeah, go buy beans.”

Tagalog pala 21 (22) a. Context: My friend is about to tell a secret of mine which I told him. I had meant to tell him not to say anything to anyone but forgot, and now that the conversation is on a related topic, I say to him: Huwag mo pala sabihin. don’t indir.2sg mir say.pv  “Don’t say anything.” b. Context: I told my friend some sensitive information. Now that the conversation is on a related topic, he sounds like he’s starting to tell my secret. I normally am happy to hear my friend talking but, since I don’t want him to tell my secret, suddenly want him to be quiet. Huwag mo (#pala) sabihin. don’t indir.2sg mir say.pv  “Don’t say anything.” In the felicitous examples of imperatives with pala thus far, the speaker’s sudden revelation has specifically been the sudden remembering of their intent to issue the imperative in question. While perhaps the most common cases where such sentences arise, it is not the only one. Such imperatives are also possible in situations where the addressee’s goals and/or knowledge state about possible means of achieving them suddenly become clear. For example, in (23), the speaker suddenly realizes that the addressee does not know where to go and therefore that the advice their imperative conveys is relevant to impart to the addressee. (23) Context: Someone who is expected to know how to get to the university already asks for directions and gets the response. Dumiretso ka nga pala dito. go.straight.imper dir.2sg part mir here “Oh yeah, go straight here.” In (23), the speaker is aware of the addressee’s goal the entire time (to get to the university) but suddenly realizes that they don’t have sufcient information to select an appropriate action to meet this goal. In contrast, (24a) represents a case where all parties are certainly aware of what actions the addressee could take to achieve the goal in question (namely, grabbing a cookie from right in front of them and eating it) but where the addressee’s goals themselves are the object of the speaker’s sudden revelation. In this case, the close paraphrase with a conditional imperative, as in (24b), can serve to make this explicit. (24) a. Context: The addressee, Jorge, typically does not like cookies, and so even though the speaker has a tray of cookies, they don’t think to ofer them to Jorge. However, Jorge is eyeing the cookies, and so the speaker suddenly realizes that he, in fact, might want one this time, and says: Kumain ka pala ng cookie. eat.imper indir.2sg mir indir cookie “Have a cookie!”

22

Scott AnderBois b. Context: [same as (24a)] Kung gusto-ng mo pala kumain ka if want-lnk indir.2sg mir eat.imper indir.2sg ng cookie indir cookie “Oh, if you want, have a cookie.”

In sum, mirative pala is felicitous in imperatives in cases where the sudden revelation concerns the relevance or utility of the imperative update, but not those where it concerns the speaker’s preferences themselves. AnderBois (2018) has established essentially the same pattern for a mirative particle in an unrelated language, Yucatec Maya bakáan MIR. Following AnderBois (2018), we take this as evidence that mirative markers in these languages encode a sudden revelation about an illocutionary update rather than a proposition per se. On the strength of these data, then, we can therefore conclude that illocutionary updates for imperatives include things like the presumption of relevance to the conversational participants’ current decision problems but do not include the speaker’s desire for the addressee to perform the imperative action (though this desire may of course arise pragmatically in many cases). That is to say that we may take this as evidence that the speaker’s own desires may not be encoded by the imperative as such (see AnderBois (2018) for a more detailed discussion of this sort of argument based on similar data in Yucatec Maya). 5.2

Interrogatives with pala

Having discussed imperatives with pala in some detail, we turn now to another non-declarative sentence type: interrogatives. As discussed at the beginning of this section, the parallel with evidentials might lead us to expect a greater range of interpretive possibilities here than for imperatives (e.g., an interrogative flip reading of some kind). For Yucatec Maya bakáan, the fact that AnderBois (2018) does not find any such uses is perhaps unsurprising, since even the reportative evidential in Yucatec Maya, bin REP, does not allow for such uses. For Tagalog, however, as we have already seen in (19), the reportative allows for both illocutionary and flip readings, and so to the extent that there is some general (i.e., not morpheme-specific) grammatical mechanism necessary to the flip interpretation, we might expect to find a diferent situation in Tagalog. However, what we in fact find is that – again parallel to Yucatec Maya bakáan – pala only seems to have illocutionary uses expressing the speaker’s sudden remembering or other realization about performing the interrogative update in question. In interrogative cases, this may mean either suddenly remembering a prior intent to ask a question or suddenly realizing the relevance or need (i.e., non-redundancy) of a given question, as seen in (25). Crucially, though, a sudden change in the speaker’s desires themselves is not sufcient, as we will see in a moment.

Tagalog pala 23 (25) a. Context: We have already started talking before I got a chance to ask your name. I deferentially ask you after a while: Ano po nga pala ang pangalan ninyo? who hon part mir dir name indir.2sg.formal “Oh, sorry I meant to ask, what’s your name?” b. Context: A child who has forgotten what nata de coco is, since it’s mostly served at special occasions, only asks: Ano nga pala ang nata de coco? what part mir dir nata de coco “Oh yeah, what is nata de coco again?” c. Context: You told me something earlier about work, but I got distracted and forgot and remember and ask: Ano pala ang sinabi mo tungkol sa proyekto? what mir dir say.pv.pfv indir.2sg about obl project “Oh, yeah, what is it that you said about the project?” As in the case of imperatives, interrogatives are subject to certain restrictions in the kinds of sudden revelations which license it. In particular, as with imperatives, the speaker suddenly deciding or realizing their desire to know something is not sufcient to license the use of pala. This gives rise to the patterns of judgments in (26), where interrogatives with pala which are felicitous in other contexts are judged inappropriate in such cases. (26) a. Context: I had been wanting to ask you who speaks Cebuano, and say: b. #Context: I hear Cebuano being spoken and suddenly am curious who is speaking it. Sino pala ang nagsasalita ng Cebuano? who mir dir speak.av.ipfv indir Cebuano “Oh (I wanted to ask you), who speaks Cebuano?” (27) a. Context A: Earlier, I had lost my keys and wanted to ask you where they were, but couldn’t. I now remember and ask you: b. #Context B: I just realized I lost my keys, and so I want to know where they are. Asan pala yung mga susi ko? where mir dem.lnk pl key indir.1sg “Oh, where are my keys?” As in the case of imperatives, we have seen that interrogatives readily allow for pala in “outside the speech act” or illocutionary uses in which the illocutionary update the utterance puts forth is itself the object of the speaker’s sudden revelation. We turn now to declarative sentences to show that there, too, we find uses with illocutionary revelations rather than propositional ones.

24

Scott AnderBois

5.3

Illocutionary uses of pala in declaratives

The illocutionary nature of pala is plain to see in imperative and interrogative examples, since it is not clear what proposition would even be the object of the sudden revelation pala encodes. In this section, we argue that even in declaratives, pala is illocutionary in nature and, as such, is usable in contexts where the speaker’s revelation concerns not the propositional content of the declarative but rather the act of asserting it at a given point in conversation. We see this illustrated in (28). In (28a), we see another example of pala being used in a case where the speaker suddenly remembers a piece of information. However, the same sentence can also be felicitously used in a context like (28b), where the speaker has no sudden revelation about the fact that Marvin left but instead suddenly remembers to tell this fact to the addressee. (28) a. Propositional Context: I forgot that my friend Maria had gone to the cafeteria, and suddenly remember. b. Illocutionary Context: I forgot that I had wanted to tell you that Maria had gone to the cafeteria, but had not forgotten that she had. Pumunta pala siya sa kainan. go.av.pfv mir dir.3sg obl cafeteria “Oh, she went to the cafeteria.” One further illustration of the illocutionary nature of (28b) is that, unlike (28a), it is more or less interchangeable with a more explicit example, like (29). (29) Illocutionary Context: I forgot that I had wanted to tell you that Maria had gone to the cafeteria, but had not forgotten that she had. Gusto ko pala-ng sabihin sayo pumunta siya sa want indir.1sg mir-lnk tell.pv obl.2sg go.av.pfv dir.3sg obl kainan cafeteria “Oh yeah, I wanted to tell you that she went to the cafeteria.” We see further examples of clearly illocutionary uses of pala in (30). In addition to cases where the speaker suddenly remembers their previous intention to inform the addressee of something, we also find cases like (30a) in which the sudden revelation concerns the relevance of the information, in this case, that Marvin left. The speaker here knows all along that Marvin left (the sentence’s propositional content), but the addressee’s confusion causes the speaker to suddenly realize that the addressee wants to know what happened to Marvin, and therefore that the assertion would be relevant in context. (30) a. Context: Marvin left to go get food. I am looking around the room, puzzled as to where he went, and you say:

Tagalog pala 25 Umalis pala si Marvin. leave.av.pfv mir dir Marvin “Oh, Marvin left.” b. Context: We are talking about birds, and so the question I had from earlier about them is suddenly relevant: Oo nga pala, may tanong ako (tungkol sa ibon). yes part mir exis question dir.1sg about obl bird “Oh yeah, I have a question (about birds).” c. Context: I suddenly realize that you might have an interest in knowing that a vase broke, and say: Nabasag pala yung vase. break.pfv mir dem.lnk vase “Oh, by the way, the vase broke.” In this section, we’ve seen that even in declaratives, there are plainly illocutionary uses in which the sudden revelation the speaker conveys does not concern the fact of the matter itself but rather some other aspect of the speaker’s assertion of that fact. At first blush, this might seem to suggest that pala in declaratives is ambiguous between an illocutionary pala and a propositional pala, with only the former possible outside of declaratives. While not an impossible state of afairs, a simpler explanation with no covert ambiguity is possible. Earlier, we have taken the infelicity of pala with imperatives in certain cases as evidence that the conventionally encoded illocutionary updates of imperatives do not include. Similarly, here we may take the felicity of pala in apparently propositional cases as evidence that declarative updates do include the speaker’s commitment to the truth of their propositional content (or at least their having adequate evidence of this) alongside the more interactional aspects found in illocutionary cases. We refer the reader to AnderBois (2018) for a fleshed-out version of such a theory along with discussion of independent support for it. 6

Interactions with other elements

We have argued that Tagalog pala does not encode surprise, counterexpectation, or new information but rather a weaker notion (i.e., one that is compatible with a superset of the situations of the others): sudden revelation. This finding is part of a growing body of work (see AnderBois (2018: §6) for recent discussion and references) concluding that many mirative morphemes do not, in fact, encode surprise or counterexpectation, even though this is often regarded as the prototypical use of mirativity. While the felicity of pala in contexts incompatible with these stronger notions provides clear support for this position, it nonetheless is the case that a great many utterances with pala seem to convey one of these stronger notions, specifically mild surprise or counterexpectation. Indeed, informal descriptions like S&O’s consistently discuss it in these terms.

26

Scott AnderBois

The claim that we make here is not that this sense of surprise is misguided, but rather that it arises from other sources together with the mirative morpheme itself. One such source, implicit in much of the discussion in §4, is world knowledge. If a speaker conveys that their assertion of a proposition p was suddenly motivated and world knowledge suggests that the speaker might have previously believed or expected something other than p, then the net effect will be to convey surprise, or at least something quite close to it. This is addedly so if the speaker expects that the addressee will have this knowledge, in which case the speaker may well intend to communicate their surprise to their interlocutor. While world knowledge plays a central role in producing surprise readings, the most interesting cases where it plays such a role are ones where it combines with other linguistic elements, as described later. We therefore do not consider its efect in isolation but only as it is engaged by diferent grammatical elements described in the following. Another factor which we will leave to future work to explore is the role that intonation plays in conveying surprise, both separately and together with pala. Intonation and prosody generally have been quite understudied compared to other aspects of Tagalog grammar (though see S&O pp.  30–54, Kaufman (2005), and Richards (2010: §3.3.3) for some discussion), and so, more background work is needed here. The most comprehensive work on Tagalog intonation by far is Schachter and Otanes (1972: 30–54), who describe a wide range of diferent meaning-bearing intonation patterns in some detail. While none of them is described as relating to surprise per se – unlike in English, where intonational expression of surprise has been widely discussed since at least Sag and Liberman (1975) – they nonetheless are likely to play at least an indirect role. Setting aside these quite general mechanisms, though, we focus here on a number of specific grammatical mechanisms which play more narrowly proscribed roles in producing more specific mirative meanings. These include other discourse particles, the syntactic expression of information structural notions like focus and topic, and exclamatory constructions, which have been claimed in other languages to be inherently mirative in some sense (e.g., Rett 2011 for English). While this large group of diferent elements all may impact the kind of mirativity an utterance expresses, they may do so in quite diferent ways. First, in §6.1, we explore elements which are themselves also plausibly miratives but which co-occur with pala. These elements include the sentence-final particle a and sentence-initial interjections ah, aba, and ay. Second, there are what we call “mirativity strategies,” elements which do not encode any of the mirative notions but which nonetheless may convey counterexpectation or another mirative notion in a suitable context. Finally, in §6.3, we discuss the particle nga, which does not itself ever appear to express mirativity but which nonetheless serves frequently to constrain the range of mirative meanings with which pala is otherwise compatible. 6.1

A preliminary look at other miratives in Tagalog

Schachter and Otanes (1972: 461–462) describe a sentence-final particle a, which they describe as follows: “a is used in sentences that express an event or situation that is contrary to expectation,” illustrated in (31).

Tagalog pala 27 (31) a. Ayun (pala) sila a! there mir dir.3pl part “(Oh,) but there they are!” b. Ma-ganda pala ito, a! Adj-beauty mir dem part “Oh, but this is pretty!”

(Schachter & Otanes 1972: 462)

(Schachter & Otanes 1972: 462)

The consultants for this study, however, do not recognize the final use of a, instead correcting such sentences (including S&O’s original examples) to instead have an initial interjection a (also spelled ah), as in (32a). S&O describe (p. 555) a range of interjections as conveying surprise, including a, and also ay and aba, illustrated in the naturalistic (32b) and the elicited (32c), respectively. While these particles may co-occur with pala, it is clear from their discussion that they may also express a mirative meaning of some sort independent from pala. (32) a. A, hindi pa pala ako! interj neg yet mir dir.1sg “Oh, not yet (I thought it was my turn already).” b. Aba! Si Mang Rudy pala! interj dir sir Rudy mir “Aha, it’s Mr. Rudy!”4 c. Ay, kasal (nga) pala si Maribel. interj marry.pfv part mir dir Maribel “Oh, Maribel got married!” While we leave a detailed exploration of what kinds of mirative meaning and other they each convey, we can note that we have already seen examples in §4.1, such as (33), repeated from (13), which suggest that sentence-initial a(h), like pala, does not encode counterexpectation or surprise. (33) Context: I am supposed to meet my friend Juan, who is very punctual, at the library. I’m checking my phone for the time and suddenly look up and see that he is there, on time as always, and say: Ah, nandito na pala si Juan. interj here now mir dir Juan “Oh, Juan’s here.” Similar to this, the most common exclamatory construction in Tagalog, formed with the direct case marker ang DIR and the bare root form of the adjective, is compatible with scenarios where no counterexpectation is found, similar to declaratives. We again leave more detailed work on this to future work, but note that here, too, we find that elements which may often convey surprise are possible in contexts where no surprise or counterexpectation is present, suggesting that sudden revelation (albeit perhaps of a diferent kind) may be a more common mirative attitude to be semantically encoded.

28

Scott AnderBois

6.2

Contrast-based mirativity strategies

One common way to express surprise with mirative pala is by combining it with other elements which contrast (some aspect of) the stated situation with some other possible way things could be or could have been. The clearest example of this is the contrastive use of focus. Whereas English encodes focus intonationally, Tagalog largely encodes it morphosyntactically, as discussed briefly in §2 (and in far more detail by Kaufman 2005 and Latrouite & Riester 2018). Thinking about focus in particular, it is hopefully clear that focus does not itself encode mirativity. Rather, it simply conveys that the stated choice was selected from a set of contextually salient alternatives. However, if world knowledge and other factors convey that those alternatives were more expected, then together with the m-performative sudden revelation of pala, we arrive at a very close approximation of surprise in the mirative sense.5 We see the efect of this combination play out across three distinct scenarios in (34). (34) a. Context: Dogs are supposed to eat dog food, but you suddenly realized that your dog Rufe was eating cake. [Yung keyk]Foc pala yung kinain ni Rufe. dem.lnk cake mir dem.lnk eat.pv.pfv indir Rufe “Oh, Rufe ate the CAKE!” b. Context: You suddenly realized that your dog Rufe is eating cake rather than doing something else altogether. Kinain pala ni Rufe yung keyk. eat.pv.pfv mir indir Rufe dem.lnk cake “Oh, Rufe ate the cake!” c. Context: You know that someone ate the cake and just realized it was your dog Rufe. [Si Rufe]Foc pala yung kumain ng keyk dir Rufe mir dem.lnk eat.av.pfv indir cake “Oh, RUFFIE ate the cake!” A whole host of other elements establishing contrast in diferent ways have similar uses in combination with pala. These include other second-position clitics, such as the temporal clitic pa “still, yet,” (35), as well as the clitic naman PART, (36), which may be used to express contrast in many contexts (see AnderBois (2016b) for a detailed discussion and analysis of naman itself).6 (35) Meron pa pala. exis yet mir “So there’s still more (I thought there’s none left).” (36) Context: A singer sings about the moment she was getting to know the addressee and suddenly realized that her preconception of the addressee as a snob was mistaken.

Tagalog pala 29 Hindi ka naman pala totoo-ng suplado neg dir.2sg part mir true-lnk snob “Oh, but you’re not a real snob after all.”

(song lyrics)7

Indeed, once we go looking for it, we find that expressions which contrast a given state of afairs with alternatives and, in some way, imply in context that the alternative was expected are quite rampant. For example, sentential negation arguably has this sort of pragmatic profile – a sentence like “My wife is not pregnant” is typically used in contexts where its opposite, “My wife is pregnant,” was expected. While none of these elements necessarily ensures counterexpectation, their tendency toward this means that when they co-occur with the pala, they readily evoke a sense of surprise or counterexpectation. In sum, we have seen that Tagalog has a wide range of diferent elements which, with varying degrees of explicitness, convey counterexpectation or surprise in suitable contexts. This counterexpectation is not necessarily rooted in the speaker’s mental evaluation at the utterance time and therefore is not by itself mirative per se. However, when combined with the sudden revelation encoded by pala, it produces the efect of conveying surprise or counterexpectation that is of the m-performative sort that characterizes miratives. The fact that elements that at least have this tendency are so pervasive – perhaps especially in sentences that contain pala – goes a long way toward explaining why so much confusion has reigned over this domain and why mirativity has often been associated with counterexpectation, even though close investigation often reveals – as it has for pala – that counterexpectation is not a necessary feature for the felicitous use of many mirative morphemes. 6.3

Interaction with nga

Thus far, we have seen cases in which other elements that co-occur with pala either themselves express mirative meanings or express some part of a particular kind of mirative meaning and therefore may convey mirativity in a particular context. In this section, we turn to examine an element which does neither of these but nonetheless influences the nature of the mirativity pala conveys: the second-position enclitic particle nga. Whereas the elements in §6.2 had the efect of conveying counterexpectation, nga has somewhat the opposite efect, constraining pala’s mirative meaning to cases of sudden remembering and ruling out new information uses and especially counterexpectational ones. We see this illustrated in (37a) for declaratives (repeated from (4) earlier) and imperatives, respectively. (37) a. Context: I knew that it was raining, but it slipped my mind. I suddenly remember and say: Umuulan nga pala. rain.ipfv part mir “Oh yeah, it’s raining (of course).”

30

Scott AnderBois b. Context: A mother is in the kitchen, cooking, and remembers that there are no beans in the house because she forgot to tell her son to go buy some, and says: Bumili ka nga pala ng monggo buy.av.imper dir.2sg part mir indir beans “Oh yeah, go buy beans.”

Stepping back to consider nga on its own, S&O give separate characterizations of nga for imperatives and declaratives.8 For imperatives, S&O describe nga as changing imperatives to polite requests. For declaratives, it has a superficially opposite efect: nga expresses “afrmation or confirmation.” The efect of nga in declaratives is typically an emphatic verum-focus-like one, sometimes even resulting in impoliteness or annoyance, as in (38). The inclusion of pala here is infelicitous, since the speaker is not experiencing any sort of revelation, having answered the question recently already. (38) Context: A younger sibling asks repeatedly where mom went and, despite already having been told the answer, keeps asking. I answer, annoyed: Pumunta nga (#pala) siya para bumili ng pagkain. go.pfv part mir dir.3sg for buy.av indir food “As you already know, she went to go buy food.” While these two efects seem quite opposite, as discussed by Lee (2018) and references therein, there are, in fact, a number of discourse particles crosslinguistically that show a similar asymmetry, including Colloquial Singapore English lah21 and Japanese yo. The basic intuition Lee (2018) develops is that this asymmetry arises because imperatives in general (following Hamblin (1987) and others) can be backed by two diferent sorts of authority: social authority and rational authority. The former is the authority to impose diferent goals on the addressee (e.g., in directive uses), while the latter is a more epistemically-based authority to propose actions in order to achieve goals the addressee already has (e.g., advice, ofers). Particles like nga, then, can be thought of as talking about the epistemic state of the hearer and/or speaker. For declaratives, this produces a “stronger” assertion in some sense, while for imperatives, the same meaning has the efect of producing a “stronger” rational authority, thereby avoiding the risk that the speaker will be perceived as flexing their social authority. Returning to cases where nga co-occurs with pala in (37), we find that these are exclusively cases like (37a) where the speaker is suddenly remembering their intent to make a given speech act, including suddenly remembering the propositional content of the declarative. Since nga signals some aspect of the illocutionary update as being previously certain/obvious/etc., its meaning is therefore incompatible with the mirative notions of new information,

Tagalog pala 31 surprise, and counterexpectation (and, arguably, sudden inference cases). The particle nga therefore constrains the sort of mirativity that pala might be taken to express in other contexts without nga, despite itself expressing an epistemic meaning of some kind rather than a mirative-related one of any sort. Finally, while we leave it to future work to flesh out exactly what kind of certainty or epistemic authority nga encodes, whose it is, etc., there is one minimal pair of contexts with nga co-occurring with pala, (39), that appears to suggest that it must involve the addressee’s epistemic state in some way rather than just the speaker’s. In these examples, consultants report that the inclusion of nga makes it sound as if I am sort of reminding both of us of a fact we already knew. Its inclusion is therefore felicitous in the context in (39b) and infelicitous in (39a). Conversely, omitting nga is strongly preferred in (39a), even though the speaker is suddenly remembering something since they are not remembering a thing that the addressee has any apparent knowledge of. (39) a. Context: We are trying to figure out where our mom is when I suddenly remember that I had seen her leave to go to the store earlier when you weren’t there, and say: Pumunta (#nga) pala siya para bumili ng pagkain. go.pfv part mir dir.3sg for buy.av indir food “Oh yeah, she went to go buy food.” b. Context: We are trying to figure out where our mom is when I suddenly remember that she had told us she went to the store, and say: Pumunta (nga) pala siya para bumili ng pagkain. go.pfv part mir dir.3sg for buy.av indir food “Oh yeah, she went to go buy food.” In this section we have seen that while pala itself expresses a quite weak mirative meaning – sudden revelation or realization – a variety of diferent elements serves to produce the efect in context of a more specific meaning. Some of these elements are themselves plausibly miratives (e.g., ah, ay, aba), while others plainly are not (e.g., contrastive focus, naman, nga). Together with world knowledge and context-driven assumptions, these elements often serve to give the impression of a stronger, more specific mirative meaning, such as surprise or counterexpectation. Beyond giving a sense of how these stronger senses are conveyed in Tagalog in the cases where they are, we hope that this serves as a general caution for researchers that any claim that a given mirative encodes a stronger mirative notion, such as surprise, necessarily relies on having controlled for a host of contextual, intonational, and other formal factors. 7

Conclusions

In this chapter, I have considered Tagalog pala against the backdrop of different conceptions of mirativity proposed in previous research on other

32

Scott AnderBois

languages. Following AnderBois (2018) and other recent work, I have taken the diferences between these distinct mirative notions to be an empirical matter to be investigated using standard tools of semantic/pragmatic research such as context-relative felicity judgment tasks. Based on this, I have proposed that pala does not encode counterexpectation, surprise, or even new information but rather the less-specific notion of sudden revelation or realization. Additionally, we show that – as AnderBois (2018) has argued for Yucatec Maya bakáan MIR – Tagalog pala has “outside the speech act” uses not only in declaratives but also in other sentence types, such as imperatives and interrogatives. The study therefore highlights a mostly unexplored dimension in the cross-linguistic study of miratives. Moreover, the striking parallels between Tagalog pala and Yucatec Maya bakáan across sentence types provides support for a particular theory of illocutionary updates as a basis for explaining the uniformity of this pattern. While mirativity is often defined informally as the linguistic encoding of surprise, we find further support for the recent body of literature arguing that many mirative morphemes do not actually semantically encode surprise or counterexpectation at all. While pala itself does not encode counterexpectation, we have surveyed a variety of diferent elements which, together with a suitable discourse context, may produce utterances which convey the speaker’s counterexpectation or surprise. These include other elements, such as interjections, which may themselves be miratives, but also a range of other elements, such as focus and other discourse particles that convey a sense of contrast and thus play a crucial role in communicating counterexpectation of the mirative sort. These elements are each independently complex, and so one clear direction for future work is to better understand these elements and their interactions in greater detail. Finally, we highlight one additional future challenge for the present account: the ability of pala to occur in certain kinds of clausal complements. While this possibility may seem unexpected, given the illocutionary account we have given, such examples are indeed possible, as illustrated in (40). (40) a. Context: I suddenly realize that you might not know that Maribel got married, and so I should double-check to make sure you do, and say: Alam mo ba pala [kinasal si Maribel] know indir.2sg interr mir marry.pv.pfv dir Maribel “Oh, did you know Maribel got married?” b. Alam mo ba [kinasal pala si Maribel] know indir.2sg interr marry.pv.pfv mir dir Maribel “Oh, did you know Maribel got married?” Although I do not have clear diferences in felicity judgments to support this at present, consultants report a subtle but consistent diference between such pairs of examples. For (40a), consultants report that the revelation

Tagalog pala 33 intuitively is more about the question about the addressee knowing, while in (40b), it seems to be more about the fact of the matter. Based on this example, then, we might speculate that the diference in the position of the particle relates to a diference in whether the matrix or embedded clause serves as the main point of the utterance (i.e., its at-issue content). It is well-known that in English sentences analogous to (40) that have a veridical attitude, its complement can be used both in discourses where the topic of conversation is about the addressee’s mental state and in ones which are about the fact of the matter itself. In this latter case, the attitudinal information serves as not-at-issue content of some kind (see Simons 2007, Hunter 2016, and AnderBois 2016c for recent accounts). A hypothesis consistent with the account in its present form, then, is that the syntactically embedded use of pala is possible only if that embedded clause has main-point/at-issue status in the discourse. Whether this speculation holds across a broad range of data remains to be tested. While embedded uses are prima facie unexpected for an illocutionary account, there remains hope that the broader pattern of embedded data will be compatible with the account here, given the illocutionary variability of such constructions. Abbreviations 1 2 3 art av clf cont dem dir mir neg obl prep part

first person second person third person article agent voice classifier continuative demonstrative direct case mirative negation oblique preposition particle

emph exis excl imper incl indir interj ipfv lnk pl pfv pv sg top

emphatic existential exclusive imperative inclusive indirect case interjection imperfective linker plural perfective patient/theme voice singular topic

Notes 1 Acknowledgments: First of, my heartfelt thanks to the Tagalog and/or Iloko consultants for sharing their language and culture with me: Alenette Ballesteros, Mariel Pacada, Amber Teng, and Gianna Uson. Thanks also to Rick Nouwen and anonymous reviewers at Journal of Semantics, where AnderBois (2018) was published, for their helpful comments and feedback on the data and ideas there, which included some preliminary data with Tagalog pala. Thanks also to the

34

2

3

4 5 6

7 8

Scott AnderBois editors of this volume and an anonymous reviewer for helpful feedback on this work. Indeed, we believe one of the difculties faced in much of the prior literature on mirativity is the lack of a clear distinction between miratives proper and mirativity strategies. For example, much of the mirativity literature has focused on mirative uses of indirect evidentials, which prima facie appear to be mirativity strategies rather than miratives proper (though see Rett and Murray 2013 inter alia for arguments to the contrary). We set aside this particular issue here, as there do not appear to be any indirect evidential or epistemic modals that play prominent roles in the expression of mirativity in Tagalog. Beyond the fairly limited empirical support for such a restriction, there is a theoretical reason to expect this to be the case as well. A rational speaker who had the best kind of evidence possible for a given claim would typically have no reason to ask their interlocutor the question. In contrast, with indirect evidentials, the speaker may reasonably expect that the addressee may have “better” evidence. Textual example from Intermediate readings in Tagalog, Philippine Center for Language Study, UC Press, 1968. Note that without pala, focus plus the prior expectedness of the alternatives do not necessarily produce surprise in the mirative sense any more than a lexical verb like English surprise does. As discussed previously in §3, this example represents a case of free indirect discourse or other similar perspective shift. While the singer is singing sometime after having established a relationship with the addressee, they are able to evoke the emotion of that moment through the use of perspective shift, in a sense “transporting” the listener back to the moment when the shift in her view of the addressee occurred. https://lyricstranslate.com/en/aegis-minahal-kita-lyrics.html. S&O also briefly discuss nga in hortatives and optatives, which we set aside here. While they do not discuss interrogatives with nga, we have already seen in (25a)– (25b) that such a combination is possible and appears to behave more or less similarly to nga in imperatives, serving to make the speaker’s request for information more polite or deferential.

References Aikhenvald, Alexandra. 2004. Evidentiality. Oxford: Oxford University Press. Aikhenvald, Alexandra. 2012. The essence of mirativity. Linguistic Typology 16. 435–485. AnderBois, Scott. 2016a. Illocutionary mirativity: The case of Yucatec Maya bakáan. In Thuy Bui & Rudmila-Rodica Ivan (eds.), Proceedings of Sula 9, 1–16. Amherst, MA: GLSA Publications, University of Massachusetts. AnderBois, Scott. 2016b. A QUD-based account of the discourse particle naman in Tagalog. In AFLA 23: The Proceedings of the 23rd Meeting of the Austronesian Formal Linguistics Association, 20–34. Canberra: Asia-Pacific Linguistics, The Australian National University. AnderBois, Scott. 2016c. Semantics and pragmatics of (not-)at-issueness in Yucatec Maya attitude reports. Semantics & Pragmatics 9(19). AnderBois, Scott. 2017. An illocutionary account of reportative evidentials in imperatives. Proceedings of SALT 27. 459–479. AnderBois, Scott. 2018. Illocutionary revelations: Yucatec Maya bakáan and the typology of miratives. Journal of Semantics 35(1). 171–206. http://doi.org/10.1093/jos/ fx019.

Tagalog pala 35 Anderson, Stephen. 2005. Aspects of the Theory of Clitics. Oxford: Oxford University Press. Bhadra, Diti. 2017. Evidentiality and Questions: Bangla at the Interface. Dissertation, Rutgers University. http://doi.org/10.7282/T3S46W30 Billings, Loren. 2005. Ordering clitics and postverbal R-expression in Tagalog: A unified analysis? In Verb First: On the Syntax of Verb-Initial Languages, 303–339. Amsterdam: John Benjamins. Billings, Loren & Abigail Konopasky. 2003. Reassessing the role of syntax inside the morphological word: Verb-adjacent clitics in Tagalog and Bulgarian. In Proceedings of AFLA 9, Cornell Working Papers in Linguistics, Cornell University. Ithaca, NY. Chen, Victoria. 2017. A Reexamination of the Philippine-Type Voice System and its Implication for Austronesian Primary-Level Subgrouping. Dissertation, University of Hawaii. http://hdl.handle.net/10125/62502 DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguistic Typology 1. 33–52. Faller, Martina. 2002. Semantics and Pragmatics of Evidentials in Cuzco Quechua. Dissertation, Stanford University. Hamblin, Charles L. 1987. Imperatives. Basil Blackwell. Hunter, Julie. 2016. Reports in discourse. Dialogue & Discourse 7(4). 1–35. Huron, David. 2006. Sweet Anticipation: Music and the Psychology of Expectation. Cambridge, MA: MIT Press. Kaufman, Daniel. 2005. Aspects of pragmatic focus in Tagalog. In The Many Faces of Austronesian Voice Systems: Some New Empirical Studies, 175–196. Research School of Pacific and Asian Studies, Australian National University. www.academia. edu/1919247/Aspects_of_pragmatic_focus_in_Tagalog. Kaufman, Daniel. 2010. The Morphosyntax of Tagalog Clitics: A Typologically Driven Approach. Dissertation, Cornell University. https://hdl.handle.net/1813/17165 Latrouite, Anja & Arndt Riester. 2018. The role of information structure for morphosyntactic choices in Tagalog. In Perspectives on Information Structure in Austronesian Languages, 247–284. Berlin: Language Science Press. Lee, Junwen. 2018. The Semantics of Emphatic Strategies in Discourse. Dissertation, Brown University. http://doi.org/10.26300/fhf6-4j58 Murray, Sarah. 2017. The Semantics of Evidentials. Oxford: Oxford University Press. Peterson, Tyler. 2016. Mirativity as surprise: Evidentiality, information, and deixis. Journal of Psycholinguistic Research 45(6). 1327–1357. Rett, Jessica. 2011. Exclamatives, degrees and speech acts. Linguistics & Philosophy 34(5). 411–442. Rett, Jessica & Sarah Murray. 2013. A semantic account of mirative evidentials. In Proceedings of SALT 23, 453–472. http://doi.org/10.3765/salt.v23i0.2687 Richards, Norvin. 2010. Uttering Trees. Cambridge, MA: MIT Press. Sag, Ivan & Mark Liberman. 1975. The intonational disambiguation of indirect speech acts. In Papers from the Eleventh Regional Meeting, Chicago Linguistic Society, 487–497. Chicago, IL: Chicago Linguistic Society. Salanova, Andrés Pablo & Javier Carol. 2016. The Guaraní mirative evidential. In Thuy Bui & Rudmila-Rodica Ivan (eds.), Proceedings of Sula 9. Amherst, MA: GLSA Publications, University of Massachusetts. Schachter, Paul & Fe T. Otanes. 1972. Tagalog Reference Grammar. Berkeley, CA: University of California Press.

36

Scott AnderBois

Simeonova, Vesela. 2015. On the semantics of mirativity. In Santa Vinerte (ed.), Proceedings of the 2015 Annual Conference of the Canadian Linguistic Association. http:// claacl.ca/actes-2015-proceedings/. Simons, Mandy. 2007. Observations on embedding verbs, evidentiality, and presupposition. Lingua 117. 1034–1056. Tawilapakul, Upsorn. 2013. Counter-Expectation in Thai. Dissertation, University of York. https://etheses.whiterose.ac.uk/9560/

2

Discourse particles in Tagalog The case of e Naonori Nagaya

1

Introduction1

Discourse particles are broadly understood in the present paper as referring to invariable linguistic items used to express the speaker’s epistemic attitude toward the propositional content of an utterance or to manipulate discourse coherence. East and Southeast Asian languages are well-known for their rich inventory of discourse particles in this sense, and accordingly, in the case of certain languages, the literature on this topic is broad and deep. For example, in their comprehensive grammar of Cantonese, Matthews and Yip (2011) identify 33 particles and provide a detailed analysis of each one. Goddard (2005: 144) observes that, in mainland Southeast Asian languages, particles are used to indicate diferent kinds of speech acts and express the speaker’s emotional responses, concluding that a “rich inventory of illocutionary particles is an areal feature of mainland Southeast Asia.” This is also the case with some languages of insular Southeast Asia. For example, Sneddon (2006) provides a chapter-long description of discourse particles in Colloquial Jakartan Indonesian. See also Nomoto (this volume) for Malay discourse particles. Unfortunately, however, little is known about discourse particles in Tagalog and other Philippine languages, and much uncertainty still exists concerning the nature of discourse particles in these languages. Schachter and Otanes (1972), the most comprehensive grammar of Tagalog, report that there are four discourse particles (or “sentence-final particles” in their terminology) in this language: a, e, ha, and o. Their description of the four particles is brief, includes few examples, and is only two pages long. This contrasts sharply with Li and Thompson’s (1981) description of Mandarin Chinese, in which about 80 pages are devoted to the discussion of five sentence-final particles. Thus, it is evident that more comprehensive research on discourse particles in Philippine languages is needed. This paper aims to bridge the aforementioned gap in the literature by examining discourse particles used in actual social interaction by Tagalog speakers. To be more specific, by observing the actual usage of discourse particles in naturally occurring conversations, this chapter ofers a general overview of discourse particles in Tagalog and presents a qualitative analysis of the DOI: 10.4324/9781351057752-3

38

Naonori Nagaya

sentence-final particle e as a case study. The particle e is the most frequently employed particle in Tagalog and displays a complex spectrum of meanings ranging from reason to negative evaluation. The rest of this chapter is structured as follows. Section 2 provides background information on Tagalog and introduces the conversational corpus employed for the purpose of the present study. Section 3 discusses the identification of discourse particles in the language and presents an overview of them. Section 4 is a case study of the sentence-final particle e. It is argued that e is best characterized as constituting a polysemy network of reason, stancejustifying, and negative evaluation uses. Section 5 then concludes the chapter. 2

Background

This section provides background information on Tagalog (Section 2.1) and the conversational corpus examined in the present study (Section 2.2). 2.1

Language profile

Tagalog belongs to the western Malayo-Polynesian subgroup of the Austronesian language family. It is spoken by more than 30 million speakers as a first language in the Republic of the Philippines. This language is also known as Filipino and, along with English, is one of the two ofcial languages of the Philippines. As a consequence of the American colonization of the Philippines in the twentieth century, most Tagalog speakers are fluent in (or, for some, native speakers of) English, and code-switching between the two languages is pervasive in the everyday conversation of Tagalog speakers. From a typological perspective, Tagalog and other Philippine languages display an array of features that are not found in most Austronesian languages of other Southeast Asian countries, such as Malay (see Nomoto, this volume). Tagalog is a verb-initial language with relatively consistent VO word order correlates. In particular, predicates appear clause-initially, while nouns are preceded by articles that also mark case relations (nom, gen, and loc). Pronouns inflect for case, person, and number. Consider the examples in (1) and (2). (1) Tas tumak-takbo daw yung babae. then run rep nom woman “Then (they say) the woman ran away.” (2) A~ayus-in ko yung damit ko. rdp~tidy-pv 1sg.gen nom clothes 1sg.gen “I will tidy up my clothes.” In (1), the clause-initial position is occupied by the verb tumakbo “run,” which is followed by the enclitic daw “reported (rep) speech, hearsay,” and then by the primary grammatical argument marked in the nominative case. In (2), the

Discourse particles in Tagalog

39

verb appears clause-initially and is followed by two arguments, the genitive pronoun ko “1sg.gen” and the primary grammatical argument in the nominative case. As seen in (1) and (2), Tagalog verbs are marked for voice: actor voice (av; mag, ), patient voice (pv; -in, ø), locative voice (lv; -an), and circumstantial voice (cv; i-). Voice selection determines the semantic role that a nominative NP plays in a clause and the syntactic transitivity of a clause, among other things. In the AV clause in (1), the nominative NP plays an actor role, and the clause is intransitive. The PV clause in (2) has a patient NP in the nominative case and is a transitive clause. Importantly, personal pronouns and some particles are second-position clitics. They occur after the first constituent of the clause, which is a verb predicate in most cases but can be a non-verbal word, such as a negator. Let us consider the examples in (3) and (4). Second-position clitics are in boldface. (3) Ta-add niya ako sa isa=ng grupo, add 3sg.gen 1sg.nom loc one=lk group “He added me to a group.” (4) Para=ng hindi ko pa rin naman alam seem=lk neg 1sg.gen yet too indeed know kung ano yung ga~gaw-in ko=ng next, whether what nom rdp~do-pv 1sg.gen=lk next “It seems that I don’t know yet what I will do next.” In (3), the clause-initial position is taken by the verb predicate inadd “added.” It is, in turn, followed by the two personal pronouns niya “3sg. gen” and ako “1sg.nom.” However, in (4), the first constituent in the complement clause of parang “seem” is the negator hindi, in which case the personal pronoun ko “1sg.gen” and the three particles pa “still, yet,” rin “too,” and naman “indeed” follow the negator rather than the predicate alam “know” of the complement clause. In Section 3.3, second-position clitics with stance-indicating and discourse-structuring functions are analyzed as sentence-medial particles. 2.2

Conversation corpus

In this chapter, naturally occurring conversations produced by Tagalog native speakers are examined to determine the actual usage of discourse particles. In particular, a corpus of video recordings of informal, everyday language usage is analyzed. The corpus examined consists of approximately three hours of spontaneous face-to-face interactions in Tagalog. The corpus contains a total of 33,713 orthographical words, including false starts and truncated words; there are 9,155 intonation units (IUs)2 in total. See Table 2.1.

40

Naonori Nagaya

Table 2.1 Conversational corpus for the present study ID

Duration

# of words

# of IUs

201603UPFCMACH 201603UPFCYEOL 201603UPFCBEEL Total

65′04 60′01 80′06 205′11

12,480 8,722 12,511 33,713

3,045 2,432 3,678 9,155

The participants in the present study were female and male college students in their 20s. Each recording involves two participants who know each other very well. The conversations were recorded in one continuous session at a location of the participants’ choosing in Metro Manila in March 2016. No prompts or conversation topics were given. The audio data were transcribed with assistance from native speakers. The transcription system used was a simplified version of Du Bois et al. (1993). The spelling convention commonly used in the Philippines has been adopted for this project. In this convention, glottal stops are not spelled out, and the diagraph ng represents a velar nasal, except that the article ng is pronounced as [naŋ] and the plural marker mga as [maŋa]. All proper names used in this chapter are pseudonyms. 3

Discourse particles in Tagalog

The definition of discourse particles calls for clarification. To begin with, there is considerable variation in what might be labeled a discourse particle. This term may be limited to those particles that express the speaker’s attitude toward the propositional content of an utterance, or it may also include those particles that are used to relate an utterance to the immediately preceding utterance (e.g., Schifrin 19.87; Fraser 1996, 1999; Schourup 1999; Aijmer 2002; Beeching 2016; Traugott 2016; Brinton 2017; Zimmermann 2019; Heine et al. 2021). The term “discourse marker” is often employed to refer to those particles with discourse-connecting functions, as distinguished from discourse particles (Fraser 1999; Schourup 1999). In addition, the labels given to what are referred to here as discourse particles vary substantially. They include “pragmatic particles,” “pragmatic markers,” and “discourse markers,” among others. In this chapter, discourse particles are broadly understood as invariable words with various communicative and interactive functions, such as expressing the speaker’s attitude toward what is being said and managing the flow and structure of discourse. By adopting this broad definition, this chapter aims to provide a broader picture of discourse particles in Tagalog. This section ofers an overview of discourse particles in Tagalog. Attention is paid to the positions of discourse particles relative to the utterance, especially at the left and right peripheries of the utterance (Beeching & Detges

Discourse particles in Tagalog

41

2014; Traugott 2015; cf. Rizzi 1997; Cinque & Rizzi 2008; Haselow 2016). Cross-linguistically, these two positions are common sites for the manifestation of diferent kinds of information structuring devices, such as topicalization, topic shift, clefting, and focus fronting.3 The distinction between the two positions matters a great deal in investigations of discourse particles, partially because discourse particles often appear in these two positions, as shown in the rest of this section, and also in part because the same discourse particle can have diferent functions in diferent positions. For example, the particle man in English has diferent meanings at the left and right peripheries (McCready 2008, 2012). Consider (5) and (6), for example. (5) The cofee is strong, man. (6) Man the cofee is strong. The sentence-final man in (5) expresses “a kind of strengthening of meaning, as if the speaker is trying to convince the hearer to accept the content of the sentence as true” (McCready 2012: 780). In contrast, the sentence-initial man in (6) indicates “that the fact that the proposition expressed by the sentence is true is noteworthy in some way, and that the speaker has some sort of emotive reaction to it” (ibid.). A similar observation is made with regard to alors “at that time, then, so” in Spoken French, which favors speaker-oriented meanings at the left periphery and addressee-oriented meanings at the right periphery (Degand & Fagard 2011). In addition to the left and right peripheries of the utterance, it is cross-linguistically not uncommon for discourse particles to appear in sentence-medial positions, as in the German discourse particles ja, doch, and wohl in (7) (Zimmermann 2019: 513). (7) a. Max ist ja auf See. b. Max ist doch auf See. c. Max ist wohl auf See. Max is particle at sea. For these reasons, in this chapter, I adopt a broader definition of discourse particles and categorize them according to the position in which they are used within an utterance. Tagalog has three classes of discourse particles, depending on the position in which they appear relative to the utterance: sentence-final particles (Section 3.1), sentence-initial particles (Section 3.2), and sentencemedial particles (Section 3.3).4 I show that Tagalog has a very elaborate system of discourse particles with regard to variation in function and position. 3.1

Sentence-final particles

Tagalog has a relatively rich inventory of sentence-final particles (sfp). Schachter and Otanes (1972: 461–463) identify four sentence-final particles

42

Naonori Nagaya

in Tagalog: e, a, ha, and o. These particles appear in the sentence-final position of declarative sentences and add diferent kinds of epistemic attitudes of the speaker or discourse-manipulative meanings to the propositional content of an utterance. See Table 2.2 for the frequency of these four sentence-final particles in the conversational corpus discussed here. Note that only e, a, ha, and o occurring in sentence-final position were counted for the purposes of Table 2.2; those in sentence-initial position were excluded (Section 3.2).

Table 2.2 Frequency of SFPs e, a, ha, and o in the conversational corpus Particle

Freq.

Freq. per 10,000 words

Freq. per 100 IUs

e a ha o

316 16 6 11

93.7 4.7 1.7 3.2

3.45 0.17 0.06 0.12

As seen in Table 2.2, e is the most frequent of these particles. It occurred 316 times during the 205 minutes of conversation. Perhaps surprisingly, it is the 25th most frequent word in the corpus, competing with the existential predicate may (324 times) and the nominative case marker ang (314 times). In the simplest case, it indicates the logical relation that the utterance which it marks has to the previous utterance. See (8) for example. Note that, for the rest of this chapter, except when their meaning is readily translatable, sentencefinal and sentence-initial discourse particles are not translated and are simply represented by their forms in boldface in the free translations (e.g., e in (8)). (8) M:

C:

M:

Wala ako bukas, neg.exs 1sg.nom tomorrow “I will not be available tomorrow,” may baito pa ako e. exs part.time.job still 1sg.nom sfp “I still have a part-time job e.” Tsaka Monday? also Monday “Also Monday?” Okay lang, okay just “It’s okay,” kasi exam~exam naman iyon e, because rdp~exam indeed dem.dis.nom sfp “because that’s just an exam e.”

Discourse particles in Tagalog

43

The example in (8) is a snippet of conversation in the corpus examined. The particle e is attached to Speaker M’s second utterance in line 2. It logically connects the second utterance to the first one, indicating that the reason the speaker will not be available on the next day is that she needs to work. This is also true of Speaker M’s last utterance: it expresses a reason for the proposition in the previous utterance. The other sentence-final particles, a, ha, and o, have less transparent and more vague meanings, which I tentatively label “counterexpectation,” “sarcasm,” and “attention-getting,” respectively (cf. Schachter & Otanes 1972: 461). See examples in (9), (10), and (11). (9) E:

B:

Na-ta~tawa lang talaga ako, av-rdp~laugh just really 1sg.nom “I just really find it funny,” kasi proud na proud kami doon. because proud lk proud 1pl.excl.nom dem.dis.loc “because we’re very proud of it.” Achievement yun a. Achievement dem.dis.nom sfp “That’s an achievement a.”

(10) C: Pati kami,  even 1pl.excl.nom “Even we,” gi~gising nang maaga rdp~wake.up(av) adv early “even we will get up early ha.”

ha. sfp

(11) (Talking about one of Y’s cats) Y: Tapos alam mo ba, then know 2sg.gen q “Then do you know,” pag-balik ko noong isa=ng linggo, ger-return 1sg.gen when one=lk week “when I returned the other week,” O: Ay okay. “Okay.” Y: Buntis na siya. pregnant already 3sg.nom “She had gotten pregnant,” ayan o! there sfp “there (you go) o!”

44

Naonori Nagaya

In (9), Speaker B admits that Speaker E and others have achieved something, which is somehow not what Speaker B was expecting. In (10), Speaker C emphasizes the fact that the entire family will get up early, in a slightly sarcastic manner. Last, in (11), by using an o-marked utterance, Speaker Y is trying to get his interlocutor’s attention. At this point, these characterizations are only sketchy. In-depth analyses of the sentence-final particles a, ha, and o must be left for future research. In addition to e, a, ha, and o, there are several items that are not analyzed as discourse particles in Schachter and Otanes’s (1972) grammar but which could count as such in light of the definition adopted in this chapter. For example, the wh-question word ano “what” has grammaticalized into a confirmation particle, as in (12) (Nagaya 2022). (12) (Talking about a picture of one of Y’s cats) O: Pa-tingin. caus-look “Let me look (at the picture of your cat).”

Y:

Ang cute. nom cute “(The cat is) cute.” Ang cute niya ano. nom cute 3sg.gen sfp “She is cute, right?”

Another confirmation marker is the tag question marker diba. This discourse particle is etymologically composed of the negative particle hindi and the question particle ba. In traditional Tagalog orthography, diba is spelled separately, as (hin)di ba, but in naturally occurring conversation, it is best analyzed as a compound tag question marker, as in (13), in which Speaker M is soliciting confirmation from Speaker C. (13) (Talking about Speaker M’s boss) M: Kasi nga nasa Italy diba? because indeed loc.exs Italy sfp “(I cannot talk to him) because he is in Italy now, right?” C: Oo. “Yes.” In terms of prosody, the aforementioned discourse particles are associated with one or more intonation patterns and are often pronounced with vowel lengthening. For example, consider the Praat analysis of (14) in Figure 2.1. The example in (14) was found in the conversational corpus, and the pitch counter shown in Figure 2.1 was reproduced by a male Tagalog speaker in his 20s for the Praat analysis. It can be observed that the pitch goes down right after the pronunciation of ha starts and then displays a prolonged rising.

Discourse particles in Tagalog (14) B:

45

Joke lang ha. joke only sfp “It’s just a joke ha.”

Figure 2.1 Prosody of ha in (14).

Some discourse particles, such as ano, have several diferent prosodic realizations, which result in diferent interpretations of them (Nagaya 2022). Unfortunately, however, it remains to be seen how the intonation of discourse particles is related to diferences in meaning. In the orthographic convention widely used in the Philippines, a comma is often inserted before a discourse particle, as in joke lang, ha. However, there is no intonation break between the discourse particle and the precedent utterance, as seen in Figure 2.1. Similarly, e, a, and o are often spelled as eh, ah, and oh in the orthographic convention, possibly because of vowel lengthening, although they are not pronounced with an [h] sound. In this chapter, these orthographic conventions are not adopted for the sake of linguistic accuracy. 3.2

Sentence-initial particles

The sentence-final particles e, a, ha, o, ano, and diba can be used as sentenceinitial particles (sip) with a slight diference in meaning. For example, the particle diba can serve as a tag question marker even when it occurs sentence-initially, while the particle e is used as a hesitation marker when it occurs sentence-initially. Consider diba in (15) and e in (16), respectively.

46

Naonori Nagaya

(15) B: Diba siya yung mag-pa~party? sip 3sg.nom nom av-rdp~have.party “She is the one who will have a party, right?” 21. “On the 21st.” E: Supposedly oo. supposedly yes “Supposedly yes.” (16) Y: Yung Rosalinda. nom Rosalinda “As for Rosalinda (a television series).” O: A. “Ah.” Y: E pa~panood ko iyon. sip rdp~watch 1sg.gen dem.dis.nom “Well, I am watching it.” In (15), Speakers B and E are talking about a mutual friend who is planning to have a party on the 21st of the month. In this snippet, Speaker B is trying to confirm the identity of the friend by means of the tag question with diba. This use of diba fulfills the same function as the sentence-final diba in (13). In (16), Speakers Y and O are talking about a television series called Rosalinda. Before Speaker Y says that she watches this television series, she uses the sentence-initial particle e as a marker of hesitation. This function of sentence-initial e is diferent from that of sentence-final e (see Section 4 for more on sentence-final e). Note that the sentence-initial particles analyzed here are treated as being the same morphemes as the sentence-final particles. Further research is needed to determine whether this analysis is correct or whether they are diferent morphemes that happen to have the same phonetic form. The former option seems promising because, cross-linguistically, discourse particles tend to occur at both the left and right peripheries (Beeching & Detges 2014). 3.3

Sentence-medial particles

In addition to sentence-final and sentence-initial discourse particles, Tagalog has a variety of sentence-medial particles: daw/raw (reported speech); pala (counterexpectation); kaya (speculation, hedging); po (addresseeoriented honorific marker); lamang/lang “only, just”; din/rin “too”; nga “indeed, exactly”; man “even”; naman “indeed”; talaga “really”; and so on. These sentence-medial particles are second-position clitics in the sense discussed in Section 2.1.5 They have less-abstract meanings and thus tend to be more connected with the propositional content of a sentence than do

Discourse particles in Tagalog

47

sentence-final and sentence-initial particles, which do not directly add meaning to the propositional content but rather express the speaker’s epistemic attitude toward it or manipulate discourse structures. This is illustrated by daw and kaya in (17). (17) (Talking about edible frogs) Y: Chicken daw ang lasa. chicken rep nom taste “(They say) the taste (of frogs) is like chicken.” O: Oo, ang sarap kaya. yes nom delicious hedge “Yes, it is delicious (although you might disagree).” Y: Ayoko, hate.1sg.gen “I don’t want it.” takot ako sa palaka e. afraid 1sg.nom loc frog sfp “I am afraid of frogs e.” In (17), two sentence-medial particles, daw and kaya, are used. They appear as the second constituent of a clause. The former indicates that the information expressed by the clause in question was obtained indirectly through hearsay, while the latter is used to hedge the speaker’s opinion.6 Another interesting example of sentence-medial particles is the frustrative sana, which marks unrealized expectations/intentions, counterfactuals, and so on. Consider (18), for example. (18) C: Tas gusto ko sana wala=ng sulat yung likod, Then want 1sg.gen frust neg.exs=lk writing nom back “I wanted there to be no writing on the back (of the exam paper).”

The sentence in (18) is used to describe a situation in which Speaker C is running out of space on an exam paper but does not want to write anything on the back of the exam paper. Without sana, the sentence simply means that the speaker wants there to be no writing on the back of the exam paper. The frustrative marker sana adds the meaning that her desire was not fulfilled. See Kroeger (2017) for an analysis of the frustrative dara in Kimaragang, a Philippine-type language in Malaysia. Sentence-medial particles cannot be employed as sentence-initial or sentence-final particles in Tagalog, and double membership is not observed. It may sometimes prove difcult to distinguish between sentence-medial and second-final particles, especially when the former is directly followed by the latter. See the sentence-medial particles din “too” and naman “indeed” and the sentence-final particle a in (19).

48

Naonori Nagaya

(19) B: Wala nama(n)=ng nag-post. neg.exs indeed=lk av.rl-post “Nobody posted (on my SNS site).” E: Masaya din naman a. happy too indeed sfp “It’s really happy, too, a.” However, the diference between them becomes clear when the sentence is negated, as in (20). Only sentence-medial particles can move to the second position. (20) Hindi din naman masaya a. neg too indeed happy sfp “It’s really not happy, either, a.” As in (20), the sentence-medial particles din “too” and naman “indeed” move to the position right after the negator hindi, while the sentence-final particle a remains in sentence-final position. Also, observe that there is a functional difference between the two kinds of discourse particles. The particles din “too” and naman “indeed” add extra meanings to the propositional content expressed by the clause, while a is used to frame the proposition as a refutation of the interlocutor’s statement. Thus, sentence-medial and sentence-final particles are formally and functionally distinguishable. 3.4

Summary

This section has demonstrated that, although little attention has been paid to Tagalog in the literature, it has a rich inventory of discourse particles in terms of number, distribution, prosody, and meanings. In particular, three kinds of discourse particles were described in terms of their position in the utterance. However, the descriptions in this section are not meant to be exhaustive, and a more detailed analysis of each particle would be beneficial to the literature on discourse particles. 4

Case study: an analysis of e

In this section, an analysis of the sentence-final particle e is presented as a case study.7 There are two reasons that e has been selected for this purpose. First, as described in Section 3.1, this sentence-final particle is the most frequently employed discourse particle in the conversational corpus. The particle e is therefore a good starting point, considering how few studies on sentence-final particles in Tagalog there are in the first place. Second, the semantics of e looks simple but is, in fact, rather complicated. At first glance, this particle seems to have a straightforward set of properties. For example, consider (21).

Discourse particles in Tagalog

49

(21) O: Gusto niya=ng mag-IT, want 3sg.gen=lk av-it “He wants to work at an IT company,” kasi magaling sa computer e. because good loc computer sfp “because he is good at computers e.” There are two utterances in (21). The first utterance states that Speaker O’s brother is planning to work at an IT company, and the second utterance provides the reason for his plan. The second utterance is marked with e. It appears that it would sufce to say that e is a marker of reason. However, there are a number of examples of the sentence-final particle e that cannot be reduced to reason-marking. For example, consider (22). (22) Y: Para=ng yung mga picture mo sa China, seem=lk nom pl picture 2sg.gen loc China “It seems . . . your pictures in China,” marami ka ba=ng nag-pict– sa many 2sg.nom q=lk loc “did you take many pict(ures) in China?” O: Oo. yes “Yes.”

China? China

Gusto ko nga=ng i-pa-print lahat e. want 1sg.gen indeed=lk cv-caus-print all sfp “(Actually) I wanted to print out all (the pictures I took in China) e.” In the snippet of conversation in (22), Speakers Y and O are talking about O’s stay in China. When Y asks if O took a lot of pictures in China, O says yes and states his intention to print out all the pictures he took in China. Crucially, his statement is followed by the sentence-final particle e. Here, it is clearly not a marker of reason: the propositional content of the e-marked utterance does not describe a reason for any utterances in (22). So what is the function of e here?8 Before exploring this question in Sections 4.2, 4.3, and 4.4, let us first look into the prosodic features of the particle e in Section 4.1.

4.1

Prosody of e

The discourse particle e has several interesting prosodic features. Consider Figure 2.2, which is the pitch analysis of the last utterance in example (23), nakakalimutan ko lagi e “I always forget things.” The Praat analysis is based on a sound file reproduced by a male speaker in his 20s.

50

Naonori Nagaya

(23) (Talking about their assignment) M: Send ko sa ’yo mamaya. send 1sg.gen loc 2sg.loc later “I will send (it) to you later.” C: O sige. sip okay “Okay.” M: Pa-alala mo, caus-remember 2sg.gen “Remind me,” na-ka~kalimut-an ko lagi e. nvol.rl-rdp~forget-lv 1sg.gen always sfp “I always forget things (i.e., I am forgetful) e.” In (23), Speaker M promises Speaker C that she will send the class material for their assignment but also asks Speaker C to remind her because Speaker M is forgetful. The sentence-final particle e is attached to the clause, indicating the reason for the reminding. As Figure 2.2 shows, e is pronounced with a noticeably prolonged vowel. In addition, the combination of an acute rise and a fall in pitch is observed. These two prosodic features of e were observed consistently throughout the corpus. Perhaps because of these prosodic features, the discourse particle e is often spelled eh and preceded by a comma in the orthography commonly used in the Philippines. For example, see the sentence in (24), which is cited from an essay written in Tagalog: Parang Kayo Pero Hindi (Capili 2013: 45).

Figure 2.2 Prosody of e in (23).

Discourse particles in Tagalog

51

(24) Gabi na, eh. night already sfp “It’s already night e.” In (24), e is spelled eh, which may be reflective of vowel lengthening. A comma is used before it, although there is no intonation break before e, as seen in Figure 2.2. In this chapter, these orthographic traditions are not followed. Instead, this discourse particle is consistently represented as e. Having touched on the prosodic features of this sentence-final particle, we are now in a position to address the issue of its functions. 4.2

Connecting utterances

In the rest of this section, I discuss the functions of the sentence-final particle e. As has already been seen in (21) and (23), an important use of e is as a discourse connective, that is, to indicate that the utterance which it marks is a reason for a previously mentioned statement, logically connecting the two utterances. For example, in (25), with the e-marked utterance, Speaker M provides a reason for her first utterance. (25) M:

C:

Hindi na ako naka-reply nun sa neg already 1sg.nom nvol.av.rl-reply at.that.time loc iyo, 2sg.loc “I couldn’t reply to you at that time,” kasi may class ako e. because exs class 1sg.nom sfp “because I had a class e.” Okay lang, okay just “It’s okay.”

Note that, as in (25) (and (21)), when it marks an utterance as a reason, the sentence-final particle e often co-occurs with the adverb of reason kasi “because.” However, this is not obligatory, as seen in (23). 4.3

Contrasting stances

Another use of the discourse particle e is to mark an act of justification when speech-act participants’ stances, such as perspectives, evaluations, and epistemic attitudes, are contrasted, negotiated, and compared (see Du Bois 2007 for more on stances). In particular, it marks utterances that are employed to justify the speaker’s own stance. To illustrate, let us consider the following snippet of conversation in (26) between Speakers Y and O. Their stances difer

52

Naonori Nagaya

regarding chicken feet, which are popular as appetizers or finger food in the Philippines. Speaker Y does not like them, while Speaker O thinks they are delicious and appears surprised at Speaker Y’s stance. (26) (Talking about eating grilled chicken feet) Y: Hindi ko kaya=ng kain-in. neg 1sg.gen can=lk eat-pv “I cannot eat (grilled chicken feet).” O: Ayaw mo? hate 2sg.gen “Don’t you like (them)?” Y: Ayoko. hate.1sg.gen “I don’t like (them).” O: Sarap~sarap e. rdp~delicious sfp “They are very delicious e.” Y: E ang ka~kain ko laman e, sip nom rdp~eat 1sg.gen meat sfp “Well, what I eat is meat e.” Ba’t ako ka~kain ng paa? why 1sg.nom rdp~eat(av) gen foot “Why should I eat feet?” In (26), two diferent stances or perspectives are contrasted with regard to eating grilled chicken feet. Speaker Y does not like eating this part of the grilled chicken, but Speaker O does. In other words, Speaker Y and Speaker O have diferent stances or evaluations toward eating this specific part of a chicken. Importantly, when they utter statements supporting their own stances, they add the sentence-final particle e to each statement. In line 4, Speaker O supports his stance by saying that this part is very delicious; in line 5, Speaker Y defends her stance by arguing that she wants to eat meat rather than feet. These utterances are marked with e. This use of e is similar to the reason marker use insofar as both are involved in giving a reason, but the two uses can be distinguished in terms of the existence or absence of an efect clause and a contrasted stance. To illustrate, consider (27). (27) (Speakers B and E are talking about their favorite movies and books. Speaker E does not have knowledge of, or interest in, most of the movies and books Speaker B recommends.) B: Alam mo yung Me Before You? know 2sg.gen nom me before you “Do you know Me Before You?” Hindi? neg “Or not?”

Discourse particles in Tagalog

53

E: Hindi rin. neg either “(I) don’t (know it) either.” Gusto ko – want 1sg.gen “I want –” B: Na-tapos ko siya=ng basa-hin e, nvol.pv.rl-finish 1sg.gen 3sg.nom=lk read-pv sfp “I have finished reading it e.” E: Me before you? “Me Before You?” B: Oo. “Yes.” In (27), Speakers B and E take diferent stances toward the book Me Before You. Speaker B likes the book, but Speaker E has not read it and does not have any favorable opinion on it. Still, Speaker B continues to talk about this book by saying the e-marked utterance natapos ko siyang basahin “I have finished reading (it).” This e is used for justifying Speaker B’s stance or favorable opinion toward the book rather than logically connecting this utterance to a previous utterance. This utterance stands alone without an efect clause, which would appear when e is used as a marker of reason. In other cases, the diference between the reason use and the stance-justifying use is only subtle, as in (28). There are two instances of e in this conversation: e in line 3 is a marker of reason, while e in line 4 is a marker of justification of stance. (28) (Speakers Y and O are talking about the culture of eating dogmeat in the Philippines.) O: Ganun siya ka~kain sa Pilipinas. like.that 3sg.nom rdp~eat loc Philippines “It (=the dog) is eaten in the Philippines in that way.” Hindi ko alam, neg 1sg.gen know “I didn’t know.” di pa naman ako naka-ka~kain ng ganun e. neg yet indeed 1sg.nom nvol.av~rdp-eat gen like.that sfp “I haven’t eaten something like that e.” Y: May tawag doon e. exs call there sfp “There is a name for it e.” Asusena, “Asusena,” Asusena yung tawag. Asusena nom call “They call it Asusena.”

54

Naonori Nagaya

In (28), Speakers Y and O are talking about the culture of eating dogmeat in the Philippines. In Speaker O’s utterance, e is used to connect the preceding efect clause and the following reason clause. Speaker O says he did not know of the existence of such a culture in the Philippines, because he has never eaten dogmeat. This reason–efect relation is explicitly marked by the discourse particle e. Crucially, neither Speaker O’s nor Speaker Y’s stance toward the dog-eating culture is relevant to this reason–efect relation. The use of e in this context can be understood only from the propositional content of the two utterances. In contrast, the stance-justifying use of e in line 4 is diferent. By the utterance may tawag doon e “there is a name for the dog-eating culture,” Speaker Y is trying to justify the prevalence of such a culture in the Philippines, insisting that the dog-eating culture is popular in the Philippines to the extent that there is a name for it. Unlike the reason-marking use in line 3, there is no efect clause to which the e-marked clause is logically connected. Instead, it is used for justifying Speaker Y’s epistemic stance that eating dogmeat is common in the Philippines, which contrasts with Speaker O’s knowledge and/or assumption. In conversation, it often happens that speakers argue from diferent perspectives. In such cases, the discourse particle e comes in handy, helping to structure the flow of conversation. See (29), in which Speakers E and B are trying to negotiate their diferent stances toward Dubsmash, a video messaging application, by means of e-marked utterances. (29) (Speaker E encourages Speaker B to post on Dubsmash, while Speaker B does not like the idea.) B: Dubsmash. “Dubsmash.” E: Si Maine diyan sikat e. p.nom Maine there become.famous sfp “As for Maine (Mendoza), she became famous for her post (on Dubsmash) e.” B: Oo dami kay– yes many “Yes, many –” E: Sikat ka. become.famous 2sg.nom “Become famous!” B: Uy grabe ka. hey severe 2sg.nom “Hey you are severe.” Hindi ko nga ma-post e. neg 1sg.gen indeed nvol.pv-post sfp “I could not post e.”

Discourse particles in Tagalog

55

Na-hiya ako. av.rl-shy 1sg.nom “I was shy.” E: Try mo lang. try 2sg.gen just “Just give it a try.” In (29), Speakers B and E have diferent perspectives on the same application, although they do not mention them explicitly. Speaker E thinks Speaker B should try this application, while Speaker B thinks it is a bad idea. They each justify their unmentioned stance with the e-marked utterances, attempting to persuade each other. Last, e is employed when the speaker spots a contradiction in their interlocutor’s statements and uses it to support their own stance. In such contexts, the stance-justifying particle e can have the nuance of irritation or accusation, as in (30). (30) (Speaker Y likes animals, but Speaker O does not.) O: Ayoko sa mga hayop. hate.1sg.gen loc pl animal “I hate animals.” Y: Na-kagat ka na ba? nvol.pv.rl-bite 2sg.nom already q “Have you already been bitten?” Na-kagat ka na ba? nvol.pv.rl-bite 2sg.nom already q “Have you already been bitten?” Ng aso? gen dog “By a dog?” O: Hindi. neg “No.” Y: O di pa naman e. sip neg yet indeed sfp “Not yet e.” Speaker O does not like animals for some reason. His stance contrasts with that of Speaker Y, who loves animals. In (30), Speaker Y is wondering if Speaker O might be afraid of animals because he was bitten by a dog at some point, but it turns out that this is not the case. Speaker Y responds to this with the e-marked utterance, supporting her positive evaluation of animals.

56

Naonori Nagaya

Another example is given in (31). Speaker C makes the utterance yaman mo na e “you are already rich,” which stands in contrast to Speaker M’s selfevaluation that she does not have enough money now, although she has a part-time job. (31) C: Magkano ba ki~kita mo diyan? how.much q rdp~earn 2sg there “How much do you earn from that (part-time job)?” (5 lines omitted) M: Feb na apat na session, February lk four lk session “For four sessions we had in February,” 2k. “Two thousand pesos.” C: 2k, “Two thousand pesos,” 2k. “Two thousand pesos.” 500 isa, 500 one “500 pesos per session.” Okay yaman mo na e. okay rich 2sg.gen already sfp “Okay, you are already rich e.” In summary, the discourse particle e can be employed as a device to manipulate speech-act participants’ stances, justifying the speaker’s stance when it is contrasted, compared, and negotiated with the hearer’s stance. This use is similar to, but distinguished from, the reason use insofar as it involves multiple stances and does not require a logically connected efect clause. 4.4

Situating reality against ideal or expected situations

The third use of the sentence-final particle e is to situate reality against ideal or expected situations or to frame an utterance as an unfortunate or unexpected deviation from ideal or expected situations, expressing a negative evaluation of the state of afairs. In other words, e is used when an unfortunate or unexpected outcome has occurred instead of a preferred or expected outcome. In (32), for example, Speaker O compares real and ideal situations. (32) Y: Di ka pa punta ng man- Mongolia? neg 2sg.nom yet go gen Mongolia “Have you been to Mongolia yet?”

Discourse particles in Tagalog

57

O: E gusto ko nga e. sip want 1sg.gen indeed sfp “Well, (actually) I wanted to (visit Mongolia) e.” Kasi wala daw visa. because neg.exs rep visa “Because they say (Filipinos) don’t need a visa (to visit Mongolia).” Y: A. “I see.” In this snippet of conversation, the speakers are talking about Speaker O’s year-long stay in China. In line 1, Speaker Y asks if Speaker O has been to Mongolia. In line 2, Speaker O does not say no to the question directly but instead produces the e-marked utterance gusto ko nga e. This utterance literally means “I want(ed) to visit Mongolia,” but it implies, “I wanted to visit Mongolia, but unfortunately I have not.” Speaker O is situating the fact of his not having visited Mongolia relative to the ideal option of having visited it. By saying so, he answers Speaker Y’s question indirectly.9 This use difers from both the reason use and the stance-justifying use. In contrast with the reason use, these e-marked utterances do not provide a reason for a previously mentioned event. In (32), the e-marked utterance is only contrasted with a fact, although the fact itself is not explicitly mentioned in the conversation. In turn, unlike the stance-justifying use, in this use, stances taken by the speaker and the hearer resonate with each other, and there is no need to negotiate them. In (32), both Speakers Y and O evaluate visiting Mongolia positively. Instead, the point of employing e in this use is to contrast two possible situations and emphasize that the unfortunate or unexpected one is the one that actually happened. Thus, typical contexts in which this use of e is employed include situations in which some favorable event the speaker is expecting to happen does not actually happen. Consider (33), for instance. (33) (Speakers M and C are talking about a soap opera in which Joong-Ki plays a main role.) M: Hindi naman daw siya na-ganda-han neg indeed rep 3sg.nom nvol.rl-beautiful-lv doon sa ano, there loc whatchamacallit “She (= their mutual friend) said she didn’t like whatchamacallit.” C: Talaga? “Really?” M: Hindi daw siya naneg rep 3sg.nom hindi daw siya ki~kilig e. neg rep 3sg.nom rdp~excite sfp “(She said) she was not excited (about the TV soap opera) e.”

58

Naonori Nagaya C: Ang pogi kaya ni ano Joong Ki, nom handsome hedge p.gen whatchamacallit Joong-ki “Joong-Ki is very handsome (although she might disagree).”

Before the snippet of conversation in (33), Speakers M and C were talking about their favorite Korean soap opera, in which Joong-Ki plays a main role. However, Speaker M reports that their mutual friend does not like their favorite soap opera. When she reports this unfavorable fact, she uses an e-marked utterance, contrasting this unexpected situation with the more agreeable and expected situation in which the friend likes it. Importantly, there is variation in the type of propositional content of utterances marked by e. The e-marked utterance expresses an ideal situation in (32) and a real situation in (33). It can also convey a negative evaluation of a real situation, as in (34) and (35). (34) (Speakers M and C are talking about Japanese grammar books; Genki is a well-known textbook for elementary Japanese.) M: May book nga kasi na bigay yung exs book indeed because lk give nom ano, whatchamacallit “It’s because there is a book given by whatchamacallit,” C: A may bigay. ah exs give “Ah, there is a (book) given (by someone to you).” M: Pero hindi naman ga~gamit usually, but neg indeed rdp~use usually “But (I) don’t use (it) usually.” Sayang nga e, waste indeed sfp “What a waste e!” Sabi ko kung Genki sana yung book, say 1sg.gen if Genki frust nom book “I said, (I would use it) if the book were Genki.” (35) (Talking about Speaker Y’s pet cat.) Y: Kain siya ng lizard. eat 3sg.nom gen lizard “It (= the cat) ate a lizard.” O: E. “Ugh.” Y: A,   “Well,”   Kadiri nga e.   disgusting indeed sfp   “It’s disgusting e.”

Discourse particles in Tagalog

59

In (34), Speaker M is reporting that she does not often use the Japanese textbook given to her, contrasting it with the ideal textbook, Genki. In (35), Speaker Y is narrating an unfavorable event in which her pet cat ate a lizard in spite of the common assumption that cats do not eat lizards. In both conversations, the e-marked phrases, sayang nga e “What a waste” in (34) and kadiri nga e “It’s disgusting” in (35), express the speakers’ negative evaluations of the unfortunate or unexpected events. In extreme cases, e may be attached to minimal expressions, like oo “yes” and hindi “no,” as in (36). (36) (Talking about popular pet birds in the Philippines.) Y: Pero hindi naman sila na-bu~buhay nang matagal e. but neg indeed 3pl.nom av.rl-rdp~live adv long sfp “But they won’t live long e.” O: Oo nga e. yes indeed sfp “Yes e.” In (36), the speakers are talking about colorful pet birds that kids can buy in a local store. Both Speakers Y and O hope to keep these birds for a long time, but unfortunately, in reality, they will not live long. Speaker Y reports this fact with the e-marked sentence in the first line. Agreeing with her stance, Speaker O says oo nga e “yes” in the second line. Both speakers share the same negative evaluation of the reality of owning these pets, contrasting it with an ideal situation by means of e. Note that, as can be seen in (36) and other examples, the sentence-final particle e frequently cooccurs with the sentence-medial particle nga “indeed, exactly,” especially when employed as a marker of negative evaluation. Some frequent combinations, such as oo nga e in (36), can even be considered to be formulaic or fixed expressions of negative evaluations. This type of combinatory use of discourse particles is ubiquitous across languages but has received only limited attention (e.g., Koops & Lohmann 2015; Lohmann & Koops 2016; Haselow 2019). This phenomenon merits further study. To summarize, the discourse particle e indicates the speaker’s negative evaluation of reality, as contrasted with ideal/desirable/expected counterparts, that is, alternate realities. This characterization is shared by all cases of this category, although e-marked utterances difer in terms of their propositional content, ranging from actual/possible situations to negative evaluations. 4.5

Polysemy approach to e

Thus far, the sentence-final particle e has been described in detail, and the three diferent uses have been identified, as listed in (37). These three uses can be distinguished in terms of the number of stances involved, the occurrence of an efect clause, and the existence or absence of negative evaluation.

60

Naonori Nagaya

(37) The three uses of e in Tagalog: a. A marker of reason b. A marker of justification of stance c. A marker of negative evaluation In light of these observations, what generalization can we make about the nature of the discourse particle e? One of the possible approaches toward such multifunctional grammatical words/morphemes is a list approach.10 In this type of approach, several diferent meanings for a single grammatical element are simply listed as in (37) and illustrated with some examples, with no further generalization provided. In fact, this is the very approach taken by Schachter and Otanes’s (1972) description of e, in which they provide a list of three uses with examples, as in the following: E has three main uses. Accompanying an explanation, it is equivalent to English “it’s because” or “you see”; accompanying a contradiction of something said by another speaker (in which case the sentence often includes the enclitic particle nga . . .), it is equivalent to “on the contrary”; in other cases (particularly in answer to questions), it expresses regret or sympathy, and may be equivalent to “unfortunately.” (Schachter & Otanes 1972: 462) Summarizing their description, Schachter and Otanes (1972) identify three uses of e, namely, (a) explanation, (b) contradiction, and (c) regret/sympathy. In this approach, no further analysis is provided beyond the list. In fact, the preceding description is everything they include about e in their lengthy reference grammar of Tagalog. It leaves quite a few unanswered questions, such as how these uses are related to each other and whether there is any single meaning that covers all the uses of e. Another possible approach is a general-meaning approach, in which a single highly abstract meaning that subsumes all the individual uses is postulated for a grammatical element at issue. This approach is taken by Malicsi (2013), for example. In his one-sentence description of the Tagalog sentence-final particles, he characterizes e as “idinidiin ang taliwas sa inaasahan (emphasizing the opposition to what is being expected)” (p. 101). This approach is elegant and pithy but fails to capture the complexity of the usage of the discourse particle e. For example, it is not clear from this characterization why e could possibly be used to mark an utterance as a reason. This chapter presents yet another approach to multifunctional grammatical words/morphemes, namely, a polysemy approach. In this approach, a grammatical element is analyzed as having multiple related meanings. In this chapter, I argue that the three diferent uses of e in Tagalog are best described as forming such a polysemous network. To begin with, the reason use is

Discourse particles in Tagalog

61

connected to the stance-justifying use in the sense that both are involved in reasoning/justification; the diference is in the number of stances involved. The former is used to support the speaker’s already-mentioned utterance, while the latter occurs when the speaker’s stance or attitude toward a state of afairs needs to be justified against the hearer’s. In addition, the stance-justifying use is related to the negative evaluation use because both uses assume that two diferent perspectives are being contrasted. The stance-justifying use involves the diferent stances of the speaker and the hearer, while the negative evaluation use invokes the contrast between reality and an ideal/expected situation. Last, the reason use and the negative evaluation use seem to lack a shared property. They are only indirectly linked to each other by means of the existence of the stance-justifying use that falls in between. Thus, in my analysis, the three diferent uses of the discourse particle e constitute a polysemy network. An interesting question, then, is whether any prototypical or basic meaning can be postulated for the aforementioned polysemy network. There seem to be at least two arguments in favor of analyzing the reason use as the basic one in the network. First, all the native speakers with whom I have consulted about e shared the intuition that the basic meaning of e is to mark a reason. Second, the reason use is less abstract than the other two uses. The reason use indicates a textual or logical relationship between two particular utterances, while the other two uses pertain to the intersubjective manipulation of stances. Considering the well-attested path of semantic change from subjective to intersubjective meanings (Traugott & Dasher 2002: Chapter 4), it seems plausible to assume that the reason use is more basic than the other uses. At the same time, however, the polysemy approach can be used even without postulating such a basic function: the polysemy network of e can also be characterized as forming a family resemblance rather than a radial category with a basic use. In either case, the multifunctionality of the sentence-final particle e can be accounted for well by the polysemy approach, rather than the list or general-meaning approaches. Importantly, this polysemy analysis of the discourse particle e supports adopting a broad definition of discourse particles, as was done in this chapter. The use of e is not limited to expression of the speaker’s epistemic stance, as this particle also serves as a discourse connective. In the literature, the latter use is sometimes excluded from the common functions of discourse particles, and this use is instead often categorized as a discourse marker, which typically refers to “a syntactically optional expression that does not afect the truthconditions associated with an utterance it introduces and is used to relate this utterance to the immediately preceding utterance” (Schourup 1999: 234). Thus, e encompasses functions realized by both discourse markers and discourse particles in a narrow sense. Focusing only on one use might risk overlooking the full scope of the polysemy network of the Tagalog sentence-final discourse particle e.11

62 5

Naonori Nagaya Conclusion

This chapter provided an overview of discourse particles in Tagalog based on conversational corpus data. Tagalog was shown to have three types of discourse particles in terms of the position relative to the utterance: sentence-initial, sentence-medial, and sentence-final particles. This chapter also presented a case study of the sentence-final particle e, which is the most frequent sentence-final particle in Tagalog. This particle has three diferent uses, namely, as a marker of reason, a marker of justification of stance, and a marker of negative evaluation. Finally, it was suggested that a polysemy approach, rather than a list or general-meaning approach, may more suitably account for the behavior of such a multifunctional particle. Taken together, this chapter has demonstrated that Tagalog, like many other East and Southeast Asian languages, has a rich inventory of discourse particles.

Abbreviations 1 2 3 adv av caus cv dem dis excl exs frust gen ger lk

first person second person third person adverb actor voice causative circumstantial voice demonstrative distal exclusive existential frustrative genitive gerund linker

loc lv neg nom nvol pl pv q rdp rep rl sfp sg sip

locative locative voice negation nominative non-volitional plural patient voice question reduplicant reported realis sentence-final particle singular sentence-initial particle

Notes 1 Acknowledgments: Part of an earlier version of this chapter was presented at the 28th meeting of the Southeast Asian Linguistics Society at Wenzao Ursuline University of Languages on May 17–19, 2018, and in my lecture at the University of the Philippines, Diliman, on August 8, 2018. I would like to thank the audience for their discussion and suggestions, especially Roelia V. Alvarez, Mark Rae C. De Chavez, Farah C. Cunanan, Shirley N. Dita, Divine Angeli P. Endriga, Maria Kristina S. Gallego, Henrison Hsieh, Jay-Ar M. Igno, Jem R. Javier, Yukinori Kimoto, Aldrin P. Lee, Michael S. Manahan, and Ria P. Rafael.

Discourse particles in Tagalog

2 3 4

5

6

7

8

9 10

11

63

I am also grateful to Florinda Palma Gil, Mai Hayashi, Hiroki Hosoba, Kyosuke Yamamoto, and anonymous reviewers for valuable comments and criticism that have helped in improving the manuscript. Naturally, all errors are my own. This work was supported by JSPS KAKENHI grant numbers JP18K12366 and JP21K00522. An intonation unit is defined as “a sequence of words combined under a single, coherent intonation contour, usually preceded by a pause” (Chafe 1987: 22). Nagaya (2007) proposes diagnostic tests to distinguish the left and right peripheries of the sentence in Tagalog. The transcription system employed in this chapter is based on intonation units, and discourse particles are named according to their relative position in an intonation unit or an utterance, rather than in a sentence in the traditionally understood sense. For this reason, the more accurate terms for these particles would be, for example, “intonation unit–final particles,” “utterance-final particles,” or “final particles” (Hancil et al. 2015). That being said, this chapter follows the traditional terminology for the sake of readability. In the literature, some of the sentence-medial particles in Tagalog have been referred to as “enclitic adverbs” (Schachter & Otanes 1972: 411) or “adverbial clitics” (Kaufman 2010: 9) and have received much more attention than sentenceinitial/sentence-final particles. The elusive nature of hedging with the particle kaya remains to be explored. In (17), it is used to indicate that the speaker is defending his stance, although it is being challenged by the hearer. The particle is also in the middle of functional change, as pointed out in Batnag (2010). See (33) for another example of kaya. The analysis presented in this section is an expanded and substantially revised version of Nagaya (2020). I would like to thank Naoki Otani for his insightful comments on the aforementioned paper. I am also grateful to Farah C. Cunanan and Michael S. Manahan for their helpful discussion on the analysis presented here. As will be discussed in Section 4.4, one of the functions of e is that of situating reality against ideal situations, thus conveying the speaker’s negative evaluation of reality. In (22), specifically, e is used to imply that Speaker Y wanted to print out all the pictures he took in China but has not yet done so, and to emphasize that Speaker Y has a negative evaluation of this reality. Interestingly, this function of the sentence-final particle e is similar to that of the sentence-medial particle sana in (18) in that both particles indicate that an expected or desired situation is not realized. The list and general-meaning approaches toward multifunctional grammatical words/morphemes are discussed in Haspelmath (2003). See also the distinction between the monosemy approach and the polysemy approach discussed in Murphy (2010: 98–104). How to describe discourse particles in one individual language is one problem; how to compare them across languages is another. In his 2007 paper, in which he argues that language comparison cannot be based on pre-established categories but instead should be substance-based, Haspelmath (2007: 128) comments: “[T]here is less reason for optimism with regard to pragmatic particles such as German doch, denn, ja, wohl, and so on, whose conditions of use are extremely subtle and which cannot be readily translated from one language to the next. The typological comparison of such particles is a much more formidable challenge than the comparison of simple grammar and lexicon.”

64

Naonori Nagaya

References Aijmer, Karin. 2002. English Discourse Particles: Evidence from a Corpus. Amsterdam: John Benjamins. Batnag, Aurora E. 2010. Sumisibol na gramatika ng wikang Filipino: Ilang obserbasyon sa mga bagong kalakaran/pagbabago sa wika [Growing grammar of the Filipino language: Some observations of new styles/changes in the language]. In Ruth Elynia S. Mabanglo & Rosita G. Galang (eds.), Essays on Philippine Language and Literature, 203–226. Pasig City: Anvil. Beeching, Kate. 2016. Pragmatic Markers in British English: Meaning in Social Interaction. Cambridge: Cambridge University Press. Beeching, Kate & Ulrich Detges. 2014. Introduction. In Kate Beeching & Ulrich Detges (eds.), Discourse Functions at the Left and Right Periphery, 1–23. Leiden: Brill. Brinton, Laurel J. 2017. The Evolution of Pragmatic Markers in English: Pathways of Change. Cambridge: Cambridge University Press. Capili, Noreen. 2013. Parang Kayo, Pero Hindi. Mandaluyong: Anvil Publishing, Inc. Chafe, Wallace L. 1987. Cognitive constraints on information flow. In Russell S. Tomlin (ed.), Coherence and Grounding in Discourse: Outcome of a Symposium, Eugene, Oregon, June 1984, 21–51. Amsterdam: John Benjamins. Cinque, Guglielmo & Luigi Rizzi. 2008. The cartography of syntactic structure. In Vincenzo Moscati (ed.), STiL – Studies in Linguistics, CISCL Working Papers on Language and Cognition 2, 42–58. Siena: University of Siena. Degand, Liesbeth & Benjamin Fagard. 2011. Alors between discourse and grammar: The role of syntactic position. Functions of Language 18(1). 29–56. Du Bois, John W. 2007. The stance triangle. In Robert Englebretson (ed.), Stancetaking in Discourse: Subjectivity, Evaluation, Interaction, 139–182. Philadelphia: John Benjamins. Du Bois, John W., Stephan Schuetze-Coburn, Susanna Cumming & Danae Paolino. 1993. Outline of discourse transcription. In Jane A. Edwards & Martin D. Lampert (eds.), Talking Data: Transcription and Coding in Discourse Research, 45–89. Hillsdale, NJ: Lawrence Erlbaum. Fraser, Bruce. 1996. Pragmatic markers. Pragmatics 6(2). 167–190. Fraser, Bruce. 1999. What are discourse markers? Journal of Pragmatics 31(7). 931–952. Goddard, Clif. 2005. The Languages of East and Southeast Asia: An Introduction. Oxford: Oxford University Press. Hancil, Sylvie, Margje Post & Alexander Haselow. 2015. Introduction: Final particles from a typological perspective. In Sylvie Hancil, Alexander Haselow & Margje Post (eds.), Final Particles, 3–35. Berlin: De Gruyter. Haselow, Alexander. 2016. A processual view on grammar: Macrogrammar and the final field in spoken syntax. Language Sciences 54. 77–101. Haselow, Alexander. 2019. Discourse marker sequences: Insights into the serial order of communicative tasks in real-time turn production. Journal of Pragmatics 146. 1–18. Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison. In Michael Tomasello (ed.), The New Psychology of Language, Vol. 2, 211–242. Mahwah, NJ: Lawrence Erlbaum.

Discourse particles in Tagalog

65

Haspelmath, Martin. 2007. Pre-established categories don’t exist: Consequences for language description and typology. Linguistic Typology 11(1). 119–132. Heine, Bernd, Gunther Kaltenböck, Tania Kuteva & Haiping Long. 2021. The Rise of Discourse Markers. Cambridge: Cambridge University Press. Kaufman, Daniel. 2010. The Morphosyntax of Tagalog Clitics: A Typologically Driven Approach. Ph.D dissertation, Cornell University. https://hdl.handle.net/1813/ 17165 Koops, Christian & Arne Lohmann. 2015. A quantitative approach to the grammaticalization of discourse markers: Evidence from their sequencing behavior. International Journal of Corpus Linguistics 20(2). 232–259. Kroeger, Paul. 2017. Frustration, culmination, and inertia in Kimaragang grammar. Glossa: A Journal of General Linguistics 2(1). 56. http://doi.org/10.5334/gjgl.146. Li, Charles N. & Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley, CA: University of California Press. Lohmann, Arne & Christian Koops. 2016. Aspects of discourse marker sequencing: Empirical challenges and theoretical implications. In Gunther Kaltenböck, Evelien Keizer & Arne Lohmann (eds.), Outside the Clause: Form and Function of Extraclausal Constituents, 417–446. Amsterdam: John Benjamins. Malicsi, Jonathan C. 2013. Gramar ng Filipino. Quezon City: Sentro ng Wikang Filipino, Unibersidad ng Pilipinas. Matthews, Stephen & Virginia Yip. 2011. Cantonese: A Comprehensive Grammar. 2nd edn. London: Routledge. McCready, Elin. 2008. What man does. Linguistics and Philosophy 31(6). 671–724. McCready, Elin. 2012. Formal approaches to particle meaning. Language and Linguistics Compass 6(12). 777–795. Murphy, M. Lynne. 2010. Lexical Meaning. Cambridge: Cambridge University Press. Nagaya, Naonori. 2007. Information structure and constituent order in Tagalog. Language and Linguistics 8(1). 343–372. Nagaya, Naonori. 2020. Kaiwa no naka no tagarogugo bunmatuzyosi e [The sentence-final particle e in Tagalog conversation]. In Toshihide Nakayama & Naoki Otani (eds.) Ninti-gengogaku to Danwa-kinoo-gengogaku no Yuukiteki Setten: Yoohoo-kiban-moderu nimotoduku Sintenkai [Toward Dynamic Interaction between Cognitive Linguistics and Discourse-functional Linguistics: New Frontiers in the Usage-based Approach to Grammar], 267–290. Tokyo: Hituzi. Nagaya, Naonori. 2022. Beyond questions: Non-interrogative uses of ano ‘what’ in Tagalog. Journal of Pragmatics 190. 91–109. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of Grammar: Handbook in Generative Syntax, 281–337. Dordrecht: Springer Netherlands. Schachter, Paul & Fe T. Otanes. 1972. Tagalog Reference Grammar. Berkeley, CA: University of California Press. Schifrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schourup, Lawrence. 1999. Discourse markers. Lingua 107(3–4). 227–265. Sneddon, James N. 2006. Colloquial Jakartan Indonesian. Canberra: Pacific Linguistics. Traugott, Elizabeth Closs. 2015. Investigating ‘periphery’ from a functionalist perspective. Linguistics Vanguard 1(1). 119–130.

66

Naonori Nagaya

Traugott, Elizabeth Closs. 2016. On the rise of types of clause-final pragmatic markers in English. Journal of Historical Pragmatics 17(1). 26–54. Traugott, Elizabeth Closs & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Zimmermann, Malte. 2019. Discourse particles. In Paul Portner, Claudia Maienborn & Klaus von Heusinger (eds.), Semantics – Sentence and Information Structure, 511– 544. Berlin: De Gruyter Mouton.

3

A Kimaragang status particle Accessible information Paul R. Kroeger

1

Introduction1

Kimaragang is an endangered Philippine-type language belonging to the Dusunic subgroup in northeastern Borneo. This chapter discusses the meaning and functions of the Kimaragang use-conditional particle gima, comparing it with the German particle ja. I propose that the core meaning of gima includes at least the following two components of meaning: (a) accessibility and (b) uncontroversiality. In other words, p gima indicates that the speaker (a) believes that p is known or knowable by the addressee (and by the speaker, of course) and (b) takes the truth of p for granted (not debatable). I discuss these components of meaning in more detail in Section 2. Gima is one of several discourse particles in Kimaragang which “indicate the status of a proposition relative to the common ground (newness, expectedness, speaker commitment etc.)” (Repp 2013). We might refer to such particles as status particles, because they mark the information status of the base proposition. The Kimaragang status particles comprise a subset of a relatively large inventory of second-position clitics, as described in Section 3. Section 4 provides evidence for the claim that gima contributes use-conditional rather than truth-conditional meaning. I use the term “use-conditional,” following Gutzmann (2015), to refer to content which is part of the conventional meaning of an expression but does not contribute to the “at issue” truth-conditional meaning of the utterance. Potts (2005) and others have proposed a number of tests for identifying use-conditional content.2 I will focus on two of these tests, which McCready (2010) identifies as being the most reliable indicators for this purpose: (a) use-conditional content is “scopeless,” meaning, that it is never interpreted within the scope of semantic operators like negation, interrogative mood, conditionals, etc., and (b) use-conditional content does not participate in denials, that is, cannot form the basis for challenging the truth of a statement. Section 5 discusses contexts where, as predicted by the proposed analysis, gima cannot be used. As we will see, many of the same restrictions are reported for unstressed ja in German. Section 6 discusses the most common uses of gima, all of which involve statements about information that is noteworthy DOI: 10.4324/9781351057752-4

68

Paul R. Kroeger

even though it is already part of the common ground, or at least accessible to the addressee. Because such statements are, in a sense, uninformative, they may appear to violate Grice’s maxim of quantity. Thus, in many of its uses, gima can be viewed as a quantity hedge, like the English phrase after all. Section 7 discusses the expressive content associated with many uses of gima, particularly when it occurs in exclamatory utterances. Many of the Kimaragang example sentences used here are taken from Jim Johansson’s dictionary, some with minor modifications. Others (including all the unacceptable examples, obviously) have been elicited specifically to test the predictions of my analysis. Information concerning acceptability in particular contexts was provided by Mr. Janama Lantubon. 2

The meaning of gima

The term “accessible” has been used in a number of diferent ways. When I say that gima signals the speaker’s belief that the proposition being asserted is equally “accessible” to both speaker and hearer, I simply mean that the information is available to both. Büring (2017), as part of his definition of German unstressed ja, expresses this condition as follows: “ja p signals that speaker and addressee are . . . in an equally good epistemic position to utter p.” As Zuferey (2014) points out, in discussing the presuppositional nature of the French conjunction puisque, there are several diferent ways in which information may be mutually accessible: it may be (a) shared knowledge, (b) previously mentioned in the current discourse, (c) observable in the utterance context, or (d) inferable by pragmatic implicature from what has been said.3 One important diference between gima and German ja is that ja stands in opposition to another similar particle, unstressed doch. Both of these particles indicate that the current proposition is true and mutually accessible. Ja further indicates that it is, in some sense, uncontroversial, whereas doch indicates the opposite. Büring (2017) expresses this contrast as follows: “ja p signals that speaker and addressee are . . . equally liable to draw joint attention to p,” whereas doch p signals that the addressee is not likely to do so. Kimaragang has no particle equivalent to doch, so the meaning of gima is not delimited by the same kind of contrast as ja. In fact, as we will see, gima can actually occur in some contexts where doch would be used in German. A slightly diferent concept of “uncontroversiality” is needed for gima. I propose the following: gima signals the speaker’s belief that the proposition being asserted is undeniable and not open to debate. 3

Second-position clitics

Like many other Southeast Asian languages, Kimaragang has a large inventory of particles. Pure expressives, such as ay “surprise” or woy “what did I tell you?” tend to occur sentence-initially, can stand alone as a complete utterance,

A Kimaragang status particle

69

and get normal word stress and (frequently) intonational prominence. Second-position (2P) particles, in contrast, can never occur on their own. They are generally unstressed4 and never get intonational prominence. These 2P particles include nominative and genitive pronouns, focus and aspect markers, at least one evidential, the frustrative marker, question particles, markers of intimacy or friendship, and status particles, which are the primary focus of the present chapter. 3.1

Defining second position

Second-position particles occur immediately after the first constituent in their clause. In a normal verb-initial clause, this means immediately after the verb, as illustrated in (1). When a negative or other adverbial element occurs in pre-verbal position, 2P clitics will also precede the verb; this is exemplified in (2)–(3). (1)

N-o-dindi nu no gaam i=wogok? pst-nvol-hog.call 2sg.gen iam Q nom=pig “Have you called the pigs?”

(2)

Sid=tana ya n-odop-on. dat=earth 1pl.excl.gen past-sleep-lv “It was on the ground that we slept (after the house burned down).”

(3)

Amu oku po dati ko-guli dot . . . neg 1sg.nom yet probably nvol.av-return comp “I probably cannot return (to work here tomorrow).”

Sentence-level conjunctions do not function as a part of the minimal clause and so do not afect clitic placement. In the following example, the minimal clause begins after the conjunction bang, and the 2P clitic pronoun occurs after the fronted location phrase id tana. (4)

Bang [id=tana ko monumpa . . .] if loc=earth 2sg.nom av:tr:swear “If it is on the ground that you swear (an oath) . . .”

In a subordinate clause, whether complement or adjunct, clitic pronouns and particles appear immediately after the first element of their minimal clause; this clearly indicates the location of sentence-internal clause boundaries. Example (5), for example, contains three subordinate clauses, as indicated by the location of the highlighted clitic elements. (5)

Ela’an ku=i’ dot [magaago yalo ] nga’ know 1sg.gen=emph comp hurry(av) 3sg.nom but

70 Paul R. Kroeger

Table 3.1 Template for 2P clitic ordering Obligatory 2P clitics GEN

pron.

ku “1sg” nu “2sg” yo “3sg” to “1du.incl” ya “1pl.excl” duyu “2pl”

NOM

Optional 2P clitics pron.

oku “1sg” ko “2sg” kito “1du.incl” tokow “1pl.incl” okoy “1pl.excl” kow “2pl”

Focus/aspect

Mood

Evaluative

Solidarity

=i’ “emphasis/ verum focus” no “iamitive” po “continuative” nogi “again” nopo “only”

ga(a)m “YNQ” ma “rhet. Q” pogi “forceful” to “empathetic”

bala’ ~ bala’ay “mirative” dara ~ dara’ay “frustrative” gima “after all” dati “likely” toomod “likely” mari “certainly” katoy “counter-assertion”

obo (man-to-man) owo (woman-to-woman)

A Kimaragang status particle

71

[n-antara-Ø ku] tu’ waro [boros pst-intercept-ov 1sg.gen because exist say ku sid=dialo]. 1sg.gen dat=3sg “I knew that he was in a hurry, but I held him up because I had something to say to him.” 3.2

Linear order of clitics

As the preceding examples illustrate, it is not uncommon to find as many as three second-position clitics within a single clause. The relative order of the clitics within this second-position cluster is, for the most part, fairly rigidly determined. This ordering can be described in terms of six position classes, as summarized in Table 3.1. The membership of each position class is determined on the basis of relative order and co-occurrence restrictions. In general, a single clause may contain at most one element from any particular class. In other words, particles assigned to the same class cannot (in general) co-occur with each other, and when two particles assigned diferent classes co-occur, they will occur in the order specified in this template. The first three classes, namely, gen, nom, and focus-aspect, obligatorily occupy the 2P clitic position. The last three classes (mood, evaluative, and solidarity) may optionally occur in clause-final position, though this seems to be relatively rare for the mood particles gaam and ma. When there are more than three particles in the same clause that could all appear in the 2P clitic position, one or more of the optional 2P particles usually appear clause-finally. Thus, clitic clusters containing more than three particles are generally avoided. 3.3

Status particles

Position class 5, containing what I have called the “evaluative” particles, is semantically somewhat heterogeneous.5 It includes one particle which does afect the truth-conditional meaning of the proposition, namely, dara “frustrative.” The meaning of the frustrative particle is discussed in Kroeger (2017). The other particles in class 5 appear to be purely use-conditional, and I refer to them as status particles. Most of these particles seem to have a range of functions, and their precise meaning depends very much on the specific context in which they are used, as is the case with the German modalpartikeln. Some preliminary examples illustrating typical usage of the more common members of this set are presented in (6). (6) a. Darun dati’ . . . “It will probably rain (this afternoon).” b. Darun katoy! “It did too rain (contrary to what you claim).” c. Ki-darun bala’ kosodoy! “Oh look, it rained last night (and I didn’t know it)!”

72

Paul R. Kroeger d. . . . ki-darun gima.

“(I didn’t go to your house because . . .) it was raining, after all/as you know.” e. . . . koo-dorun-an mari. “(At this time of year) it rains a lot (certainly/ generally).” I tentatively identify dati’ “probably,” toomod “probably,” and mari “certainly” as validational markers, expressing the strength of the speaker’s commitment to the truth of the current proposition, rather than markers of modality in the narrow sense. Mari is often used to indicate knowledge shared by the whole community, or certainty based on prior knowledge of someone’s characteristic properties or behavior, but these particles have not been investigated in detail, and I will not have much to say about them here. Further examples illustrating core uses of gima, bala’ and katoy are presented in (7)–(9). (7)

Isos-on nu gima banar ino mato nu, sagay aragang no. rub-ov 2sg gima really that eye 2sg.gen reason red iam “After all, you keep rubbing your eye hard, that is why it is all red.”

(8)

Wiwidsing-o ku it=rangalaw nga’ napapasa=i’ dup.peel-atemp.ov 1sg nom=rambutan but pst.dup.rotten=emph bala’ iri. mir this “I peeled the rambutan, but (I discovered) it was rotten.”

(9)

Yalo katoy ot minanakaw, okon.ko yoku po. 3sg.nom katoy nom pst.av.steal not 3sg.emph foc “It was him that stole it, not me (contrary to your assertion).”

The status particles are a common feature of conversational speech but, generally speaking, do not occur in narrative monologue, apart from direct quotations. (The one exception is the mirative particle bala’, which can occur in narratives with a shift in perspective to indicate surprise on the part of some central participant.) This restriction of usage has a grammatical correlate. “Main-line” events in narrative are expressed using the atemporal form of the verb (Kroeger 1991).6 This atemporal form functions as a “narrative tense” and serves as a genre marker to distinguish the act of telling a story from simply describing an event or sequence of events. Most of the status particles seem to be strongly dispreferred in clauses inflected for narrative tense, as illustrated in (10). The exception is katoy, which can occur in such clauses in its expressive function, marking disapproval. (10) Piniutap/?*piutapo’ dot=manuk om sada mangarakan bala’ iti. mix.ov.pst/atemp acc=chicken and fish av.boil prtcl this “He/she boiled fish and chicken together (I am surprised to learn).”

A Kimaragang status particle 4

73

Use-conditional rather than truth-conditional meaning

As noted in the introduction, the meaning contributed by gima does not seem to be part of the “at issue,” truth-conditional content of the sentence. One reason for making this claim is the fact that the particle cannot be questioned or negated. In fact, this seems to be true for all of the Kimaragang status particles. They do not seem to occur in questions at all, not even rhetorical questions.7 They may occur in another clause of a sentence that contains a question (11a), but not within the interrogative clause itself (11b): (11) a. Siongo mat kisakot ilo’ togilay yo dot8 where rq grassy that corn 3sg comp pigamasan yo gima. clear.repeatedly 3sg gima “How could there be grass growing in his cornfield, when (as you know) he always clears/cuts the grass there?” b. ?*Siongo mat kisakot ilo’ togilay yo gima? Moreover, status particles always take scope over clausal negation. Examples (12)–(13) illustrate this for gima. (12) Amu gima owo sinuput nu i=paip noputut, not gima prtcl pst.connect.ov 2sg nom=pipe broken intaay aso weeg tokow. look.dv.imp not.exist water 1pl.incl “You didn’t fix the broken pipe, as you well know; now look, we don’t have any water.” (cannot mean: “It is not known to you that you fixed the broken pipe . . .”) nu ino kuuy, ino (13) Amu gima notongkuban 2sg that cake that not gima pst.cover.dv bala’ ot=kororogis dino. mir nom=dup.reason.sandy that “You failed to cover the cakes, as you well know, and that is why they got all sandy.” (cannot mean: “It is not known to you that you covered the cakes . . .”) A second reason for analyzing gima as use-conditional rather than truthconditional in nature is that the presence of gima cannot form the basis for challenging the truth of a statement. Example (14) illustrates an appropriate challenge based on the truth-conditional content of a statement. Example (15, B1) shows that lack of familiarity or accessibility is not sufcient grounds for challenging the truth of a statement which contains gima. An acceptable way of challenging the appropriateness of gima in a particular context is illustrated in (15, B2).

74

Paul R. Kroeger

(14) A: Yokoy diti, musikin okoy, aso 1pl.excl this poor 1pl.excl neg.exist ya. 1pl.excl.gen “As for us, we are poor, we have no wealth.” B: Momudut katoy, amu babanar; akaya yalo’ av.lie katoy neg dup.true rich 3sg “He is lying, that is not true; he is rich.”

tarata property

dilo’. that

(15) A: Pi-ilang-o yoalo tu’ sompusasawo recp-eat.together-ov.imp 3pl.nom because married.couple gima.9 gima “Have them eat together, because they are husband and wife after all.” B1: #Momudut katoy, amu babanar; a=ku nela’an! av.lie katoy neg dup.true neg=1sg knew “You are lying, that is not true; I did not know that.” (odd in this context) B2: Ay? A=ku nela’an! prtcl neg=1sg knew “Oh? I did not know that.” (appropriate response) 5

Contexts where gima cannot be used

In Section 1, I proposed that gima indicates that the speaker (a) believes that p is known or knowable by the addressee and (b) takes the truth of the base proposition for granted, not controversial or debatable. Kratzer and Matthewson (2009 ms.) have suggested that only the second of these components is part of the conventional meaning of unstressed ja, but both seem to be necessary to account for the behavior of gima. Consider the following example: (16) (Situation: B is attending a meeting in the state capital. There he meets A for the first time. A has no prior knowledge about B’s family.) A: Ki-sawo ko=no oy? exist-spouse 2sg=iam q “Are you married?” (lit: “Do you have a spouse already?”) B: Waro, nga’ amu yalo nakawaya. exist but neg 3sg accompanied Okodok po (#gima) it=tanak ya. small yet gima nom=child 1pl.excl “Yes, but she didn’t come with me. Our children are still small.”10 Clearly, B takes the content of his utterance to be uncontroversial, but because he cannot reasonably expect the information to be accessible to A, the use of gima in his response would be infelicitous.

A Kimaragang status particle

75

Unstressed ja may be diferent from gima in this regard. Kratzer and Matthewson state: “German ja can also be used felicitously in cases where the information given is unlikely to be known to the addressee, or even known to be unknown to the addressee.”11 They provide an example which is very similar to (16), in which ja is felicitous. (17) Context: You are talking to a new colleague who doesn’t know you and talking about how it’s hard to go on holiday. You say: “We always have to find someone who takes care of our animals.” Wir haben ja zwei Katzen. “We have ja two cats.” In other respects, however, the limitations on gima are very similar to those reported for ja. For example, a number of authors have observed that unstressed ja occurs only in declarative clauses, and the same is true for gima.12 And as we would expect, gima is infelicitous in most contexts if the addressee does not have prior knowledge of the relevant facts. The particle would be unnatural in (18) if the addressee does not already know who cleaned the fish, and in (19) if the addressee does not already know that the person in question was drunk at topic time. (18) Ololonsi no iti tunturu ku, yoku gima o=minonobuk dup.stink iam this finger 1sg 1sg(emph) gima nom=av.pst.stab di=sada. acc=fish “My fingers stink, (because) I was the one after all who cleaned the fish.” (19) Songkoboroso dialo dot asot tatantu, owukan gima. speak.wildly 3sg comp neg.exist dup.certain drunk gima “He was saying crazy things, after all he was drunk.” Zimmermann (2011) lists the following specific types of contexts which are incompatible with the use of unstressed ja, and these same restrictions apply to gima as well, as illustrated in Sections 5.1–5.3: In contrast, ja is illicit whenever the truth of the propositional content of an utterance is not known to be shared by the addressee, or even known to be controversial. This is typically the case in breaking news . . ., in answers to questions, which denote a set of controversial alternatives to be resolved by the addressee . . ., or in corrections of previous assertions. 5.1

Breaking news/out-of-the-blue statements

When a speaker conveys new information which the addressee would have no way of knowing, especially when that information is unexpected, the particle gima cannot felicitously be used. A striking example of this type, in which the

76 Paul R. Kroeger news comes literally “out of the blue,” is found in the beginning of St. Luke’s gospel, when the angel Gabriel appears to the Virgin Mary with some unexpected news. As (20b) shows, the use of gima in this context would be highly unnatural. (20) a. Monon-tiyan ko nôono dino om monusu dot kusay av.wear-stomach 2sg prtcl that and av.give.birth comp male ot=tanak. nom=child “You will become pregnant and give birth to a son.” (Luke 1:31) b) #Monontiyan ko gima om monusu dot kusay ot tanak. (impossible in this context) 5.2

Answering a question

We have said that the use of gima indicates the speaker’s belief that the base proposition is known or knowable by the addressee and is not controversial (i.e., the truth of the proposition is not open for discussion). If the base proposition is presented as the answer to a question which the addressee has just asked, then the information is normally not known by the addressee and the truth of that proposition is, in fact, the current issue under discussion. Under these circumstances, the use of gima would again be highly unnatural, as illustrated in (21). (21) Q: Nunu ot=tonomon daalo ad gopu yo dilo’? what nom=plant.ov 3pl in garden 3gen that.dist “What will they plant in their garden plot?” A. Togilay dati’/mari’ /#gima ot=tonomon daalo. maize probably/certainly/gima nom=plant.ov 3pl “Probably/naturally/#as you know they will plant corn/maize.” 5.3

Contradictions of previous assertions

As discussed later (Section 6.2), gima can be used to highlight accessible information which is relevant to current purposes but seems to be ignored or overlooked by the addressee or some other salient person. However, gima is not appropriate when the speaker directly contradicts something that has just been stated. In such contexts, the truth of the proposition is very much under discussion, and katoy would be used instead of gima. In (22), for example, if the speaker has just been told that he was accused of stealing by the owner of the coconuts, katoy would be appropriate, but gima would not.13 A similar example is seen in (23), where the owner of a certain chicken is reported to have claimed (mistakenly or falsely) that he bought it.

A Kimaragang status particle

77

(22) Yalo katoy ot minangangat dogon manganu di=niyuw 3sg prtcl nom av.pst.invite 1sg av.take nom=coconut doalo. 3pl “He was the one who invited me to take some of their coconuts.” (23) A: Minomoros i=Jim dot “At=manuk dilo’ av.pst.say nom=Jim comp nom=chicken that.dist binoli ku,” ka. bought 1sg quot “Jim said, ‘I bought that chicken.’” B: Doo maantad do manuk ilo’, okon.ko’ 3sg.dat originally lnk chicken that neg binoli yo katoy/*gima. bought 3sg prtcl “It was his chicken in the first place; he didn’t buy it (contrary to what he now claims).” 6

Uses of gima

Grice explained why we do not normally tell people what they already know: it would be uninformative and thus a violation of the maxim of quantity. Gima typically functions as a quantity hedge, like the English phrase after all (Levinson 1983: 162): a signal to the hearer that the current utterance may not be informative. This function is illustrated in examples (24)–(25), in which the clause containing gima expresses information which must clearly be known to the addressee at the time of speaking. (24) Gumu nu gima monorimo, orubat nopo ami=i’ much 2sg gima cook.rice waste only neg=emph naawi mangakan. finished av.eat “After all, you cooked a lot of rice; it is a shame that it didn’t all get eaten.” (25) Subay.ko ipag-on nu yalo dilo’ tu’ should brother.in.law-ov 2sg.gen 3sg.nom that because tobpinee di=sawo nu yalo gima. sibling gen=spouse 2sg.gen 3sg.nom gima “You should/must call him ipag (‘brother-in-law’), because after all, he is your wife’s brother.” Like unstressed ja, gima is expected when the speaker is stating something that is already obvious (Kaufmann & Kaufmann 2012). Example (26) comes

78

Paul R. Kroeger

again from the gospel of Luke. In this context, Peter’s statement would be unnatural without the particle gima. (26) (Context: Jesus is walking along a road surrounded by a large crowd of people.) Jesus: “Ay, isay minonudu dogon?” hey who av.pst.touch 1sg “Hey, who touched me?” Peter: “Guru, ogumu gima o tongo tulun kotobping om teacher many gima nom pl person bump and muurudun id dikaw.” crowd/jostle loc 2sg “Rabbi, there are many people crowding and bumping you.” (Luke 8:45) Even when a quantity hedge is used, the assertion of information which is already available to the addressee is generally somewhat odd, apart from special motivating circumstances. In my data, the most commonly attested types of circumstances which license such statements, and thus uses of gima, seem to belong to one of the following types. First, gima frequently occurs in reason clauses. In this construction, the content of the reason clause itself may already be known, but the assertion of a causal relation between the two clauses could still be informative. A second common use of gima is for reminding the addressee of information which is already accessible but which the addressee seems to have forgotten or ignored, as in example (26). Third, gima frequently occurs with certain kinds of expressive meaning, in particular with expressions of surprise, scolding, and certain types of exclamatory utterance. In these cases, the truth-conditional, at-issue content of the utterance may be known or accessible to the addressee, but the expressive content may be new. On the other hand, expressive content does not seem to be governed by the maxim of quantity in the same way as descriptive content; speakers all too often express their feelings even when this information is well known to the addressee.

6.1

Reason clauses

Examples (18) and (19) illustrated the use of gima in unmarked reason clauses, which are simply juxtaposed to the main clause. Another such example is presented in (27). (27) Nopuunan ko bo dino, winajak nu gima hexed 2sg prtcl that pst.spoke.clearly.ov 2sg gima momoros yalo dot pangansakon. av.speak 3sg comp cause.to.cook.ov

A Kimaragang status particle

79

“You have brought a hex on yourself; after all, you asked her directly to cook food for you.” More often, however, reason clauses are marked with the conjunction tu’ “because,” as seen in examples (15-A) and (25). Further examples of this type are presented in (28)–(31). As noted earlier, gima appearing in the “because” clause indicates that the reason is shared or accessible information. Example (28), for example, would be unnatural if the addressee does not already know that the deceased woman was a priestess (shaman). (28) Imboluan yalo’ dilo’ tu’ boboliyan gima. toll.gong.dv 3sg.nom that because priestess gima “They will toll the funeral gong for her, because after all she was a priestess.” (29) Isot babatang nga’ a=ku elaan tu’ a=ku one dup.letter also neg=1sg know because neg=1sg nokosikul gima owo. av.pst.nvol.attend.school gima prtcl “I don’t know even one letter, because after all I never went to school.” (30) Munaru po yalo dilo’ tu’ omulok po gima. grow.longer yet 3sg that because young yet gima “He/she will grow taller, because after all he/she is still young.” (31) Amu needu bâanar iti bongkuris diti wagas tu’ neg removed truly this rice.hull this uncooked.rice because tinutu gima. pst.pound.ov gima “The hulls were not completely removed from this rice, because after all it was pounded (in wooden mortar and pestle, rather than being milled).” Another way of marking causal relations is with the conjunction sagay “reason.” This conjunction is used to introduce clauses expressing a result, with gima frequently occurring in the reason clause, as illustrated in (32). (32) Sagay nelaan ku ot=wayaan mongoy sid=Kudat, reason pst.know.dv 1sg.gen nom=way av.go dat=Kudat babaya nokoongoy oku gima. dup.previously av.pst.nvol.go 1sg.nom gima “The reason I knew the way to Kudat was because I’ve been there before, after all.”

80

Paul R. Kroeger

6.2

Correction

One way in which mutually accessible information might be worthy of mention is if the addressee (or some other salient person) has failed to access that information when it would be relevant to current purposes. In German, the particle doch would be used in such contexts, but since there is no equivalent to doch in Kimaragang, gima is sometimes used here as well. (33) Kukuro yoalo’ misasawo, miobpipinee gima. how 3pl.nom recp.spouse recp.dup.sibling gima “How can they marry each other, after all, they are siblings.” (34) Siongo mat kisakot ilo’ togilay yo dot pigamasan where rq grassy that corn 3sg comp clear.repeatedly yo gima. 3sg gima “How could there be grass growing in his cornfield, when after all he always clears/cuts the grass there?” However, gima is not used to directly contradict something that has just been stated. As illustrated in (22)–(23) earlier, only katoy and not gima can appropriately be used for this purpose. 6.3

Surprise

Another situation in which information that is already mutually accessible might be considered newsworthy is if the information has been newly discovered by the speaker. Examples (35)–(37) involve information which is known to the hearer but new and surprising to the speaker. However, this new information is treated as being accessible to the speaker because it is observable in the immediate speech context. German unstressed ja can also be used in contexts of this type, for example, “Oh, you have ja green eyes” (noticed for the first time; Grosz 2014). (35) Kawantang no diri ilot tanak nu momoros gima fluent iam this that child 2sg av.speak gima dot okodok po om.14 comp small yet and “Your child can already speak clearly/well, even though it is still small!” (36) Nakaganaru ko=no dîiri gima. grew.longer 2sg=iam this gima “You have gotten taller (since I last saw you)!” (37) Sabat po om a=ku notutunan ika, little yet and neg=1sg pst.recognize.dv 2sg.nom

A Kimaragang status particle

81

orurungut ko=no dino bongit gima. dup.overgrown 2sg.nom=iam that beard gima “I almost didn’t recognize you; your beard has gotten so long and shaggy.” (lit: “You have been overgrown by that beard.”) In these contexts, gima may be interchangeable with the mirative particle bala’. In other similar contexts, either particle may be possible, but with a subtle diference in meaning between the two. In examples (38)–(39), bala’ merely indicates that the information is newly acquired by the speaker, while gima is described as indicating that the information is contrary to prior expectations. I do not have an explanation for this diference. (38) a. Nokosogurut no dîiri a=paray duyu gima. became.lush iam this nom=rice.plant 2pl gima “Your rice plants have grown well (contrary to my expectation).” b. Nokosogurut no dîiri a=paray duyu bala’. became.lush iam this nom=rice.plant 2pl mir “(I see that) your rice plants have grown well (which I didn’t know before).” (39) a. Ririnumangkama=i’ do gima i=kangkung tinanom ku.15 dup.pst.creep=emph lnk gima nom=kangkung planted 1sg “The kangkung (swamp spinach) that I planted has crept/spread out (contrary to my expectation).” b. Ririnumangkama=i’ bala’ i=kangkung tinanom ku. dup.pst.creep=emph mir nom=kangkung planted 1sg “The kangkung (swamp spinach) that I planted has crept/spread out (I see now; previously unaware of this development).” This surprise use of gima, as with German ja, is only possible when the information being asserted is observable in the immediate utterance context. For this reason, gima cannot be used for describing past discoveries on the part of the speaker; only bala’ is possible for such statements, as illustrated in (40)– (41). This restriction follows from the meaning of gima: if the information being reported as a new discovery is not observable in the utterance context, it cannot be assumed to be accessible to the addressee. (40) a. Powurilongo ku it=takod ku sid=luwang nga’ put.into.hole 1sg nom=foot 1sg dat=hole but aralom bala’ iri. deep mir this “I stuck my foot into the hole, and it turned out to be deep.” b. ?*Powurilongo ku it takod ku sid luwang nga’ aralom gima iri.

82

Paul R. Kroeger

(41) a. Tantaman ku sompusasawo yoalo, miobpipinee bala’ay. thought 1sg married.couple 3pl dup.recp.sibling mir “I thought they were husband and wife, but they turned out to be siblings.” b. ?*Tantaman ku sompusasawo yoalo, miobpipinee gima. In addition to marking the information status of the current proposition in these examples, both gima and bala’ seem (impressionistically) to convey expressive meaning as well. Malay translations of such sentences frequently include the adverbial particle pula “also,” which in this context seems to have a purely expressive function, indicating surprise. 6.4

Scolding

Yet another reason for stating information that is already known to the addressee is to express displeasure with something the addressee has done. Gima is frequently used in scolding and complaints about the behavior of the addressee, as seen in examples (12)–(13) earlier. Additional examples are presented in (42)–(45). (42) Unanawon ku=no itit paray, monuu ko=po gima. dup.crush.ov 1sg=iam this rice av.order 2sg=yet gima “Here I am already crushing the rice seed (e.g., to feed chickens), and you still tell me to do it gima!” (43) Monigagang ko gima, sodoy om muli ko nogi. av.frighten 2sg gima night and av.return 2sg only.then “You frightened/worried me gima, coming home so late at night!” (44) Osorulakan nu manganit ino kulit do=kayu gima. backwards 2sg av.peel that skin gen=tree gima “You peeled that bark of against the grain gima!” (45) Ad=susut gima ot=pinangalaasan loc=space.below.house gima nom=place.of.splitting nu dino suduwon, 2sg that fire.wood intaay pogi nakawawantuk no=ino kapak look.imper prtcl dup.snagged iam=that axe dilot tontom. that floor.joist “You chose to split the firewood under the house gima; now look, the axe has caught on the floor joist!” The use of gima for complaints about the behavior of the addressee is consistent with its status-marking function, since this information is presumably

A Kimaragang status particle

83

already known by the addressee. But in scolding and complaints, as in expressions of surprise, gima seems to contribute not only the presumption of accessibility but also some afective or expressive content, expressing the speaker’s anger or frustration.16 In the next section, we will consider expressive functions of gima in more detail. 7

Expressive meaning and exclamatory use

Gima frequently occurs in exclamatory statements, that is, declarative sentences which not only assert a proposition but also express the speaker’s feelings or attitude toward that proposition.17 Intonation plays an important role in distinguishing exclamatory statements from other declarative sentences, but exclamations can also be identified by the presence of certain sentence-initial expressive particles, as in (46)–(47), or other formulaic elements. (46) Woy obo, nakaabir at=takanon, osongow prtcl prtcl scattered nom=cooked.rice rough ko gima monook! 2sg gima av.scoop “Now look what happened! The rice is scattered all over because you scooped it out so roughly/carelessly!” (47) Woy obo oleed om nitutup nu nogi’ gima it=tuunuson, prtcl prtcl long.time and closed 2sg then gima nom=gate nokosuwang it=karabaw doalo sid=paray tokow. have.entered nom=bufalo 3pl dat=rice.plant 1pl.incl “Now look what happened! You were too slow in closing the gate, and now their bufalo have gotten into our rice field!” The expressive particle woy by itself generally conveys the sense of “What did I tell you?” or “I told you so.” Obo18 can occur by itself as an interjection of surprise, but the sequence of particles observed in (46)–(47) seems to be an exclamation formula meaning something like “Now look! What do you expect?” and often rendered in Malay translations as Itu=lah! “that=foc.” Another formulaic marker of exclamations was seen in (12) and (45) earlier, intaay (pogi) “just look!” which occurs frequently in scolding and complaints. An additional example is provided in (48). (48) Bibinuak nu gima i=weeg owo, intaay pogi dup.pst.waste.ov 2sg gima nom=water prtcl look foc asot pomoog da=pinggan. neg.exist wash.iv acc=plate “You wasted the water; now look, we don’t have any to wash the plates with!”

84 Paul R. Kroeger In addition to scolding, complaints, and expressions of surprise, gima appears in other types of exclamations as well. The exclamatory formula Sagay gima “No wonder!” typically rendered in Malay as Patut=lah! “appropriate=foc,” introduces exclamations about causal relations. Sagay by itself is used to introduce clauses expressing a result, as illustrated in (32) earlier. When the two words sagay and gima appear together, as seen in (49)–(52), they indicate exclamatory force. One indication of the formulaic nature of this combination is that the particle gima seems to occur in the “wrong” clause: in this construction, it marks the result rather than the reason. On the other hand, perhaps gima occurs here because the result is known (old information), whereas the reason is newly discovered. (49) Sagay.gima aso no=ot=weeg siti=id=dagay, no.wonder neg.exist iam=nom=water here=loc=1pl.excl nonus i=paip. pulled.out nom=pipe “No wonder we don’t have any water; the pipe has been pulled out!” (50) Sagay.gima dumarun nopo owo, no.wonder raining only prtcl urarangkadon dialo at=lobong da=tulun. dup.dig.up.ov 3sg nom=grave gen=person “No wonder it just keeps raining; he broke open/is breaking open someone’s grave!” (51) Sagay.gima nakalabus no i=sada owo, nayangat i=pangat. no.wonder has.escaped iam nom=fish prtcl bent nom=hook “No wonder the fish got away; the hook bent!” (52) Sagay.gima nokosuwang no=ilo’ sapi, amu nokoolit no.wonder has.entered iam=that cow neg restore ilo’ lalawangan. that gate “No wonder the cows got in; the gate did not get closed!” Exclamations are utterances that convey expressive meaning (frequently in addition to descriptive meaning). The fact that gima often occurs in such utterances suggests an association between gima and expressive meaning. A further indicator of this association comes from expressive reduplication, a pattern of partial reduplication in Kimaragang described by Kroeger and Johansson (2016). That paper illustrates a wide range of semantic functions associated with expressive reduplication and discusses a number of criteria by which expressive reduplication can be distinguished from aspectual reduplication.

A Kimaragang status particle

85

In a number of the examples presented earlier, gima is reinforced by the use of expressive reduplication: baya in (32), miobpinee in (33), orungut in (37), nakawantuk in (45), and binuak in (48). The reduplicated form batang “letter” is used in (29) to emphasize total illiteracy (“not one single letter!”), occurring in the same sentence as gima, but not in the same clause. The same is true for olonsi “stink” in (18) and tantu “certain” in (19).19 Since both gima and expressive reduplication are frequently observed in exclamations, it is not surprising that they should frequently co-occur. In some contexts, however, this co-occurrence seems to be obligatory (or at least strongly preferred). My primary informant stated that examples (53) and (54a) would be unnatural if the expressive reduplication is omitted but gima is retained.20 The simple declarative example (54b), in contrast, which contains neither gima nor expressive reduplication, is fully acceptable. (53) Amu gima si-sinobut dialo it=gopu yo neg gima dup-pst.visit.ov 3sg nom=field 3sg.gen sampay notowunan do=sakot i=togilay. until pst.nvol.cover.dv gen=grass nom=corn/maize “He never went to check on his field gima, so his corn got overgrown with grass.” (54) a. Ri-rinumangkama=i’ do gima i=kangkong dup-pst.av.creep=emph lnk gima nom=water.spinach tinanom ku. (=39a) pst.plant.ov 1sg.gen “The kangkung (water spinach) that I planted has spread out (I am surprised to see)!” b. Rinumangkama no i=kangkong tinanom ku. pst.av.creep iam nom=water.spinach pst.plant.ov 1sg.gen “The kangkung that I planted has spread out.” (neutral statement) Clearly, there is a strong association between certain uses of gima and certain kinds of expressive meaning, for example, surprise, annoyance, disapproval, etc. It may be possible to provide a purely pragmatic explanation for this association, or it may turn out that these expressive meanings have become grammaticalized or conventionalized to some extent. I leave this question as a topic for future research. 8

Conclusion

In some ways, it seems curious for a language to have a grammatical morpheme which indicates that the information being expressed is already available to the addressee, since this should be a somewhat abnormal kind of utterance.

86

Paul R. Kroeger

In fact, as noted by Zimmermann (2011) and Grosz (2016 ms.), such morphemes have been reported in a number of languages. The motivation for using such markers is summarized by Crone (2017: iv–v) as follows: It so happens that redundant utterances .  .  . are quite often explicitly marked as redundant. . . . The puzzle is why a speaker would ever explicitly mark an utterance as redundant, when this is unnecessary for achieving the speaker’s goals. It is argued here that speakers do so in order to ensure that their listeners are well-informed with respect to the speakers’ beliefs about their listeners. Put diferently, if I don’t tell you that I know you know, you might conclude that I don’t know you know. To ensure that you know that I know you know, I tell you that I know you know. Gima, like other status particles, helps speaker and hearer to manage the common ground by signaling the speaker’s awareness of the hearer’s knowledge. I have noted a number of similarities of usage between gima and German unstressed ja, but also some diferences. As Grosz (2016 ms.) points out, this is a common situation in comparing the discourse particles of one language with those of another language: Nevertheless, from a cross-linguistic perspective, the issue of the discourse particles’ individual contributions is precarious. While other closed-class items, such as modal auxiliaries, exhibit a certain degree of equivalence across unrelated languages, it appears to be rather difcult to establish one-to-one correspondences between a particle α in one language and a particle α′ in another language. . . . Nevertheless, tentative correspondences can be established. . . . Moreover, on a pretheoretic level, we observe that, in particular, the uncontroversiality component of ja . . . and the contrast component of doch . . . surface as “semantic atoms” in many languages (where the term “semantic atoms” informally refers to a part of the meaning contribution of an abstract functional element). For example, Wouk (1998: 403) describes the core sense of the Indonesian discourse particle kan as “presupposition of conjoint knowledge.” Like gima, kan is often used as a quantity hedge and indicates that the information being expressed is already accessible to the addressee.21 Clearly, the two particles share much of their core meaning. However, the range of uses is not identical. Kan can be used for several functions where gima is never used, for example, to extend the common ground by inviting the addressee to accommodate new information, or as a topic marker (activating information that is already in the common ground). Similarly, based on Wouk’s description, it appears that kan is not used in some contexts where gima frequently occurs, for example, to indicate the speaker’s surprise about something newly discovered and observable in the utterance context. The Mandarin sentence-final particle ma in declarative sentences also appears to have essentially the same core meaning as gima but again exhibits a slightly diferent range of uses (Chappell 1991; Chu 1998).

A Kimaragang status particle

87

Of course, part of the challenge in comparing such particles across languages is that the number of languages for which detailed information is available concerning the meanings and functions of these particles is still relatively small. This case study is ofered as a modest contribution toward enriching the empirical basis for further investigation. Abbreviations 1 2 3 acc atemp av caus comp dat dist du dv exist excl emph exclm gen hab iam imper incl intr irr

first person second person third person accusative case atemporal actor voice causative complementizer dative case distal (demonstrative) dual dative voice existential exclusive emphatic particle exclamation genitive case habitual iamitive (“already”/perfect) imperative inclusive intransitive irrealis mode

iter iv lnk loc lv mir neg nmlz nom nvol ov pl pfv prtcl pst q quot recp red rel rq sg tr

iterative aspect instrumental voice linker locative case locative voice mirative negator nominalizer nominative case non-volitive objective voice plural perfective particle past interrogative quote marker reciprocal reduplication relativizer rhetorical question singular transitive

Notes 1 Acknowledgments: I wish to express my thanks to Jim Johansson for making his draft dictionary available to me; to Jim and Nelleke Johansson for sharing drafts of translated materials; and to Janama Lantubon, my primary language consultant. Part of the research for this study was supported by NEH-NSF Documenting Endangered Languages fellowship no. FN-50027–07. 2 Potts extends Grice’s term “conventional implicature” to cover essentially the same range of phenomena which Gutzmann identifies as use-conditional. 3 Zuferey describes these types of information as being “mutually manifest,” in the sense of Sperber and Wilson (1986). 4 Apparent exceptions include toomod, dara’ay, and bala’ay. Another exception is the emphatic particle nôono, which carries a pitch accent on the first syllable; its usage is illustrated in example (20). 5 Class 3 is also somewhat heterogeneous, but most of its members are polysemous between aspectual and focus-marking functions. 6 The atemporal form is also used for imperatives and for main verbs following an auxiliary. 7 The only exception I have found to this generalization is an example involving bala’ay, which seems to be an exclamatory form of the mirative particle. I was

88

Paul R. Kroeger told that bala’ay here indicates that the speaker has just realized there isn’t any viand. (i) Tongo bala’ay ot=rinapa? what mir nom=viand “What are we going to eat with our rice?”

8 The particle dot can function as a linker or relativizer within NP or as an all-purpose complementizer which introduces both complement and adjunct clauses. In this case, it introduces a circumstantial adverbial clause. 9 Of course, the main clause in this example, as an imperative, has no truth value to challenge. But the reason clause marked with tu’ is not subordinated and counts as a separate (but secondary) assertion. It is this secondary assertion which B’s reply is intended to challenge. 10 A reviewer points out that within Kimaragang culture, the fact that a married couple has children is probably predictable, or at least statistically probable. Shouldn’t this kind of inference make gima at least possible in this context? Two points can be made in this regard. First, the assertion marked with gima is not simply about the existence of children but about their ages, which is clearly not predictable. Second, in answer to the question “Are you married?” an afrmative reply probably does not, apart from some special circumstance, trigger a conversational implicature that the speaker has children, even in traditional Kimaragang culture. The mention of inference-based accessibility in Section 2 was specifically limited to conversational implicatures. 11 This statement does not seem entirely consistent with the generalizations of Zimmermann (2011) quoted directly after (19). I do not know whether there are special factors involved in (17) that make it acceptable. 12 For example, Grosz (2016 ms.) states: “At the matrix level, ja is restricted to declarative assertions and banned from interrogatives.” 13 In addition to marking a contradiction, katoy can also be used as an expressive particle to indicate disapproval. Perhaps both functions are intended in (22), and also in (14). 14 The coordinating conjunction om “and” normally occurs between the two conjoined clauses. Here it is used in sentence-final position as a concessive particle, “even though.” This is somewhat reminiscent of the current usage in spoken English of certain conjunctions as sentence-final particles, especially so. Some English speakers use sentence-final but in a similar way. 15 The phrase do gima seems to be interchangeable with gima in some contexts, as here, but not in others. The diferences between the two are not yet understood. 16 Wouk (1998) and Östman (1981) also argue that a quantity hedge may serve as a marker of information status (its primary function) and simultaneously express secondary stylistic or emphatic functions. 17 I follow Rett (2011) in distinguishing these declarative exclamations from exclamatives. An exclamative is formed from something other than a declarative sentence and does not count as an assertion of its propositional content, for example, “How very beautiful she was! Was he ever mad! The nerve of some people!” No investigation has been attempted as yet on exclamatives in Kimaragang. 18 The expressive particle obo occurs in initial position and is distinct from the 2P solidarity particle obo listed in Table 3.1. 19 Unawon in (42) and urangkadon in (50) are ambiguous between expressive reduplication and aspectual reduplication. Examples of expressive reduplication occurring with the mirative particle bala’ were seen in (8) (widsingo “peel” and napasa “rotten”) and (41a) (miobpinee “related as siblings”).

A Kimaragang status particle

89

20 My informant made a similar comment about binuak “wasted” in example (48), but he also stated that this root is rarely used without reduplication. Perhaps a reference to someone wasting something usually involves expressive as well as descriptive meaning. 21 I am ignoring here the other major use of kan, as a confirmation tag, which is distinguished by intonation and (fairly often) word order.

References Büring, Daniel. 2017. Ja Doch! In Clemens Mayr & Edwin Williams (eds.), Festschrift für Martin Prinzhorn. Wiener Linguistische Gazette 82. 23–35. Chappell, Hilary. 1991. Strategies for the assertion of obviousness and disagreement in Mandarin: A semantic study of the modal particle me. Australian Journal of Linguistics 11.1. 39–65. Chu, Chauncey. 1998. A Discourse Grammar of Mandarin Chinese. New York & Bern: Peter Lang Publishing. Crone, Philip. 2017. Redundancy and Awareness in Discourse. Doctoral dissertation, Stanford University. Grosz, Patrick G. 2014. German doch: An element that triggers a contrast presupposition. Proceedings of the Chicago Linguistic Society 46. 163–177. Grosz, Patrick G. 2016 ms. Discourse particles. To appear in Lisa Matthewson, Cécile Meier, Hotze Rullmann & Thomas Ede Zimmermann (eds.), The Companion to Semantics. Oxford: Wiley. Gutzmann, Daniel. 2015. Use-Conditional Meaning: Studies in Multidimensional Semantics. Oxford: Oxford University Press. Kaufmann, Magdalena & Stefan Kaufmann. 2012. Epistemic particles and performativity. Proceedings of SALT 22. 208–225. Kratzer, Angelika & Lisa Matthewson. 2009. Anatomy of two discourse particles. In Elizabeth Bogal-Allbritten (ed.), Proceedings of SULA 6: The Semantics of UnderRepresented Languages in the Americas, 1–16. Amherst, MA: GLSA. Kroeger, Paul. 1991. The event line in Kimaragang narrative. In Stephen Levinsohn (ed.), Thematic Continuity and Development in Languages of Sabah, 93–104. Canberra: Pacific Linguistics C-118. Kroeger, Paul. 2017. Frustration, culmination, and inertia in Kimaragang grammar. Glossa: A Journal of General Linguistics 2(1). 56, 1–29. Kroeger, Paul & Jim Johansson. 2016. Descriptive vs. expressive reduplication in Kimaragang. In Hiroki Nomoto, Takuya Miyauchi & Asako Shiohara (eds.), AFLA 23: The Proceedings of the 23rd Meeting of the Austronesian Formal Linguistics Association, 1–19. Asia-Pacific Linguistics 34. Canberra: Australian National University. Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press. McCready, E. 2010. Varieties of conventional implicature. Semantics and Pragmatics 3(8). 1–57. Östman, Jan. 1981. ‘You Know’: A Discourse-Functional Approach. Amsterdam: John Benjamins. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford: Oxford University Press. Potts, Christopher. 2007. Conventional implicatures, a distinguished class of meanings. In Gillian Ramchand & Charles Reiss (eds.), The Oxford Handbook of Linguistic Interfaces. Studies in Theoretical Linguistics, 475–501. Oxford: Oxford University Press.

90

Paul R. Kroeger

Repp, Sophie. 2013. Common ground management: Modal particles, illocutionary negation and verum. In Daniel Gutzmann & Hans-Martin Gärtner (eds.), Beyond Expressives: Explorations in Use-Conditional Meaning, 231–274. Leiden: Brill. Rett, Jessica. 2011. Exclamatives, degrees and speech acts. Linguistics and Philosophy 34(5). 411–442. Wouk, Fay. 1998. Solidarity in Indonesian conversation: The discourse marker kan. Multilingua 17. 379–406. Zimmermann, Malte. 2011. Discourse particles. In Paul Portner, Claudia Maienborn & Klaus von Heusinger (eds.), Handbook of Semantics/Handbücher zur Sprachund Kommunikationswissenschaft, 2012–2038. Berlin: Mouton de Gruyter. Zuferey, Sandrine. 2014. Givenness, procedural meaning and connectives: The case of French puisque. Journal of Pragmatics 62(1). 121–135.

4

A syntactic universal in a contact language The story of Singlish already Michael Yoshitaka Erlewine

1

Introduction1

The colloquial lingua franca of Singapore, Singlish, is a contact language born of British English together with other languages of colonial Singapore, including a number of Chinese languages and Malay.2 Singlish has long been recognized as exhibiting various grammatical and lexical diferences from familiar “Standard” Englishes. One such feature is the sentence-final use of already exemplified in (1). As has been noted previously, especially by Bao (1995, 2005a), the semantics of Singlish sentence-final already difers from the meaning of the adverb already of “Standard” Englishes. This is reflected in the translation in (1). (1) The wall white already. “The wall turned white.”

(Bao 2005a: 239)

This chapter concerns the fine-grained syntax and semantics of Singlish sentence-final already and how this lexical item came to be. Bao (2005a) has proposed that already is the result of a process of relexification (Muysken 1981; Lefebvre 1998), combining the syntax and semantics of the perfective verbal sufx -le and sentence-final particle (SFP) le from Mandarin Chinese with the English phonological form “already.” I will adopt Bao’s hypothesis that already is the result of a process of relexification, but will argue for a somewhat-simpler history: the SFP le – or, more likely, its cognates liau/laa in southern Chinese languages – is the sole substrate source for already. Building on recent work on the semantics of Mandarin SFP le (in particular, Soh & Gao 2006, 2008, and Soh 2009), I propose a uniform semantics for already which derives its distinct interpretational efects from the telicity of event description it modifies. This relexification story leads us to expect the syntax of Singlish already to mirror the syntax of its cognate SFPs in substrate Chinese languages, both in linear and hierarchical position. Although Singlish already has indeed inherited the sentence-final linear position of its Chinese cognate SFPs le/liau/laa, evidence from semantic scope will show that they difer in their hierarchical DOI: 10.4324/9781351057752-5

92

Michael Yoshitaka Erlewine

position. Singlish already unambiguously attaches high, scoping over the entire clause, whereas its substrate cognate SFPs unambiguously attach to a clause-medial position:3 (2) a.

Singlish already: TP

b.

Chinese cognate SFPs: TP

already subject

subject

T T

VP

VP

le/liau/laa

What accounts for this diference in the hierarchical position of Singlish already vs. its cognate Chinese SFPs? This problem comes into sharper focus when we consider the acquisition of these items. Singlish already and its cognate particles in Chinese languages are essentially identical in their linear position in the input that children are exposed to. Why do Singlish-acquiring children uniformly target a high structural position for already, despite the same semantics in Chinese languages being associated with a lower structural position? I propose that the solution to this puzzle comes from the consideration of the Final-over-Final Constraint (FOFC), a proposed universal on structure-building and linearization (Holmberg 2000; Biberauer et al. 2008, 2014; Biberauer et al. 2009; Sheehan et al. 2017; a.o.). As head-final heads in an otherwise head-initial clausal spine, the position of already and le/ liau/laa will be severely constrained by FOFC. In particular, in Erlewine (2017), I proposed that such SFPs will be limited to the edges of SpellOut domains. In Chinese languages, which lack verbal inflection, the lower phase of the clause (approximately VP) can undergo Spell-Out, allowing for the low attachment of SFPs at this boundary, as in (2b). In contrast, Singlish exhibits verbal inflection, reflecting a morphological dependency between T and V. This dependency will suspend the independent SpellOut of the lower phase (VP), forcing already to occupy a higher hierarchical position, as in (2a). I begin in Section 2 with a brief introduction to the theory of relexification and its application to the development of Singlish. In Section 3, I propose a uniform semantics for sentence-final already and argue for my revision to the Bao (2005a) relexification theory for already. I then discuss the hierarchical position of already and its Chinese cognate SFPs in Sections 4 and 5, establishing the contrast in (2). I then discuss the Final-over-Final Constraint and its efect on the development of Singlish already in Section 6. 2

The grammar of Singlish and relexification

Scholars have debated the status of Singlish within the typology of contact languages, for example, as an Outer Circle English variety with contact-induced grammatical changes, or as an English-based “creoloid” (Platt 1975).

A syntactic universal in a contact language

93

Regardless, it is widely recognized that the grammar of Singlish is distinct from that of so-called “standard” Englishes. Consider, for example, the Singlish expression in (3). Although this sentence is made up of recognizable English lexical forms, its grammatical structure and interpretation take advantage of a number of grammatical features absent from many standard Englishes. (3) Cannot also must can. “Even if (you think/say) you cannot, you have to be able to.” Immediately recognizable from this sentence is the fact that Singlish is a pro-drop language (Tan 2007; Sato 2011)4 with bare conditionals (Bao 2005b). Recognition of these features helps us translate (3) into a standard English variety as “If you cannot, you also must be able to,” but this does not fully convey the correct semantics of (3). To fill this gap, we must additionally recognize that Singlish also with a leftward focus (here, focus on cannot) has a scalar, “even” like interpretation rather than an additive one (Quek & Hirsch 2017). It also helps to know that must in Singlish has a predominantly deontic interpretation (Bao 2010), perhaps due to competition with the prevalent use of confirm as an epistemic necessity modal (Kang 2016). The grammar of Singlish, then, difers from standard Englishes as a product of the complex contact situation in the history of colonial and modern Singapore (see, for example, Leimgruber 2013). The question now is how such grammatical features developed in the language. Here I will introduce one such theory – relexification, as pioneered by work such as Muysken (1981) and Lefebvre (1998) – as it is invoked in the most sophisticated previous description of Singlish already (Bao 2005a).5 Relexification has been proposed as a model for creole genesis whereby comparable lexical items from substrate and superstrate lexicons are recombined in order to produce the lexicon of the resulting contact language. More specifically, this process pairs the syntactic and semantic specifications of a lexical item in the substrate language with a corresponding phonological form from the superstrate language, as schematized in (4). (4) Relexification: Substrate: 2 4

SYN

SYNi

SEM

SEM i

PHON

PHON i

3

Result: 2

! 5 ! 4

SYN

(based on Muysken 1981:61; Lefebvre 1998:16; a.o.) Superstrate/lexifier: 3 2 3 SYN i

SEM

SEM i

PHON

PHON j

4

5 Ã

SYN

SEM j

SEM

SYN j

PHON

PHON j

5

An independent and compelling example of the applicability of relexification theory to the grammar of Singlish is found in Lee et al.’s (2009) study of Singlish invariant got. Singlish has inherited the English lexical verb get with

94

Michael Yoshitaka Erlewine

its past-tense form got, but it also has a number of other uses of the form got, where it does not covary with get. Lee et al. describe these uses of Singlish invariant gots, some of which are illustrated in the following. (5) Some Singlish invariant gots (Lee et al. 2009: 301–303): a. That cat got so many kittens. possessive “That cat has so many kittens.” b. Here got many nice houses. existential “There are many nice houses here.” c. You got play tennis (before)? habitual “Do you play tennis?”/“Did you used to play tennis?” d. Where got!? expression of surprise/incredulity; see also Koh (2018) Lee et al. observe that the range of got’s functions in (5) neatly parallels that of the existential/possessive verb u in Hokkien, a southern Chinese language which was one of the dominant languages of Singapore during the development of Singlish. Hokkien examples corresponding to the Singlish examples in (5) are reproduced in (6). (6) Some uses of Hokkien u (ibid.: 301–303): a. Hi jiak ngeow u ani zway ngeow knia. possessive that clf cat u int many cat dim “That cat has so many kittens.” b. Jit dao u jin zway swee e cu. existential this place u int many nice mod house “There are many nice houses here.” c. Li (yee zeing) u pha bang giu bo? habitual you before u hit tennis neg “Do you play tennis?”/“Did you used to play tennis?” d. Dolo/Mana6 u!? expression of surprise/incredulity where u Lee et al. therefore propose that Singlish invariant got is the result of a process of relexification. As the expression of possession is a primary function of Hokkien u, the English lexical form (has) got was identified as an appropriate corresponding English form. Relexification produced new lexical items which pair the syntactic and semantic functions of u in (6) with the invariant English form got, resulting in the modern Singlish constructions observed in (5). Such examples motivate relexification as one of the processes involved in the development of modern Singlish grammar, in particular by pairing English superstrate forms with Chinese substrate syntax/semantics. 3

The semantics and development of Singlish sentence-final already

I now turn to the description of Singlish sentence-final already. Singlish descriptively has two homophonous lexical items with the surface form

A syntactic universal in a contact language

95

already: one is an adverb with the same “earlier than expected” meaning as in Standard Englishes, and the other is the sentence-final form which concerns us here. The existence of these two forms can be observed in the following contrast. (7) Context 1: Huimin was supposed to come here tomorrow. a. She today already come. b. She today come already. (8) Context 2: Huimin was supposed to come here yesterday, but she didn’t. a. #She today already come. b. She today come already. In the (a) examples, already is in a pre-verbal position and must be interpreted as the familiar standard English adverb already. Just as in English, then, the felicity of this already is sensitive to whether or not Huimin’s arrival today counts as earlier than expected or not. In contrast, the sentence-final already in the (b) sentences is insensitive to this distinction and instead appears to express a diferent temporal relation. We are interested here in the semantics and development of this Singlish sentence-final already. The most extensive description of the uses and efects of Singlish sentencefinal already is in Bao (2005a). Bao observes that the contribution of already systematically difers based on the aspectual class of the predicate. With a stative predicate such as white in (9), the addition of already signals a change of state. With an eventive predicate such as wash my hand in (10), the addition of already signals completion of the event.7 (9)

The wall white already. (Bao 2005a: 239) “The wall turned white.”  being white just started

(10) I wash my hand already. “I washed my hands.”  washing hands just ended

(ibid.: 239)

There are also cases where the contribution of already appears to be ambiguous, as in (11). Bao (2005a) notes that these two interpretations of already in (11) correspond to the use of two diferent, homophonous morphemes le in Mandarin Chinese: the perfective verbal sufx -le in (12a) and the sentencefinal particle le in (12b). (11) It rain already. (ibid.: 241) a. “It rained/has rained.”  the raining has ended b. “It has started/is about to rain.”  the raining has begun (12) a. Xià-le yǔ . down-pfv rain “It rained/has rained.”

b. Xià yǔ le. (ibid.: 242)  down rain le “It has started/is about to rain.”

96

Michael Yoshitaka Erlewine

Bao therefore proposes that Singlish already is the result of a complex relexification pairing both Mandarin Chinese verbal -le and SFP le with the phonological form of the English adverb already. According to Bao, event completion readings as in (10) and (11a) reflect the perfective aspect semantics of Mandarin verbal -le (12a),8 whereas the change of state and event start readings in (9) and (11b) correspond to Mandarin SFP le (12b). An immediate complication for Bao’s complex relexification account is that Singlish already never surfaces in an immediately post-verbal position (see (13)), despite one of its substrate sources being the verbal sufx -le, which linearly intervenes between the verb and any post-verbal arguments, as in (12a). Bao proposes that the results of relexification are adjusted so that they “conform to the (surface) structural requirements of the lexical-source language” (p. 258). In this case, because English adverbs never intervene between a verb and object, the resulting Singlish already also avoids this position, even in cases where – according to Bao – its semantics is due to relexification of verbal -le. (13) *I wash already my hand. I argue that this relexification hypothesis for the development of Singlish already can be simplified, with the Mandarin Chinese SFP le – or, more likely, its cognate particles in southern Chinese languages of Singapore, such as Hokkien liau and Cantonese laa9 – being the sole substrate source for already. The key to this proposal is the observation that SFP le can itself also serve to ensure completion of an event – that is, the function of Singlish already that Bao attributes to the semantics of Mandarin verbal -le – if the event description is itself telic. It has been observed that “no predicate in [Mandarin Chinese] is inherently telic” (Sybesma 1997: 227) but that telicity can be introduced by the explicit encoding of an end state, as in verb compounds. For example, the verb compound “arrive-reach” in (14) together describes a telic accomplishment. The addition of the SFP le in this case enforces completion of the event at the reference time. (14) Wǒ men dào-dá shān-dǐng le. we arrive-reach mountain-top le “We have reached the top of the mountain.”

(Soh & Gao 2006: 107)

Here I propose to adopt the basic semantics for Mandarin SFP le developed in Soh and Gao (2006, 2008) and Soh (2009) and apply it for all uses of Singlish sentence-final already. The proposed semantics introduces a presupposition that the prejacent proposition did not hold at a prior time.10 (15) already (p ): a. asserts: p is true at the reference time R b. presupposes: p is false before the reference time R

A syntactic universal in a contact language

97

Let’s see how this uniform semantics for already derives the meanings previously described. First, consider the stative predicate white in (9). With already taking the proposition p = “that the wall is white” as its argument, the combination of the assertion of p and the presupposition that p was false before now entails that there was a change of state: the wall turned white. (16) already (state p ) in (9): p = the wall is white (state); R = now a. asserts: the state “the wall is white” is true now b. presupposes: the state “the wall is white” was false before now  the wall just started being white, that is, “The wall turned white (just now).” In contrast, suppose we take the predicate wash my hands in (10) to itself describe a telic accomplishment. The accomplishment p will be true at time t if its designated point has been reached by t. The use of already in (10) results in the expression of the recent completion of the event. (17) already (accomplishment p ) in (10): p = I wash my hands (accomplishment); R = now a. asserts: the accomplishment “I wash my hands” is true now b. presupposes: the accomplishment “I wash my hands” was false before now  the end state of “I wash my hands” was just met, that is, “I just finished washing my hands.” However, as I noted earlier, this same sentence may also mean “I wash my hands now (habitually) but didn’t used to.” This is predicted by the application of already to a habitual interpretation of I wash my hands, which is stative. (18) already (habitual state p ) in (10): p = I habitually wash my hands (state); R = now a.   asserts: the state “I habitually wash my hands” is true now b.   presupposes: the state “I habitually wash my hands” was false before now     the speaker habitually washes their hands now, but did not before. Similarly, the predicate rain in (11) may be naturally interpreted as either an atelic state description or as a telic (perfective) description.11 The application of already to these two interpretations of the underspecified rain in (18a, b) will result in an inchoative rain-starting reading (11a) or a completive rainending reading (11b), respectively. As the preceding examples with wash and rain make clear, Singlish bare verbs may allow for both telic and atelic interpretations with no change in

98

Michael Yoshitaka Erlewine

overt morphosyntax. This is in contrast to Mandarin and other Chinese languages, for which it has been claimed that all verbs are inherently atelic; see, for example, Sybesma (1997), quoted earlier. But Singlish also has verb forms with English inflections, such as -ing and -ed. Such forms serve to overtly specify the aspectual interpretation, leading to disambiguation of the efect of already. Compare the unambiguous and (20) to the ambiguous It rain already (11) earlier. (19) V-ing  atelic description  “inchoative already ” Outside raining already.12 a. *“It rained/has rained outside.” b. “It has started/is about to rain outside.” (20) V-ed  telic (perfective) description  “completive already ” It rained already. a. “It rained/has rained.” b. *“It has started/is about to rain.” My explanation for the correct unambiguous interpretations for already in (19)–(20) and its multiple interpretations in (11) with bare rain serve as an argument for the proposal here, where already has the uniform semantics in (15), in turn based on Soh and Gao’s proposal for the Mandarin SFP le. The prediction of this interaction with bare vs. inflected verb forms must be stipulated under the approach of Bao (2005a). Having established that Singlish sentence-final already has the uniform semantics in (15), we are now in a position to reconsider its development. Recall that Bao (2005a) proposes that already is the result of relexification based on two substrate morphemes: the Mandarin verbal sufx -le and SFP le. I argue instead that Mandarin SFP le – and/or its cognate particles in southern Chinese languages, liau and laa (see footnote 10) – is the single substrate source for the process of relexification which led to Singlish sentence-final already. As schematized in (21), relexification predicts a new lexical item in the Singlish lexicon which pairs the semantics of Chinese SFP le/liau/laa in (15) with the phonological form of the English adverb already. (21) The relexification of Singlish sentence-final already , modified from Bao (2005a) Substrate languages: Result: Hokkien, Cantonese, ... Singlish 2 3 2 3 SYN SFP SYN SFP ! 4 SEM (15) 5 ! 4 SEM (15) 5 PHON le/liau/laa PHON already Ã

Superstrate/lexifier: English 2 3 SYN adverb 4 SEM ¼ earlier... 5 PHON already

A syntactic universal in a contact language

99

Based on the theory of relexification developed by Muysken, Lefebvre, Lumsden (Lumsden 1999), and colleagues, the relexification process as in (21) also predicts the resulting lexical item in Singlish to inherit the syntactic specification of its source Chinese substrate SFPs. On the level of surface word order, this prediction is correct: like its substrate cognate particles, Singlish already is necessarily sentence-final, as observed in the contrasts in (7)–(8). In particular, it is ungrammatical between a verb and object (13), which complicated the relexification account of Bao (2005a) that takes the verbal sufx -le to also be a substrate sources for already. My proposal, which identifies the Chinese SFPs le/liao/laa to be the sole substrate source for already, immediately predicts this strict sentence-final position. We now have a concrete analysis for the semantics of Singlish already and how it came to be paired with its phonological form and linear position, based on the process of relexification schematized in (21) earlier. But the syntactic specification of a lexical item is not simply its linear position. A further question is that of the hierarchical position of Singlish sentence-final already. The remainder of this chapter will address this question. In the next section, I first consider the hierarchical position of the Chinese substrate cognate SFPs le/ liao/laa, and then I turn to the hierarchical position of Singlish already itself in Section 5. 4

The syntax of Chinese le/liau/laa

The syntax of Chinese SFPs, especially Mandarin Chinese le, has been the subject of significant previous research. We concentrate on two major questions in this section: the syntactic category of these SFPs and their hierarchical position. SFPs in Chinese languages are a closed class of grammaticalized “particles” defined by their strict sentence-final position. Common functions for SFPs are clause typing (e.g., in Mandarin, polar question ma, imperative ba), the expression of speaker attitude, or temporal marking. Here I will adopt the common view that these particles are head-final heads in the clausal spine. Given that Chinese clause structure is otherwise rigidly head-initial, Chinese SFPs have recently received broader theoretical interest as potential exceptions to the Final-over-Final Constraint, a proposed syntactic universal, which will be the topic of Section 6. For previous work on Chinese SFPs, see especially Lee (1986); Cheng (1991); Paul (2014); Pan and Paul (2015, 2016); and Erlewine (2017) on Mandarin and Tang (1998); Law (2002); and Sybesma and Li (2007) on Cantonese. A challenge for the study of Chinese SFPs by linguists – as well as their acquisition by children – is that their linear position is under-informative as to their syntactic position. For example, le/liau/laa could be attached to the entire clause (e.g., TP) or to a lower, clause-medial position (adjoined to VP or an extended projection of VP):

100

Michael Yoshitaka Erlewine

(22) Two hypotheses for the position of Chinese SFPs le/liau/laa a.

b. TP subject

le/liau/laa

TP subject T

T

VP

VP

le/liau/laa

As many Chinese SFPs encode clause type or speaker attitude, many scholars have assumed that all SFPs necessarily occupy a high position in the clause, as in (22a). In Erlewine (2017), I investigated this question through evidence from the semantic scope of SFPs. In particular, I argued that the previously identified “low” class of SFPs in Mandarin Chinese – including sentence-final le as well as the sentence-final “only” éryǐ – unambiguously occupy a clause-medial position between TP and VP, as roughly illustrated in (22b). Here I will reproduce one of the arguments presented in Erlewine (2017) for the clause-medial position of Mandarin SFP le. Consider the contrast in (23) reproduced from Soh and Gao (2006). These two sentences difer only in the choice of negator: the low negator bù and the higher bú-shì, which incorporates a copular verb. (23a) expresses that the speaker no longer misses home, whereas (23b) expresses that the speaker didn’t miss home before and still does not miss home. Note that the two sentences are identical in the relative linear order of negation and le. (23) The scope of Mandarin le vs. two negators, based on Soh and Gao (2006: 119) a. “bù ... le”  le > neg: b. “búshì ... le”  neg > le: Wǒ bù xiǎng jiā le. Wǒ bú-shì xiǎng jiā le. I neg miss home le I neg-cop miss home le “I don’t miss home anymore.” “I still don’t miss home.” Asserts: “I do not miss home Asserts: “I do not miss home now.” now.” Presupposes: “I did miss home Presupposes: “I did not miss before.” home before.” Soh and Gao argue that this diference can be explained as a scope ambiguity. In (23a), le takes scope over the low negator bù, applying the semantics in (15) to the proposition “I don’t miss home.” This results in the presupposition that “I don’t miss home” was false before, that is, that the speaker did miss home before. In contrast, in (23b), le takes scope under the high negator búshì. The application of le to “I miss home” yields the assertion that “I miss home,” together with the presupposition that “I miss home,” was false before. The higher negation then negates the assertion, but not the presupposition, resulting in the semantics described in (23b), which asserts that the speaker does not miss home and presupposes that they did not miss home before either. I refer the interested reader to Erlewine (2017) for additional discussion of this argument. Erlewine (2017) also presents additional arguments for the

A syntactic universal in a contact language

101

clause-medial height of Mandarin SFP le as well as another low SFP in Mandarin Chinese, the sentence-final “only” éryǐ.13 But recall that Mandarin was most likely not a dominant substrate influence in the development of Singlish; see footnote 10. It is therefore important to also check the structural height of cognates to Mandarin le in other Chinese languages of Singapore – specifically, liau in Southern Min languages (e.g., Hokkien, Teochew, Hainanese) and laa in Cantonese. Examples (24) and (25), which follow, are modeled on (23) earlier and reproduce the previous core finding for Hokkien and Cantonese. The cognate SFPs liau/laa pattern with Mandarin Chinese sentence-final le in attaching unambiguously to a clause-medial position, taking scope between the higher and lower negators.14 (24) The scope of liau in Hokkien (Southern Min) vs. two negators15 a. “bô ... liáu”  liau > neg: b. “m̄sī ... liáu”  neg > liau: I bô suka li  liáu. I m̄-sī  suka li liáu. s/he neg like you liau s/he neg-cop like you liau “S/he doesn’t like you “S/he still doesn’t like you.” anymore.” Asserts: “S/he doesn’t like you Asserts: “S/he doesn’t like you now.” now.” Presupposes: “S/he liked you Presupposes: “S/he didn’t like you before.” before.” (25) The scope of laa in Cantonese vs. two negators16 a. “m ... laa”  laa > neg: b. “mhai ... laa”  neg > laa: Keoi2 m4  zung1ji3 nei5 laa3 Keoi2 m4-hai6 sik6-zo2  faan6 laa3. s/he neg like you laa I  neg-cop eat-asp rice laa “S/he doesn’t like you “S/he hasn’t already eaten rice.” anymore.” Asserts: “S/he doesn’t like you Asserts: “S/he hasn’t eaten rice now.” now.” Presupposes: “S/he did like Presupposes: “S/he hadn’t eaten you before.” rice before.” A reviewer poses a question regarding the scope of subject quantifiers. The discussion here predicts that subject quantifiers would uniformly take scope above le/liao/laa, but as the reviewer notes, negative subject quantifiers appear to take scope under le/liao/laa. I illustrate this point with (26) in Mandarin, based on the reviewer’s example. (26) Negative subject quantifier appears to scope under le 17 le > no one    le. Méi-yǒu rén yòng zhè zhǒng xiyī ̌ jī neg-have person use this type washing machine le Asserts: “No one uses this kind of washing machine now.” Presupposes: “Someone used this kind of washing machine in the recent past.”

102

Michael Yoshitaka Erlewine

The attested interpretation of example (26) involves the negative quantifier scoping under le: it asserts that no one uses this kind of washing machine now (p) and presupposes that p was false at a prior time, together expressing a change of state – some used this kind of washing machine previously, but now no one does.18 However, such data does not challenge our conclusion regarding the height of le/liao/laa when we consider the nature of such negative quantifiers in Chinese languages. Apparent negative quantifiers such as Mandarin méi-you ̌ rén in (26) are limited to subject position (see, for example, Huang 2003) and have been analyzed by many authors as a biclausal presentational structure, with the negative morpheme – méi-you ̌ in the Mandarin (26) earlier – being the negative existential/possessive verb “not have” (see, for example, Huang 1987; Fang & Lin 2008; Paul 2021). The sentence-final le in examples such as (26) may then occupy a clause-medial position in the matrix clause, above the verb méi-yǒu. Given this possibility, such examples do not counterexemplify the analysis here, following Erlewine (2017), that the Chinese SFPs le/liao/laa uniformly realize a fixed, clause-medial position between TP and VP (22b). Recall that Chinese SFPs have traditionally been analyzed as head-final heads in an otherwise head-initial clausal spine. The existence of head-final projections of this form in a clause-medial position has important implications for the Final-over-Final Constraint, as I will discuss in Section 6 later. This understanding of the syntax of these Chinese SFPs le/liau/laa will be an important point of comparison for our investigation into Singlish sentence-final already, to which we return next. As discussed in Section 3, these Chinese SFPs are the substrate sources in the relexification process which led to Singlish sentencefinal already. From a straightforward application of the relexification analysis as schematized in (21) previously, we predict Singlish sentence-final already to have inherited the syntactic specification of its substrate sources le/liau/laa, not only in its linear position, but specifically also as a head-final head in a low, clause-medial position. 5

The syntax of Singlish already

We now return to the description of Singlish sentence-final already. As described earlier in Section 3, Singlish already has the semantics of Chinese le/ liau/laa and linearly occupies the same sentence-final position. The question now is what hierarchical position it takes: Does already exhibit the clausemedial attachment exhibited by its substrate cognates? In stark contrast to the uniform behavior of le/liau/laa observed earlier, Cheong (2016) argues that Singlish sentence-final already unambiguously scopes over the entire clause. Her evidence also comes from semantic scope, modeled in part on the tests I presented in Erlewine (2017), and shows that already unambiguously scopes over the entire clause. First, consider the scope of sentence-final already with respect to negation. Example (27) expresses a change of state: the habitual “I wash my hands” was true before but no longer true now. This meaning can be

A syntactic universal in a contact language

103

modeled using the semantics for already in (15) with already taking scope over negation. (27) I don’t wash hand already. “I don’t wash my hands anymore.” Asserts: “I do not wash my hands now.” Presupposes: “I used to wash my hands before.”

(Cheong 2016: 18) already > neg

If negation took scope over already in (27), already would introduce the presupposition that “I wash my hands” was false before, and this presupposition would then project through the higher negation. The resulting meaning – that “I wash my hands” was false before and also false now – is not possible in (27) but is possible with a biclausal negation.19,20 (28) Is not that I wash hand already. “I still haven’t washed my hands.” Asserts: “I have not washed my hands now.” Presupposes: “I had not washed my hands before.”

(ibid.: 18) neg > already

Consider also the scope of already with respect to subject quantifiers. Example (29) asserts that no one goes to school now but that was false before, that is, at least one person used to go to school in the past. This, too, is captured by our semantics for already in (15) if already takes scope over no one. (29) No one go school already. (ibid.: 18)   “It is now the case that no one goes to school.” already > no one Asserts: “No one goes to school now.” Presupposes: “Someone used to go to school before.” Note also that, unlike with the Chinese negative subjects as in (26) earlier, there is no reason to suspect that this Singlish structure in (29) is biclausal. Its interpretation transparently shows us that already must attach above the subject position. Suppose instead that already attached to a clause-medial, VP-peripheral position. Assuming a VP-internal base position for the subject no one, represented by the variable x in (30), already (15) would take the proposition “x goes to school” in its scope and introduce the presupposition that “x did not go to school before.” (30) Hypothetical LF for (30) with no one > already [TP [no one] λx [[VP x go school] already]] already à “x did not go to school before” There is a question now of how this variable x is interpreted in the content of the presupposition of already – the so-called “projection problem” for

104

Michael Yoshitaka Erlewine

presuppositions (Heim 1983). Previous work has established that presuppositions under negative quantifiers such as no one project over their entire domain of quantification. For example, Chemla (2009) reports experimental results which show that the factive presupposition of know in (31) must hold for all students. (31) No studenti knows that hei’s lucky. Presupposes: “Every student is lucky.” If no one took scope over already as in (29), we would then predict (29) to express a stronger presupposition that everyone went to school before, but this is not the interpretation of (29). We can conclude with Cheong (2016) that already unambiguously scopes over subjects. In conclusion, Singlish sentence-final already unambiguously takes scope over the entire clause, even scoping over subject quantifiers, as in (29). This leaves us with a puzzle. As discussed earlier, we hypothesize that the SFPs le/ liau/laa in Chinese languages of Singapore were the substrate sources for Singlish sentence-final already, and these SFPs attach to a clause-medial position between VP and TP. The position of Singlish sentence-final already thus difers drastically from that of its substrate cognate SFPs, despite appearing in the same sentence-final linear position. (32) Positions for already and substrate cognates, determined by semantic scope (= 2) a.

Singlish already: TP

b.

Chinese cognate SFPs: TP

already subject

subject

T T

VP

VP

le/liau/laa

Again, I note that this contrast in structural position is not only a puzzle for the historical development of Singlish but is also a puzzle for the synchronic acquisition of these languages. With the same semantics as in (15), learners of Chinese languages robustly converge on a clause-medial structural position for le/liau/laa, whereas learners of Singlish robustly converge on a high, clauseperipheral position for sentence-final already. The contrast reported earlier between the structural positions of already and Chinese le/liau/laa is even true of the individual grammars of Singlish–Mandarin bilingual speakers that I have consulted. As one first possibility for explaining this puzzle, I will briefly discuss the syntactic category of Singlish sentence-final already. Recall that already in “standard” Englishes is an adverb, whereas the relevant Chinese substrate particles have been analyzed in previous work as head-final heads in the clausal spine. Perhaps the diference in (32) can be explained if Singlish already is

A syntactic universal in a contact language

105

synchronically an adverb rather than a head-final head. This hypothesis is schematized in (33). (33) Relexification of Singlish sentence-final already as an adverb (to be rejected) Substrate languages: Hokkien, Cantonese, ... 2

3

6 6 4

6 6 5 ! 4

SYN

SEM PHON

head-final head, 7 below TP 7

(15) le/liau/laa

Result: Singlish 2 SYN

SEM PHON

Superstrate/lexifier: English 3 2 adverb SYN adverb à 7 6 7 6 5 4 SEM ¼ earlier... (15) already PHON already Ã

3 7 7 5

This mode of explanation faces multiple challenges. First, if the relevant Singlish already were an adverb, we might expect it to not be limited to sentence-final position. In fact, Singlish does have an adverb already which exhibits freer word order but instead has the “earlier than expected” semantics of standard English already, as we saw in examples (7)–(8) earlier. The contrasts in (7)–(8) support the idea that the Singlish already which we are studying here cannot simply be specified as an adverb, as in (33). Suppose, then, that we let already be an adverb but (somehow) independently restrict its appearance to a sentence-final surface position. Could this help explain the unambiguously high position of sentence-final already (32b)? For this mode of explanation to go through, it would have to be the case that Singlish independently interprets adverbs in sentence-final position as unambiguously structurally high, taking scope over the whole clause. A preliminary look at other Singlish adverbs in sentence-final position shows that this approach is also untenable. Here, too, I draw on observations reported in Cheong (2016) regarding the scope of sentence-final only in Singlish. The frequent appearance of only in sentence-final position is a notable feature of Singlish and Hong Kong English (Hiramoto 2012), also found in Indian English (Parviainen 2012). Consider the scope of sentence-final only with respect to the negation don’t in (34). Sentence-final only can take scope over or under negation, in contrast to the strictly wide scope of sentence-final already in (27). (34) Context: The relevant languages under discussion are French, Spanish, and Japanese. John don’t speak [French]F only. (Cheong 2016: 33) a. “John does not only speak French.” neg > only    He speaks French and speaks another language as well. b. “John only does not speak French.” only > neg    He does not speak French but speaks both French and Spanish.

106

Michael Yoshitaka Erlewine

A similar ambiguity is observed with the scope of only with respect to the subject quantifier no student. This, too, contrasts with the behavior of sentencefinal already in (29). (ibid.: 29) (35) No student study [Spanish]F only. a. “There is no student who studies only Spanish.” no student > only    Every student studies Spanish and another language. b. “It is only the case that no student studies only > no student Spanish.”    No student studies Spanish, but there are students that study French and Japanese. This behavior of only in sentence-final position in Singlish shows that it is not the case that adverbs in sentence-final position necessarily take wide scope, as would be required to explain the strict wide scope of sentence-final already as an adverb. The diference between already and only in Singlish is also apparent in their surface distribution. In contrast to the already described here which is limited to sentence-final position (7)–(8), only can occur in Singlish in pre-verbal position as well, with no diference in meaning. In other words, only in Singlish has the expected, relatively free distribution (both in linear position and structural height) of an English adverb – with the sole diference being that only is much more common in sentence-final position in Singlish (and Hong Kong English and Indian English) than in standard British and North American Englishes (Hiramoto 2012; Parviainen 2012). This comparison with the adverb only in Singlish thus motivates a description of Singlish sentence-final already as instantiating a distinct syntactic category. I propose that Singlish sentence-final already is a head-final head in the clausal spine with a fixed hierarchical position.21 My proposed lexical specification for Singlish sentence-final already is given in (36). (36) The development of Singlish sentence-final already (final, to be explained) Substrate languages: Hokkien, Cantonese, ... 2 6 6 4

SYN

SEM PHON

3

Result: Singlish 2

head-final head, ! SYN 7 6 below TP 7 6

(15) le/liau/laa

5!4

SEM PHON

3

head-final head, 7 above TP 7

(15) already

5

Superstrate/lexifier: English 2 SYN adverb 6 6 4 SEM ¼ earlier... Ã PHON already

3 7 7 5

From the point of view of the relexification hypothesis for the development of already (Section 3), already being a head-final head would be the expected result, as it matches the syntactic description of its Chinese substrate cognate SFPs as head-final heads rather than adverbs.22 The one diference, however, is its hierarchical position: Singlish sentence-final already unambiguously

A syntactic universal in a contact language

107

attaches above TP, whereas its substrate cognate SFPs occupy a fixed clausemedial position below TP. What led to this fixed, high hierarchical position for Singlish already? It cannot be attributed to the relevant substrate and superstrate lexical specifications, also in (36). And as noted earlier, there are no observable diferences in linear position which would have led to such a change. This innovation in the lexical specification of Singlish already must be due to an independent pressure in the grammar of Singlish. 6

Reanalysis due to a syntactic universal

As we have seen earlier, the fine-grained syntactic behavior of Singlish sentence-final already shows that it is not the straightforward result of a process of relexification but instead reflects an additional change. I propose that already underwent reanalysis triggered by independent pressure in the grammar of Singlish – but not in substrate Chinese languages – to avoid the placement of head-final heads in a clause-medial position. Already thus changed from a head-final head in the VP periphery, like its Chinese cognate SFPs, to a headfinal head in the CP periphery.23 I furthermore identify the relevant constraint which triggered this reanalysis as the Final-over-Final Constraint (Holmberg 2000; Biberauer et al. 2008, 2009, 2014; Sheehan et al. 2017: a.o.), a proposed universal on structure-building and linearization. The Final-over-Final Constraint (FOFC) is a constraint on the shape of syntactic structures, stated in (37). Informally, FOFC allows for three types of local syntactic configurations, schematized in (38): (a) strictly head-initial (HI), (b) strictly head-final, and (c) structures with a head-initial projection dominating head-final projection; (d) structures with a head-final projection dominating a head-initial projection are claimed not to exist. (37) The Final-over-Final Constraint (FOFC) (Holmberg (2000: 124)) If a phrase α is head-initial, then the phrase β immediately dominating α is head-initial. If α is head-final, β can be head-final or head-initial. (38) Predictions of the Final-over-Final Constraint a.

X HF

over HF: βP

αP XP

β α

b.

X HI

over HI: βP

β

αP α

XP

c.

X HI

β

over HF: βP

d. * HF over HI: βP

αP

αP

XP

α

α

β

XP

FOFC was first formulated to describe certain curious gaps in Finnish word order patterns. Consider the order of the auxiliary, verb, and object in the Finnish wh-questions in (39). Suppose the auxiliary occupies a head immediately selecting the verb phrase: α = V and β = Aux.24 The verb may take its

108

Michael Yoshitaka Erlewine

complement object to its left or right, and the auxiliary can precede or follow the VP, with one exception: the V-O-Aux order (d), which would require a head-final Aux taking a head-initial VP as its complement. Biberauer et al. (2008, 2014) show that this *V-O-Aux gap is also observed across a range of modern and historical Germanic languages. (39) Word orders in Finnish wh -questions (Holmberg 2000: 128) a. Aux-V-O: Milloin Jussi olisi kirjoittanut romaanin?  when Jussi would.have written a novel  “When would Jussi have written a novel?” b. Aux-O-V: Milloin Jussi olisi romaanin kirjoittanut?  when Jussi would.have a novel written c. O-V-Aux: Milloin Jussi romaanin kirjoittanut olisi?  when Jussi a novel written would.have d. V-O-Aux: *Milloin Jussi kirjoittanut romaanin olisi?  when Jussi written a novel would.have Subsequent work by Biberauer, Holmberg, Roberts, Sheehan, and colleagues has shown that this constraint (37) accurately explains word order gaps for various types of syntactic structures in a wide range of languages, motivating FOFC as a syntactic universal. At the same time, FOFC cannot hold over entire utterances. Consider the German example in (40). The German VP is head-final but dominates a headinitial DP. (40) Head-final VP over head-initial DP in German

(Biberauer et al. 2008: 99)

Johann hat [VP [DP den Mann] gesehen]. John has the man seen “John has seen the man.” A common intuition for accounting for such data is that FOFC holds only over certain spans of syntactic structure rather than the entire utterance. I refer to such domains of FOFC enforcement as FOFC domains. Biberauer et al. (2009, 2014) and Biberauer and Sheehan (2012) propose that FOFC domains are extended projections (Grimshaw 2000). FOFC is not enforced between the DP and VP projections in (40) because the DP and VP are not in the same extended projections. In contrast, the absence of V-O-Aux order in Finnish and Germanic earlier is due to FOFC being enforced across the entire clausal spine – the extended projection of V, which includes the heads V, T,

A syntactic universal in a contact language

109

C, and others, depending on the inventory of functional heads assumed in the clausal spine. As Chinese SFPs have been analyzed as head-final heads in the clausal spine, they have received attention as potential exceptions to FOFC (Biberauer et al. 2008, 2009; Bailey 2010; Chan 2013; Paul 2014: a.o.). Much of this literature has focused on high SFPs with clause-typing or speaker attitude functions, which can reasonably be assumed to be in the extended CP periphery (see especially Paul 2014). In this case, a relatively straightforward resolution can be found by making reference to the notion of FOFC domains. FOFC is enforced only over extended projections, according to Biberauer et al., and the clause edge is a natural break point in extended projections. Suppose an SFP is a head-final head which takes the full verbal extended projection as its complement. This situation would not violate FOFC as the head-final SFP, and the head-initial heads in the verbal extended projection would be in diferent FOFC domains and thus not be evaluated together for FOFC compliance. However, the existence of so-called “low” Chinese SFPs in a clause-medial position – as I’ve argued for Mandarin le and “only” éryǐ in Erlewine (2017) and we saw for Hokkien liau and Cantonese laa in Section 4 – is more challenging to reconcile with FOFC. This structural configuration is sketched in (41). (41) Chinese SFPs in two positions (Erlewine 2017: 39) CP high SFP Ã head-final

TP subject

.. .

T . .. VP V

low SFP

head-initial à head-final head-initial

object

If the entire verbal extended projection is a single FOFC domain, as Biberauer et al. (2008) suggest, we predict that no head-final head could appear in the middle of this clausal spine, given the clearly head-initial lower verbal projections. FOFC (37) claims that no head-initial projection can be dominated by a head-final projection. This suggests that we should abandon the analysis of low SFPs as head-final heads or abandon FOFC itself. In order to resolve this tension, in Erlewine (2017) I argued for a new characterization of FOFC domains which allows for principled cross-linguistic variation in the size of FOFC domains: (42) FOFC domains = Spell-Out domains (Erlewine 2017: 67) FOFC holds only within individual Spell-Out domains.

110

Michael Yoshitaka Erlewine

Phase theory (Chomsky 2000, 2001) posits that syntactic structure is built in chunks, phase-by-phase, with CP, vP, and DP being classical examples of phases. Once one full phase is built, the complement of the head of the phase may undergo Spell-Out. The determination of word order and surface morphophonological forms then takes place during Spell-Out. Under this view of cyclic structure building, it is natural to conceive of FOFC – a constraint on the linearization of syntactic structures – as operating over structures which undergo Spell-Out together, but not across larger structures. In Mandarin Chinese, I proposed that the lower phase head – which I label “SFP1” and take to be the locus of the low SFPs, such as Mandarin le – and the higher phase head C each triggers Spell-Out of their complements, as illustrated in (43). FOFC is enforced over the heads in each of these Spell-Out domains, which are predominantly head-initial. If any heads in the clausal spine are head-final, they must be at these boundaries. The lowest head or heads in each Spell-Out domain can be head-final without violating FOFC, even though their complement is head-initial, because their complement is in a separate Spell-Out domain. This model accurately explains the very limited structural distribution of Chinese SFPs: SFPs are head-final heads and thus limited to these break points in FOFC enforcement, which may be at the edge of the lower clausal phase of the clausal spine (traditionally called vP) and the edge of the higher clausal phase, at the clause periphery. (43) Spell-Out domains of the clausal spine in Chinese languages (hierarchical) phase head

phase head

[CP C(=SFP2 ) [TP T ... [SFP1P SFP1 ... [vP v [VP V ... Spell-Out domain

Spell-Out domain

I furthermore proposed that Spell-Out can be suspended if there are morphological dependencies between the potential Spell-Out domain and higher functional heads (see also Bobaljik & Wurmbrand 2013). For example, in languages with verbal inflection, the morphological realization of a verb in the lower phase of the clause is dependent on the ϕ-features on T or the presence or absence of auxiliaries, which is information in the higher phase. In this case, the complement of the lower phase head cannot independently undergo Spell-Out and fix its surface morphophonological forms. The structure in the lower phase (XP in (44)) will instead undergo Spell-Out together with higher material, once the higher (CP) phase is built.25 The inflectional dependency is indicated schematically in as a link between T and v.

A syntactic universal in a contact language

111

(44) Spell-Out domains in languages with verbal inflection (hierarchical) phase head

phase head

[CP C [TP T ... [XP X ... [vP v [VP V ... Spell-Out domain

The efect of such morphological dependencies, then, is to combine the lower and higher portions of the clause into one Spell-Out domain and thus one domain of FOFC enforcement (42). This suspension of lower Spell-Out does not occur in Chinese languages, because there is no verbal inflection, resulting in two Spell-Out domains and thus the opportunity for head-final heads in a clause-medial position (41/43). Recall that one of the original motivations for FOFC is the lack of V-O-Aux word orders in Finnish and across modern and historic Germanic languages. If the auxiliary in such cases occupies T, the FOFC-based explanation of this VO-Aux gap necessarily relies on FOFC being enforced across both the higher and lower phases of the clause. Under my proposal, this is a consequence of the fact that these languages have verbal inflection, unlike Chinese languages. A prediction of my account is that a head-final head in an otherwise headinitial clausal spine – a potentially FOFC-violating structure – may occur in a clause-medial position only if the language lacks morphological dependencies between the higher and lower phases of the clause, such as verbal inflection. This prediction is supported by the existence of V-O-Aux word order in various languages without verbal inflection. Simpson (2001) shows that the apparently FOFC-violating V-O-Aux order is attested by an ability modal in Middle Chinese and a number of modern languages of mainland Southeast Asia. An example from Thai is reproduced in (45). Importantly, these languages discussed by Simpson lack verbal inflection. (45) Predicate-final possibility modal in Thai Khun [[VP pai kap khaw phrung-nii]   dai]. you go with him  tomorrow can “You can go with him tomorrow.”

(Simpson 2001: 94)

A similar connection between the availability of apparent FOFC-violating structures and the lack of verbal inflection has been noted previously by Philip (2013), citing Matthew Dryer (p.c.): “for many of the VO languages exhibiting final uninflected tense or aspect particles, there is simply no verbal inflection in the language at all” (p. 206). With this background in mind, we now return to Singlish already. We have motivated earlier that Singlish already is the result of relexification pairing the surface form of the English adverb already with the syntax and semantics of the le/liau/laa SFPs in Chinese substrate languages. This predicts already to be a head-final head at the lower phase edge, as made possible by independent Spell-Out of the complement of the lower phase. But we have determined that this is not the accurate synchronic position for Singlish already.

112

Michael Yoshitaka Erlewine

Unlike Chinese languages, Singlish exhibits verbal inflection. Many previous works note the optionality of Singlish past tense -ed and third singular -s (Ho & Platt 1993; Gupta 1994; Lai et al. 2013: a.o.), as also demonstrated in examples (11) vs. (20) earlier, but what’s relevant here is the availability of verbal inflection, which reflects a morphological dependency between the lower and higher phases of the clause.26 (46) Wait lah, John say [∅ speak(s) Hokkien].

(Sato 2014: 371)

This morphological dependency is indicated as a link between T and V in (47b) in the following. As noted earlier, this dependency blocks the independent Spell-Out of the complement of the lower phase, making a larger portion of the clause subject to FOFC evaluation together. This blocks already from appearing in its predicted structural position, as a head-final head in a clausemedial position, forcing its reanalysis to a clause-peripheral position. These two steps of relexification and reanalysis are illustrated in (47). (47) Relexification and FOFC-driven reanalysis in the development of Singlish already a. Chinese le/liau/laa:

b. Singlish already:

TP

TP .. .

T

.. le/liau/laa . VP V

7

FOFC domain

T

.. .

reanalysis

.. *already . VP

relexification FOFC domain

already

FOFC domain

V

Conclusion

The close study of the syntax/semantics of Singlish already and its substrate cognates leads to a puzzle. Building on the earlier work of Bao (2005a), we identify the Chinese sentence-final particles le/liau/laa as the substrate sources of already in a process of relexification. These items all share a common semantics, distinct from that of the adverb already in standard Englishes, with the same sentence-final linear position. However, evidence from semantic scope in Cheong (2016) and Erlewine (2017) and above shows that these items do not share a uniform hierarchical position: Singlish sentence-final already unambiguously occupies a high, clause-peripheral position, whereas its cognate SFPs le/liau/laa attach to a clause-medial position. I argue that the consideration of the Final-over-Final Constraint (FOFC) can help explain this puzzling diference. In Erlewine (2017), I proposed that

A syntactic universal in a contact language

113

FOFC holds over Spell-Out domains and that morphological dependencies (Agree relationships) can suspend the independent Spell-Out of material in lower phases. Practically, the existence of verbal inflection forces structure in the lower and higher phases of a clausal spine to undergo Spell-Out together and therefore constitute a single domain for FOFC enforcement. Singlish allows for verbal inflection, unlike Chinese languages but like standard Englishes. This led to a diference in regions of FOFC enforcement between Singlish and its substrate Chinese languages, forcing a reanalysis of sentencefinal already to a clause-peripheral position. This work, in turn, supports the relevance of FOFC for language development and change, as also discussed in Biberauer et al. (2010, 2014) – and the relevance of grammatical constraints on language contact more generally (see, for example, Bowern 2008) – as well as the characterization of FOFC as enforced over spell-out domains, as proposed in Erlewine (2017).

Abbreviations asp clf cop dim FOFC

aspect classifier copula diminutive Final-over-Final Constraint

int mod neg pfv SFP

intensifier modifier negation perfective sentence-final particle

Notes 1 Acknowledgments: This project grew out of conversations with Phoebe Cheong and builds on the findings in her BA honors thesis, Cheong (2016). For discussion and comments, I thank her as well as Bao Zhiming, Kenyon Branan, Hadas Kotek, E-Ching Ng, Yosuke Sato, Hooi Ling Soh, Rebecca Starr, audiences at the 2017 Society for Pidgin and Creole Linguistics summer meeting (especially J. Clancy Clements), the 2018 Linguistic Society of America meeting, Chulalongkorn University, and the National University of Singapore, and an anonymous reviewer. I also thank Len Wanyan, Lim Junjie, Hannah Lin, and Ngui Jiangang for assistance with additional data collection, and Keely New and Tan Yan Er for further discussion of judgments. Errors are mine. This research is supported by the National University of Singapore under grant WBS A-007220–00–00. 2 In academic literature, Singlish is also often referred to as vernacular or colloquial Singapore English. These terms all refer to the basilectal variety spoken in Singapore, which coexists with Standard Singapore English. See, for example, Platt (1975) and Bao and Hong (2006). The Singlish judgments which I report on here reflect the grammar of speakers in their early to mid-20s while this project was conducted in 2016–2020. 3 Anders Holmberg (p.c.) points to other attested cases of contact-induced borrowing of linear patterns without identical syntactic structures. This includes the Estonian V2 patterns reported by Holmberg et al. (2020), hypothesized as resulting from Germanic V2 contact in their ongoing work, as well as embedded questions

114

Michael Yoshitaka Erlewine

with inversion in Hiberno-English and Welsh English due to Celtic substrate influence, as discussed in Roberts (2007: ch. 5). 4 However, it generally resists matrix subject-drop. See Branan and New (to appear). 5 But see also DeGraf (2002) for criticisms of Lefebvre’s original motivation for the relexification hypothesis from Haitian Creole. 6 Mana is a borrowing from Malay also meaning “where.” See Lee et al. (2009: note 5). 7 Example (10) can also be interpreted as “I wash my hands now (habitually) but didn’t use to.” This same ambiguity may be even more salient in the contrast between I eat already, which regularly means that the speaker has finished eating, versus Baby eat already, which can be taken to mean that the baby has recently begun eating solid foods. This ambiguity will be explained by the following account. 8 Bao (1995) notes that this connection to Chinese perfective aspect markers is made earlier by Kwan-Terry (1989) in her description of the bilingual acquisition of English and Cantonese by a child in Singapore. 9 The Chinese substrate influences on Singlish were likely a range of southern Chinese languages, not Mandarin Chinese, as the discussion in Bao (2005a) seems to suggest. See, for example, Gupta 1994; Siegel 2012; Wong 2014; Lee et al. (2009) for discussion and support. Note also that the form liao is also used for SFP le in Singapore Mandarin. 10 Soh and Gao’s proposal is, in turn, informed by the analysis of German schon and related expressions, such as English already in Löbner (1989); see also discussion in Krifka (2000). Soh and Gao (2008: 470) and Soh (2009: 652) observe that standard English already also appears to encode the Löbnerian presupposition that the prejacent was false at a past time (15b), perhaps in addition to the “earlier than expected” meaning observed in (7)–(8) above. 11As an anonymous reviewer notes, this approach predicts the availability of bare predicates such as rain with both for- and in-phrase temporal modifiers. Although bare predicates with an atelic interpretation are indeed compatible with for-phrases, I have been unable to produce clear examples of bare predicates with the predicted telic perfective interpretation with in-phrases. Bare predicates with in-phrases are possible in certain circumstances, but with inchoative achievement semantics instead: (i) A: Why sky so dark? B: Maybe will rain in one hour.  “Maybe it will start to rain in an hour.” Here I will leave open the puzzle of the apparent unavailability of in-phrases with telic perfective interpretations of bare verbs. 12 The speakers I consulted judge It raining already to sound unnatural, although Outside raining already or even Raining already with no subject is acceptable with the inchoative reading. 13 However, see also Pan (2018) and Zhang (2019) for critical discussion of my account there. 14 A reviewer asks about the precise positions of these higher and lower negators in Chinese languages. As discussed in Erlewine (2017) (see especially note 5), what we know for certain is (a) that both the higher and lower negations are monoclausal constructions and (b) that the low class of SFPs, including Mandarin le and éryǐ, occupies a fixed, clause-medial position between them, as also reflected here. In Erlewine (2017), I conclude that the lower negators must be within the lower phase of the clause (traditionally, vP), and the higher negators must be outside of

A syntactic universal in a contact language

15 16

17 18 19 20

21 22

23

115

the lower phase. See discussion in Section 6 for the relevance of this phase boundary. A more precise identification of their positions is not necessary for our purposes here. I thank Ting-Chun Chen for the transcription of Hokkien data. Superscripts indicate tones. The examples in (25) are not parallel, due to a difculty with identifying appropriate predicates which can co-occur with the diferent negators and laa3. Jess Law and Leslie Lee also note that there appear to be other examples where laa scopes over the higher negator mhai. Further work on laa and its interaction with these negators and predicates of diferent aspectual classes is warranted. I thank Leslie Lee (Singapore) and Elaine Lau, Jess Law, and Joanna Sio (Hong Kong) for discussion of this and related Cantonese data. I thank Zheng Shen for discussion of this Mandarin example. See discussion around (30) on what the hypothetical no one > LE scope interpretation would be. Unlike the Chinese languages discussed earlier, Singlish appears to lack distinct, higher and lower monoclausal negations. As noted earlier, the predicate wash hand is ambiguous between a habitual activity reading and an accomplishment reading. The dominant reading here seems to be episodic, reporting the lack of the wash hand accomplishment in the current situation. This can be thought of as negation applying to (17), that is, the accomplishment reading of wash hand with already applied to it. This is in contrast to example (27), where the dominant reading is habitual, modeled as in (18), but with negation applying before already. I thank Keely New for discussion of this point. Lan (2017) corroborates and expands upon this analysis in his expansive study of the syntax of Singlish sentence-final particles. Hiramoto (2012) proposes that both sentence-final only and already are the result of transfer from cognate SFPs in Chinese substrate languages. Indeed, just as the relevant Chinese languages have cognate SFPs for Singlish sentence-final already, as presented earlier in Section 4, they also have SFPs with the semantics of only. See example (10’) in Hiramoto (2012: 644). In Erlewine (2017), I show that the Mandarin sentence-final “only” éryǐ unambiguously takes clause-medial scope, at a height equal to that of Mandarin SFP le, also observed earlier. The striking diference in the synchronic syntactic behaviors of Singlish sentence-final only and already may suggest that they did not, in fact, develop in the same manner, contrary to Hiramoto’s suggestion. One possibility is that multiple influences may have led to the development of Singlish sentence-final only, rather than being a simple relexification of Chinese “only” SFPs. Parviainen (2012) shows that frequent sentence-final use of only is also observed in Indian English (without Chinese substrate influence) and hypothesizes that this feature of Indian English may have also been an influence in the development of Singlish. In contrast, Indian English lacks the frequent sentence-final already described here (Hanna Parviainen, p.c.), making its use in Singlish (and Hong Kong English) more directly attributable to Chinese substrate influence. These synchronic and historical diferences between Singlish sentence-final already and sentence-final only (and also) warrant further investigation, which I will leave for future work. I do not make a claim as to whether these two processes – relexification and reanalysis – took place over time or in one step.

116

Michael Yoshitaka Erlewine

24 This assumption simplifies the illustration but is not strictly necessary to explain the pattern in (39). FOFC bans a head-final projection above a head-initial one at any level, so once the VP is head-initial, FOFC predicts that all higher projections will necessarily be head-initial – so long as the verb and auxiliary are both in the same “FOFC domain.” See discussion of exceptions to the *V-O-Aux constraint and FOFC domains in the following. 25 To make the comparison to the Chinese structure in (43) fully parallel, the head X above vP is taken to be the lower phase head. This head X in (44) corresponds to SFP1 in (43), the hypothetical locus for a low SFP. 26 Karlos Arregi and Kenyon Branan ask whether the presence or absence of verbal inflection changes the availability of already taking scope in a lower position. It appears that it does not. In particular, already takes high scope even in cases where no verbal inflection is realized, as in examples (27) and (29). We can think of this in one of two ways. First, we may hypothesize that the relevant inflectional dependency (an Agree relationship) is always established between T and the verb, with optionality reflecting only whether or not this information is later expressed on the verb. Alternatively, we may hypothesize that this dependency (Agree) is only established when verbal inflection is expressed, but that the existence of sufcient examples with verbal inflection in the input nonetheless forced the reanalysis of already to specifically target a clause-peripheral position, and this same high position is used even in cases where the lower portion of the clause could constitute an independent Spell-Out domain.

References Bailey, Laura R. 2010. Sentential word order and the syntax of question particles. In Newcastle Working Papers in Linguistics 16, 23–43. Newcastle upon Tyne: Newcastle University. Bao, Zhiming. 1995. Already in Singapore English. World Englishes 14(2). 181–188. Bao, Zhiming. 2005a. The aspectual system of Singapore English and the systemic substratist explanation. Journal of Linguistics 41. 237–267. http://doi.org/10.1017/ S0022226705003269. Bao, Zhiming. 2005b. Systemic transfer, topic prominence, and the bare conditional in Singapore English. Journal of Pidgin and Creole Linguistics 20(2). 269–291. Bao, Zhiming. 2010. Must in Singapore English. Lingua 120. 1727–1737. Bao, Zhiming & Huaqing Hong. 2006. Diglossia and register variation in Singapore English. World Englishes 25(1). 105–114. Biberauer, Theresa, Anders Holmberg & Ian Roberts. 2008. Structure and linearization in disharmonic word orders. In Proceedings of WCCFL 26, 96–104. Somerville, MA: Cascadilla. Biberauer, Theresa, Anders Holmberg & Ian Roberts. 2014. A syntactic universal and its consequences. Linguistic Inquiry 45(2). 169–225. Biberauer, Theresa, Glenda Newton & Michelle Sheehan. 2009. Limiting synchronic and diachronic variation and change: The final-over-final constraint. Language and Linguistics 10(4). 701–743. Biberauer, Theresa & Michelle Sheehan. 2012. Disharmony, antisymmetry, and the final-over-final constraint. In Myriam Uribe-Etxebarria & Vidal Valmala (eds.), Ways of Structure Building, 206–244. Oxford: Oxford University Press. Biberauer, Theresa, Michelle Sheehan & Glenda Newton. 2010. Impossible changes and impossible borrowings: The final-over-final constraint. In Anne Breitbarth,

A syntactic universal in a contact language

117

Christopher Lucas, Sheila Watts & David Willis (eds.), Continuity and Change in Grammar, 35–59. Amsterdam: John Benjamins. Bobaljik, Jonathan David & Susi Wurmbrand. 2013. Suspension across domains. In Ora Matushansky & Alec Marantz (eds.), Distributed Morphology Today, 185–197. Cambridge, MA: MIT Press. Bowern, Claire. 2008. Syntactic change and syntactic borrowing in generative grammar. In Gisella Ferraresi & Maria Goldbach (eds.), Principles of Syntactic Reconstruction, 187–216. Amsterdam: John Benjamins. Branan, Kenyon & Keely New. To appear. Pronominal paradigms in two varieties of English. In Proceedings of WCCFL 38. Somerville, MA: Cascadilla. Chan, Brian Hok-Shing. 2013. Sentence-final particles, complementizers, antisymmetry, and the final-over-final constraint. In Theresa Biberauer & Michelle Sheehan (eds.), Theoretical Approaches to Disharmonic Word Orders, 445–468. Oxford: Oxford University Press. Chemla, Emmanuel. 2009. Presuppositions of quantified sentences: Experimental data. Natural Language Semantics 17(4). 299–340. Cheng, Lisa Lai-Shen. 1991. On the Typology of Wh-Questions. Massachusetts Institute of Technology dissertation, Cambridge, MA. Cheong, Phoebe Si En. 2016. Sentence-Final Already and Only in Singapore English. BA Honors thesis, National University of Singapore. Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David Michaels & Juan Uriagereka (eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–156. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A Life in Language, 1–52. Cambridge, MA: MIT Press. DeGraf, Michel. 2002. Relexification: A reevaluation. Anthropological Linguistics 44(4). 321–414. Erlewine, Michael Yoshitaka. 2017. Low sentence-final particles in Mandarin Chinese and the final-over-final constraint. Journal of East Asian Linguistics 26(1). 37–75. Fang, Huilin & Tzong-Hong Jonah Lin. 2008. The Mandarin yǒu existential: A verbal analysis. In University System of Taiwan Working Papers in Linguistics, vol. 4, 43–56. Hsinchu: National Tsing Hua University. Grimshaw, Jane. 2000. Locality and extended projection. In Peter Coopmans, Martin B. H. Everaert & Jane Grimshaw (eds.), Lexical Specification and Insertion, 115– 134. Amsterdam: John Benjamins. Gupta, Anthea Fraser. 1994. The Step-Tongue: Children’s English in Singapore. Clevedon: Multilingual Matters. Heim, Irene. 1983. On the projection problem for presuppositions. In Michael Barlow, Daniel P. Flickenger & Nancy Wiegand (eds.), Proceedings of WCCFL 2, 114–125. Stanford, CA: Stanford University. Hiramoto, Mie. 2012. Pragmatics of the sentence-final uses of can in Colloquial Singapore English. Journal of Pragmatics 44. 890–906. Ho, Mian-Lian & John T. Platt. 1993. Dynamics of a Contact Continuum: Singapore English. Oxford: Oxford University Press. Holmberg, Anders. 2000. Deriving OV order in Finnish. In Peter Svenonius (ed.), The Derivation of VO and OV, 123–152. Amsterdam: John Benjamins.

118

Michael Yoshitaka Erlewine

Holmberg, Anders, Heete Sahkai & Anne Tamm. 2020. Prosody distinguishes Estonian V2 from Finnish and Swedish. In Speech Prosody 2020, 439–443. http://doi. org/10.21437/SpeechProsody.2020-90 Huang, C.-T. James. 1987. Existential sentences in Chinese and (in)definiteness. In Eric J. Reuland & Alice G. B. ter Meulen (eds.), The Representation of (In)definiteness, 226–253. Cambridge, MA: MIT Press. Huang, C.-T. James. 2003. The distribution of negative NPs and some typological correlates. In Yen-hui Audrey Li & Andrew Simpson (eds.), Functional Structure(s), form, and Interpretation, 262–280. London: RoutledgeCurzon. Kang, Grace Yixuan. 2016. Ways of Expressing Necessity: A Study of Singlish Must and Confirm. BA Honors thesis, National University of Singapore. Koh, En Hui Lauren. 2018. An Analysis of Colloquial Singapore English “Where Got” Constructions. BA Honors thesis, National University of Singapore. Krifka, Manfred. 2000. Alternatives for aspectual particles: Semantics of still and already. In Proceedings of BLS 26, 401–412. Berkeley, CA: University of California Berkeley. Kwan-Terry, Anna. 1989. The specification of stage by a child learning English and Cantonese simultaneously: A study of acquisition processes. In Hans W. Dechert & Manfred Raupach (eds.), Interlingual Processes, 33–48. Tübingen: Gunter Narr Verlag. Lai, Eric Yong Ming, Liling Tan, Vincent Wong, Lenny Teng Tao Loke & Francis Bond. 2013. The OPT-ional phenomenon in Singapore English: A corpus-based approach using time annotated corpora. Procedia – Social and Behavioral Sciences 95. 431–441. Lan, Yingjie. 2017. Being Particularly Singlish: The Syntax of Sentence-Final Particles in Singaporean English. MA thesis, University of Cambridge. Law, Ann. 2002. Cantonese sentence-final particles and the CP domain. In UCL Working Papers in Linguistics. London: University College London. Lee, Hun-tak Thomas. 1986. Studies on Quantification in Chinese. Dissertation, University of California at Los Angeles. Lee, Nala Huiying, Ling Ai Ping & Hiroki Nomoto. 2009. Colloquial Singapore English got: Functions and substratal influences. World Englishes 28(3). 293–318. Lefebvre, Claire. 1998. Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Leimgruber, Jakob R. E. 2013. Singapore English: Structure, Variation, and Usage. Cambridge: Cambridge University Press. Löbner, Sebastian. 1989. German schon – erst – noch: An integrated analysis. Linguistics and Philosophy 12(2). 167–212. Lumsden, John S. 1999. The role of relexification in creole genesis. Journal of Pidgin and Creole Linguistics 14(2). 225–258. Muysken, Pieter. 1981. Halfway between Quechua and Spanish: The case for relexification. In Arnold Highfield & Albert Valdman (eds.), Historicity and Variation in Creole Studies, 52–78. Ann Arbor: Karoma. Pan, Victor Junnan. 2018. Derivation of the apparent narrow scope of sentence-final particles in chinese: A reply to Erlewine (2017). Studies in Chinese Linguistics 39(2). 99–126.

A syntactic universal in a contact language

119

Pan, Victor Junnan & Waltraud Paul. 2015. What you see is what you get: Chinese sentencefinal particles as head-final complementisers. In Josef Bayer & Volker Struckmeier (eds.), Discourse Particles: Formal Approaches to their Syntax and Semantics, 49–77. Berlin: de Gruyter. Pan, Victor Junnan & Waltraud Paul. 2016. Why Chinese SFPs are neither optional nor disjunctors. Lingua 170. 23–34. Parviainen, Hanna. 2012. Focus particles in Indian English and other varieties. World Englishes 31(2). 226–247. Paul, Waltraud. 2014. Why particles are not particular: Sentence-final particles in Chinese as heads of a split CP. Studia Linguistica 68(1). 77–115. Paul, Waltraud. 2021. Nobody there? On the non-existence of nobody in Mandarin Chinese and related issues. Canadian Journal of Linguistics 66(3). 279–316. Philip, Joy. 2013. (Dis)harmony, the Head-Proximate filter, and linkers. Journal of Linguistics 49(1). 165–213. Platt, John T. 1975. The Singapore English speech continuum and its basilect ‘Singlish’ as a ‘creoloid’. Anthropological Linguistics 17. 363–374. Quek, Yihui & Aron Hirsch. 2017. Severing focus form and meaning in Standard and Colloquial Singapore English. In Andrew Lamont & Katerina Tetzlof (eds.), Proceedings of NELS 47, vol. 3, 15–25. Amherst, MA: University of Massachusetts Amherst. Roberts, Ian. 2007. Diachronic Syntax. Oxford: Oxford University Press. Sato, Yosuke. 2011. Radical pro drop and fusional pronominal morphology in Colloquial Singapore English: Reply to Neeleman and Szendroi.́ ́ Linguistic Inquiry 42. 356–365. Sato, Yosuke. 2014. Argument ellipsis in Colloquial Singapore English and the AntiAgreement Hypothesis. Journal of Linguistics 50. 365–401. Sheehan, Michelle, Theresa Biberauer, Ian Roberts & Anders Holmberg (eds.). 2017. The Final-over-Final Condition: A Syntactic Universal. Cambridge, MA: MIT Press. Siegel, Jef. 2012. Constraints on substrate transfer revisited. Journal of Linguistics 48(2). 473–478. Simpson, Andrew. 2001. Focus, presupposition and light predicate raising in Southeast Asian languages and Chinese. Journal of East Asian Linguistics 10(2). 89–128. Soh, Hooi Ling. 2009. Speaker presupposition and Mandarin Chinese sentence-final le: A unified analysis of the ‘change of state’ and the ‘contrary to expectation’ reading. Natural Language & Linguistic Theory 27. 623–657. Soh, Hooi Ling & Meijia Gao. 2006. Perfective aspect and transition in Mandarin Chinese: An analysis of double le sentences. In Proceedings of the 2004 Texas Linguistics Society Conference, 107–122. Somerville, MA: Cascadilla. Soh, Hooi Ling & Meijia Gao. 2008. Mandarin sentential le, perfect and English already. In Johannes Dölling, Tatjana Heyde-Zybatow & Martin Schäfer (eds.), Event Structures in Linguistic form and Interpretation, 447–473. Berlin: Walter de Gruyter. Sybesma, Rint P. E. 1997. Why Chinese verb-le is a resultative predicate. Journal of East Asian Linguistics 6. 215–261. Sybesma, Rint P. E. & Boya Li. 2007. The dissection and structural mapping of Cantonese sentence final particles. Lingua 117. 1739–1783. Tan, Ludwig. 2007. Null Arguments in Singapore Colloquial English. Dissertation, University of Cambridge.

120

Michael Yoshitaka Erlewine

Tang, Sze-Wing. 1998. Parametrization of Features in Syntax. Irvine dissertation, University of California. Wong, Jock O. 2014. The culture of Singapore English. Cambridge: Cambridge University Press. Zhang, Niina Ning. 2019. Sentence-final aspect particles as finite markers in Mandarin Chinese. Linguistics 57.5. 967–1023.

5

On the discourse marker dah in Colloquial Malay (and sudah in Sabah Malay) Hooi Ling Soh

1

Introduction1

The morpheme dah in Colloquial Malay (CM) spoken in West Malaysia has not received much attention in studies of Malay. To the extent that dah is mentioned, it is usually referred to as the colloquial counterpart of the aspectual auxiliary sudah in Standard Formal Malay (SFM) (e.g., Gan 1991: 29). Sudah in SFM has been noted to express completive or perfective aspect (Omar 1970; Salleh 1989; Soh 1994) and is often translated into English as already (e.g., Kader 1981: 36). Koh (1990: 202) attributes the same meanings to dah in Colloquial Malay. However, unlike sudah in SFM, dah in CM has a wider syntactic distribution. It may appear in three surface linear positions: pre-verbally, post-verbally, and sentence-finally. CM and SFM examples are given in (1) and (2), respectively. (1)

Colloquial Malay (West Malaysia)2 a. Aku dah sampai kat KLCC. 1sg dah arrive at KLCC “I have arrived at KLCC.” b. Aku sampai dah kat KLCC. 1sg arrive dah at KLCC “I have already arrived at KLCC.” c. Aku sampai kat KLCC dah. 1sg arrive at KLCC dah “I have already arrived at KLCC.”

(2)

Standard Formal Malay a. Saya sudah sampai di KLCC. 1sg sudah arrive at KLCC “I have arrived at KLCC.” b.*Saya sampai sudah di KLCC. 1sg arrive sudah at KLCC c.*Saya sampai di KLCC sudah. 1sg arrive at KLCC sudah DOI: 10.4324/9781351057752-6

122 Hooi Ling Soh In addition to preceding a prepositional phrase as in (1b), post-verbal dah can also precede an object noun phrase, as shown in (3) and (4). (3)

Colloquial Malay (West Malaysia) a. Dia dah tahu perkara ini. 3sg dah know matter this “S/he knows about this matter now (and didn’t before).” b. Dia tahu dah perkara ini. 3sg know dah matter this “S/he already knows about this matter.” c. Dia tahu perkara ini dah. 3sg know matter this dah “S/he already knows about this matter.”

(4)

Colloquial Malay (West Malaysia) a. Ali dah tegur anak-nya. Ali dah scold child-3sg “Ali has scolded his child.” b. Ali tegur dah anak-nya. Ali scold dah child-3sg “Ali has already scolded his child.” c. Ali tegur anak-nya dah. Ali scold child-3sg dah “Ali has already scolded his child.”

Preliminary observations regarding the meaning of dah in diferent positions do not reveal a readily accessible diference in meaning. For example, the sentences with dah in (1) all express that the arrival event has completed and imply that the speaker is at KLCC.3 There is also no contrast between dah in diferent positions in their ability to appear in wh-questions. (5)

a. Siapa dah sampai kat KLCC? who dah arrive at KLCC “Who has arrived at KLCC?” b. Siapa sampai dah kat KLCC? who arrive dah at KLCC “Who has already arrived at KLCC?” c. Siapa sampai kat KLCC dah? who arrive at KLCC dah “Who has already arrived at KLCC?”

Despite these observations, I argue in this chapter that one must distinguish two uses of dah: one as an aspectual marker (perfective/completive, following previous works) and one as a discourse marker that shares components of meaning with English already and Mandarin Chinese sentence-final particle -le.

On the discourse marker dah in Colloquial Malay

123

Whether dah is used as an aspectual marker or a discourse marker depends on its syntactic positioning. Specifically, I claim that while dah in a pre-verbal position serves as an aspectual marker, dah in a post-verbal or sentence-final position serves as a discourse marker.4 The argument is based on new observations about (i) patterns in the restrictions in the co-occurrence of dah in different linear positions and (ii) contrasts in the ability of dah in diferent linear positions to appear with downward-entailing quantifiers. I ofer preliminary analyses of the syntax of dah in diferent positions. I propose an analysis of the discourse dah as a marker of change, extending our analysis of Mandarin Chinese sentence-final particle -le (Soh & Gao 2008; Soh 2008, 2009) to CM post-verbal/sentence final dah. I show how the proposed analysis accounts for the restriction in the co-occurrence of post-verbal/sentence-final dah with downward-entailing quantifiers. I ofer additional support for the analysis from new observations about the distribution of dah in yes/no questions. I then compare the use of CM dah with sudah in Sabah Malay (SM) spoken in East Malaysia, which also may appear pre-verbally, post-verbally, and sentence-finally. SM has been noted to exhibit a preference for sudah to appear post-verbally (Hoogervorst 2011: 64–65), raising question about whether post-verbal sudah is in fact the aspectual counterpart of (pre-verbal) sudah in SFM. I show that like their counterparts in CM, post-verbal and sentence-final sudah pattern diferently from pre-verbal sudah. The results have implication on the cross-linguistic properties distinguishing “already” from perfect aspect (Vander Klok & Matthewson 2015; Nomoto & Mohd. Farez Syinon 2019). The current analysis connects the distribution of CM dah and SM sudah with Mandarin particle -le, which is known to have an aspectual function or a discourse function, depending on its syntactic positioning, whether post-verbally or sentence-finally (e.g., Li & Thompson 1981). As the aspectual and discourse -le have been analyzed as sharing a core meaning, with diferent semantic efects arising from their distinct syntactic positions (e.g., Huang & Davis 1989; Soh 2008), the current study adds two varieties of Malay to the group of languages that are fruitful to consider when determining how the syntactic positioning of a grammatical item may afect its meaning. This chapter is organized as follows: In Section 2, I show patterns in the co-occurrence restrictions of CM dah in diferent linear positions that support distinguishing pre-verbal dah from post-verbal and sentence-final dah. In Section 3, I show that CM pre-verbal dah contrasts with post-verbal and sentence-final dah with respect to their ability to appear with downwardentailing quantifiers. In Section 4, I present assumptions and preliminary analyses of the syntax of dah in diferent positions. In Section 5, I propose an analysis of the semantics of post-verbal and sentence-final dah in CM that accounts for their restricted distribution with downward-entailing quantifiers. I ofer support for this analysis from the appearance of dah in yes/no questions. In Section 6, I present new descriptions about the distribution of sudah in SM and show that it patterns like CM dah in the relevant respects. In Section 7, I discuss implications of the analysis.

124 2

Hooi Ling Soh

Patterns of co-occurrence restrictions with Colloquial Malay dah

More than one instance of dah may appear in a sentence in CM. However, there are restrictions on which instances of dah may co-occur. While pre-verbal dah can co-occur with post-verbal and sentence-final dah, post-verbal dah cannot co-occur with sentence-final dah, as shown in (6) and (7). (6)

a. Aku dah sampai kat KLCC dah. 1sg dah arrive at KLCC dah “I have already arrived at KLCC.” b. Aku dah sampai dah kat KLCC. 1sg dah arrive dah at KLCC “I have already arrived at KLCC.” c.*Aku sampai dah kat KLCC dah. 1sg arrive dah at KLCC dah

(7)

a. Dia dah tahu perkara ini dah. 3sg dah know matter this dah “S/he already knows about this matter.” b. Dia dah tahu dah perkara ini. 3sg dah know dah matter this “S/he already knows about this matter.” c.*Dia tahu dah perkara ini dah. 3sg know dah matter this dah

A sentence cannot contain three instances of dah, as shown in (8) and (9). (8)

*Aku dah sampai dah kat KLCC dah. 1sg dah arrive dah at KLCC dah

(9)

*Dia dah tahu dah perkara ini dah. 3sg dah know dah matter this dah

These restrictions suggest that pre-verbal dah is to be distinguished from post-verbal and sentence-final dah in terms of their syntax and/or semantics. In terms of syntax, assuming that a syntactic position can only be occupied by one dah, the pattern can be explained if we assume that pre-verbal dah occupies a distinct syntactic position from post-verbal and sentencefinal dah, with the latter instances of dah occupying the same syntactic position. The fact that a sentence may not contain three instances of dah is expected, given that only two syntactic positions exist for dah. With respect to semantics, the restrictions may be explained if we assume that items with identical meaning cannot co-occur and that pre-verbal dah has a distinct meaning from post-verbal and sentence-final dah, while the latter share the same meaning.

On the discourse marker dah in Colloquial Malay 3

125

Patterns of co-occurrence restrictions between dah and quantifiers

In this section, I show that pre-verbal dah contrasts with post-verbal and sentencefinal dah with respect to their ability to occur with downward-entailing quantifiers in specific sentences. The relevant sentences describe telic situations where the quantifiers specify information relating to the end point of the situation. 3.1

Downward-entailing versus non-downward-entailing quantifiers

Downward-entailing quantifiers (e.g., less than three students, at most three students, no student(s)) are quantifiers that license inferences from supersets to subsets.5 Thus, in the example in (10), where the set of people who voted is a superset of the set of people who voted for Bill, it is valid to infer that if less than three students voted is true, then it must be true that less than three students voted for Bill. (10) Less than three students voted. ––>Less than three students voted for Bill Downward-entailing quantifiers do not license inferences from subsets to supersets. Thus, it is not valid to infer from the truth of less than three students voted for Bill that less than three students voted. (11) Less than three students voted for Bill. ––>Less than three students voted. In CM, downward-entailing quantification expressions include those introduced by kurang dari(pada) “less than,” tak sampai “less than,” tak cukup “less than,” and phrases modified by baru “just” and (sa)je “only” and baru . . . (sa)je “only.”6,7 Non-downward-entailing quantifiers, on the other hand, do not license inferences from supersets to subsets. They include upward-entailing quantifiers (e.g., at least three students, some students), which license inferences from subsets to supersets, and non-entailing quantifiers (e.g., exactly three students), which do not license inferences in either direction. (12) a. At least three students voted for Bill. ––> b. At least three students voted.

––>

(13) a. Exactly three students voted for Bill. ––> b. Exactly three students voted.

––>

At least three students voted. At least three students voted for Bill. Exactly three students voted. Exactly three students voted for Bill.

In CM, non-downward-entailing quantifiers include bare numeral expressions and ones introduced by lebih dari (pada) “more than,” sekurang-kurangnya “at least,” lebih kurang “roughly,” dekat “close to,” as well as those introduced by lebih “more.”

126 3.2

Hooi Ling Soh Pre-verbal dah versus post-verbal and sentence-final dah

Pre-verbal dah may appear with both downward-entailing and non-downward-entailing quantifiers, as shown in (14) and (15). (14) a. Dia dah makan dua biji ubat (tu). 3sg dah eat two cl pill that “S/he has taken two (of those) pills.” b. Dia dah kerja lebih dari(pada) dua bulan. 3sg dah work more than two month “S/he has worked for more than two months.” c. Dia dah kahwin sekurang-kurangnya dua bulan. 3sg dah marry at least two month “S/he has been married for at least two months.” (15) a. Dia dah makan kurang dari(pada) dua biji ubat (tu). 3sg dah eat less than two cl pill that “S/he has taken less than two (of those) pills.” b. Dia dah kerja tak sampai dua bulan. 3sg dah work not reach two month “S/he has worked for less than two months.” c. Dia dah kahwin baru dua bulan je. 3sg dah marry just two month only “S/he has been married for only two months.” Unlike pre-verbal dah, sentence-final dah may appear with non-downwardentailing quantifiers, but not with downward-entailing quantifiers, as shown in (16) and (17). (16) a. Dia makan dua biji ubat (tu) dah. 3sg eat two cl pill that dah “S/he has taken two (of those) pills already.” b. Dia kerja lebih dari(pada) dua bulan dah. 3sg work more than two month dah “S/he has worked for more than two months already.” c. Dia kahwin sekurang-kurangnya dua bulan dah. 3sg marry at least two month dah “S/he has been married for at least two months already.” (17) a. #Dia makan kurang dari(pada) dua biji ubat (tu) dah. 3sg eat less than two cl pill that dah “#S/he has taken less than two (of those) pills already.” b. #Dia kerja tak sampai dua bulan dah. 3sg work not reach two month dah “#S/he has worked for less than two months already.”

On the discourse marker dah in Colloquial Malay

127

c. #Dia kahwin baru dua bulan je dah. 3sg marry just two month only dah “#S/he has been married for only two months already.” Like pre-verbal and sentence-final dah, post-verbal dah may appear with nondownward-entailing quantifiers, as shown in (18). (18) a. Dia makan dah dua biji ubat (tu). 3sg eat dah two cl pill that “S/he has taken two (of those) pills already.” b. Dia kerja dah lebih dari(pada) dua bulan. 3sg work dah more than two month “S/he has worked for more than two months already.” c. Dia kahwin dah sekurang-kurangnya dua bulan. 3sg marry dah at least two month “S/he has been married for at least two months already.” With its occurrence with downward-entailing quantifiers, post-verbal dah exhibits variable behaviors that are prosodically conditioned. It may not occur with a downward-entailing quantifier within the same intonational phrase but may do so when the quantifier and dah appear in separate intonational phrases (intonational phrase boundary marked by //). (19) a. #Dia makan dah kurang dari(pada) dua biji ubat (tu). 3sg eat dah less than two cl pill that “#S/he has taken less than two (of those) pills already.” b. #Dia kerja dah tak sampai dua bulan. 3sg work dah not reach two month “#S/he has worked for less than two months already.” c. #Dia kahwin dah baru dua bulan je. 3sg marry dah just two month only “#S/he has been married for only two months already.” (20) a. Dia makan dah // kurang dari(pada) dua biji ubat (tu). 3sg eat dah less than two cl pill that “S/he has taken (something) already, less than two (of those) pills.” b. Dia kerja dah // tak sampai dua bulan. 3sg work dah not reach two month “S/he has worked already, for less than two months.” c. Dia kahwin dah // baru dua bulan je. 3sg marry dah just two month only “S/he has been married already, for only two months.” Thus, when prosody is taken into account, post-verbal dah patterns like sentence-final dah in exhibiting a restricted distribution with downward-entailing quantifiers.

128 3.3

Hooi Ling Soh Sentences with more than one dah

When more than one dah appears in a sentence, as when pre-verbal dah cooccurs with either post-verbal or sentence-final dah, the same restriction as post-verbal and sentence-final dah sentences is found with respect to the occurrence of downward-entailing quantifiers. Sentences with pre-verbal dah and sentence-final dah may contain non-downward-entailing quantifiers but may not contain downward-entailing quantifiers, as shown in (21) and (22). (21) a. Dia dah makan dua biji ubat (tu) dah. 3sg dah eat two cl pill that dah “S/he has taken two (of those) pills already.” b. Dia dah kerja lebih dari(pada) dua bulan dah. 3sg dah work more than two month dah “S/he has worked for more than two months already.” c. Dia dah kahwin sekurang-kurangnya dua bulan dah. 3sg dah marry at least two month dah “S/he has been married for at least two months already.” (22) a. #Dia dah makan kurang dari(pada) dua biji ubat (tu) dah. 3sg dah eat less than two cl pill that dah “#S/he has taken less than two (of those) pills already.” b. #Dia dah kerja tak sampai dua bulan dah. 3sg dah work not reach two month dah “#S/he has worked for less than two months already.” c. #Dia dah kahwin baru dua bulan je dah. 3sg dah marry just two month only dah “#S/he has been married for only two months already.” Similarly, sentences with pre-verbal dah and post-verbal dah may contain nondownward-entailing quantifiers, as shown in (23), but may not contain downward-entailing quantifiers within the same intonational phrase, as shown in (24) and (25). (23) a. Dia dah makan dah dua biji ubat (tu). 3sg dah eat dah two cl pill that “S/he has taken two (of those) pills already.” b. Dia dah kerja dah lebih dari(pada) dua bulan. 3sg dah work dah more than two month “S/he has worked for more than two months already.” c. Dia dah kahwin dah sekurang-kurangnya dua bulan. 3sg dah marry dah at least two month “S/he has been married for at least two months already.” (24) a. #Dia dah makan dah kurang dari(pada) dua biji ubat (tu). 3sg dah eat dah less than two cl pill that “#S/he has taken less than two (of those) pills already.”

On the discourse marker dah in Colloquial Malay

129

b. #Dia dah kerja dah tak sampai dua bulan. 3sg dah work dah not reach two month “#S/he has worked for less than two months already.” c. #Dia dah kahwin dah baru dua bulan je. 3sg dah marry dah just two month only “#S/he has been married for only two months already.” (25) a. Dia dah makan dah // kurang dari(pada) dua biji ubat (tu). 3sg dah eat dah less than two cl pill that “S/he has taken (something) already, less than two (of those) pills.” b. Dia dah kerja dah // tak sampai dua bulan. 3sg dah work dah not reach two month “S/he has worked already, for less than two months.” c. Dia dah kahwin dah // baru dua bulan je. 3sg dah marry dah just two month only “S/he has been married already, for only two months.” The distribution of dah with downward-entailing quantifiers further supports the claim that post-verbal dah and sentence-final dah are distinct from preverbal dah. 4

4.1

Assumptions and preliminary analyses of the syntax of dah in diferent positions The syntax of pre-verbal dah

There is evidence that pre-verbal dah occupies an aspectual position above the temporal auxiliary nak and the negative marker tidak/tak.8 The evidence involves the relative ordering and scope between pre-verbal dah and these elements. Preverbal dah can appear before the temporal auxiliary nak and the negative marker tak/tidak, but not after them, as shown in (26) and (27). (26) a. Aku dah nak sampai kat KLCC. 1sg dah fut arrive at KLCC “I am about to arrive at KLCC now (and didn’t before).” b. * Aku nak dah sampai kat KLCC. 1sg fut dah arrive at KLCC (27) a. Aku dah tak ingat nama dia (dah). 1sg dah not remember name 3sg dah “I don’t remember his/her name now (but did before).” b. * Aku tak dah ingat nama dia (dah). 1sg not dah remember name 3sg dah The interpretations suggest that the aspectual marker takes scope over these elements. For example, (27) expresses the idea that the speaker does not

130

Hooi Ling Soh

remember the individual’s name but used to, rather than that it is not the case that the speaker used to remember the individual’s name. These facts support placing pre-verbal dah in an aspectual position above the temporal auxiliary nak and the negative marker tak, as shown in (28). (28) a. . . . [ASPP dah [AuxP nak [vP sampai kat KLCC]]] b. . . . [ASPP dah [NegP tak [vP ingat nama dia]]] 4.2

The syntax of post-verbal and sentence-final dah

I assume that post-verbal dah and sentence-final dah occupy the same syntactic position. Following Erlewine (2017) and Soh (2019, Volume I), I assume the availability of two syntactic areas where discourse-related particles may appear: one in the CP edge above items associated with sentence force (see, for example, Paul (2014)) and one in the clause-medial area between TP and vP. I assume that the syntactic position of post-verbal and sentence-final dah parallels that of discourse markers like Mandarin sentence-final -le (Soh & Gao 2006; Erlewine 2017) and Mandarin sentence-final de (Soh, Volume I) and appears in a position between TP and a modal projection associated with non-epistemic modals. The projection is right-headed, and I label it dah-P for convenience. I further assume that post-verbal instances of dah are derived via the extraposition of the preposition phrase or the object noun phrase to rightadjoin to TP. (1c), with sentence-final dah, and (1b), with post-verbal dah (repeated in (29)), are assumed to have the derivations given in (30). (29) a. Aku sampai kat KLCC dah. 1sg arrive at KLCC dah “I have already arrived at KLCC.” b. Aku sampai dah kat KLCC. 1sg arrive dah at KLCC “I have already arrived at KLCC.” (30) a. . . . [TP aku [dahP [vP sampai kat KLCC] dah]] b. . . . [TP [TP aku [dahP [vP sampai ] dah]] [PP kat KLCC]] I assume that examples in (20) involve two independent clauses (with possible elided materials in the second clause), separated by the prosodic boundary.9 (20a) is repeated in the following in (31a) with the assumed structures in (31b). (31) a. Dia makan dah // kurang dari(pada) dua biji ubat (tu). 3sg eat dah less than two cl pill that “S/he has taken (something) already, less than two (of those) pills.” b. . . . [TP dia [dahP [vP makan] dah]] // . . . [TP [DP kurang dari(pada) dua biji ubat (tu)]]

On the discourse marker dah in Colloquial Malay 5

131

The semantics of post-verbal and sentence-final dah

In this section, I present an analysis of the semantics of post-verbal and sentencefinal dah that extends our analysis of Mandarin sentence-final particle -le (Soh & Gao 2008; Soh 2008, 2009) to CM post-verbal/sentence-final dah.10 5.1

Assumptions

The proposed analysis relies on the notion of presupposition and common ground in the sense of Stalnaker (1998, 1999, 2002), where common ground refers to what is accepted among the participants in the conversation, and presuppositions refer to the speaker’s beliefs about the common ground. Note that the definition of common ground makes use of the notion of acceptance (Stalnaker 2002), which is broader than the notion of belief. Acceptance of a proposition includes belief, presumption, assumption, and acceptance of a proposition for the purposes of an argument or an inquiry. Thus, one may accept a proposition for the purpose of the conversation without believing that the proposition is true. (32) It is common ground that  in a group if all members accept (for the purpose of the conversation) that , and all believe that all accept that , and all believe that all believe that all accept that , etc. (Stalnaker 2002: 716) I assume that what a speaker asserts is taken as proposals to change the common ground. The common ground changes as the participants in a conversation accept what the speaker asserts and include the relevant proposition to the subsequent common ground (Stalnaker 1999: 86). 5.2

Proposal

The proposed analysis for post-verbal and sentence-final dah in CM is as follows: (33) The speaker using a sentence with a post-verbal or sentence-final dah (i) asserts a proposition p at speech time (ts), (ii) presupposes [¬p before speech time (ts)], and (iii) accepts the inclusion of the presupposition in the subsequent common ground. As in Soh (2009), two common grounds at two distinct times are relevant for the analysis. The first is the common ground characterizing the context for the time (shortly) before, during, and (immediately) after the assertion is made, but before the acceptance or rejection of the assertion (Stalnaker 1998: 8). I refer to this common ground as common groundi at the interval ti, with ts

132

Hooi Ling Soh

included in ti. The second is the common ground subsequent to speech time if what is asserted at speech time is accepted by the participants of the conversation. I refer to this common ground as common groundj at the interval tj, with tj after ti. The relations among the times relevant for the analysis are diagrammed in the time line in (34). ti, the time that common groundi holds, precedes tj, the time that common groundj holds; and ts, speech time, is included in ti. (34) Time Line ts --------[------------[------------]---]-------------[-----------------]--------------ti tj (33ii) makes reference to the common groundi at ti. The speaker believes that [¬ p before speech time] is part of the common groundi at ti. (33i) and (33iii) have an efect on the common groundj at tj, if what is asserted in (33i) and accepted in (33iii) by the speaker is accepted by the participants of the conversation. The table in (35) is a schematic representation of the speaker’s beliefs about the common grounds with the use of post-verbal or sentence-final dah. P1, P2, and P3 refer to the propositions involved. P3 has its source from (33i), P1 from (33ii), and P2 from (33iii). (35) Using post-verbal or sentence-final dah common groundi at ti

common groundj at tj

P1: ¬ p before ts. [presupposition]

P2: ¬ p before ts. [acceptance of presupposition] P3: p at ts. [assertion]

5.3

Explaining the distribution of post-verbal and sentence-final dah with quantifiers

The restricted distribution of post-verbal and sentence-final dah with downwardentailing quantifiers follows from the speaker’s presupposition and how negation interacts with downward-entailing quantifiers. Consider the restricted distribution of post-verbal and sentence-final dah with the downwardentailing quantifier tak sampai dua bulan “less than two months” in what follows: (36) a. #Dia kerja dah tak sampai dua bulan. 3sg work dah not reach two month “#S/he has worked for less than two months already.”

On the discourse marker dah in Colloquial Malay

133

b. #Dia kerja tak sampai dua bulan dah. 3sg work not reach two month dah “#S/he has worked for less than two months already.” In uttering either (36a) or (36b), the speaker asserts that the proposition s/he has worked for less than two months is true at speech time. I assume that the negation of [S/he has worked for less than two months] has the same meaning as [S/he has worked for two months or more]. Thus, the use of sentence-final dah indicates that the speaker believes that the common ground (common groundi) includes the proposition s/he has worked for two months or more before ts (P1). The speaker includes the presupposition in the subsequent common ground by adding P2. However, the inclusion of the presupposition in the subsequent common ground is impossible because P2 contradicts the assertion P3, which is also to be included in the same subsequent common ground. The speaker uttering either (36a) or (36b) thus expresses a belief about the truth of two propositions, P2 and P3, that are incompatible given our conception of time. The sentences are thus considered anomalous. (37) Using post-verbal or sentence-final dah common groundi at ti

common groundj at tj

P1: S/he has worked for two months or more before ts. [presupposition]

P2: S/he has worked for two months or more before ts. [acceptance of presupposition] P3: S/he has worked for less than two months at ts. [assertion]

As shown earlier, there is no problem for post-verbal and sentence-final dah to appear with non-downward-entailing quantifiers: (38) a. Dia kerja dah lebih dari(pada) dua bulan. 3sg work dah more than two month “S/he has worked for more than two months already.” b. Dia kerja lebih dari(pada) dua bulan dah. 3sg work more than two month dah “S/he has worked for more than two months already.” In uttering either (38a) or (38b), the speaker asserts that the proposition s/ he has worked for more than two months is true at speech time. I assume that the negation of [S/he has worked for more than two months] has the same meaning as [S/he has worked for less than two months]. Thus, the use of sentence-final dah indicates that the speaker believes that the common ground

134

Hooi Ling Soh

(common groundi) includes the proposition s/he has worked for less than two months before ts (P1). In this case, the inclusion of the presupposition in the subsequent common ground is possible because P2 does not contradict the assertion P3, which is also to be included in the same subsequent common ground. (39) Using post-verbal or sentence-final dah common groundi at ti

common groundj at tj

P1: S/he has worked for two months or less before ts. [presupposition]

P2: S/he has worked for two months or less before ts. [acceptance of presupposition] P3: S/he has worked for more than two months at ts. [assertion]

5.4

Further support

In this section, I ofer new empirical generalizations on the distribution of post-verbal/sentence-final dah in a type of yes/no questions formed with the particle ke that provide further support for the proposed analysis. 5.4.1

Yes/no questions with the particle ke

The question particle ke in CM can occur in a pragmatically neutral yes/no question (with a sentential focus) or a yes/no question with a marked focus (with the focus of the question associated with a constituent within the sentence) (Kader 1981; Cantor 2010; Nomoto 2006; Fortin 2009). In pragmatically neutral questions, the question particle ke appears at the end of the sentence, as in (40a), or after the verb, as in (40b).11 (40) a. Kau dah sampai kat KLCC ke?12 2sg dah arrive at KLCC q “Have you arrived at KLCC?” b. Kau dah sampai ke kat KLCC? 2sg dah arrive q at KLCC “Have you arrived at KLCC?” In a question with a marked focus, the question particle ke follows the constituent marked for focus, whether it is a nominal or a prepositional phrase. The focused constituent marked with ke may appear in a left periphery position or remain in situ (see Kader (1981) for some restrictions). When the focused constituent occupies a left periphery position, the appearance of yang, which I assume to be a complementizer, is optional.

On the discourse marker dah in Colloquial Malay

135

(41) a. Buku tu ke (yang) kau beli semalam? book that q yang 2sg buy last.night “Was it that book you bought last night?” b. Kau beli buku tu ke semalam? 2sg buy book that q last.night “Was it that book you bought last night?” When the focused constituent is a subject, the appearance of yang is preferred. The in situ option is considered marginal.13 (42) a. Kau ke yang dah sampai kat KLCC? 2sg q yang dah arrive at KLCC “Was it you that had arrived at KLCC?” b. ?? Kau ke dah sampai kat KLCC? 2sg q dah arrive at KLCC 5.4.2

The distribution of post-verbal/sentence-final dah in yes/no questions with ke

Focusing on the distribution of ke in pragmatically neutral yes/no questions, I show in the following that sentence-final and post-verbal dah can appear with ke in these questions, but only if dah precedes ke. (43) shows examples with both dah and ke at the end of the sentence, while (44) shows examples of both these items appearing post-verbally. The accepted order is one of dah preceding ke.14 (43) a. Kau (dah) sampai kat KLCC dah ke? 2sg dah arrive at KLCC dah q “Have you already arrived at KLCC?” b. *Kau   (dah) sampai kat KLCC ke dah? 2sg dah arrive at KLCC q dah (44) a. Kau (dah) sampai dah ke kat KLCC? 2sg dah arrive dah q at KLCC “Have you already arrived at KLCC?” b. *Kau   (dah) sampai ke dah kat KLCC? 2sg dah arrive q dah at KLCC Sentences with a post-verbal ke followed by a sentence-final dah are not acceptable, as shown in (45). (45) * Kau (dah) sampai ke kat KLCC dah?  2sg dah arrive q at KLCC dah For sentences with a sentence-final ke and a post-verbal dah, there are variations among speakers in their acceptability.

136

Hooi Ling Soh

(46) (*) Kau (dah) sampai dah kat KLCC ke? 2sg dah arrive dah at KLCC q “Have you already arrived at KLCC?” 5.4.3

Assumptions regarding the syntax of the particle ke

Many previous researchers have considered kah (the formal counterpart of ke) to be an interrogative marker with a focus feature. Some have analyzed kah in Malay as attached to a focused constituent (e.g., Kader 1981). Others have suggested that it is located in C (e.g., Cantor 2010) or Foc (e.g., Nomoto 2006), assuming that the CP layer is further split into distinct functional projections in accordance with Rizzi (1997). Fortin (2009) argues that kah in Indonesian is located in Foc, while Malay kah (though related to Foc) is likely not in Foc. This is because unlike Indonesian, Malay allows kah to appear in non-periphery positions, when the focused constituent with kah remains in situ (see (41b)). Following Kader (1981) and extending Fortin (2009) on the diference between Indonesian and Malay, I assume that ke is adjoined to the right of the focused constituent rather than occupying Foc. I assume that CM has an overt Foc with an EPP feature, namely, yang, and two types of null Foc, one with an EPP feature and one without. A Foc with an EPP feature forces movement of the focused constituent to its specifier, while a Foc without an EPP feature allows the focused constituent to remain in situ. I further assume that ke bears a [+Q feature] that raises to Foc, and the scope of the yes/no question is the sister node of ke.15 The structures assumed for (41a) and (41b) (repeated here) are given here: (47) a. Buku tu ke (yang) kau beli semalam? book that q yang 2sg buy last.night “Was it that book you bought last night?” b. [FOCP [DP [DP buku tu] ke] [FOC’ (yang) [TP kau beli semalam]]] (48) a. Kau beli buku tu ke semalam? 2sg buy book that q last.night “Was it that book you bought last night?” b. [FOCP [TP kau beli [DP [DP buku tu] ke] semalam]] In cases where the question particle ke appears at the end of the sentence as in (40a) (repeated here) and the sentence is interpreted with a sentential focus, I assume that the entire TP is focused and the focused constituent remains in situ. (49) a. Kau dah sampai kat KLCC ke? 2sg dah arrive at KLCC q “Have you arrived at KLCC?” b. [FOCP [TP [TP kau dah sampai kat KLCC] ke]]

On the discourse marker dah in Colloquial Malay

137

The focused TP does not have the option of raising, given that an overt Foc with an EPP feature, namely, yang, cannot appear following the focused TP. (50) a. *Kau dah sampai kat KLCC ke yang? b. [FOCP [TP [TP kau dah sampai kat KLCC] ke] [FOC’ yang ]] With (40b), repeated in the following, which also is interpreted as having sentential focus, I assume that the prepositional phrase has been extraposed to the right and adjoined to TP. (51) a. Kau dah sampai ke kat KLCC? 2sg dah arrive q at KLCC “Have you arrived at KLCC?” b. [FOCP [TP [TP [TP Kau dah sampai ] ke] [PP kat KLCC]] ] 5.4.4

Explaining the ordering/scope restriction between ke and dah

The pattern of restrictions resembles the ones found with Mandarin sentencefinal le (Soh & Gao 2006; Erlewine 2017) as well as discourse marker de (Soh, Volume I) with the A-not-A form of yes/no questions. I extend our analysis in Soh and Gao (2006) and Soh (Volume I) to account for the restriction in CM. I suggest that dah cannot scope over a constituent with a [+Q feature], given the assumption that the question particle ke bear a [+Q-feature]. Consider the appearance of dah and ke at the end of the sentence. (43a), repeated in (52a), has the structure given in (52b) (ignoring the optional pre-verbal dah in the following discussion). (52) a. Kau (dah) sampai kat KLCC dah ke? 2sg dah arrive at KLCC dah q “Have you already arrived at KLCC?” b. [FOCP [TP [TP kau [dahP [vP sampai kat KLCC] dah]] ke]] (43b), repeated in (53a), has the structure in (53b). The ordering can be derived with ke adjoined to vP (rather than TP). Even so, the sentence is unacceptable, as dah scopes over a [+Q feature]. (53) a. *Kau (dah) sampai kat KLCC ke dah? 2sg dah arrive at KLCC q dah b. [FOCP [TP kau [dahP [vP [vP sampai kat KLCC] ke] dah]]] The same situation obtains with post-verbal ordering of dah and ke. The prepositional phrase extraposes and adjoins to TP and appears after dah and ke. The structures for (44a) and (44b) are given in the following.

138

Hooi Ling Soh

(54) a. Kau (dah) sampai dah ke kat KLCC? 2sg dah arrive dah q at KLCC “Have you already arrived at KLCC?” b. [FOCP [TP [TP [TP kau [dahP [vP sampai ] dah]] ke] kat KLCC]] (55) a. *Kau (dah) sampai ke dah kat KLCC? 2sg dah arrive q dah at KLCC b. [FOCP [TP [TP kau [dahP [vP [vP sampai ] ke] dah]] kat KLCC]] A post-verbal ke cannot be followed by a sentence-final dah, as in (45), repeated in the following, because dah would scope over a [+Q feature]. Also, the ordering would involve the prepositional phrase extraposed to adjoin to vP, rather than TP, as assumed, to appear after ke and before dah. (56) a. *Kau (dah) sampai ke kat KLCC dah? 2sg dah arrive q at KLCC dah b. [FOCP [TP kau [dahP [vP [vP [vP sampai ] ke] kat KLCC] dah]]] As noted earlier, there are variations among speakers in the acceptability of sentences with a sentence-final ke and a post-verbal dah. The structure observes the constraint against having a [+Q feature] within the scope of dah. The variation in judgments may be due to whether individuals allow an extraposed constituent to be within the domain of the question particle. (57) a. (*) Kau (dah) sampai dah kat KLCC ke? 2sg dah arrive dah at KLCC q “Have you already arrived at KLCC?” b. [FOCP [TP [TP [TP kau [dahP [vP sampai ] dah]] kat KLCC] ke]] The restriction against having a [+Q-feature] within the scope of dah supports the proposed analysis for dah. This is because the use of dah involves the speaker making assertions and presuppositions that involve propositions rather than questions.16 It is interesting to note that the restriction exhibited by sentence-final dah contrasts with sentence-final discourse particle punya in CM, which I analyze in Soh (2019) as a modal evidential heading a projection in the CP domain, namely, CPattitude (following Paul (2014)). Unlike dah, punya cannot appear in questions (whether wh-questions or yes/no questions). The contrast between dah and punya further supports the availability of two possible areas for discourse markers and shows the utilization of both syntactic areas for discourse markers within CM.

On the discourse marker dah in Colloquial Malay 6

139

Sabah Malay sudah

While there has not been much work on Sabah Malay (SM) spoken in East Malaysia, the syntactic positioning of sudah in Sabah Malay has sometimes been noted to be distinctive of this variety of Malay. Like CM dah, SM sudah may appear in three linear positions: pre-verbally, post-verbally, and sentencefinally, as shown in (58). (58) Sabah Malay a. Saya sudah sampai di 1B. 1sg sudah arrive at 1B “I have arrived at 1B.” b. Saya sampai sudah di 1B. 1sg arrive sudah at 1B “I have already arrived at 1B.” c. Saya sampai di 1B sudah. 1sg arrive at 1B sudah “I have already arrived at 1B.” Hoogervorst (2011: 64–65) notes a preference for the aspectual marker sudah to appear post-verbally in SM. This raises the question of whether postverbal sudah is, in fact, the aspectual counterpart of sudah in SFM. I show in what follows that like CM dah, post-verbal and sentence-final sudah pattern distinctly from pre-verbal sudah with respect to the pattern of co-occurrence restrictions of sudah in diferent linear positions, and with respect to their occurrence with quantifiers. 6.1

Patterns of co-occurrence restrictions of sudah in Sabah Malay

More than one sudah may appear in a sentence in SM, though there may be more individual variations with the acceptance of sentences with two instances of sudah in Sabah Malay compared to that of dah in Colloquial Malay. For those who accept more than one sudah in a sentence, the patterns of cooccurrence are the same as that of dah in CM. While pre-verbal sudah can cooccur with post-verbal and sentence-final sudah, as shown in (59a) and (59b), post-verbal sudah cannot appear with sentence-final sudah, as shown in (59c). (59) a. Saya sudah sampai di 1B sudah. 1sg sudah arrive at 1B sudah “I have already arrived at 1B.” b. Saya sudah sampai sudah di 1B. 1sg sudah arrive sudah at 1B “I have already arrived at 1B.” c. *Saya sampai sudah di 1B sudah. 1sg arrive sudah at 1B sudah

140

Hooi Ling Soh

A sentence also cannot contain three instances of sudah, as shown in (60). (60) *Saya sudah sampai sudah di 1B sudah. 1sg sudah arrive sudah at 1B sudah The patterning suggests that pre-verbal sudah is to be distinguished from post-verbal and sentence-final sudah, either in its syntax, semantics, or both. 6.2

Co-occurrence restrictions of sudah with quantifiers in Sabah Malay

Sudah may appear with non-downward-entailing quantifiers, regardless of its linear position in the sentence. In SM, non-downward-entailing quantifiers include bare numeral expressions and ones introduced by lebih (dari) “more than,” paling kurang “at least,” sekurang-kurangnya “at least,” and lebih kurang “roughly.” Examples involving a bare numeral phrase and quantifier phrase introduced by paling kurang “at least” are given in the following: (61) a. Dia sudah makan dua biji ubat. 3sg sudah eat two cl pill “S/he has taken two (of those) pills.” b. Dia makan dua biji ubat sudah. 3sg eat two cl pill sudah “S/he has already taken two (of those) pills.” c. Dia makan sudah dua biji ubat. 3sg eat sudah two cl pill “S/he has already taken two (of those) pills.” (62) a. Dia sudah makan paling kurang dua biji ubat. 3sg sudah eat at least two cl pill “S/he has taken at least two (of those) pills.” b. Dia makan paling kurang dua biji ubat sudah. 3sg eat at least two cl pill sudah “S/he has already taken at least two (of those) pills.” c. Dia makan sudah paling kurang dua biji ubat. 3sg eat sudah at least two cl pill “S/he has already taken at least two (of those) pills.” With downward-entailing quantifiers such as tidak sampai “less than,” tidak cukup “less than,” kurang “less than,” baru “only,” saja “only,” and baru . . . saja “only,” there is a contrast between pre-verbal and sentence-final sudah. While pre-verbal sudah may appear with these quantifiers, sentence-final sudah may not. Examples with quantifiers containing saja “only” and tidak sampai “less than” are given in the following:

On the discourse marker dah in Colloquial Malay

141

(63) a. Dia sudah makan dua biji ubat saja. 3sg sudah eat two cl pill only “S/he has taken only two (of those) pills.” b. #Dia makan dua biji ubat saja sudah. 3sg eat two cl pill only sudah (64) a. Dia sudah makan tidak sampai dua biji ubat. 3sg sudah eat less than two cl pill “S/he has taken less than two (of those) pills.” b. #Dia makan tidak sampai dua biji ubat sudah. 3sg eat less than two cl pill sudah “S/he has already taken less than two (of those) pills.” In terms of occurring with downward-entailing quantifiers, post-verbal sudah exhibits variable behavior that is conditioned by prosody. It may not occur with a downward-entailing quantifier within the same intonational phrase but may do so when the quantifier and sudah appear in separate intonational phrases. (65) a. #Dia makan sudah dua biji ubat saja. 3sg eat sudah two cl pill only b. Dia makan sudah // dua biji ubat saja. 3sg eat sudah two cl pill only “S/he has taken (the pills) already. Only two (of those) pills.” (66) a. #Dia makan sudah tidak sampai dua biji ubat. 3sg eat sudah less than two cl pill b. Dia makan sudah //tidak sampai dua biji ubat. 3sg eat sudah less than two cl pill “S/he has taken (the pills) already. Less than two (of those) pills.” Thus, when prosody is taken into account, post-verbal sudah patterns like sentence-final sudah in exhibiting a restricted distribution with downward-entailing quantifiers. The distribution of sudah with downward-entailing quantifiers supports the claim that post-verbal sudah and sentence-final sudah are distinct from pre-verbal sudah. Given that the distribution of SM sudah patterns like that of CM dah, I assume that the same analysis for post-verbal and sentence-final dah in CM can be extended to post-verbal and sentence-final sudah in SM. 7

Implications and conclusions

The current descriptions and analyses ofer evidence that pre-verbal dah and sudah in CM and SM, respectively, are to be distinguished from their post-verbal and sentence-final counterparts. Questions remain regarding how precisely pre-verbal dah/sudah and post-verbal/sentence-final dah/sudah in CM/SM

142

Hooi Ling Soh

are to be distinguished semantically and how their meanings compare to perfect aspect and English already. In Soh and Gao (2008), we show that English already may not appear in a sentence with a downward-entailing quantifier, as in (67a), a restriction not shared by English perfect aspect, as in (67b). (67) a. #I (have) already lived there for less than three years. b. I have lived there for less than three years. Post-verbal and sentence-final dah/sudah in CM/SM thus pattern like English already in their restricted distribution with downward-entailing quantifiers, while pre-verbal dah/sudah in CM/SM patterns like perfect and unlike English already in the relevant respect. The patterning suggests that post-verbal and sentence-final dah/sudah in CM/SM share a certain component of meaning with English already that is not available in pre-verbal dah/sudah in CM/SM. The results have implication on the cross-linguistic properties distinguishing “already” from perfect aspect (Vander Klok & Matthewson 2015). Vander Klok and Matthewson (2015) identify several cross-linguistic properties that distinguish “already” from the perfect aspect, and examine the semantics of wis in Javanese, which has been connected to sudah in Indonesian and has been variously glossed as already, a (present) perfect, a past, or a perfective marker. They argue that wis is not a marker of perfect aspect but rather is a focus operator that presupposes that the focus is a maximal element among a set of ordered alternatives, following Krifka’s (2000) analysis of English already. Nomoto and Mohd. Farez Syinon (2019) apply the diagnostics in Vander Klok and Matthewson (2015) to Malay pre-verbal sudah/dah and reach the same conclusion, that they have the semantics of “already” (following Krifka’s (2000) analysis) rather than perfect aspect. These results may be reconciled with the current study by distinguishing diferent types of “already” uses (e.g., Löbner 1989, Erlewine (this volume)), separating those that are incompatible with downward-entailing quantifiers in specific contexts from those that do not exhibit such restrictions. The current descriptions and analyses connect dah/sudah in CM/SM with studies on Mandarin Chinese particle -le. Like dah/sudah in CM/SM, the particle -le in Mandarin may appear in more than one position in a sentence: immediately after the verb (verbal -le) or at the end of the sentence (sentential -le), as in (68). It also may appear in both positions simultaneously, as shown in (69). (68) a. Tamen daoda-le shan-ding. 3pl reach-le mountain-top “They reached the top of the mountain.” b. Tamen daoda shan-ding le. 3pl reach mountain-top le “They reached the top of the mountain (which they hadn’t done/ contrary to what one may expect).”

On the discourse marker dah in Colloquial Malay

143

(69) Tamen daoda-le shan-ding le. 3pl reach-le mountain-top le “They reached the top of the mountain, (which they hadn’t before/ contrary to what one may expect).” The particle -le has been the subject of much discussion in the literature. A central issue is whether verbal -le and sentential -le are related and, if so, how. Many authors have taken the position that they are unrelated, though they difer in their specific proposals (e.g., Chao 1968; Li & Thompson 1981; Ross 1995; Smith 1997; Sybesma 1999). For example, Li and Thompson (1981) argue that verbal -le is a perfective aspect marker, while sentential -le marks “current relevant state,” similar to perfect in English. Others argue that verbal -le and sentential -le share a core meaning, with diferences between them either lexically specified (Lin 2003) or specified through a diference in scope (e.g., Huang & Davis 1989; Soh 2008). For example, in Soh (2008), I propose an analysis of -le as a marker of transition/change, while distinguish three types of transitions: transitions involving propositions (P-transitions), transitions involving eventualities (E-transitions), and transitions involving values on a scale (V-transitions). Under the proposed analysis, the syntactic position of -le partially determines the type of change that can be expressed by -le. By connecting the study of dah/ sudah in CM/SM to Mandarin le, the current study adds two diferent varieties of Malay to the group of languages that are fruitful to consider when determining how the syntactic positioning of a grammatical item may afect its meaning. Abbreviations 1 2 3 cl

first person second person third person classifier

fut pl q sg

future plural question singular

Notes 1 The research reported here received support from a Grant-in-Aid and an Imagine Fund Annual Faculty Award from the University of Minnesota. I am grateful to speakers of Colloquial Malay and Sabah Malay I consulted with for this project, and to Hiroki Nomoto for assistance with the project and for discussions. I would like to thank Asako Shiohara for the workshop on Sabah Malay in Kota Kinabalu, Malaysia, in August 2016, supported by the Research Institute for Languages and Cultures of Asia and Africa (ILCAA), Tokyo University of Foreign Studies (TUFS), where I began my research on Sabah Malay. I would also like to thank Jozina Vander Klok for helpful comments about the chapter. All errors are mine. 2 The following abbreviations are used: 1 first person; 2 second person; 3 third person; sg singular; pl plural; cl classifier; fut future; q question.

144

Hooi Ling Soh

3 An in-depth investigation comparing the semantic contributions of dah in diferent positions, with careful considerations of the aspectual situation types of the sentences, is left for future work. 4 In Ambonese Malay, the corresponding pre-verbal marker is su, while the sentencefinal particle appears as suda (van Minde & Tjia 2002). It is not clear if there is a post-verbal variant in Ambonese Malay. Interestingly, van Minde and Tjia (2002) claim that while su is a tense-aspect marker, suda is a discourse marker. In Indonesian, sudah seems to be able to appear in both pre-verbal and sentence-final position (van Minde & Tjia 2002). It is also not clear if there is a post-verbal variant in Indonesian. 5 The discussion here is based on Horn (1996: 9). 6 The literal translations for kurang dari(pada), tak sampai, and tak cukup are “less from,” “not reach,” and “not enough,” respectively. 7 See von Fintel (1999) for the way in which only phrases are considered downward-entailing. 8 In addition to its function as a temporal auxiliary indicating “(imminent) future” (e.g., apa nak jadi, jadila “what will happen will happen,” nak in CM may also function as a verb meaning “want” (e.g., nak beli apa “want to buy what”) or a complementizer (e.g., tak ingat nak balik “not think of going home” (Koh 1990: 209)). Besides these functions, Koh (1990) also includes the following as functions of nak: (i) as an auxiliary linking a modal auxiliary, such as boleh “can,” sanggup “willing,” or takkan “surely not,” to the rest of the verbal predicate, and (ii) as a subordinating adverbial clause marker. The former function may be subsumed under its function as a temporal auxiliary. The latter may be analyzed as instances of nak as a complementizer. 9 I thank Jozina Vander Klok for comments about the structure of such sentences and for encouraging me to spell out my assumptions about the syntax of dah. 10 Like CM dah, Mandarin sentence-final -le also may not appear with downwardentailing quantifiers in sentences describing telic situations where the quantifiers specify information relating to the end point of the situation (Soh & Gao 2008; Soh 2008, 2009). 11 The acceptability of ke between the auxiliary and the verb is variable among speakers for unclear reasons. (i)

(*) Kau dah ke sampai kat KLCC? 2sg dah Q arrive at KLCC “Have you arrived at KLCC?”

Such sentences will not be considered further in the chapter. 12 (40a) may be interpreted as having a neutral sentential focus or a marked focus on the locational phrase. See following for discussion of ke and marked focus. 13 Sentences with the question particle kah immediately after the subject are considered unacceptable in Standard (Formal) Malay in Kader (1981). As noted earlier, we find sentences with ke after the subject marginal in Colloquial Malay, but clearly contrasting with ke in non-subject positions. Some speakers find the sentences acceptable with a pause after ke. Sentences with and without yang may involve diferent structures. See Cole et. al (to appear) and Kroeger (2009) for arguments that (wh)-questions with yang contain a headless relative clause. 14 Among the five speakers consulted, one accepted sentences with the ke-dah order, noting that such sentences involve enhanced stress on dah, used in a marked context related to “unexpectedness” and “afrmation.” Further investigation is needed to clarify the contexts of use that license this order for the speaker. 15 The assumptions here follow the analysis of A-not-A questions in Mandarin Chinese (Huang 1982, 1988; Ernst 1994; Soh & Gao 2006; Soh 2018). 16 dah (including post-verbal and sentence-final dah) may appear in wh-questions (as shown in (5)). I assume that the interrogative feature resides in Foc in these questions.

On the discourse marker dah in Colloquial Malay

145

References Cantor, Sara. 2010. Yes/no questions in Malay: A multiple C0-based account. Paper Presented at the 14th International Symposium on Malay/Indonesian Linguistics, April/May 2010, University of Minnesota, Twin Cities, MN. Chao, Yuen Ren. 1968. A Grammar of Spoken Chinese. Berkeley, CA: University of California Press. Cole, Peter, Gabriella Hermon & Norhaida Aman. To appear. Clefted questions in Malay. In David Gil & James Collins (eds.), Malay/Indonesian Linguistics. London: Curzon Press. Erlewine, Michael Yoshitaka. 2017. Low sentence-final particles in Mandarin Chinese and the final-over-final constraint. Journal of East Asian Linguistics 26. 37–75. Erlewine, Michael Yoshitaka. This volume. A syntactic universal in a contact language: The story of Singlish already. In Hiroki Nomoto & Elin McCready (eds.), Discourse Particles in Asian Languages, Volume II, 91-120 London: Routledge. Ernst, Thomas. 1994. Conditions on Chinese a-not-a questions. Journal of East Asian Linguistics 3. 241–265. Fortin, Catherine R. 2009. On the left periphery in Indonesian. In I. Paul, S. Chung, D. Finer & E. Potsdam (eds.), Proceedings of Austronesian Formal Linguistics Association XVI, 29–43. London: University of Western Ontario. Gan, Kok Siong. 1991. Kata kerja bantu dalam sintaksis Bahasa Malaysia [Auxiliary Verbs in Malay Syntax]. Kuala Lumpur: Dewan Bahasa dan Pustaka. Hoogervorst, Tom G. 2011. Some introductory notes on the development and characteristics of Sabah Malay. Wacana 13. 50–77. Horn, Larry. 1996. Exclusive company: Only and the dynamics of vertical inference. Journal of Semantics 13. 1–40. Huang, C.-T. James. 1982. Logical Relations in Chinese and the Theory of Grammar. Ph.D dissertation, Massachusetts Institute of Technology. http://hdl.handle. net/1721.1/15215 Huang, C.-T. James. 1988. Wo pao de kuai and Chinese phrase structure. Language 64. 274–311. Huang, Lillian Meei Jin & Philip W. Davis. 1989. An aspectual system in Mandarin Chinese. Journal of Chinese Linguistics 17. 128–166. Kader, Mashudi. 1981. The Syntax of Malay Interrogatives. Kuala Lumpur: Dewan Bahasa dan Pustaka. Koh, Ann Sweesun. 1990. Topics in Colloquial Malay. Ph.D dissertation, University of Melbourne. Krifka, Manfred. 2000. Alternatives for aspectual particles: Semantics of ‘still’ and ‘already’. In L. J. Conathan, Jef Good, D. Kavitskaya, A. B. Wulf & A. C. L. Yu (eds.), Proceedings of the 26th Annual Meeting of Berkeley Linguistics Society: General Session and Parasession on Aspect, 401–412. Berkeley: Berkeley Linguistics Society. Kroeger, Paul. 2009. Malagasy clefts from a Western Malayo-Polynesian perspective: Commentary on the paper by Hans-Martin Gaertner. Natural Language and Linguistic Theory 27. 817–838. Li, Charles N. & Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Lin, Jo-wang. 2003. Temporal reference in Mandarin Chinese. Journal of East Asian Linguistics 12. 259–311. Löbner, Sebastian. 1989. German schon-erst-noch: An integrated analysis. Linguistics and Philosophy 12. 167–212.

146

Hooi Ling Soh

Nomoto, Hiroki. 2006. A Study on Complex Existential Sentences in Malay. Master’s thesis, Tokyo University of Foreign Studies. Nomoto, Hiroki & Mohd. Farez Syinon bin Masnin. 2019. Aspect in Malay. Journal of the Institute of Language Research 24. 417–430. Tokyo University of Foreign Studies. Omar, Asmah. 1970. Bahasa Malaysia kini 1 [Malay Today 1]. Kuala Lumpur: Federal Publications. Paul, Waltraud. 2014. Why particles are not particular: Sentence final particles in Chinese as heads of a split CP. Studia Linguistica 68. 77–115. Rizzi, Luigi. 1997. The fine structure of the left periphery. In L. Haegeman (ed.), Elements of Grammar: Handbook in Generative Syntax, 281–337. Dordrecht: Kluwer. Ross, Claudia. 1995. Temporal and aspectual reference in Mandarin Chinese. Journal of Chinese Linguistics 23. 87–135. Salleh, Ramli. 1989. Fronted Constituents in Malay: Base Structures and Move Alpha in a Configurational Non-Indo-European Language. Kuala Lumpur: Dewan Bahasa dan Pustaka. Smith, Carlota S. 1997. The Parameter of Aspect, 2nd edn. Dordrecht: Kluwer. Soh, Hooi Ling. 1994. Aspect and the Organization of Argument Structure and Phrase Structure: Evidence from Malay. Master’s thesis, University of Calgary. Soh, Hooi Ling. 2008. The syntax and semantics of change/transition: Evidence from Mandarin Chinese. In Susan Rothstein (ed.), Theoretical and Cross-Linguistic Approaches to the Semantics of Aspect, 387–419. Philadelphia: John Benjamins. Soh, Hooi Ling. 2009. Speaker presupposition and Mandarin sentence-final – le: A unified analysis of the ‘change of state’ and ‘contrary to expectation’ reading. Natural Language and Linguistic Theory 27. 623–657. Soh, Hooi Ling. 2019. Colloquial Malay discourse particle punya as a modal evidential. Oceanic Linguistics 58. 386–413. Soh, Hooi Ling. Volume I. 2023 Mandarin Chinese sentence final particle de as a marker of private evidence. In Elin McCready & Hiroki Nomoto (eds.), Discourse Particles in Asian Languages, Volume I, 67-82. London: Routledge. Soh, Hooi Ling & Meijia Gao. 2006. Perfective aspect and transition in Mandarin Chinese: An analysis of double -le sentences. In Pascal Denis, Eric McCready, Alexis Palmer & Brian Reese (eds.), Proceedings of 2004 Texas Linguistics Society Conference: Issues at the Semantics-Pragmatics Interface, 107–122. Somerville: Cascadilla Press. Soh, Hooi Ling & Meijia Gao. 2008. Mandarin sentential -le, perfect and English already. In Johannes Dölling, Tatjana Heyde-Zybatow & Martin Schäfer (eds.), Event Structure in Linguistic Form and Interpretation, 447–473. Berlin: Mouton de Gruyter. Stalnaker, Robert. 1998. On the representation of context. Journal of Logic, Language, and Information 7. 3–19. Stalnaker, Robert. 1999. Context and Content: Essays on Intentionality in Speech and Thought. Oxford: Oxford University Press. Stalnaker, Robert. 2002. Common ground. Linguistics and Philosophy 25. 701–721. Sybesma, Rint. 1999. The Mandarin VP. Dordrecht: Kluwer. Vander Klok, Jozina and Lisa Matthewson. 2015. Distinguishing already from perfect aspect: A case study of Javanese wis. Oceanic Linguistics 54. 172–205. van Minde, Don & Johnny Tjia. 2002. Between perfect and perfective: The meaning and function of Ambonese Malay su and suda. Bijdragen tot de Taal-, Land- en Volkenkunde 158. 283–303. von Fintel, Kai. 1999. NPI licensing, Strawson entailment, and context dependency. Journal of Semantics 16. 97–148.

6

On the apparently nonadditive use of Malay additive pun Hiroki Nomoto

1

Introduction1

The particle pun in Standard Malay (hereafter simply “Malay”) is polyfunctional (Asmah 2009; Mintz 2002; Goddard 2001; Hoogervorst 2018; Chambert-Loir 2019).2 Two of its meanings are easy to describe, namely, its use as simple (i.e., non-scalar) and scalar additive markers. The simple additive pun corresponds to too or also in English, whereas the scalar additive pun is often translated as even. Furthermore, pun also has a final particle use, whose meaning is elusive, as is often the case with final particles cross-linguistically. Asmah (2009: 192) refers to this use as kepastian, which Goddard (2001) translates as “determinative.” The main focus of the present chapter is yet another use of pun. Previous authors have described this use as having to do with time (e.g., Asmah 2009; Cumming 1991; Mintz 2002; Goddard 2001). Asmah (2009: 190) and Goddard (2001) call the relevant use as the waktu “temporal” and the “event-sequence” focus pun, respectively. However, I show that the relevant use is, in fact, more general and is not restricted to temporal relations between events. I argue that this meaning of pun is essentially a type of additive, specifically, the additive whose alternatives difer among one another at the level of propositions (“propositional additive”) as opposed to individuals or properties thereof (“individual additive”). In my analysis, the temporal meaning is not encoded by the pun itself but arises from the context in which it occurs. The additive semantics common to both the simple additive and the so-called event-sequence pun can be represented by means of discourse trees based on the Questions Under Discussion (QUD) framework (Roberts 2012; Büring 2003). This chapter is organized as follows. In Section 2, I will first overview various uses of pun reported in the literature. The existing descriptions are generally satisfactory, though little has been done on the concrete formalization of the various meanings conveyed by pun. I will thus put forth a rough formal analysis for each meaning, which will become a basis for more sophisticated analyses in future research. Section 3 discusses the event-sequence pun. I will point out the problems of the previous studies and propose an alternative analysis in terms of QUD. The proposed analysis reveals a close relationship between the eventsequence and the simple additive uses, which previously remained unclear. In DOI: 10.4324/9781351057752-7

148

Hiroki Nomoto

Section 4, I will show that there are examples that can be best classified into the event-sequence category but whose meanings are clearly not temporal, and demonstrate that the proposed analysis is general enough to deal with such examples as well. The concluding Section 5 points out that similar, apparently non-additive uses of the additive particles are also found in other languages. 2

Meanings of pun

This section classifies diferent uses of pun and summarizes their meanings. I will provide a rough formal analysis for each meaning as a first attempt at the formal semantic analysis of pun. The various uses of pun can be divided into two major categories by means of a syntactic criterion, namely, the second-position pun and the final pun. The tree in Figure 6.1 summarizes the classification of various uses of pun adopted in this chapter.

second position focus marker

event-sequence

final modality/evidential marker

additive scalar

Figure 6.1 Classification of various uses of pun.

Sentences (2) and (3), respectively, illustrate the pun occurring in the second and final positions, based on the basic sentence in (1). As will be discussed shortly, other word orders are also possible for both positions. (1) Saya tak tahu nama dia. I not know name 3sg “I don’t know his/her name.” (2) Second-position pun

Saya pun tak tahu nama dia. I pun not know name 3sg (i) “I also don’t know his/her name.” (ii) “Even I don’t know his/her name.” (3) Final-position pun Saya tak tahu nama dia pun. I not know name 3sg pun “I really don’t know his/her name.”

In what follows, I will examine each use one by one.

(ADDITIVE) (SCALAR)

On the apparently non-additive use of Malay additive pun 2.1

149

Second-position pun

The second-position pun includes two kinds of focus-sensitive uses, which I simply refer to as “focus marker.” The so-called event-sequence pun also occurs in the second position. The focus marker pun adds a scalar or non-scalar additive meaning. I will refer to the two meanings as “scalar” and “additive,” respectively. 2.1.1

Additive

The additive pun is typically translated into English as too or also. Asmah (2009: 189) calls this category persamaan “sameness.”3 Examples are given in (4) (= (2)) and (5). In these and subsequent examples, [ ]F indicates the constituent in focus, and only the relevant interpretation is provided. (4) [Saya]F pun tak tahu nama dia. I pun not know name 3sg “I also don’t know his/her name.” [There is at least one individual other than the speaker who does not know his/ her name.] (5) a. Mulyadi dari Indonesia. [Yusrita]F pun dari Indonesia. Mulyadi from Indonesia Yusrita pun from Indonesia “Mulyadi is from Indonesia. Yusrita is also from Indonesia.” (Nomoto 2020: 23) b. . . . mak saya dulu sibuk kerja ha [sekarang ni]F pun kerja juga. . . .4 mum I before busy work yeah now this pun  work too “[M]y mum was busy with work before, yeah, she’s working now too.”

Focused constituents may vary in terms of grammatical relations and syntactic categories. The focused constituent is the subject DP in (4) and (5a). However, this is not necessarily the case. In (5b), the subject of the verb kerja “to work” is not expressed overtly. The focused constituent sekarang ni “now” is a temporal adjunct. In (6), the constituent in focus is a fronted object. (6) [Nama dia]F pun saya tak tahu. name 3sg pun I not know “I also don’t know his/her name.” [There is at least one thing other than his/her name that the speaker does not know.]

Structurally, the focused constituent must occur immediately before pun. It is not possible to insert an overt subject in (5b) unless the inserted subject becomes the focused constituent, as shown in (7). (7)

*. . . mak saya dulu sibuk kerja ha [sekarang ni]F dia pun mum I before busy work yeah now this she pun kerja juga . . . . work too but Intended: “[M]y mum was busy with work before, yeah, she’s working now too.”

150

Hiroki Nomoto

Moreover, an object cannot bear focus without fronting. Thus, the basic order counterpart of (6) in (8) cannot have the same interpretation as (6) (cf. (4)).5 (8) * Saya pun tak tahu [nama dia]F.  I pun not know name 3sg   Intended: “I also don’t know his/her name.”   [There is at least one thing other than his/her name that the speaker does not know.]

The ungrammaticality of (8) cannot be remedied by placing pun immediately after the focused constituent, as shown in (9). (9) * Saya tak tahu [nama dia]F pun.  I not know name 3sg pun  Intended: “I also don’t know his/her name.”  [There is at least one thing other than his/her name that the speaker does not know.]

These facts suggest that the additive pun does not simply adjoin to the focused element but it is a functional head on its own, and the focused constituent occurs in its specifier: FocP

(10)

Foc0

XPF Foc pun

TP . . . tXP . . .

I assume that the semantics of the additive pun is essentially the same as too or also in English. Thus, x pun P conveys that the property P holds for individual x and at least one alternative of x.6 I take the meaning component concerning the alternatives to be a presupposition. Assuming that XP and TP in (10), respectively, denote an individual x and a property λx.P(x),(10) will receive the following interpretation: (11) Semantics of (10) with punadditive a. At-issue content: P(x) b. Presupposition: y[y  ALT(x)P(y)], where ALT(x) is a function yielding the set of x’s alternatives (excluding x itself) in a given context.

The semantics in (11) in the preceding text should be straightforward for cases in which the focused constituent is an argument, as in (5a) and (6). To capture cases in which the focused constituent is an adjunct, as in (5b), repeated in the following, in a uniform fashion, neo-Davidsonian event semantics (Parsons

On the apparently non-additive use of Malay additive pun

151

1990) can be adopted, as illustrated in (13). The constants m and t stand for the denotations of mak saya “my mum” and sekarang ni “now,” respectively. (12) . . . mak saya dulu sibuk kerja ha [sekarang ni]F pun kerja juga . . . . mum I before busy work yeah now this pun work too “[M]y mum was busy with work before, yeah, she’s working now too.” (13) Semantics of (12) a. At-issue content: b. Presupposition:

λx.e[working(e)Agent(e,m)AT(e,x)](t) = e[working(e)Agent(e,m)AT(e,t)] y[y  ALT(t)e[working(e)Agent(e,m)AT(e,y)]]

Before moving on to the next section, I would like to point out that the relevant alternatives are not always immediately clear. Example (14) from Goddard (2001) exemplifies the case in point. (14) Tuk Wan mahu beri peluang pada Nit makan sebanyak-banyak yang mahu. Dan [Nit]F pun tahu itu. and Nit pun know that “Tuk Wan wanted to give Nit the chance to eat as much as he wanted. And Nit knew it.” (Goddard 2001: (12))

Nowhere in the preceding discourse (including the parts not included in the preceding excerpt) is information about other individuals who knew about Tuk Wan’s desire discussed. Thus, Goddard presents (14) as an example that lacks a “parallelism context.” However, one can infer from the first sentence that there is, in fact, one such individual, namely, Tuk Wan himself. The fact that the desire-holder normally knows about his own desire is so trivial that it is not mentioned at all. This analysis is corroborated by the fact that the pun sentence cannot be negated. (15)

Tuk Wan mahu beri peluang pada Nit makan sebanyak-banyak yang mahu. #Tetapi [Nit]F pun tidak tahu itu. but Nit pun not know that “Tuk Wan wanted to give Nit the chance to eat as much as he wanted. But Nit didn’t know it.”

In passing, Goddard (2001) proposes the Natural Semantic Metalanguage explication in (16) for the additive and scalar uses of pun, which he collectively refers to as “topic-comment pun.” (16) XTOPIC pun (Comment) = I want to say something about this person/thing/place (i.e., X) not about someone/something/somewhere else

However, this explication fails to account for the meanings of pun, as it is more apt to the contrastive topic marker pula. The second sentence in (17)

152

Hiroki Nomoto

can be paraphrased using the very pattern in (16): I am saying something about this person, but not about that person. (17) Itu Dr. Nur Fathima. Dan ini pula Encik Abd. Hadi. that Dr. Nur Fathima and this pula Mr. Abd. Hadi “That is Dr. Nur Fathima. And this (on the other hand) is Mr. Abd. Hadi.” (Nomoto 2020: 23) 2.1.2

Scalar

The scalar pun can normally be translated into English as even. Asmah (2009) sets up two categories for what I call “scaler pun” here: tolak ansur “concessive” and tokokan “additive.” Mintz (2002: 387–390) follows Asmah in this respect. Asmah distinguishes the two categories because they are paraphrased with diferent expressions. The concessive pun can be paraphrased with concessive conjunctions, such as meskipun and walaupun “even though,” as in (18), whereas the “additive” pun is paraphrasable with additive conjunctions such as bahkan and malah “moreover, on the contrary,” as in (19). (18) a. Bunga ini tahan lama. Tidak di-beri air pun boleh hidup flower this last long not pass-give water pun can live berminggu-minggu. for.weeks b. =... Meskipun tidak di-beri air, boleh hidup berminggu-minggu. even.though not pass-give water can live for.weeks “This flower lasts for a long time. Even if you don’t water it, it will be alive for weeks.” (Asmah 2009: 191) (19) a. Jangan marah dia  kerana kesilapan itu. Orang  dewasa pun don’t scold 3sg because mistake that person adult pun tidak sepi dari kesilapan. not free from mistake b. =... Malah orang dewasa tidak sepi dari  kesilapan. moreover person adult not free from mistake “Don’t scold him/her for the mistake. Even adults are not free from mistakes.” (Asmah 2009: 191–192)

I regard these two categories as one because they share a common core meaning, which is reflected in the English translations as even. The two difer in what constitutes the focus associated with pun, namely, an eventuality in the former (“the flower not being watered” in (18a)) and an individual in the latter (“adults” in (19a)). Consequently, the pun in the former helps establish a clause linkage in a multi-clausal complex sentence, as conjunctions do. I say “helps” because the context alone, if sufciently strong, can establish the same inter-clausal relation, as in (20) (see Nomoto & Aznur Aisyah 2015 for details).

On the apparently non-additive use of Malay additive pun

153

(20) Matahari belum terbit lagi, saya sudah bangun. sun not.yet rise yet I already wake.up “Even though the sun hasn’t risen, I’m already awake.” (Faridah & Kondo 2005: 46)

I assume that a paratactic structure like (20) involves a null complementizer indicating that the clause introduced by it modifies the main clause in some way, though not as specifically as overt complementizers, such as kalau “if” or walaupun “even though,” specify. Thus, in (18a), the concession relation does not stem from pun but from this null complementizer. This is confirmed by the fact that an overt complementizer may occur in (18a). (21) Kalau tidak di-beri air pun boleh hidup berminggu-minggu. if not pass-give water pun can live for.weeks “Even if you don’t water it, it will be alive for weeks.”

The syntax of the scalar pun is basically the same as that of the additive pun. That is, the constituent in focus occurs in the specifier of the phrase projected by pun, as in (10). The constituent in focus must occur immediately before pun, as shown by the contrast between (22a) and (22b). (22) a. [Kalau tidak di-beri air]F pun bunga itu boleh hidup if not pass-give water pun flower that can live berminggu-minggu. for.weeks “Even if you don’t water it, it will be alive for weeks.” b. *Bunga itu pun boleh hidup berminggu-minggu [kalau tidak flower that pun can live for.weeks if not di-beri air]F. pass-give water Intended: “Even if you don’t water it, it will be alive for weeks.” (23) shows the structure of (18a) and (22a). FocP

Foc0

CPF Ø/Kalau tidak diberi air

Foc pun

TP tCP

TP

boleh hidup berminggu-minggu

154

Hiroki Nomoto

Note that there is an exception to this analysis, specifically cases involving indefinites, such as interrogatives and “one (X).” The pun in these cases seems to express the same meaning, but its distribution is rather free compared to the other cases seen earlier. Pun is not necessarily in the second position but occurs wherever its associate is.7 (24a) involves the interrogative apa “what,” whereas (24b–c) involve the numeral satu “one.” In all cases, pun occurs in(side) the object position. (24) a. Sedangkan Lisa tidak membuat [kesalahan apa]F pun.8 meanwhile Lisa not do guilt what pun “Meanwhile Lisa did not do any guilt.” b. Kami tak dapat [satu sen upah]F pun dengan menulis di we not get one sen wage pun by write at blog ini.9 blog this “We don’t get paid even a penny for writing on this blog.” c. Hingga ke saat ini, Anwar masih belum menerima [satu]F till to moment this Anwar still not.yet receive one pun salinan dokumen untuk pembelaan.10 pun copy document for defense “Up until this moment, Anwar has not received even one copy of the documents for the defense.”

I argue that this type of pun adjoins to the focused constituent/word and is syntactically distinct from the pun as the Focus head in (10). This exception aside, both the additive and the scalar pun occur in the same syntactic position, giving rise to a systematic ambiguity between the two readings, as in the following examples. (25) Saya pun tak tahu nama dia. I pun not know name 3sg (i) “I also don’t know his/her name.” (ADDITIVE) [There is at least one individual other than the speaker who does not know his/her name.] (ii) “Even I don’t know his/her name.” (SCALAR) [The speaker is the least likely individual that does not know his/her name.] (26) Nama dia pun saya tak tahu. name 3sg pun I not know (i) “I also don’t know his/her name.” (ADDITIVE) [There is at least one thing other than his/her name that the speaker does not know.] (ii) “I even don’t know his/her name.” (SCALAR) [His/her name is the least likely thing that the speaker does not know.]

The scalar reading can be ensured by adding a boundary-specifying expression, such as sampaikan and sehinggakan “to the extent,” as in (27) (Nomoto & Aznur Aisyah 2017).

On the apparently non-additive use of Malay additive pun

155

(27) Sampaikan/Sehinggakan kanak-kanak kecil pun di-suruh tolong buat to.the.extent/to.the.extent children small pun pass-ask help do kerja itu. work that “Even small children were asked to help the work.” (Nomoto & Aznur Aisyah 2017: 124)

Comparing pun with the other additive particle juga, Mintz (2002: 386– 387) states that they are diferent in that while “[j]uga is the more neutral of the two,” “[p]un has an additional meaning conveyed in English by ‘even.’” However, I do not think that the additive pun encodes the “even” meaning as well. Consider (5a), repeated in the following, for example. This sentence can be used without an “even” meaning, for example, when simply listing Indonesians in a group of international students. (28) Mulyadi dari Indonesia. [Yusrita]F pun dari Indonesia. Mulyadi from Indonesia Yusrita pun from Indonesia “Mulyadi is from Indonesia. Yusrita is also from Indonesia.” (Nomoto 2020: 23)

Thus, the additive pun is as neutral as juga and is distinct from the scalar pun. Mintz’s remark is thought to have resulted from the systematic ambiguity between the two readings. Whenever the additive reading is available, the scalar reading is also possible, given an appropriate context. Regarding the semantics of the scalar pun, I assume that it is essentially the same as that of even in English. In the simplest case, where x is an individual, x pun P roughly means that x is the least likely individual that P, compared to its alternatives. The relevant likelihood is based on a contextually determined scale. As with the case of the additive pun (11), I take the meaning component concerning the alternatives to be a presupposition. If XP and TP in (10), respectively, denote an individual x and a property λx.P(x), (10) will receive the following interpretation: (29) Semantics of (10) with punscalar a. At-issue content: P(x) b. Presupposition: y[[y  ALT(x)P(y)] P(y)> P(x)], where ‘p > q’ indicates that the likelihood of p is higher than that of q on a contextually determined likelihood scale.

For the use of punscalar to be informative, ALT(x) cannot be empty. Hence, (29b) implicates that the relevant property holds true of at least one alternative of x, which is exactly what the additive pun presupposes (cf. (11b)). The additive and the scalar pun are related to each other in this way. Again, neo-Davidsonian event semantics enables a unified analysis for cases in which the focused constituent is an adjunct, as in (30). The constants a and

156

Hiroki Nomoto

t in (31) stand for the denotations of Abuya (personal name) and hari-hari cuti “holidays,” respectively. (30) Sebab itu [hari-hari cuti]F pun Abuya pergi bermain di sekolah.11 because that day.pl holiday pun Abuya go play at school “Because of that, Abuya went playing at school even on holidays.” (31) Semantics of (30) a. At-issue content: x.e[going_play_at_school(e) Agent(e,a) AT(e,x)](t) = e[going_play_at_school(e) Agent(e,a) AT(e,t)] b. Presupposition: y[[y  ALT(t) e[going_play_at_school(e) Agent(a,e) AT(e,y)]]  e[going_play_at_school(e) Agent(e,a) AT(e,y)] > e[going_play_at_school(e) Agent(e,a) AT(e,t)] 2.2

Final pun

Unlike the additive and scalar pun, it is difcult to come up with an English equivalent applicable to all instances of the final pun. Asmah (2009: 192) refers to this type of pun as kepastian “determinative.”12 Nomoto (2020: 160) notes the modal and evidential aspects of the final pun as follows: “this use of pun is used when the speaker makes a definitive statement without leaving a room for objection from the hearer because s/he is certain of the factuality of the situation, for example, by having witnessed it.” As Goddard (2001: 40) points out, the final pun often conveys “the potential existence of an expectation to the contrary.” Most instances of the final pun occur in negative clauses, as in (32)–(33). (32) a. Saya tak tahu

nama dia

pun.

I not know name 3sg pun b. Saya tak tahu pun nama dia. I not know pun name 3sg “I really don’t know his/her name.” (33) Sepanjang hari, kami tidak menjejak tanah pun. whole day we not step land pun “We didn’t set foot on the land at all for the whole day.” (Asmah 2009: 192) However, it can also occur in afrmative clauses: (34) a. Seorang cendiakwan Pilipina, Kardinal Sin berkata; “Tempat yang paling berbahaya di dunia adalah dalam rahim.” Memang benar pun.13 indeed true pun “A Filipino scholar, Kardinal Sin, said: ‘The most dangerous place in the world is inside the womb.’ That’s indeed true.”

On the apparently non-additive use of Malay additive pun

157

b. Maya, semua-nya dah terjadi pun.14   Maya all-3 already happen pun   “Maya, all of them have already happened, really.” c. hehe tahu pun kau.15  haha know pun you “Ha ha you do know (the reason why I called you).” d. Nah, itu pun dia (Zafwan datang menuju ke tempat Ahmad dan alright that pun 3sg Zafwan come head to place Ahmad and Daud).16 Daud “Alright, there he is (Zafwan came towards Ahmad and Daud).”

In (35), pun occurs in a rhetorical question. (35) Mana ada ketawa lebih-lebih pun.17 where be laugh excessively pun “When did I laugh too much? (I’ve never done such a thing.)” [lit. Where is it that I laughed excessively?]

The final pun does not seem to occur in genuine information-seeking questions and imperatives. Regarding its position, the final pun most often occurs in the sentence-final position. However, it is found in other “final” positions too. Consider examples (32b), (34c), and (34d) earlier. In (32b), pun separates the verb from its object. In (34c) and (34d), pun appears to be attached to a fronted predicate phrase, with the subject being left behind. Since pun is accompanied by the falling intonation characteristic of a clause boundary, I argue that these orders are derived by right-dislocation and that pun occurs in the same clause-final position as in other examples. I posit the structure in (36) for the final pun, whereby it is a functional head taking a TP in its specifier. I identify the relevant functional category as Part(icle). (36) [PartP TP [Partʹ pun t tp]]

The trees in (37) show the structures of (32b) and (34c), in which the object and subject DPs have moved out of TP before the TP moves up to Spec, PartP, respectively. (37) a. (32b)

b. (34c) PartP

Part pun

tDP tahu

TP tTP

Part0

TP

Part0

TP Saya tak tahu tDP

PartP

DP nama dia

Part pun

TP tTP

DP kau

158

Hiroki Nomoto

Regardless of the position in which pun occurs, it modifies the proposition described by the TP. Following Grosz’s (2021) formulation of the semantics of German discourse particle ja, given in (39), I propose the semantics of the final pun in (38), where p stands for the proposition denoted by the TP in (36). The notion of active consideration of a proposition employed in both formulations is defined as in (40). (38) Semantics of (36) (punfinal) a. At-issue content: p b. Presupposition: Speaker believes that p is so true that the addressee should not actively consider the possibility of ¬p. (39) German ja (Grosz 2021)

ja

g

= λ phs ;ti : g(speaker0) believes that g(addressee0) does not actively consider

the possibility of ¬p in g(w0).p (40)

x actively considers the possibility of φ if x believes that φ or x tries to resolve the questions of whether φ or ¬φ.

The main diference between the Malay pun and German ja lies in the strength of assertion.18 The German ja only ensures the addressee’s information state so that the proposition in question becomes a shared belief by the speaker and the addressee, that is, a part of the common ground. It thus functions as a marker of confirmation. The Malay counterpart of ja is kan.19 The efect of the Malay pun is stronger. It can force the addressee to have a particular belief; a response like “Wait a minute. I’m not sure about it” is very difcult to make, if not completely disallowed (cf. von Fintel & Matthewson’s 2008 tests for identifying presuppositions). If the speaker thinks that the addressee may cast doubt on p before p pun is uttered, pun helps to urge the latter to update his/her belief with p so that the presupposition will be satisfied. It is indeed in such situations that the final pun is most efectively used. Otherwise, the weaker particle kan, which I see as the Malay counterpart of the German ja (39), is more felicitous. This will explain the fact that the final pun often conveys “the potential existence of an expectation to the contrary” (Goddard 2001: 40). The final pun is thought to have developed from the scalar pun. Semantically, the scalar pun presupposes that the focused element x is the least likely individual of which the property P holds true (cf. (29)). Asserting that P holds for x should then require stronger confidence on the part of the speaker than asserting P with any other alternative of x. This kind of strong confidence, through repeated uses of the scalar pun, could eventually develop into a distinct use of pun, namely, the final pun, which now conventionally encodes strong confidence as a presupposition. On the syntactic side, one of the prerequisites for the final pun is the movement of TP to the left periphery, more specifically to the left of pun. I claim that structures like (23), in which

On the apparently non-additive use of Malay additive pun

159

the scalar pun can take a CP in its specifier, played a role in the emergence of the structure in (36) involving the final pun. 2.3

Initial pun

Before turning to the problematic event-sequence pun, I would like to point out instances of pun occurring at the beginning of a clause, namely, the initial pun. Two types of usages can be identified, both of which I claim have developed from the scalar pun. In the first type, pun occurs on its own as a discourse connector, as in (41). It can be substituted by (walau) bagaimanapun “however.” (41) Pun, dalam kunjungan kedua saya ini, pun in visit second my this dan bercerita apa sahaja yang berlaku and narrate anything rel happen “However, in my second visit this time, s/he that happens to her/him.”20

dia menjadi lebih terbuka 3sg become more open pada-nya. to-3 became opener and told anything

Two analyses are conceivable for this type of pun. First, it has the same syntax and semantics as the scalar pun, except for the fact that the focused constituent is not overt. One can posit a null pronoun anaphoric to a salient proposition in the preceding context. (42) [FocP pro [Foc' pun TP ]] (cf. (10), (23))

I believe that such a construction existed at least at some stage in the history of the language. Secondly, it is also possible that the stand-alone pun is no longer a focus marker but has become a full-fledged conjunction. The second type of the initial pun is illustrated by (43). I regard this type of pun as a subordinate conjunction and a part of the pun concessive conjunction series, namely, walaupun , meskipun , sungguhpun , biarpun , and kendatipun “although.” (43) Pun begitu, setiap tahun beliau sentiasa menantikan ketibaan bulan pun so every year 3sg always await arrival month penuh kerahmatan iaitu bulan Ramadan. full blessing namely month Ramadan “However, every year she always looks forward to the arrival of the month full of blessings, namely the month of Ramadan.”21

I argue that these subordinate conjunctions have emerged as a result of the head movement of the scalar pun, followed by the reanalysis of the resultant complex head as a single lexical item. The hypothesized syntactic change is summarized in (44). After the reanalysis, the pun is no longer a focus head, which causes the loss of focus movement and focus interpretation.

160

Hiroki Nomoto

(44) a. Scalar pun in situ ForceP FocP 8 Force 9 > < Ø > = walau Foc0 begituF > : ... > ; Foc TP pun . . . tbegitu b. Head movement to Force ForceP Force 8 Force 9 Foc > < Ø > = pun walau > : ... > ;

FocP Foc0

begituF tpun

c. Reanalysis as a single head ForceP Force pun walaupun > .. : .

8 >
=

TP . . . tbegitu

TP . . . begitu

> ;

The so-called event-sequence pun Previous descriptions and their problems

Syntactically, the event-sequence pun occurs in the second position (Figure 6.1). However, its meaning does not appear to pertain to focus in the traditional sense. Rather, it typically expresses clausal relations, particularly a temporal one. This led previous authors to characterize the relevant use as temporal (e.g., Asmah 2009; Cumming 1991; Mintz 2002; Goddard 2001), using labels such as waktu “temporal” (Asmah 2009: 190) and event-sequence (Goddard 2001). Illustrative examples are given in (45). (45) a. Dia memanggil saya. Saya pun menyahut. 3sg call me I pun respond “S/he called me. So, I responded.” Not: “. . . . So, {I also/even I} responded.”

On the apparently non-additive use of Malay additive pun

161

b. Apabila dia selesai membaca surat itu, dia pun mengoyak-ngoyak-nya.   when 3sg finish read letter that 3sg pun tear.into.pieces-3  “When s/he finished reading the letter, s/he tore it into pieces.” Not: “. . ., {s/he also/even s/he} tore it into pieces.” (Asmah 2009: 190)

Regarding this use of pun, Asmah (2009: 190) states the following: Pun dalam subgolongan ini merujuk kepada waktu sesuatu perbuatan dilakukan; tegasnya, waktu sesudah perbuatan dalam ayat pertama dilakukan. “Pun of this subtype refers to a time of something being done; in particular, to the time just after the completion of the action carried out in the first clause.”22 She suggests paraphrases using kemudian “subsequently” and lalu “then.” As for the English equivalent of this use of pun, Goddard (2001: 38) notes that “[s]tylistically, the efect of pun is roughly comparable to verb-particle/adverb inversion in English, as found in sentences like Of he went,” which is reflected in his translation of the sentence in (46). (46) “Aku boleh bersahabat dengan mereka,” kata rusa. Ia pun pergi menghampiri anjing-anjing itu. Apabila anjing-anjing 3sg pun go approach dog.pl that when dog.pl ternampak rusa, mereka pun mengepung-nya. see deer they pun surround-3 “‘I can make friends with them,’ thought (lit. said) the deer. Of he went towards the dogs. When the dogs spotted the deer, they surrounded him.” (Goddard 2001: (20))

Mintz (2002: 394) writes that “the presence of pun indicates a related sequence of events that are not coincidental, but expected.” In English translations, he diferentiates the meanings of sentences with and without pun by means of the presence or absence of so: (47) a. Hari sudah gelap, dan saya pun pergi.  day already dark and I pun go  “It was already dark, and so I left.”23 b. Hari sudah gelap, dan saya pergi.  day already dark and I go  “It was already dark and I left.”

(Mintz2002: 394)

Although these descriptions successfully capture typical examples of the relevant uses of pun, they also pose questions. First, why is it that only this use is temporal? Furthermore, why does it express sequence rather than other temporal relations, such as concurrence? It is puzzling that only this use has such a concrete meaning when the other uses have more abstract meanings pertaining to focus and modality.

162

Hiroki Nomoto

Relating to this point, how does this use relate to the focus marker pun? All the uses we have seen so far are either a focus marker or somehow related to the focus marker pun. However, it is not obvious how the temporal meaning as described by the previous authors relates to the focus marker pun. One may wonder if this use of pun is simply a homonym of the other uses of pun. However, as I will point out in Section 5, at least two languages unrelated to Malay, namely, Japanese and Vietnamese, exhibit very similar polysemy. Therefore, homonymy is not an insightful way to solve the issue. 3.2

Alternative analysis: individual and propositional additives

I argue that the so-called event-sequence use is related to the additive use in that both require the existence of at least one alternative proposition in the current discourse. The alternative proposition is often explicitly stated, but its existence may also be implicitly assumed. When it is explicitly stated, the relevant statement precedes the pun sentence. The two uses difer with respect to how the relevant alternative propositions are calculated. In the additive use, the alternative propositions difer from one another at the individual level, whereas in the event-sequence use, they difer from one another at the proposition level. To put it diferently, the alternatives of additive pun sentences share a common property, but the property applies to diferent individuals. In the case of event-sequence pun sentences, the alternatives share a higher-level property, something that can be called a discourse-structural property, which applies to diferent propositions. Thus, the two uses are in fact both kinds of additives. The traditional additive can be termed “individual additive,” whereas the event-sequence use can be termed “propositional additive.” Regarding the individual additive pun, the preceding characterization largely restates the discussion in Section 2.1.1. A slightly updated version of the semantics of x pun P is given in (48), which has a discourse structure condition on top of the earlier version in (11). (48) Semantics of x punindividual_additive P a. At-issue content: P(x) b. Presupposition: y[y  ALT(x) P(y)], where ALT(x) is a function yielding the set of x’s alternatives (excluding x itself) in a given context. c. Discourse structure: P(y) < P(x), where φ < ψmeans that proposition φ precedes proposition ψ.

The semantics of the propositional additive pun can be formulated in a parallel fashion, as shown in (49). P is a property of propositions (type s,t,t). It specifies the property that the proposition denoted by the sentence x pun P has in the discourse (hence P(P(x))). (49) Semantics of x punpropositional_additive P a. At-issue content: P(x)

On the apparently non-additive use of Malay additive pun

163

b. Presupposition: q[q  ALT(P(x)) P(q)], where ALT(p) is a function yielding the set of p’s alternatives (excluding p itself) in a given context. c. Discourse structure: q < P(x), where φ < ψ means that proposition φ precedes proposition ψ.

Now, the question is what exactly the discourse-structural property P is. I argue that it is a property of being an answer to the immediate QUD in the current discourse in the sense of Roberts (2012). (50) P = λ p.p  Ans(QUD), where Ans(QUD) is a set of answers to the immediate question under discussion.

According to the QUD theory, discourse is hierarchically organized by often implicit questions and assertions that answer the questions. Let us consider the toy discourse in Japanese from Nomoto et al. (2020) in (51). They analyze its discourse structure (QUD tree) as in (52). (51) a. Tikokusite sumimasen. be.late sorry “I’m sorry for being late.” b. (i) Zitensha-ga pankusitesimai, (ii) densha mo okurete . . . bicyle-nom puncture train too be.delayed “(i) My bicycle had a flat tire, the train too was delayed. . . .” (52) QUD tree of (51) A1 : I’m sorry for being late. Q2 : {Why were you late?}

A2 0 : My bicycle had a fat tire.

A2 00 : The train was delayed.

The discourse in (51) has no explicit question. Hence, an implicit question Q2 is posited. Implicit questions are shown in { }. To avoid implicit questions from being arbitrary, I adopt the following three constrains proposed by Riester (2019). (53) a. Q-A Congruence QUDs must be answerable by the assertion(s) that they immediately dominate. b. Q-Givenness Implicit QUDs can only consist of given (or, at least, highly salient) material, except for function words, in particular wh-pronouns. c. Maximize-Q-Anaphoricity Implicit QUDs should contain as much given (or salient) material as possible.

Q2 captures the relationship between the two sentences in (51). The second sentence (51b) gives the reasons for the first sentence (51a). (51b) consists of

164

Hiroki Nomoto

two propositions, as is represented by two sister nodes in (52), namely, A2 and A2. Here, Q2 is the immediate QUD when sentence (51b) is uttered. Turning back to the propositional additive pun in Malay, given the proposed semantics in (49)–(50), a pun sentence occurs in a discourse structure, as shown in (54). Q

(54) A0

(. . . )

A00 : x pun P

An event-sequence reading arises when the immediate QUD is “What happened to X?” (55) The event-sequence reading of the propositional additive pun: Q: {What happened to x?}

A0 : x R (event 1)

(. . . )

A00 : x pun P (event n)

Note that the temporal sequence meaning arises neither from pun itself nor from the discourse structure it imposes. Instead, it arises from the context by virtue of the Gricean conversational maxim of manner, particularly “Be orderly” (Grice 1975). When x R is uttered before x pun P, a cooperative conversation participant will understand that the event denoted by x R (event 1) happened before that denoted by x pun P (event n). This efect is still available even when event 1 is not actually verbalized, because the relevant Gricean maxim is concerned with meanings, but not specific linguistic forms. Hence, uttering only x pun P gives the impression that the time has moved forward. This impression is reinforced in certain genres, such as narratives, in which events are typically described in the order of occurrence by convention. 3.3

Examples

I would like to demonstrate how the proposed analysis works with two examples from Section 3.1. Example 1

The first example is (45a).

(45a) Dia memanggil saya. Saya pun menyahut. 3sg call me I pun respond “S/he called me. So, I responded.”

(Asmah 2009: 190)

In this discourse, the alternative propositions are not overtly expressed. However, one can come up with multiple events that must have taken place before the speaker actually answered the call. For example, s/he wondered whether s/he should answer the call, s/he decided to do so, s/he

On the apparently non-additive use of Malay additive pun

165

grabbed the phone, s/he pressed the answer button, and so on. All these events are compressed into one preparatory event in the following discourse structure. A1 : S/he called me.

(56)

Q2 : {What happened to me?}

A2 0 : {I prepared to answer the call.}

A2 00 : I pun responded.

Example 2 The second example is (46), which is more complex than the first example. (46) “Aku boleh bersahabat dengan mereka,” kata rusa. Ia pun pergi menghampiri anjing-anjing itu. Apabila anjing-anjing 3sg pun go approach dog.pl that  when  dog.pl ternampak rusa, mereka pun mengepung-nya. see deer they pun surround-3 “‘I can make friends with them,’ thought (lit. said) the deer. Of he went towards the dogs. When the dogs spotted the deer, they surrounded him.” (Goddard 2001: (20))

Although (46) is obviously a part of a longer story, I treat it as if it were a story on its own for the sake of exposition here. To make it a more natural story, I add an introductory sentence at the beginning: “There lived a deer and some dogs” (A1). One of its possible discourse structures is shown in (57).24 Q1 : {What’s the story about?}

(57)

A1 : There lived a deer and some dogs. Q2 : {What happened to the deer and the dogs?}

Q2:1 : {What happened to the deer?}

Q2:2 : {What happened to the dogs?}

A2:1 0 : The deer said “. . . .”

A2:2 0 : The dogs A2:2 00 : They pun spotted the deer. surrounded the deer.

A2:1 00 : He pun went toward the dogs.

Here I follow Riester (2019) and allow for questions that are triggered by assertions. Van Kuppevelt (1995) calls such question-engendering assertions “feeders.” In Roberts’s (2012) system, all questions (but the very initial Big Question) arise from a question as its subquestions. In (57), Q2 is triggered by the feeder assertion A1. Q2.1 and Q2.2 are subquestions of Q2, each having two answers. Notice that pun occurs in the second answer in both cases, matching the discourse structure in (54). Because the immediate QUDs they answer are “What happened to X” questions, the sentence with pun receives an event-sequence reading (cf. (55)).

166

Hiroki Nomoto

It is worth discussing why pun does not occur in the clause Apabila anjing-anjing ternampak rusa “When the dogs spotted the deer.” The reason is straightforward in the proposed analysis: it is not preceded by any alternative proposition that answers the same immediate QUD. By contrast, previous studies, which described pun as lexically encoding the event-sequence meaning, have predicted that pun can felicitously occur there. This is because the event of the dogs spotting the deer takes place after the event of the deer approaching the dogs. The clause could have contained a pun if it were not for the conjunction apabila “when.” The conjunction strengthens the connection between the last two propositions and keeps it separate from the previous two propositions. Without it, the discourse could have the structure in (58), where the four propositions are sisters and answer the same immediate QUD. (58)

... Q2 : {What happened to the deer and the dogs?}

A2 0 : The deer said “. . . .”

A2 00 : He pun went toward the dogs.

A2 000 : The dogs pun spotted the deer.

A2 0000 : They pun surrounded the deer.

Note that it is possible for pun to occur in succession, as shown in (59). (59) Lepas puas mata pun mengantuk dan waktu tu pun telah after satisfied eye pun sleepy and time that pun pfv pukul 3.00 pagi aku pun mengajak dia pulang ke hotel. o’clock 3.00 morning I pun invite 3sg return to hotel “After getting satisfied, I got sleepy, and the time too was 3.00 a.m.; I invited him/her to return to the hotel.”25

4

Non-event-sequence uses of the propositional additive pun

In the previous section, I claimed that the so-called event-sequence use is a special case of the more general propositional additive use, specifically, the case in which the immediate QUD is “What happened to X?” This section discusses other meanings ascribed to the propositional additive pun. They differ from the event-sequence use in that the immediate QUD that the pun sentence answers is something other than “What happened to X?” Moreover, they do not seem to fit any of the categories in Figure 6.1. Nomoto (2020) regards (60) as an example of the scaler pun. (60) Takpe, mak cik. Saya pun tengah free sekarang. it’s.ok auntie I pun prog free now “It’s ok, auntie. I’m free anyway.”

(Nomoto 2020: 160)

On the apparently non-additive use of Malay additive pun

167

With the second sentence with pun, the speaker reassures the auntie in a humble manner by giving a reason that she does not need to worry. Nomoto attempts to explain this meaning with the paraphrase “I am someone who is most likely to be free.” However, the semantics of the scalar pun in (29) actually yields the opposite meaning: the speaker is the least likely individual who is free now, or “I am someone busier than anyone else.” Therefore, the pun in (60) is not the scalar pun. This leaves the possibility of an individual additive reading (“I am also free”). However, (60) is still natural even when the speaker is the only person who is free at the moment. Hence, there is yet another reading, namely, a reading based on the propositional additive pun. I posit the discourse structure in (61). (61)

A1 : It’s ok, auntie. (= The auntie does not need to worry.) Q2 : {Why is it that the auntie does not need to worry?}

A2 0 : {The Auntie didn’t do anything bad.}

A2 00 : I’m pun free now.

The immediate QUD here is “Why is it that X?” which captures the fact that the second sentence gives a reason for the first sentence. The implicit alternative answer A2' need not be the one given in this tree. However, it must be a proposition that addresses Q2 more directly than A2'' to abide by the Gricean conversational maxim of relevance (“Be relevant”): more-relevant answers occur before less-relevant ones. Consequently, the pun sentence gives one of the least direct answers, accounting for the humbleness attached to the use of pun. Example (62) from Goddard (2001) has a similar discourse structure, which is clear from the conjunction sebab “because.” A more direct reason for the parents raising the issue of expenses for the children would be that they did not have money at the moment. (62)

Pernah ibu bapa saya bersuara tentang duit belanja kami anak-beranak, sebab keluarga kami pun bukan orang senang. because family our pun not person well-of “There are times my parents raise the issue of expenses for the children, because our family isn’t well-of.” (Goddard 2001: (14))

Although “What happened to X?” and “Why is it that X?” questions can account for most cases of propositional additive pun, other QUDs also seem possible. I claim that (63) involves the QUD “What happened as a result of X?” though one could also interpret this pun as a scalar use (“even the house”). (63) Aku harap-harap yang datang banci rumah aku nanti adalah makwe cun. Aku dah siap-siap minyak wangi paling mahal. Baju berkolar kain lembut. Seluar paling sedap-sedap pakai.

168

Hiroki Nomoto Rumah pun dah siap-siap kemas macam kilang Ikea.26 house PUN already ready tidy like factory Ikea “I hope the one who’s coming for the census to my house will be a cute girl. I’ve already put on the most expensive perfume. I wore a collared shirt made of soft cloth. I wore my coolest trousers. My house, too, was already ready and tidy like an Ikea factory.”

(64) QUD tree of (63) A1 : I hope the one who’s coming for the census to my house will be a cute girl. Q2 : {What happened as a result of me hoping the one . . . will be a cute girl?}

A2 0 : I’ve already put on the most expensive perfume.

A2 00 : I wore a collared shirt made of soft cloth.

A2 000 : I wore my coolest trousers.

A2 0000 : My house pun was already ready and tidy like an Ikea factory.

The conversation in (65) involves another kind of QUD: “What else should/can be added to X?” This QUD (partially) deals with the relation called “elaboration” in discourse theories capitalizing on a list of predefined rhetorical relations (e.g., Mann & Thompson 1988; Asher & Lascarides 2003). (65) “Abang Hisyam dah makan ke, Mak Teh?,” . . . “Tak. Sikit sangat dia jamah nasi tadi. Tak ada selera katanya. Muka pun macam tak bermaya saja.27 face pun like not vigorous just “‘Has Brother Hisyam eaten yet, Mak Teh?’ . . . ‘No. He only had very little just now. He said he didn’t have an appetite. His face also looked weak.’” (66) QUD tree of (65) Q1 : Has Brother Hisyam eaten yet, Mak Teh? A1 : No. Q2 : {What else should/can be added to “No”?} A2 : He only had very little just now. Q3 : {What else should/can be added to the fact that he only had very little just now?}

A3 0 : He said he had no appetite.

A3 00 : His face pun looked weak.

Finally, I would like to comment on the resemblance that can be observed among the alternative propositions, even though it is sometimes very weak. For instance, in (65), the pun sentence (A3") and its alternative (A3') both describe Hisyam’s bad health conditions. The resemblance here follows from the semantics of the propositional additive pun in (49) and (50), that is, the alternatives

On the apparently non-additive use of Malay additive pun

169

answer the same immediate QUD. Crucially, a contrastive statement, whether implicitly or by means of overt items, such as tetapi “but” and the contrastive topic marker pula, requires a discourse structure diferent from that which licenses pun (see Büring 2003 regarding the discourse structure for contrastive topic). Therefore, a special stipulation about the resemblance is not necessary. 5

Conclusion

This chapter has presented a formal analysis of the particle pun in Malay. I argued that the so-called event-sequence pun should not be seen as a category independent from the focus pun (i.e., normal/non-scalar additive and scalar additive) but as being related to the latter closely. Specifically, it is a special case of what I call “propositional additive” as opposed to “individual additive.” A sentence with the propositional additive pun conveys that there is at least one alternative proposition resembling the proposition denoted by it. The resemblance is that the alternative propositions answer the same immediate QUD as the latter proposition does. The idea of propositional additive made it possible to account for many instances of pun that do not fit into any category proposed in the previous literature, namely, non-event-sequence uses of pun. Moreover, it explains why the event-sequence pun and other propositional additive uses of pun share the same form as the focus pun. The polysemy involving individual and propositional additives is not limited to Malay; it is also found in the additive markers of other languages, including Japanese mo (cf. (51)), Vietnamese cũng, and Persian ham (Okubo 2019: 109–110). Furthermore, acute readers must have noticed that some propositional additive pun sentences can be translated into English with too.28 An interesting research question remaining for future research to address is how the additive markers are similar or diferent with respect to their propositional additive usage cross-linguistically. Close attention to discourse structure is indispensable for such investigation. Abbreviations 3 nom pass pfv

third person nominative passive perfective

pl prog rel sg

plural progressive relativizer singular

Notes 1 Acknowledgments: The research reported in this chapter was supported by the JSPS “Program for Advancing Strategic International Networks to Accelerate the Circulation of Talented Researchers” grant ofered to Tokyo University of Foreign Studies titled “A Collaborative Network for Usage-Based Research on Less-Studied Languages” and JSPS KAKENHI grant number JP18K00568. 2 The Malay variety discussed in this chapter is the standard variety, both formal and colloquial, used in Malaysia. However, most (but not all) of the descriptive generalizations and theoretical analyses proposed in this chapter are also valid in the related Malay variety, Indonesian.

170

Hiroki Nomoto

3 The English translations of Asmah’s terms here and elsewhere are due to Goddard (2001). 4 This sentence was taken from the VAR STD-L subcorpus of Korpus Variasi Bahasa Melayu (Nomoto 2018) using MALINDO Conc (Nomoto et al. 2018). The spelling in this and other examples from this corpus has been standardized. https:// github.com/matbahasa/Melayu_Standard_Lisan/blob/master/ KL201726.txt. 5 The other additive particle juga difers from pun in this respect. It can take a focused constituent to its right. It can also be placed in the clause-final position. (i) a. Saya juga tak tahu [nama dia]F. cf. (8) I juga not know name 3sg b. Saya tak tahu [nama dia]F juga. cf. (9) I not know name 3sg juga “I also don’t know his/her name.” [There is at least one thing other than his/her name that the speaker does not know.] 6 The other focus particle, juga, but not pun, is used for predicate focus, which conveys that P and at least one alternative of it hold for x. (i) Bahasa Melayu bahasa kebangsaan Malaysia. Bahasa Melayu language Malay language national Malaysia language Malay juga/*pun bahasa kebangsaan Singapura. juga/pun language national Singapore “Malay is the national language of Malaysia. Malay is also the national language of Singapore.” (Nomoto 2020: 23) 7 In this connection, Mintz (2002) regards the pun occurring with interrogatives as a distinct category. He does not discuss similar cases involving “one (X).” 8 http://daliaz.blogspot.com/2010/08/seutas-rindu-buat-kamu.html. This sentence was taken from the ZSM WEB2012 subcorpus of the Leipzig Corpus Collection (LCC Goldhahn et al. 2012). All examples from LCC presented in this chapter were obtained using MALINDO Conc (Nomoto et al. 2018). 9 http://arifomar.blogspot.com/2009_12_01_archive.html. ZSM WEB2012, LCC. 10 http://pahangdaily.blogspot.com/2010/02/misteri-imej-baru-tanda-lekuk-hi tam-di.html. ZSM MXD2012, LCC. 11 http://pemimpinpalingajaib.blogspot.com/2007/09/bab-4-persekolahan-seki tar-tahun-1945.html. ZSM MXD2012, LCC. 12 Goddard (2001) refers to it as “predicate-focus pun.” This term could be confusing, as there are genuine cases of “predicate focus,” in which the predicate of a sentence bears focus, as discussed in footnote 6. 13 http://aziziabdullah.blogspot.com/2010/04/jangan-bunuh-kami.html. ZSM WEB2012, LCC. 14 http://damiamyra.blogspot.com/2011/02/hey-mr-bad-guy_1395.html. ZSM WEB2012, LCC. 15 http://muzri.wordpress.com/2009/10/page/44/. ZSM MXD2012, LCC. 16 http://akubdksiabanyahoocom.blogspot.com/. ZSM WEB2012, LCC. 17 http://arminua.blogspot.com/2011/06/hanya-milikmu-14.html. ZSM WEB 2012, LCC. 18 It is interesting to note that a similar (but not exactly identical) diference is found with the particles yo and ne in Japanese. McCready (2009) proposes semantics along the line of (38) and (39) for the two particles. While McCready introduces a special kind of assertion called “strong assertion” to realize the information state update of the addressee, the update occurs only as a consequence of the presupposition in (38). 19 Grosz (2021) speculates that kan in Manado Malay (Stoel 2005: 79–80) is similar to the German ja in its meaning and its occurrence in embedded clauses. I agree

On the apparently non-additive use of Malay additive pun

171

with him regarding the similarity in meaning. However, I do not think that the example he cites (given in the following) illustrates the embeddability of kan. In this example, kalu, glossed as “if,” functions as a topic marker and does not form a conditional clause in the normal sense. As the context and Stoel’s translation suggest, the sentence is about what the speaker did not do at night, in contrast with what s/he did at daytime. (i) Manado Malay (Stoel 2005: 79) (At daytime I helped the workers to dry the cloves in the sun.) kalu malam kan ɲandaʔ ada matahari, mo jumur if night part not have sun asp dry.in.sun bagimana le ciŋe. how part clove “At night there is no sun; how could we dry cloves in the sun?” 20 http://pekanrabu.yforum.biz/t976p550-berita-dunia-hiburan-part-4.ZSM MXD2012, LCC. 21 Kosmo!, http://kosmo.com.my/kosmo/content.asp?y=2017&dt=0530&pub=Ko smo&sec=Stailo&pg= st_01.htm#ixzz4uc1VMcet, accessed 5/10/2017. 22 The English translation is due to Goddard (2001). 23 The boldface here is Mintz’s. 24 Just as some sentences have more than one syntactic parsing, discourse can have more than one structure, leading to various interpretations by diferent individuals. 25 http://saksibisu.blogspot.com/2009_03_01_archive.html. ZSM MXD2012, LCC. 26 blog.en.sani, http://hairussani.blogspot.com/2010/07/banci-penduduk.html, accessed 3/1/2022. 27 Manisnya Kenangan, http://manisnyakenangan.blogspot.com/2011/01/kuseru-rindu-itu-31.html, accessed 3/1/2022. 28 I learned about this use of too from Christopher Tancredi and Elin McCready.

References Asher, Nicholas & Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Asmah Haji Omar. 2009. Nahu Melayu mutakhir [Contemporary Malay Grammar], 5th edn. Kuala Lumpur: Dewan Bahasa dan Pustaka. Büring, Daniel. 2003. On d-trees, beans, and B-accents. Linguistics and Philosophy 26(5). 511–545. http://doi.org/10.1023/A:1025887707652. Chambert-Loir, Henri. 2019. The particle pun in modern Indonesian and Malaysian. Archipel 98. 177–238. http://doi.org/10.4000/archipel.1361. Cumming, Susanna. 1991. Functional Change: The Case of Malay Constituent Order. Berlin: Mouton de Gruyter. Faridah Mohamed & Yumi Kondo. 2005. CD ekusupuresu mareego [CD Express: Malay]. Tokyo: Hakusuisha. Goddard, Clif. 2001. The polyfunctional Malay focus particle pun. Multilingua 20(1). 27–59. Goldhahn, Dirk, Thomas Eckart & Uwe Quasthof. 2012. Building large monolingual dictionaries at the Leipzig Corpora Collection: From 100 to 200 languages. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC’12), 759–765. Istanbul: European Language Resources Association. www.lrec-conf.org/ proceedings/lrec2012/pdf/327_Paper.pdf. Grice, H. Paul. 1975. Logic and conversation. In Peter Cole & Jerry L. Morgan (eds.), Syntax and Semantics: Speech Acts, vol. 3, 41–58. New York: Academic Press.

172

Hiroki Nomoto

Grosz, Patrik G. 2021. Discourse particles. In Daniel Gutzmann, Lisa Matthewson, Cécile Meier, Hotze Rullmann & Thomas Ede Zimmermann (eds.), The Wiley Companion to Semantics. Oxford: John Wiley & Sons. http://doi. org/10.1002/9781118788516.sem047. Hoogervorst, Tom G. 2018. Utterance-final particles in Klang Valley Malay. Wacana 19(2). 291–326. http://doi.org/10.17510/wacana.v19i2.704. Mann, William C. & Sandra A. Thompson. 1988. Rhetorical structure theory: Toward a functional theory of text organization. Text 8(3). 243–281. http://doi. org/10.1515/text.1.1988.8.3.243. McCready, E. 2009. Particles: Dynamics vs. utility. In Yukinori Takubo, Tomohide Kinuhata, Szymon Grzelak & Kayo Nagai (eds.), Japanese/Korean Linguistics, vol. 16, 466–480. Chicago, IL: University of Chicago Press. Mintz, Malcolm W. 2002. An Indonesian & Malay Grammar for Students, 2nd edn. Perth: Indonesian/Malay Texts and Resources. Nomoto, Hiroki. 2018. Korpus variasi bahasa Melayu: Standard lisan [Corpus of Malay Varieties: Colloquial Standard]. https://github.com/matbahasa/Melayu_Standard_ Lisan. Nomoto, Hiroki. 2020. Mareego no kyoukasho: Shoukai bunpou [Textbook of Malay: Comprehensive Grammar]. Tokyo: Next Publishing Authors Press. Nomoto, Hiroki & Aznur Aisyah Abdullah. 2015. Mareeshiago no renyoushuushokuteki hukubun [Clause combining in Malay]. Gogaku Kenkyuujo Ronshuu 20. 253–276. http: //hdl.handle.net/10108/84567. Nomoto, Hiroki & Aznur Aisyah Abdullah. 2017. Mareeshiago no toritate hyougen to hutei hyougen [Focus-sensitive particles and indefinites in Malay]. Gogaku Kenkyuujo Ronshuu 22. 121–131. http://doi.org/10.15026/93726. Nomoto, Hiroki, Shiro Akasegawa & Asako Shiohara. 2018. Building an open online concordancer for Malay/Indonesian. Paper Presented at the 22nd International Symposium on Malay/Indonesian Linguistics (ISMIL). http://www.tufs.ac.jp/ts/personal/ nomoto/ismil2018-slides.pdf Nomoto, Hiroki, Wataru Okubo & Yuta Sakon. 2020. Nihongo koopasu no danwakouzou anoteeshon nimuketa yobitekikenkyu [A preliminary study for discourse structure annotation in Japanese corpora]. In 161st LSJ Meeting Handbook, 333– 339. Linguistic Society of Japan. www.ls-japan.org/modules/documents/LSJpapers/meeting/161/handouts/f/F-4_161.pdf. Okubo, Wataru. 2019. The Focus Particles Ham and Ke in Persian. MA thesis, Tokyo University of Foreign Studies. Parsons, Terence. 1990. Events in the semantics of English: A study in subatomic semantics. In Current Studies in Linguistics Series No. 19. Cambridge, MA: MIT Press. Riester, Arndt. 2019. Constructing QUD trees. In Malte Zimmermann, Klaus von Heusinger & V. Edgar Onea Gaspar (eds.), Questions in Discourse, vol. 2, 163–193. Leiden: Brill. http://doi.org/10.1163/9789004378322_007. Roberts, Craige. 2012. Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics & Pragmatics 5. 1–69. http://doi. org/10.3765/sp.5.6. Stoel, Ruben. 2005. Focus in Manado Malay: Grammar, Particles, and Intonation. Leiden: CNWS Publications. van Kuppevelt, Jan. 1995. Discourse structure, topicality and questioning. Journal of Linguistics 31. 109–147. http://doi.org/10.1017/S002222670000058X. von Fintel, Kai & Lisa Matthewson. 2008. Universals in semantics. The Linguistic Review 25. 139–201. http://doi.org/10.1515/TLIR.2008.004.

7

A unified analysis of (some) discourse particles in Thai Upsorn Tawilapakul and Elin McCready

1

Introduction

While the term particle is used to cover a broad range of kinds of linguistic objects with diferent kinds of functions (McCready 2008; Davis 2009; Zimmermann 2011), discourse particles manifest a specific intention of a speaker and her attitude toward the propositional content of the information. According to Zimmermann (2011), they represent both the speaker’s own attitude toward the content of the information and her assumption about the hearer’s attitude toward it. Zeevat (2005) links the characteristics of discourse particles to speech acts. The compatibility of discourse particles and speech acts calls for proper conditions and settings which are mutually shared by the interlocutors of the discourse. The discourse particle lɛ́ɛw in Thai, for example, denotes the perfective aspect while, at the same time, functioning as both a presupposition trigger and a discourse particle which suggests a contrast against the previous belief mutually shared by the interlocutors (Tawilapakul 2013). The influences of these roles of the particle on the interpretation are illustrated in (1): (1) Danai finished his report lɛ́ɛw Interpretation: Danai finished his report, and this contrasts with the prior belief that he is still writing it. Basically, once a discourse particle is added to a sentence, it provides an extra meaning to the descriptive content of the proposition, contributing presupposition without any efect on the truth conditions of the sentence. It also shows the speaker’s assumption toward the status of the information in the common ground, that is, whether it is part of the common ground or shared by the speaker and the addressee. Discourse particles are used for various purposes. The selection of a particle depends on the speaker’s assumption and attitudes and the intentions she wants to express via her utterance. The functions of particles vary from simply asking for agreement or disagreement to giving a suggestion and showing sarcasm. DOI: 10.4324/9781351057752-8

174

Upsorn Tawilapakul and Elin McCready

The difculties concerning the comprehension and interpretation of discourse particles lie in the detection of the particles’ meanings and the complexity of their information-related operation when they appear in an utterance. Some discourse particles like lɛ́ɛw carry particular meanings which still relate to the meanings and uses they had at earlier stages of historical development. Hence, the investigation of such discourse particles faces less difculties even though their current uses might not resemble how they were used in the past. On the other hand, a number of discourse particles, for example, the Thai lâʔ, ná, and khá, do not carry any obvious propositional meanings of their own that can be expressed in terms of changes in truth-conditions. Besides meaning, their efects on intonation are complicated, especially in tone languages like Thai. Due to the speculative characteristics of discourse particles as discussed earlier, it can be extremely difcult to determine the meaning and felicitous use of a given particle which marks information status, expectations about responses, or special discourse moves. As exemplified in (2a), a particle can mark politeness, while the utterance in (2b) calls for a response from the hearer thanks to the presence of a question particle. (2) a. chan ̌ mây chɔ̂ɔp khun khà 1-f-mid neg like 2-mid/high pol “I don’t like you” (and the speaker is being quite formal) b. yàak cà khuy tɔ̀ɔ máy want mod talk continue q “Do you want to continue talking?” c. wanníi nǎaw ná today cold d.prt “It’s cold today, isn’t it?” Our focus is on items like (2c) in which a discourse particle is used to ofer an additional meaning intended by the speaker (cf., for example, yo, ne in Japanese, sentence-final man in English: McCready 2008; Davis 2010). This research presents an attempt to tackle the operations performed by discourse particles, with a focus on Thai discourse particles làʔ, nàʔ, nìʔ, and ná. Recently, a number of methods for the analysis of discourse particles have been proposed within formal pragmatics. Taking the status of information in the common ground and information management by the speaker into account, in this research we propose an approach which incorporates the mechanism of Question Under Discussion (henceforth QUD). We connect QUD-based approaches with other non-ofcially QUD-based particle analyses and exemplify its application with new analyses of the particles in Central Thai mentioned in the previous paragraph. The relation between discourse particles and intonation will also be briefly discussed through the analysis of the particle ná.

A unified analysis of (some) discourse particles in Thai

175

The rest of the chapter is organized in the following way. Section 2 addresses the basic concept of QUD and its use. Two main approaches to implement QUD to account for the operations of discourse particles are also discussed in this section. Then, Section 3 introduces Thai discourse particles. The QUD-based analyses of the operations of the discourse particles lâ, laʔ, and nâʔ are proposed in Sections 4, 5, and 6, respectively. The potential application of the QUD-based approach to other Thai discourse particles is ofered in Section 7. Lastly, the conclusion of this study and the outlook concerning how relevance is settled between interlocutors and how QUD accounts for it are stated in Section 8. 2

Questions under discussion

Conversations work on one level as mechanisms for information transmission. Consider (3): (3) A: Peter is touring Europe, isn’t he? How is his trip so far? B: He said the trip has been excellent. He’ll leave Germany for Austria tonight and will stay there for a couple of days before travelling further to Slovakia. Speaker A asks a question in order to urge Speaker B to provide the information she wants to know. Speaker B, in a response to Speaker A’s question and intention, ofers the specific information which is relevant to the question, reliable, and of proper amount. This example of a response which represents a typical situation in any daily conversations follows the key idea of Gricean pragmatics, according to which the information provided by a speaker should be relevant, sufcient, not excessive, and clear. Following the concept of QUD (Roberts 1996), any statement is uttered as a response to a question. In other words, the content of the utterance made by a speaker is designated by a specific question she has on her mind at the time of utterance. The conversation in (3) illustrates the essential idea of the Question Under Discussion (QUD): a conversation is structured around a question which establishes the issue to which the rest of the conversation is required to relate. Considering the utterances of Speakers A and B, the question-based conversation shows the intention of H or the hearer (Speaker A) to acquire a piece of information answering his question while, at the same time, designating the role for S or the speaker (Speaker B) to provide that information. Information irrelevant to the question is, following Gricean pragmatics (Grice 1975), not transmitted. The preceding example conversation reflects the standard view of QUDs as information questions. However, it is possible to view a QUD more generally as a decision problem, that is, a problem that determines how interlocutors plan their conversational grids in order to provide a satisfactory resolution to the problem which represents the goal of the conversation (Jefrey 1983;

176

Upsorn Tawilapakul and Elin McCready

Davis 2009). The conversation in (3) thus showcases the resolution to the question asking about Peter’s trip, which requires the hearer to resolve the problem by giving updates about the trip. 2.1

Uses

QUDs can be analytically useful for particles in two ways. They can be used to examine how relevance is set up in a sentence with a particle as opposed to an alternative sentence without it. Moreover, changes in QUD can help analyze shifts in conversational direction. As part of a conversation, an utterance with a particle necessarily obeys the Gricean pragmatic rule of relevance. Utterances, including certain particles such as the Thai ná, like other kinds of utterances, must not introduce another discussion issue which diverts from the issue being discussed. (4) A: chan cà chuan thaanii pay phuukèt ̌  1-f-mid mod invite Thani go Phuket  “I will ask Thani to travel to Phuket with me.” B: (i) tɔɔnníi khaw mây sàbaay ná ̌  now 3-m-mid neg well d.prt  “Now he’s unwell.” (ii) tɔɔnníi khaw chɔɔp chiaŋmày ná ̌ ̂  now 3-m-mid like Chiangmai d.prt  “Now he likes Chiangmai.” (iii) khwaamkít dii ná idea good d.prt “That’s a good idea.” The addition of the discourse particle ná to the three alternative utterances in (4B) is felicitous. Despite different contents, these three utterances still relate to the utterance (4A) and are possible responses to it. In the preceding case, the conversation does not start and is formed by a particular question, and thus, how relevance is set and how responsive utterances are constructed are not explicitly displayed. Rather, the QUD is implicit, and the relevance of utterances to it can be computed by examining the entropy of particular bits of propositional content (i.e., answers to the implicit question) with respect to the issue the question raises. In other words, under a QUD analysis, an assertion like the assertion in (4A) can be viewed as relevant to the extent that it is responsive to the QUD. Following van Rooij (2003), relevance for the entire conversation is definable in terms of entropy resolution. The relevance in these various responsive utterances in (4Bi)–(4Biii) can be accounted for by checking the question entropy resulting from the various possible answers to (4A), that is, the degree to which those possible answers resolve the question at issue. Further, decision problems can be seen in terms of change in optimal actions. As exemplified in (4), the responsive actions for the assertion in (4A) obviously vary because

A unified analysis of (some) discourse particles in Thai

177

the purpose of the speaker, that is, asking for agreement, can lead to either agreement or disagreement from the hearer. Relevance identified through the mechanism of QUD would explain how the problem is formed through the purpose of the speaker and how it is resolved through diferent optimal actions which are determined in accordance with how the speaker would like to act as a response to the assertion. Generally, an assertion can address a sub-QUD like “Who came to the party? Did John?” However, the assertion in (4A) has apparently shown that an assertion can also induce diferent questions which guide the speaker to diferent optimal actions representing changes in QUD. As the example conversation shows, the diferent optimal actions as well as the constructions of responsive utterances are assumed to be directed by diferent sets of possible responses and their consonant entropic levels. These sets of entropies are built based on the view of the speaker on the relations between the assertion and the mutually shared information in the common ground. In the preceding case, the relation is not constrained only to the single possibility that the entire information of the asserted proposition is shared by the interlocutors, as exhibited in the response in (4Biii). In contrast, other possible relations can be proposed if new or contrastive information is not mutually shared and is considered to give a great impact on the question or decision problem formed by the assertion in (4A). This leads to the possible responses in (4Bi) and (4Bii) which imply disagreement caused by two diferent pieces of information that are not shared by the speaker of (4A): (1) that Thani is unwell and thus cannot join the trip and (2) that Thani likes Chiangmai and might not have an interest in a trip to Phuket. The whole structure of super-QUD and sub-QUDs is provided in the following diagram. “I will ask Thani to travel to Phuket” Should I ask Thani to travel to Phuket?

a. How is Thani’s health now?

b. Which province does Thani like?

i. Is Thani well? Ans. = No ii. Is Thani unwell? Ans. = Yes

i. Does Thani like Phuket? Ans. = No ii. Does Thani like Chiangmai? Ans. = Yes

“Now he’s unwell.” (You should not ask him)

“Now he likes Chiangmai.” (You should not ask him)

c. Which kind of place is best for this season? i. Are mountains best for this season? Ans. = No ii. Are beaches best for this season? Ans. = Yes

“That’s a good idea.” (You should ask him)

Figure 7.1 The structure of super-QUD and sub-QUDs of (4).

178

Upsorn Tawilapakul and Elin McCready

Investigating the nature of the sub-QUDs, we hope that the assumed partial resolution of these sub-QUDs in a treelike structure on the QUD domain will provide us a clear picture of shifts in conversational direction. 2.2

QUDs and particles

Several lexical items, including particles, reference operations on QUDs. In the studies on these references, QUD is implemented in at least two ways. Firstly, QUD is used to indicate the (non)resolution of the current QUD or the efects of current utterance. The key point is in parallel with the standard view of QUD, that is, QUD is an information question or at least can be modeled as such a question when implicit. The operation of a lexical item facilitates the resolution of the question being discussed. This, in a way, emphasizes how the resolution of QUD is carried out in compliance with the current utterance. Simply speaking, this approach focuses on the upper level of QUD: the resolution to a question. At a higher level, analyses are mainly aimed at investigating shifts in QUD. Corresponding to realistic conversations, conversational moves do not progress with only straightforward pairs of questions and answers. Due to the fact that an assertion comprises of an eclectic mix of information, shifts in QUD are not always simple. Instead of revealing the resolution to the QUD, this approach explores the grounds for the formation of the QUD and the methods by which it can be manipulated, altered, or shifted. The two approaches of QUD implementation have been taken in the literature both directly and indirectly, as discussed in the following sections. 2.3

Direct approaches

QUD has been used directly to account for the operations of particles. One good example is Rojas-Esponda’s (2014: 13) analysis of the German particle überhaupt. Consider the following restated conversation: (5) A: Möchtest du ein Glas Wein? (Would you like a glass of wine?) B: Nein, Danke. (No, thank you.) A: Hättest du gerne ein Bier? (Would a beer appeal to you?) B: Nein. Ich trinke überhaupt keinen Alkohol. (No. I drink überhaupt no alcohol.) The accumulation of the lines of utterance in (5) shows how the particle überhaupt is normally used. Speaker A tries to find out what kind of alcoholic drink Speaker B would like to drink. After Speaker B has refused two types of alcohol, it seems that Speaker B expects that Speaker A will further ofer other alcoholic drinks and will stop ofering once she is satisfied with the answer.

A unified analysis of (some) discourse particles in Thai

179

However, actually, alcoholic drinks are not Speaker B’s favorites. So in order to conclude the conversation, she just informs Speaker A that she does not drink alcohol at all. The issue under discussion is not explicitly addressed in this conversation, but the interlocutors are well aware that alcohol and Speaker B’s alcoholic preference are being discussed. In the last line, Speaker B added the particle überhaupt to the sentence “I drink no alcohol,” which should already be a perfect response to complete Speaker A’s intention. The occurrence of überhaupt stresses the termination of this conversation. According to Rojas-Esponda, when überhaupt is attached to an assertion, it indicates that the assertion is a move responding to a higher QUD. As mentioned earlier, the cascades of questions asked by Speaker A can be traced back to the super-QUD, “What alcohol would you (Speaker B) like to drink?” Relevance is determined by the issue in this super-QUD, and at the same time, it designates relevant issues for the rest of the conversation. The questions asked by Speaker A as well as the final utterance of Speaker B are responses to the super-QUD. In the preceding case, the assertion with überhaupt, “I drink no alcohol,” calls the super-QUD into question and resolves an answer to it. Rojas-Esponda’s approach successfully unifies a very complex range of data through which a super-QUD or the real key issue of discussion can be retrieved only indirectly. 2.4

Indirect approaches

An implicit reference to QUD is made in the analyses on the two varieties of intonation of the Japanese sentence-final particle yo. Here, QUD comes into play at pragmatic level, in motivations for the discourse moves. According to Davis (2009: 344), yo with rising intonation can be analyzed as marking high relevance. Adapting Grice’s (1975) situation in which the utterances in a conversation are not clearly responding to each other, as shown in (6), Davis proposes that the response given by B indirectly assists Speaker A to solve the problem of petrol running out. What’s more, relevance is defined in terms of question resolution. The addition of yo to the sentence in (7) is intended by Speaker B, who wants to stress that the information she has given is highly relevant to the issue under discussion, that is, that the petrol is running out, in the sense that it suggests a solution to the problem. Yo reminds the hearer that the assertion attached to it is of high importance and must be taken into account. (6) Context: A is standing by an obviously immobilized car and is approached by B. A: I am out of petrol. B: There is a garage around the corner. (7) B: kono this

miti-o road-acc

massugu straight

it-ta go-pst

tokoro place

ni at

180

Upsorn Tawilapakul and Elin McCready gasorinsutando-ga ari-masu yo gas.station-nom be-hon “There’s a gas station straight down the road yo.”

The falling intonation of yo, on the other hand, has been analyzed as “forcing” the hearer to accept the asserted content by inducing a revision in hearer beliefs (McCready 2008). 3

Thai particles

Thai, like other isolating Southeast Asian languages, has a large number of particles. The exact number is hard to state, and individuation depends on theoretical choices, such as intonation and the role they play with interlocutors. Tawilapakul (2016) categorized particles in Thai according to their general functions: signaling gender and politeness, expressing speaker’s reaction, and calling for hearer’s reaction. In the conversation in (8), three politeness particles are used to show the speakers’ respect to the hearers. They also indicate the genders of the speakers. Khráp is used by a male-presenting speaker, while a female-presenting speaker uses khâ and khá. (8) A: dɔ̀ɔkmây raakhaa chɔ̂ɔ lá thâwrày khráp flower price bouquet a/per how much pol-m “How much is a bouquet?” B: sɔ̌ɔŋrɔ́ɔy bàat khà mii lǎay sǐi ná khá two hundred baht pol-f have many color d.prt pol-f “Two hundred baht. There are many colors.” Speaker B in the conversation in (9) uses the particle lɛ́ɛw to express her reaction to the assertion stated by Speaker A. She denies the past expectation about Sutha’s traveling to Pattaya implied in Speaker A’s statement. The expectation was true before the utterance time but has changed otherwise. In (10), Speaker A uses the particle là to inform Speaker B about the activity she did the previous night, and also to urge her to say something to react to it. Speaker B takes her turn and uses the particle or particle sequence là sì. Unlike là, là sì requires the hearer’s confirmation about the speaker’s thought. (9)

A: sùthaa cà pay pháttháyaa Sutha mod go Pattaya “Sutha will go to Pattaya.” B: khaw ̌ plìancay pay phuukèt lɛ́ɛw 3-m-mid change one’s mind go Phuket d.prt “He has changed his mind. He will go to Phuket.”

(10) A: mʉ̂akhʉʉn pay paatíi maa là last night go party come d.prt “I was at the party last night.”

A unified analysis of (some) discourse particles in Thai

181

B: rə̌ə sànùk là sì q fun d.prt d.prt “You were? You had a great time, didn’t you?” A: sànùk mâak ləəy fun a lot d.prt “Yes, I did.” Some Thai discourse particles are also naturally analyzed using a QUD model. Tawilapakul (2013) proposes a solution for the controversial issue concerning the role of the Thai discourse particle lɛ́ɛw using QUD. The conventional claim that lɛ́ɛw functions as a perfective marker is countered by the data from the daily use, which clearly shows that lɛ́ɛw is actually used as a discourse particle to contrast the expectation about the issue. QUD is very efcient in the sense that it paves the way to the reasons of the use of lɛ́ɛw and how the super- and sub-QUDs are stacked with relevance to the issue under discussion in order that the speaker achieve her intended counterexpectation. In this chapter, our focus is on the discourse particles lâ, laʔ, nâʔ, nîʔ, and ná. Our analyses of these particles follow the descriptive characterizations of Cooke (1989) in which Thai particles are categorized into four main types in accordance with their semantic and contextual functions. These include:

• • •

Signaling the relationships between the speaker and the hearer (e.g., khá and khráp) Asking for the hearer’s response (e.g., ná) Signaling a response from a hearer (e.g., là) Signaling the setting of the context (e.g., lâ, laʔ, nâʔ, nîʔ)

4

Lâ

•

Normally, lâ is used to signal “a shift of focus from one question or concern to another directly related one” (Cooke 1989: 9). This is shown in the example from Cooke (1989: 10), restated in (12) here: (11) A: phrûŋníi thəə cà pay nǎy tomorrow 2-mid mod go where “Where will you go tomorrow?” B: mâydây pay nay̌ lɔ̀k lɛ́ɛw thəə lâ neg go where d.prt conn 2-mid d.prt “I’m not going anywhere. And how about you?” Lâ is added to the final position of the utterance in (11B), signaling Speaker B’s intention to shift from the information about Speaker B to the question asking for the same information of Speaker A. Following Cooke’s characterization of lâ, the shift of question or issue signaled by lâ as exhibited in (11B) resembles a shift in the current QUD. That

182

Upsorn Tawilapakul and Elin McCready

is to say, a shift from the current question about the place Speaker B will go the next day to the question about the travel plan of Speaker A is proposed. Nonetheless, Cooke’s notable characteristic of a new question as a “directly related issue” to the current one leaves room for the explication of its notion. In response to the challenge about the relevance of the new issue to the current issue, we propose that given the current QUD Q, there must exist a super-QUD Q. The new QUD, Q′, must also be a subquestion of Q such that Q ≠ Q′. 4.1

Question semantics

Following Groenendijk and Stokhof (1984), the basic semantics of a question Q induces a set of propositions such that each is an answer to Q. The Q-denotation partitions the space of possibilities. This is applied to yes/no and wh-questions in the way shown in (12) and (13). The denotation of the yes/ no question in (12) limits the possibilities in the set only to the propositions “John came” and the negated version “John did not come.” As for the whquestion in (13), the space of possibilities is far larger, as it is an information question which allows all salient propositions relevant to the context to be possible answers. (12) Yes/no question: [[ Did John come to the party?]]={came(j), ¬ came(j)}. (13) Wh-question: [[ Who came to the party?]]={came(a), came(b), came(c), came(ab), . . .} for x  C, where C is a set of salient individuals 4.2

QUD shift

The analysis about the QUD shift, as observed in the use of lâʔ in (11), is based on the definition of contextual subquestion proposed by Djalali et al. (2011). The definition, which is developed from Groenendijk and Stokhof’s (1984) semantics of questions as discussed earlier, is presented in (14). Here,  and  are propositions in the questions Q and Q′, and  is the current information state: (14) Definition of contextual subquestion: Q is a subquestion of Q′ relative to information state , written Q  Q′, if   Q,   Q′   (  ) = . Following the definition, for each answer to Q, there is an incomplete answer to Q′. Thus, the question “Did John come to the party?” is a subquestion of the question “Who came to the party?” as shown in (15). Both questions are relative to the information state determined by context.

A unified analysis of (some) discourse particles in Thai

183

(15) Q = {came(j), ¬ came(j)}, Q′ = {¬xcame(x), came(j), came(m), came(j  m)}. Applying the definition in (15), lâʔ(?) indicates that: (16) i.  QUD(C) = Q ii. QUD(C[?]) = Q′ where Q′′[Q, Q′  Q′′]. The question triggered by lâ, which is a new question, is relative to the information state, as suggested by the current question. Both questions are subQUDs of the super-QUD. Accordingly, further shifts of question are possible, as new questions relating to the same issue can also be discussed. The shift of QUD shown in the restated conversation in (17) can be explained through the schematic semantics provided in (18): (17) A. phrûŋníi thəə cà pay nǎy tomorrow 2-mid mod go where “Where will you go tomorrow?” B. mâydây pay nay̌ lɔ̀k lɛ́ɛw thəə lâ neg go where d.prt conn 2-mid d.prt “I’m not going anywhere. And how about you?” (18) Schematic semantics: • A: ?x(go(B, x)). Call this Q. • B: x¬(go(B, x)); ?x(go(A, x)). Call this Q′ Let Q′′ = x  C?y(go(x, y)). Then Q, Q′  Q′′ (assuming s, h are salient) The use of lâ in the conversation is thus appropriate as it indicates the speaker’s awareness of the discourse structure, that is, that the question B asks is part of the super-QUD “Who will go where?” which establishes the context of the conversation. Conversely, the use of lâ in the conversation in (19) is unacceptable as the question attached to lâ, the progress of A’s report, does not correspond to the super-QUD and the context of the conversation. (19) A. phrûŋníi thəə cà pay nǎy tomorrow 2-mid mod go where “Where will you go tomorrow?” #B. mâydây pay nay̌ lɔ̀k lɛ́ɛw raayŋaan khɔ̌ɔŋ thəə lâ neg go where d.prt conn report of 2-mid d.prt “I’m not going anywhere. And how about your report?” 5

Laʔ

Laʔ is a discourse particle that is used to signal that “a decisive or critical point has now been reached or has already been passed” (Cooke 1989: 11). An example

184

Upsorn Tawilapakul and Elin McCready

displaying the use of laʔ is given in (20) and (21). The speaker of the sentence in (20) has reached the point where she finally made a decision to go home. This marks the termination of any activity or any event that had proceeded previously. The critical point in this case thus refers to the call to end one line of inquiry and start another. Likewise, in (21) the decision for the action “not going to eat” is not made as a plan for the immediate future. It is made after the speaker had eaten something which was possibly not nice for some time. This decision indicates that a critical point or the point where the speaker could no longer stand the food has passed and the action “eating something” has been terminated. (20) klàp bâan laʔ go home d.prt “Well, I’m going home now.” (21) wáay! mây kin laʔ eek! neg eat d.prt “Eek! I’m not going to eat that.” The notion of resolution operative in the function of làʔ as shown earlier can be formalized as follows, where a and a’ are actions and U(a) represents the utility of action a for the agent whose utilities are under discussion: (22) • Q resolved in  if   Q s.t.  = . • Decision problem: U, A, A a set of possible actions and U a function assigning a value to those actions. • A decision problem d is resolved wrt  if a  Ad such that a, a′  Ad[Ud(a) ≥ Ud(a′)] given . • QUD is resolved if (a)QUD = Q is resolved in  or (b)QUD = d and d is resolved in . We take the particle contribution to be expressive (e.g., McCready 2008; Gutzmann 2012). The presence of làʔ in the previous two sentences signals the resolutions of the issues under discussion. Làʔ() indicates that  resolves QUD(C). The QUDs which can be recovered from the contexts here rather than linguistically explicit. In other words, the QUD for (20) involves whether the speaker wanted to terminate the action he had been doing for some time. And if he did, he would do it after the termination. In (21), the speaker has to deal with his attitude toward the food and whether he should eat it up or stop eating it. Accordingly, QUDs again are considered decision problems (cf. Jefrey 1983; Davis 2009; McCready 2012). 6

Nâʔ

The addition of nâʔ to a sentence, according to Cooke (1989: 16), “signals that some . . . consideration is . . . of minor or passing importance, something

A unified analysis of (some) discourse particles in Thai

185

of incidental or low-key relevance.” This descriptive analysis comes from the use of the particle in sentences like (23), which is the example taken from Cooke (1989: 17): (23) mây rúucàk sàpráy lə̌. kɔ̂ náam ʔàtlom nâʔ neg know Sprite q conn water carbonated d.prt “You’re not familiar with Sprite? It’s just a carbonated drink.” Nâʔ-marked utterance in the preceding example has low relevance for the current issue and thus is of little importance. Though the utterance gives new information as to the question “What is Sprite?” the information is regarded by the speaker as peripheral and not really part of the current issue under discussion. Given a QUD-based story about relevance, this is straightforward. Consider the definition of relevance in (24). (24) • A piece of content  is Q-relevant if it reduces the entropy of Q (van Rooij 2003; McCready 2009). • A piece of content  is decision problem (DP)-relevant if a  A such that UC(a) ≠ UC[](a) (cf. McCready 2012). • Learning  changes the utilities the speaker assigns to a (often by changing probabilities). •  is relevant if it is Q-relevant or DP-relevant. • Rel() can be defined as the degree to which  is relevant: 1. Entropy, for Q-relevance (so resolution = max-Q-relevance), 2. Utilities, for DP-relevance (cf. McCready 2012). Then, nâʔ() indicates that  has low relevance in C and is thus not an essential issue for the discussion. Whether the topic in the nâʔ-marked utterance will motivate a QUD shift depends on the attitude of the hearer toward the new information. Sprite might become the new issue under discussion, or the whole issue might be discarded right away. 7 7.1

Other Thai particles Nîʔ

Unlike nâʔ, nîʔ-marked utterances contain important information, as Cooke (1989: 21) points out that nîʔ “signals that some fact, consideration, factor, event is of striking, critical . . . relevance to the matter at hand.” His descriptive analysis is based on the following example conversation: (25) A: pay wâaynáam kan máy go swim together q “Want to go swimming together?”

186

Upsorn Tawilapakul and Elin McCready B: mây dây lɔ̀k. mây mii chút ʔàapnáam nîʔ neg can d.prt neg have suit bathing d.prt “No, I can’t. I don’t have a bathing suit.”

In many cases like (25), nîʔ is used to suggest that the utterance attached to it serves as a reason which responds directly to the current issue under discussion. Another example which illustrates how nîʔ is used is provided in (26): (26) nân khruu khon mày nîʔ that teacher clf new d.prt “That’s the new teacher.” Nîʔ in the preceding utterance is uttered out of the blue and does not correspond to any explicit question. The use of nîʔ signals that the presence of the new teacher has become an important issue. Accordingly, it is very likely that QUD will be shifted to any fact about the teacher which will designate the direction of the following part of the conversation. In the earlier examples, it is obvious that nîʔ indicates a high level of relevance in C, which can be defined in a way opposite to the definition of low relevance immediately above. Conversely, a replacement of nîʔ with nâʔ gives a reverse efect. The presence of the new teacher is only what the interlocutors are experiencing and does not call for any serious attention of the hearer. 7.2 Ná

The discourse particle ná is widely used in Thai. However, its analysis is complex because of intonational factors which result in several phonological variants of the particle including ná, nâ, náa, nâa, and naa. The uses of these variants are exemplified in (27)–(31). Note that additional variants might be found in spoken Thai, and thus, the number of the particles is actually not definitive; it is also controversial whether these should be viewed as diferent particles or as a single particle which interacts with phonology in a more or less systematic manner. (27) phrûŋníi cà pay phuukèt ná. bɔ̀ɔk mɛ̂ɛ dûay tomorrow mod go Phuket d.prt tell mother also “Tomorrow I’ll go to Phuket. (Don’t forget to) Please tell mum.” (28) pay nâ yàa yûŋ go d.prt don’t mess with “Go away! Don’t mess with me!” (29) phrûŋníi mii sɔ̀ɔp náa. pay ʔàan naŋs ̌ ʉ̌ʉ tomorrow have exam d.prt go read book

A unified analysis of (some) discourse particles in Thai

187

“There’s an exam tomorrow. Go study!” (30) ráan nán ʔàrɔ̀ɔy nâa. pay lɔɔŋ kanthə̀ shop that delicious d.prt go try let’s “That restaurant serves delicious food. Let’s try!” (31) nǎŋsʉ̌ʉ lɛ̂m níi dii naa. mây lɔɔŋ ʔàan rə̌ə book clf this good d.prt neg try read q “This book is a good. Don’t you try reading it?” The optimal approach to account for the phonological variants is to define the core meaning of ná and then add the meanings for intonation and lengthening (Cooke 1989; McCready 2015). The basic idea is that ná indicates speaker’s expectations about the hearer’s behavior post-utterance. As seen in the utterances in the preceding text, these expectations vary, and they result in the variants of ná which are conditioned by phonological variations in tone (ná and nâ) and length (naa, náa, and nâa). Nevertheless, the whole project is difcult and complex, and we will not attempt it in this brief chapter. 8

Conclusion

This chapter has briefly introduced the Question Under Discussion model of discourse structure. We have connected the existing analyses of some Japanese particles to the QUD model. We have proposed new analyses of several Thai particles within this model. It remains to be seen how far such analyses can be extended to other particles in Thai and other languages. Our analyses have set immediate future directions for the investigations about particles. In Thai, a combination of multiple discourse particles is possible, as shown earlier in (10) and in (32) and (33) here: (32) phrûŋníi danay cà pay pháttáyaa lɛ́ɛw sì ná tomorrow Danai mod go Pattaya d.prt d.prt d.prt “Tomorrow Danai is traveling to Pattaya, right?” (33) phrûŋníi danay cà pay pháttáyaa lɛ́ɛw nà sì tomorrow Danai mod go Pattaya d.prt d.prt d.prt “The point is, tomorrow Danai is traveling to Pattaya.” It would be interesting to implement the mechanism of QUD to account for the multiple-layered operation arising from the combination. There is also a relation between particles and intonation which can be seen from the cases of the Thai ná and the Japanese yo and variants of those discussed here. This requires an investigation of the phonology of the particles as well (Pittayaporn & Chulanon 2012) and the relation between their tones and sentential intonation.

188

Upsorn Tawilapakul and Elin McCready

Moreover, in some cases, utterances with discourse particles are not responses to questions but to statements in which current questions under discussion are formed through context. This pragmatic factor determines the accessibility of the background information and how the question under discussion is formed. An illustration of the impact of the accessibility of background information is given through the operation of the Thai discourse particle lɛ́ɛw in Tawilapakul (2013: 107–108). Examples are provided in (34)–(36): (34) Sutha: cam máa thîi pùay dây máy remember horse REL sick can Q “Do you remember the sick horse?” Sunan: cam dây. thammay rə̌ə remember can why Q “Yes, I do. Why?” Sutha: tɔɔnní man dəən dây lɛ́ɛw now it walk can d.prt “It can walk now (previously it was sick and could not walk, and we did not expect that it would be able to walk).” (35) Sutha: wanníi chaawbâan diicay mâak today villager delighted very “Today the villagers were very delighted.” Sunan: thammay rə̌ə why Q “Why?” Sutha: khonráay thùuk tamrùat càp dây lɛ́ɛw thief pass police capture successfully d.prt “The thief was captured by the police (previously this thief was on the run and was not expected to be captured by the police).” (36) Sutha: câwnâathîi laaʔɔ̀ɔk càak ŋaan lɛ́ɛw ofcer resign from job d.prt “The ofcer quit his job (previously he was expected to continue doing it).” Sunan: pɛnpaymâydây impossible “That’s impossible!” Generally, the operation of lɛ́ɛw, as proposed by Tawilapakul, is similar to that of the Chinese sentence-final particle le, which, as Soh (2009) proposes, marks change of state and contrast to expectation. Nevertheless, the issue concerning the accessibility of background knowledge is not raised in the case of le. Therefore, it will be interesting to find out how relevance is reached by the interlocutors and how the QUDs are resolved in cases like those in the earlier examples.

A unified analysis of (some) discourse particles in Thai

189

Abbreviations acc conn d.prt hon mid mod nom pol q

accusative connective discourse particle honorific mid-level formality modal nominative politeness particle question particle/marker

clf f high m neg pass pst rel

classifier feminine high-level formality masculine negative/negation passive past relativizer

References Cooke, Joseph. 1989. Thai Sentence Particles and Other Topics. Volume A-80 of Pacific Linguistics. Canberra: Australian National University. Davis, Christopher. 2009. Decisions, dynamics and the Japanese particle yo. Journal of Semantics 26. 329–366. Davis, Christopher. 2010. Constraining Interpretation: Sentence Final Particles in Japanese. Ph.D. thesis, University of Massachusetts at Amherst. Djalali, Alex, David Clausen, Sven Lauer, Karl Schultz & Christopher Potts. 2011. Modeling expert efects and common ground using Questions Under Discussion. In Proceedings of the AAAI Workshop on Building Representations of Common Ground with Intelligent Agents. Washington, DC: Association for the Advancement of Artificial Intelligence. Grice, Herbert Paul. 1975. Logic and conversation. In Peter Cole & Jerry Morgan (eds.), Syntax and Semantics, 3, Speech Acts, 41–58. New York: Academic Press. Groenendijk, Jeroen & Martin Stokhof. 1984. Studies on the Semantics of Questions and the Pragmatics of Answers. Ph.D. thesis, Centrale Interfaculteit, Amsterdam. Gutzmann, Daniel. 2012. Use-Conditional Meaning: Studies in Multidimensional Semantics. Ph.D. thesis, Universität Frankfurt. Jefrey, Richard. 1983. The Logic of Decision. Chicaco: University of Chicago Press. McCready, E. 2008. What man does. Linguistics and Philosophy 31. 671–724. McCready, E. 2009. Particles: Dynamics vs. utility. In Yukinori Takubo, Tomohide Kinuhata, Szymon Grzelak & Kayo Nagai (eds.). Japanese/Korean Linguistics vol. 16, 466–480. McCready, E. 2012. Determining questions. Talk given at the 13th Texas Linguistics Society Conference, University of Texas at Austin. McCready, E. 2015. Toward Particle Universals. Talk given at Particles Workshop, Thammasat University. Pittayaporn, Pittayawat & Pirachula Chulanon. 2012. Syntactically naughty?: Prosody of final particles in Thai. In Tadao Miyamoto, Naoyuki Ono, Kingkarn Thepkanjana & Satoshi Uehara (eds.), Typological Studies on Languages in Thailand and Japan, 13–28. Tokyo: Hituzi Syobo Publishing. Roberts, Craige. 1996. Information structure: Towards an integrated formal theory of pragmatics. In The Ohio State University Working Papers in Linguistics 49, 91–136. Columbus, OH: Ohio State University.

190

Upsorn Tawilapakul and Elin McCready

Rojas-Esponda, Tania. 2014. A discourse model for überhaupt. Semantics and Pragmatics 7(1). 1–45. Soh, Hooi Ling. 2009. Speaker presupposition and Mandarin Chinese sentence-final -le: A unified analysis of the ‘change of state’ and the ‘contrary to expectation’ reading. Natural Language and Linguistic Theory 27. 623–657. Tawilapakul, Upsorn. 2013. Counter-Expectation in Thai. PhD thesis, University of York. Tawilapakul, Upsorn. 2016. Discourse Particles of Thai and Their Prominent Features. Talk given at the Workshop on Formal, Probabilistic and Typological Approaches to Discourse Particles and Modal Adverbs, European Summer School in Logic, Language and Information (ESSLLI 2016), Bolzano-Bozen, Italy. van Rooij, Robert. 2003. Questioning to resolve decision problems. Linguistics and Philosophy 26. 727–763. Zeeva, Henk. 2005. A dynamic approach to discourse particles. In Kerstin Fischer (ed.), Approaches to Discourse Particles, Studies in Pragmatics 1, 133–148. Amsterdam: Elsevier. Zimmermann, Malte. 2011. Discourse particles. In Paul Portner, Claudia Maienborn & Klaus von Heusinger (eds.), Handbook of Semantics (Handbücher zur Sprach und Kommunikationswissenschaft), 2011–2038. Berlin: Mouton de Gruyter.

8

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai conversation Kiyoko Takahashi

1

Introduction1

“Pragmatic particles,” broadly defined, encode non-propositional, relational meanings in discourse context, be they discourse-oriented (discursive, textprocedural), speaker-oriented (subjective, cognitive), or interlocutor-oriented (intersubjective, social). They function as a metalinguistic operator for, say, situating a spoken utterance or a written passage in a discourse context, indicating the speaker/author’s attitude or commitment with regard to the truth of a proposition, suggesting the speaker/author’s communicative intention, mitigating or reinforcing the speaker/author’s emotive attitude toward the interlocutor/reader, indexing the degree of formality of the communicative situation, and so forth. The Thai language is known to abound in pragmatic particles (cf. Bhamoraput 1972; Cooke 1989; Peyasantiwong 1981; Sa-anwong 1981; Singhabhandhu 1983, among others). Especially in oral communication, Thai speakers favor the use of a variety of pragmatic particles to make their utterances as expressive and appropriate to the speech situation as possible. Thai pragmatic particles in general fall into three main types in terms of their syntactic positions: a. Interjectional type occurring in isolation, for example, /hə́y/ in (2) b. Phrase-final type occurring at the end of a prosodic unit called an “intonational phrase” (Selkirk 1984 cited in Pittayaporn & Chulanon 2012: 16), for example, /ŋay/, /pàɂ/ in (1) and /ɂàɂ/ in (2)2 c. Predicate-initial type occurring immediately before a verb- or noun-predicate and after the subject noun phrase, if any, for example, /thɯ̌ŋ/, /kɔ̂ɔ/ in (1) (1). . . sǒŋsay̌ wâa thammay man thɯŋ pùat ŋay ̌ doubt comp why pron prtcl ache prtcl kheen rúu pàɂ Ken(proper.name) know prtcl chán kɔ̂ɔ mây rúu pron prtcl neg know DOI: 10.4324/9781351057752-9

192

(2)

Kiyoko Takahashi “(I) wonder why it aches. Ken, do (you) know (the reason)? I don’t know (the reason).” hə́y hây lâw ciŋ ɂàɂ prtcl make narrate be.serious prtcl kɔ̂ɔ tɔɔn rɛ̂ɛk  dooreemii naaŋ pay rooŋ rian . . . prtcl at.the.beginning Doremi(proper.name) pron go school “Hey, (you will) make (me) tell (you the story of the movie), (are you) serious? Then, at the beginning, regarding Doremi (= a character in the movie), she went to school.”

Common members of the predicate-initial type include /cɯŋ/, /thɯ̌ŋ/, /ləəy/, and /kɔ̂ɔ/. In present-day Thai, the latter two may also appear at the clause-initial position (the beginning of a clause that consists of a predicate with its overt or covert subject noun phrase), for example, /kɔ̂ɔ/ in (2), that is, they are hybrids of the predicate-initial and the clause-initial types.3 The basic function these predicate-initial particles have in common is a text-procedural one; they organize discourse by instructing the interlocutor/reader to find a certain information structure in discourse and thereby facilitate discourse cohesiveness. Additionally, the predicate-initial particles have a cognitive function. Using them helps the interlocutor/reader construe that there exists an implicit relation between the propositional content of the predicate marked by the particles and the propositional content of the preceding predicate(s). Specifically, they signify the speaker/author’s construal of causal relation between the two propositional contents, such that a cause produces an efect or that a reason accounts for a result. The most frequent predicate-initial particle of Thai is /kɔ̂ɔ/. It is assumed that /kɔ̂ɔ/ originates in a sequential indicator for marking the chronological succession of actions and events in narrative discourse (Burusphat 2008). However, the meaning of /kɔ̂ɔ/ in modern Thai is not merely text-procedural but attitudinal as well. The use of /kɔ̂ɔ/ normally alludes to cognitive attitudes on the part of the speaker/author; it may add modal senses, such as conclusive judgment and concessive evaluation, to propositional contents. What is more, /kɔ̂ɔ/ has extended its functional domains not only into the subjective domain (“subjectification”)4 but further into the intersubjective one (“intersubjectification”).5 Its additional interpersonal functions set it apart from another common predicate-initial particle /cɯŋ/ whose functions are confined to text-procedural and cognitive ones (Takahashi 2004). The present chapter focuses on the interactional nature of the Thai pragmatic particle /kɔ̂ɔ/. It aims to present an account of social interactive uses of the particle in conversation. So far, the particle’s dynamic communicative functions have not been seriously examined. This study, therefore, attempts to identify typical communicative functions of the particle through a close examination of instances gathered from corpus data of spoken discourse. It also tries to delineate how the particle’s diferent uses – that is, text-procedural uses in the discursive domain, logical and modal uses in the subjective domain,

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 193 and interpersonal and social uses in the intersubjective domain – are related to one another. The remainder of this chapter is organized as follows. Section 2 explains main characteristics of the particle /kɔ̂ɔ/. Section 3 analyzes various uses of the particle. The findings of the analysis show that uses of the particle can be categorized into several diferent speech-act types. Of particular interest is that in a pragmatically ambiguous context, the use of /kɔ̂ɔ/ concerns a hybrid speech act between an assertive and a directive (description of logical outcome and hortative suggestion at the same time). This hybrid sense is not directly connected with /kɔ̂ɔ/’s original, text-procedural, assertive sense (description of phenomenal consequence or temporal sequence), but it would appear to arise from the subjective assertive sense (description of logical outcome) that is derived from the text-procedural assertive sense. That is to say, the subjective assertive sense (description of logical outcome) links the text-procedural assertive sense (description of phenomenal consequence) with the intersubjective directive sense (hortative suggestion). Thus, it is likely that the usage of /kɔ̂ɔ/ first extends from the discourse-oriented domain into the speaker-oriented domain, and further into the interlocutor-oriented domain, as indicated in Table 8.1. Section 4 concludes this study. Table 8.1 Plausible extension of the usage of /kɔ̂ɔ/ Discourse-oriented domain

 

Speaker-oriented domain

 

Interlocutor-oriented domain

Text-procedural assertive sense (e.g., description of phenomenal consequence)

>

Subjective assertive sense (e.g., description of logical outcome)

> 

Intersubjective directive sense (e.g., hortative suggestion)

2

Characteristics of the pragmatic particle /kɔ̂ɔ/

Burusphat (2008) posits that /kɔ̂ɔ/ in Thai has its origin in Khmer. She regards /kɔ̂ɔ/ and its corresponding lexical items in other Southeast Asian languages (such as /kɔ/ in Khmer, /thì/ in Vietnamese, /lɛː/ in Burmese, /pun/ in Malay, etc.) as a pre-verbal temporal auxiliary or sequential indicator, which, she argues, is an areal feature in the Southeast Asian linguistic area. As mentioned earlier, however, /kɔ̂ɔ/ in present-day Thai is not a mere sequential indicator functioning only in the discursive domain; its functions have expanded into the subjective and the intersubjective domains. The plausible links between its original and derived functions will be discussed in detail in the next section. The basic syntactic and semantic properties of /kɔ̂ɔ/ have not changed since 13–14 CE, the earliest period of the documented history of the Thai language (Sa-anwong 1981). Generally, /kɔ̂ɔ/ places emphasis on the propositional content represented by a verb- or noun-predicate it precedes (Singhabhandhu 1983). In narrative discourse, the predicate-initial /kɔ̂ɔ/ is used

194

Kiyoko Takahashi

to highlight backbone information describing informative and central events which advance the storyline (Chodchoey 1986). From the perspective of clause linkage, /kɔ̂ɔ/ can be considered a linker of two clauses: a prior subordinate/ supporting clause and a posterior main/focal clause marked by /kɔ̂ɔ/ (Iwasaki & Ingkaphirom 2005: 173–174, 274). As such, it is used for the marking of the main/focal clause, but not of a subordinate/supporting clause.6 See (3) for an example.7 (3)

. . . phɔɔ klàp pay as.soon.as return go kɔ̂ɔ nân lɛ̀ɂ prtcl that prtcl kɔ̂ɔ cháy chiiwít naaŋ pokatìɂ . . . prtcl spend life pron normally “As soon as (she = a character in the movie) got back, (it was) that; (she) spent her normal life.” [predicate-initial, highlighting, narration]

Example (3) comprises the prior adverbial clause (phɔɔ klàp pay “as soon as (she) got back”) and the two posterior main clauses serialized. The first main clause contains a noun-predicate (nân lɛ̀ɂ “it was that”), and the second one contains a verb-predicate (cháy chiiwít naaŋ pokatìɂ “(she) spent her normal life”), which is an enlarged paraphrase of the noun-predicate. Being led by /kɔ̂ɔ/, the two predicates each convey focal (as opposed to supporting) and foreground (as opposed to background) information. Furthermore, when /kɔ̂ɔ/ occurs clause-initially in narrative discourse, it functions as a conjunction associated with such assertive senses as additive description, adequate interpretation, and others (see Section 3.3.2). In (4A.ii), for example, a clause-initial /kɔ̂ɔ/ is used for the sense of additive description. (4)

A.i: . . . pə̀ət pratuu khâw pay cəə maacɔɔrikâa open door enter go meet Majorica(proper.name) maacɔɔrikâa yaŋ pen mɛ̂ɛ mót yùu ləəy Majorica still cop witch cont prtcl “(Doremi = a character of the movie) goes in through the door and meets Majorica. Majorica still remains a witch.” B:  maacɔɔrikâa ɂɔ̌ɔ pen mɛ̂ɛ mót Majorica prtcl cop witch tɔɔn laˇŋ pen lûuk ɂɔ́ɔt afterwards become tadpole “Majorica, oh, (she) is a witch. Afterwards (she will) become a tadpole.” A.ii: yaŋ pen mɛ̂ɛ mót yùu ləəy still cop witch cont prtcl kɔ̂ɔ dooreemii pay chíi nâa . . . prtcl Doremi(proper.name) go point face “(She) still remains a witch. Then, Doremi points at (her) face.” [clause-initial, additive description, narration]

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 195 In conversational discourse, a clause-initial /kɔ̂ɔ/ may appear at the opening of a responsive turn by one of conversational coparticipants. The particle /kɔ̂ɔ/ that occupies the outset position of a responsive turn (for short, “turninitial /kɔ̂ɔ/”) is ordinarily followed by a clause with an overt or covert subject noun phrase. Iwasaki and Ingkaphirom (2005: 175–176) call it a “response marker.” According to them, the use of /kɔ̂ɔ/ as the response marker characteristically signals that the response may not completely satisfy the questioner. Further, they claim that /kɔ̂ɔ/ has developed into a hedging device for avoiding a halt of the flow of conversation. Speaker B’s utterance with /kɔ̂ɔ/ in (5) illustrates this hedging usage.8 (5)

A: thîi nay baa tham nâathîi ɂaray bâaŋ háɂ at in bar do duty what any prtcl “What are some of (your) duties at the bar?” B: kɔ̂ɔ pen wéttrèet prtcl cop waitress “Well, (I) am a waitress.” [hedging reply]

To summarize, uses of /kɔ̂ɔ/ can be classified from both a discursive and a syntactic perspective, as shown in Table 8.2. Table 8.2 Main types of the pragmatic particle /kɔ̂ɔ/ Discursive types

Syntactic types

1. Narrative type or conversational turninternal type, for example, (1), (2), (3), (4) 2. Conversational turn-initial type, for example, (5)

1. Predicate-initial type, for example, (1), (3) 2. Clause-initial type, for example, (2), (4), (5)

The present chapter focuses on the turn-initial type of the clause-initial subtype, for example, (5). To my knowledge, no studies have ofered an in-depth analysis of communicative functions of the turn-initial /kɔ̂ɔ/, and only a few studies (Chodchoey 1986; Iwasaki & Ingkaphirom 2005; Noss 1964; Peinukachon 2002) have briefly mentioned such turn-opening usages. Employing a corpus-driven approach, this study explores what illocutionary forces can be set forward by the use of a turn-initial /kɔ̂ɔ/ in talk-in-interaction. 3

Corpus-based analyses

The corpus data of this study comes from the recording and transcription of a 32-minute dialogue between a female and a male college student in Bangkok. It is one of the 11 conversations which comprise the TUFS Spoken Thai Corpus (see footnote 1). The two college students engaged in a casual conversation; they largely talked about their favorite things, such as movies and foods. In total, 128 tokens of /kɔ̂ɔ/ were found in the data. Out of 128, 13 tokens

196

Kiyoko Takahashi

(about 10%) are of the turn-initial type, and 115 tokens (about 90%) are of the turn-internal type. Although the number of instances of the turn-initial /kɔ̂ɔ/ was only 13, several diferent illocutionary-force types pertaining to its use could be identified.

3.1

Attested usage types of turn-initial /kɔ̂ɔ/

The working definition of “turn-initial /kɔ̂ɔ/” in this study is as follows. If a token of /kɔ̂ɔ/ (or idiomatic coalescent unit containing /kɔ̂ɔ/)9 appears at the start of a turn that immediately follows the previous turn (or if it follows an interjectional particle that immediately follows the previous turn)10 and the two serial turns represent two distinct speech acts, then the token is categorized as a turn-initial /kɔ̂ɔ/. Put diferently, the two speakers involved in the dialogue must carry out diferent, though closely related, speech acts in the respective turns. Table 8.3 shows diferent usage types of the turn-initial /kɔ̂ɔ/ that I found in the corpus data. Table 8.3 Usage types of the turn-initial /kɔ̂ɔ/ 1. Primary event-participant(s) 1.1. Sharing, for example, (6) 1.2. Non-sharing, for example, (7) 2. Illocutionary forces 2.1. Response 2.1.1. Reply, for example, (6), (7) 2.1.2. Skew reply, for example, (8) 2.2. Association 2.2.1. Concurring, for example, (9) 2.2.2. Contesting, for example, (10)

In terms of the “primary event-participant(s)” in propositional contents, the attested turns linked with a turn-initial /kɔ̂ɔ/ can be classified into two types: (a) the type with the sharing of the primary event-participant(s) among the two propositional contents, and (b) the type without the sharing (for short, “sharing” and “non-sharing” types). In the sharing type, the two propositional contents of the two speakers’ turns before and after /kɔ̂ɔ/ share the primary event-participant(s). In (6), for instance, the place called “SCB” is the topic and the primary event-participant in both the two turns’ propositional contents. In contrast, the two propositional contents of the non-sharing type do not share the primary eventparticipant(s), as shown in (7). The primary event-participant of (7A) is the movie that the two speakers are talking about, while that of (7B) is the characters of the movie. (6)

A: thîi ɂés sii bii nîi rabə̀ət rɯ̌ɯ plàaw at SCB(proper.name) prtcl explode prtcl rɯ̌ɯ rabə̀ət dâan or explode fail.to.explode “At the SCB, did an explosion occur? Or, (it) failed to explode?”

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 197 B: kɔ̂ɔ rabə̀ət yùu náɂ hěn phâap ɂàɂ prtcl explode cont prtcl see picture prtcl “Well (I will tell you), (at the SCB) an explosion has occurred. (I) saw a picture (of the SCB).” [sharing] (7)

A: man pen rɯ̂aŋ kìaw kàp ɂaray pron cop story link with what “What kind of story is it (= the movie)?” B: kɔ̂ɔ kháw khɛ̀ŋ rót kan prtcl pron race car recp “Well (I will tell you), they (= the characters in the movie) did a car race.” [non-sharing]

With the same primary event-participant(s), the two propositional contents of the two turns before and after /kɔ̂ɔ/ of the sharing type (6) are more directly linked with each other, compared to those of the non-sharing type (7). The point to observe is that the use of a turn-initial /kɔ̂ɔ/ does not depend on whether the primary event-participants in the two propositional contents are the same or not. The use of it signals that more talk is forthcoming, which is suggestive of the turn-taker’s collaborative stance in conversational interaction. On this basis, I consider the turn-initial /kɔ̂ɔ/ as an indicator of reactive assertion (response or association) (see following). In terms of speech acts by interlocutors, the illocutionary-force types of a pair of turns connected by a turn-initial /kɔ̂ɔ/ can be categorized into two combinational types: (a) the type of one’s response to the other’s question, and (b) the type of one’s association with the other’s assertion (for short, “response” and “association” types). Each type, in turn, subsumes two contrastive types. The response type encompasses (a) the “reply” type, which directly answers the conversational coparticipant’s question, and (b) the “skew reply” type, which does not directly answer the question. The association type comprises (a) the “concurring” type, with a harmonious coordination of the two turns, and (b) the “contesting” type, with a disharmonious apposition of the two. Examples of the response type include (6) to (8). Speaker B of the reply type (6) and (7) gives a proper answer to Speaker A’s question, while Speaker B of the skew reply type (8) evades doing so. In (8), Speaker B does not give an answer to the question about movie stories but just says that she barely has seen Thai movies. (8)

A: lɛ́ɛw mii rɯ̂aŋ naˇy ɂìik ɂàɂ then exist story which else prtcl “Then, what else (= an another movie story) do (you) have?” B: kɔ̂ɔ naˇŋ thay mây khɔ̂y dây duu náɂ prtcl movie Thai neg rather rea watch prtcl “Well (I will tell you), regarding Thai movies, (I) hardly saw (them).” [skew reply]

198

Kiyoko Takahashi

(9) and (10) are two contrastive examples of the association type: concurring and contesting types, respectively. The latter turn by Speaker B of the concurring type (9) (“you must try to go to Japan for eating Japanese food there just one time”) is a comment in favor of, or a hortative suggestion in support of, Speaker A’s desire to eat Japanese food in Japan. By contrast, the latter turn by Speaker B of the contesting type (10) (“I still want to eat Japanese food”) ofers an opinion opposed to Speaker A’s explanation that there are also choices to eat Western food in Japan. (9)

A: . . . kɔ̂ɔ dii ɂàɂ prtcl be.good prtcl “(To eat Japanese food in Japan) is good.” B: kɔ̂ɔ tɔ̂ŋ lɔɔŋ pay sák khráŋ nɯŋ duu prtcl must try go just time one look “Well (if you think so), (you) must try (to go to Japan for eating Japanese food there) just one time.” [concurring]

(10) A: mii hɛɛmbəəkoŋ hɛɛmbəəkə̂ə sataabák yaŋ mii yùu exist hamburgers.and.the.like Starbucks also exist cont tɛ̀ɛ yùu thîi tookiaw náɂ but be.located at Tokyo prtcl “There are hamburgers and the like. There are also Starbucks cofee shops, but (they) are located in Tokyo.” B: kɔ̂ɔ yàak thaan yùu prtcl want eat cont “Well (though you suggest so), (I) still want to eat (Japanese food).” [contesting, explanation] (11) provides an additional example of the contesting type. Speaker B’s turn in (11) starts with the compound (idiomatic coalescent unit) /tɛ̀ɛ kɔ̂ɔ/ “even so,” which is composed of the concessive linkage morpheme /tɛ̀ɛ/ “but” and /kɔ̂ɔ/.11 With this explicit concessive marker, the turn is readily interpreted to express an evaluation contrary to the expected evaluation for the movie story being described. (11) A: com nám taay mây chây . . . sink water die neg “(They = the characters in the movie) were drowned, but not . . .” B: tɛɛ sanùk dii ̀ kɔ̂ɔ but prtcl be.fun be.good “But (even so) (it) is fun.” [contesting, concessive description] It is noteworthy that all the attested usage types of the turn-initial /kɔ̂ɔ/ show afnities across a dialogically juxtaposed pairing of utterances before and after /kɔ̂ɔ/. Put another way, the two utterances of either type hold a more or less

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 199 cohesive relation, irrespective of whether or not the turn-taker straightforwardly replies to the interlocutor’s question, whether or not the turn-taker’s associative utterance gets into line with the interlocutor’s expectations, and whatever emotion the turn-taker bears in mind while taking the turn. By means of /kɔ̂ɔ/, the turn-taker indicates his/her intention of reacting to the interlocutor in a positive and cordial manner and also of cooperating with the interlocutor to sustain the conversation. Moreover, all the usage types make a reactive assertion with a modal flavor, such as emphatic description, conclusive judgment, concessive evaluation, and the like. This implication of subjective meanings is similarly involved in the text-procedural usage of the predicate-initial /kɔ̂ɔ/, as mentioned in Section 1. Table 8.4 summarizes all attested usage types of the turn-initial /kɔ̂ɔ/ with the number of tokens of each type. To recapitulate, a couple of turns before and after /kɔ̂ɔ/ may or may not share the primary event-participant(s). With the sharing of the primary event-participant(s), the two propositional contents of the prior and posterior turns are more directly linked with each other (sharing type); without it, the two are less directly linked (non-sharing type). The two turns each exhibit a distinct speech act. There are two combinational speech-act types. One is the combination of a question and its response (response type); the other is the combination of an assertion and its association (association type). These principal illocutionaryforce types are further categorized into two antithetical types. The response type is either conducive (reply type) or deviatory (skew reply type). The association type is either conjunctive (concurring type) or disjunctive (contesting type). Table 8.4 Usage frequencies of each type of the turn-initial /kɔ̂ɔ/ Primary event-participant(s) Sharing [2 tokens] Illocutionary forces Response [7 tokens] Association [6 tokens]

Non-sharing [11 tokens]

Reply

[2 tokens], for [4 tokens], for example, (7) example, (6) Skew reply [1 token], for example, (8) Concurring [4 tokens], for example, (9) Contesting [2 tokens], for example, (10)

The number of tokens is put in brackets. Because the overall number of turn-initial instances is only 13, the number of each subtype is very small, ranging from 4 to 1. The most frequent were the type of “non-sharing and reply” [4 tokens] and the type of “non-sharing and concurring” [4 tokens]; the second most were the type of “sharing and reply” [2 tokens] and the type of “non-sharing and contesting” [2 tokens]; the third most was the type of “non-sharing and skew-reply” [1 token]; and there were no instances of the types of “sharing and skew-reply” and “sharing and association (neither concurring nor contesting)” in the corpus. I admit that the amount of data is a limitation of this study. There might be missing types that are actually in use but not attested due to the limitation of the data.

200

Kiyoko Takahashi

Nevertheless, I believe that the diferences in the number of the attested subtypes tell us something about Thai speakers’ disposition in the usage of the turn-initial /kɔ̂ ɔ/. The figures in Table 8.4, though small in number, may allow for the following interpretations. First, regarding the types of sharing and non-sharing of the primary event-participant(s), instances of the non-sharing type [11 tokens] outnumber those of the sharing type [2 tokens]. From this, we can see that, often, a turn-taker beginning with /kɔ̂ɔ/ introduces his/her novel perspective involving diferent primary event-participant(s) into the conversation, that is, the turn-taker does not simply take the same angle as the dialogic partner’s previous turn, but he/she tends to say something viewed from a diferent vantage point. Second, when comparing the two main illocutionary-force types, the number of instances of the response type [7 tokens] does not very much difer from that of the association type [6 tokens]. This means that /kɔ̂ɔ/ is used for the marking of response (reaction to a question) and of association (reaction to an assertion) rather impartially. In this regard, the nomenclature “response marker” seems inadequate, since it refers to only one side of principal speech acts put forward by the turn-initial /kɔ̂ɔ/. Rather, we may call the turn-initial /kɔ̂ɔ/ more generally a “reaction marker” instead. Third, as for the response type, instances of the reply type [6 tokens] outnumber those of the skew-reply type [1 token]. The skew-reply type gives an explanation rather than an answer. The explanation is motivated by what is said in the previous turn. Only one token of the skew-reply type is found in the corpus data. It follows that in a responsive turn, /kɔ̂ɔ/ is likely to be associated with a collaborative response (response proper), which allows development of a good rapport with the dialogic partner and helps the dialogue proceed in concord. Fourth, with respect to the association type, the number of instances of the concurring type [4 tokens] is larger than that of the contesting type [2 tokens]. The concurring type marks a harmonious association, which meets the dialogic partner’s suggestions, desires, reasoning, inferences, expectations, purposes, or the like. The contesting type may allow the turn-taker to make a clear, concessive description or rather to explanatorily present an opinion opposed to the partner’s. Anyhow, the corpus data show that a turn-initial /kɔ̂ ɔ/ used in an associative turn tends to be of the concurring type. The important point to note is that the turn-initial /kɔ̂ɔ/ may be employed not only as the marker of consonant reaction (reply and concurring types) but also of dissonant reaction (skew-reply and contesting types). Whether the reaction is consonant or dissonant with the previous turn, the turn-taker’s act of reaction to the dialogic partner, in itself, reflects his/her wish to continue the dialogue with the partner.

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 201 3.2

Frequent usage types

Table 8.4 earlier shows that 13 tokens of the turn-initial /kɔ̂ɔ/ can be classified into the following four speech-act types: (i-a) reply, (i-b) skew reply, (ii-a) concurring, and (ii-b) contesting types. The reply type has six tokens, which is the most among the four, and the concurring type has four tokens, which is the second most. This section discusses exactly what illocutionary forces these two frequent types are compatible with. Section 3.2.1 explicates particular speech acts of the reply type, and Section 3.2.2 those of the concurring type. 3.2.1

Reply types

The reply type constitutes two subcategories: (a) straight reply (i.e., reply proper) and (b) hedging reply. The corpus data contains four tokens of the former type, for example, (6), (7), and (12), and two tokens of the latter type, for example, (13). (12) A: lɛ́ɛw mii ɂaray ɂìik ɂàɂ then exist what else prtcl “Then, what else is there?” B: kɔ̂ɔ mây rúu ɂàɂ prtcl neg know prtcl “Well (though you ask so), (I) don’t know.” [straight reply] (13) A: lɛ́ɛw yùu yîipùn dây ŋay then stay Japan pos prtcl “Then, how can (you) stay in Japan?” B: kɔ̂ɔ hěn wâa yùu yâak prtcl see comp stay be.difcult “Well (as you may think), (I) think that it is difcult (for me) to stay (in Japan).” [hedging reply] Straight and hedging replies both willingly provide the conversational coparticipant with relevant information according to the coparticipant’s request for a response, despite the fact that sometimes the degree of relevance of the information given by a hedging reply is low. Compare (12) and (13). In the straight-reply usage (12), Speaker B gives quite a straight answer (“I don’t know”) to Speaker A’s question (“What else is there?”). Even though the content of Speaker B’s reply is negative, this negative reply is no less relevant to the question than an afrmative one. In the hedging-reply usage (13), on the other hand, the content of Speaker B’s reply (“It is difcult for me to stay in Japan”) does not have a direct connection with Speaker A’s question (“How can you stay in Japan?”), though it is not irrelevant to the question at all.12

202

Kiyoko Takahashi

3.2.2

Concurring types

All four tokens of the concurring usage similarly make a reactive assertion, but they difer in their detailed illocutionary forces. The following two subtypes of the concurring usage are conceivable from the corpus data: (a) the description of phenomenal consequence, for example, (14), and (b) the description of logical outcome, for example, (15) and (16). Let us consider each of examples (14) to (16). (14) A: man phə̂əm khɯ̂n rɯ̂ay rɯ̂ay pron increase inc continually “It (= the reported intensity-value of the earthquake in the Chiangmai district) increased continually.” B: kɔ̂ɔ ləəy ŋoŋ prtcl therefore be.puzzled “Well (that being the case), therefore (we) were puzzled.” [description of phenomenal consequence] Example (14) exemplifies the first type: the description of phenomenal consequence. Speaker A speaks of a strange thing about the recent large-scale earthquake in the Chiangmai district (“The reported intensity-value of the earthquake increased continually”). Quickly taking a turn, Speaker B describes the situation at that time, that the strange information about the intensityvalue made people, including her, feel puzzled. Evidently, the content of Speaker B’s turn is in accord with that of Speaker A’s turn, and furthermore, the two contents have a cause-and-efect relationship. That is, a series of reports on the varying intensity-value of the earthquake (the cause) led to the resultant confusion among people (the efect). (15) A: dây rɔ́ɔy hâa sìp tɛ̂m obtain 150 score “(If they can catch the Golden Snitch, they will) get 150 points.” B: ɂɔ̌ɔ kɔ̂ɔ khɯɯ chanáɂ ŋíi rə̌ə prtcl prtcl namely win like.this prtcl “Oh, well (that being the case), that is, (they) won like this, yes?” [description of logical outcome] Example (15) illustrates the second type of the concurring usage: the description of logical outcome. On the basis of Speaker A’s explanation about the rules of the game Quidditch, as played by the characters in the Harry Potter movies (“If they can catch the Golden Snitch, they will get 150 points”), Speaker B reflects a logic implied by Speaker A’s explanation and eventually understands that the characters won the game because they could catch the Golden Snitch and so got 150 points. The contents of the two conversational

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 203 coparticipants’ utterances of (15), like those of (14), have a causal relation. However, the causal relation involved in (14) and that in (15) difer in the level of abstraction. The causal relation of (14) exists in the physical space, that is, the cause-situation brings about the consequence-situation. It is a phenomenological, truth-conditional relation, whose level of abstraction is relatively low. In contrast, the causal relation of (15) is highly abstract since it involves a logic, viz. the reason accounts for the result. It requires mental calculations, inferences, suppositions, etc. on the part of the speaker. In (15), Speaker B is told of the game’s rule that one should get 150 points when catching the Golden Snitch, from which he draws an inference to explain why the persons in question won the game. (16) A: . . . kɔ̂ɔ dii ɂàɂ prtcl be.good prtcl “(To eat Japanese food in Japan) is good.” B: kɔ̂ɔ tɔ̂ŋ lɔɔŋ pay sák khráŋ nɯŋ duu prtcl must try go just time one look “Well (if you think so), (you) must try to go (to Japan for eating Japanese food there) just one time.” [description of logical outcome and hortative suggestion at the same time] Example (16)=(9), too, is of the second type: the description of logical outcome. However, it can be considered a hybrid between an assertive and a directive speech act, for it makes a resultative description and a hortative suggestion at the same time. Going along with Speaker A’s opinion that it is good to eat Japanese food in Japan, Speaker B claims that if she thinks so, she must try doing so. This claim is concurrently an advice for her to try doing so. The hybrid type comprises a speaker-oriented assertive (description of logical outcome) and an interlocutor-oriented directive (hortative suggestion). On the other hand, the original speech-act type of /kɔ̂ɔ/ as a sequential indicator is of the discourse-oriented assertive type: the description of phenomenal consequence. This speech act inherently has nothing to do with genuinely interlocutor-oriented directives, such as suggestion and advice. It is likely that /kɔ̂ɔ/’s speech-act types have expanded in a step-by-step manner, as in the following. /kɔ̂ɔ/ first gained an additional assertive sense associated with the speaker-oriented, cognitive domain (description of logical outcome), and then it further enlarged the range of speech acts into the interlocutor-oriented, social domain (hortative suggestion) (cf. Table 8.1 in Section 1). To better comprehend this step-by-step expansion process of speech-act types bearing on uses of /kɔ̂ɔ/, it is necessary to distinguish the extended type (description of logical outcome, for example, (15)) from the original type (description of phenomenal consequence, for example, (14)).

204

Kiyoko Takahashi

Table 8.5 Subtypes of “reaction” speech act pertinent to uses of the turn-initial /kɔ̂ɔ/ i. Response to question

ii. Association with assertion

i-a. Reply i-a.1. Straight reply (or reply proper), for example, (6), (7) and (12) i-a.2. Hedging reply, for example, (13) i-b. Skew reply (Explanation), for example, (8) ii-a. Concurring ii-a.1. Description of phenomenal consequence, for example, (14) ii-a.2. Description of logical outcome, for example, (15) ii-a.3. Description of logical outcome and Hortative suggestion, for example, (9)=(16) ii-b. Contesting ii-b.1. Contesting explanation, for example, (10) ii-b.2. Concessive description, for example, (11)

To sum up, the results of the corpus-based research reveal that turn-initial uses of the reaction marker /kɔ̂ɔ/ are associated with two main illocutionaryforce categories: (i) response to question and (ii) association with assertion. Further, they embrace at least eight subcategories as listed in Table 8.5: (i-a.1) straight reply, (i-a.2) hedging reply, (i-b) skew reply (explanation), (ii-a.1) description of phenomenal consequence, (ii-a.2) description of logical outcome, (ii-a.3) description of logical outcome and hortative suggestion at the same time, (ii-b.1) contesting explanation, and (ii-b.2) concessive description. 3.3

Less-frequent-usage types

As discussed in the preceding subsections, the turn-initial /kɔ̂ɔ/ of the clauseinitial subtype mainly functions as a reaction marker. It marks the turn it accompanies as a certain reaction to the immediately preceding turn that expresses a question or assertion. The reaction may or may not match up to the questioner’s expectations, and it may or may not support the conversational coparticipant’s views. The turn-internal /kɔ̂ɔ/ of the predicate-initial subtype functions in the text-procedural domain as a highlighting marker, as mentioned in Section 2. It makes the proposition it accompanies stand out, thereby marking the proposition as the foreground consequence-situation in contrast with the background cause-situation described by the immediately preceding propositional unit. In other words, it retrospectively connects the proposition to the preceding propositional unit, creating an implicit causal relation between them. Apart from these typical uses (“reaction” and “highlighting”), /kɔ̂ɔ/ has other less-frequent uses. In the following, the use for bringing about “collaborative completion” in conversations (the turn-initial type of the predicate-initial subtype) is explained in Section 3.3.1, and “conjunction” uses in narrations (the turn-internal type of the clause-initial subtype) are introduced in Section 3.3.2.

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 205 3.3.1

Collaborative completion

The turn-initial /kɔ̂ɔ/ has two subtypes in terms of its syntactic position: (a) predicate-initial type and (b) clause-initial type (Table 8.6). This section explicates the nature of the former predicate-initial type. Table 8.6 Syntactic subtypes of the turn-initial /kɔ̂ɔ/ 1. Predicate-initial type: Collaborate completion, for example, (18) 2. Clause-initial type: Reaction, for example, (6) to (16)

A typical text-procedural use of the predicate-initial /kɔ̂ɔ/ is exemplified in (3) earlier. Similarly, the constructed example (17) in the following is a narration by a single speaker, and its second clause includes a predicate-initial /kɔ̂ɔ/, which highlights the content of the predicate preceded by /kɔ̂ɔ/ (/lîan/ “be oily”). (17) tɛ̀ɛ thâa thaan thoŋkhátsɯ̀ɂ thúk wan man kɔ̂ɔ lîan but if eat breaded.pork every day pron prtcl be.oily “But if (we) eat breaded pork everyday, then it is oily.” [predicate-initial, narration] (18) A: tɛ̀ɛ thâa thaan thoŋkhátsɯ̀ɂ thúk wan kɔ̂ɔ but if eat breaded.pork every day prtcl “But if (we) eat breaded pork everyday, then . . .” B: kɔ̂ɔ lîan prtcl be.oily “Well (if so), (it) is oily.” [turn-initial, conversation] By contrast, example (18) illustrates an interpersonal use of the turn-initial /kɔ̂ɔ/ in a dialogue. Speaker B’s turn contains a turn-initial /kɔ̂ɔ/, and so it is easily interpreted as a reactive assertion. Speaker B kindly finishes Speaker A’s ongoing turn, assuming that she should wind up her turn as such or that she might want to refrain from saying a negative thing. It looks as if the two speakers jointly produce a single utterance. Speaker B does not intend to interrupt Speaker A’s turn. Rather, he contributes to the completion of her turn. The pairing of the two turns linked by /kɔ̂ɔ/ in (18) forms a “compound turnconstructional unit” (Lerner 1991: 441), that is, a single turn-constructional unit that is produced across the talk of two speakers. Wilkinson and Kitzinger (2011: 31–34) call this kind of phenomenon observed in conversational interaction a “collaborative completion.” What is common between the clause-initial use (14) and the predicateinitial use (18) is that Speaker B takes into consideration Speaker A’s previous turn. Hence, they are both interpersonal. But the intention of Speaker B is divergent between the two, namely, he will make a description of phenomenal

206

Kiyoko Takahashi

consequence, as in (14), or he will construct a collaborative utterance together with the dialogic partner, as in (18). 3.3.2

Conjunction types

Like the turn-initial /kɔ̂ɔ/, the turn-internal /kɔ̂ɔ/ has two subtypes: (a) predicate-initial type and (b) clause-initial type (Table 8.7). This section accounts for diverse assertive speech acts pertaining to uses of the latter clauseinitial subtype. Table 8.7 Syntactic subtypes of the turn-internal /kɔ̂ɔ/ 1. Predicate-initial type: Highlighting, for example, (1), (3), (17) 2. Clause-initial type: Conjunction, including (i) description of phenomenal consequence, for example, (19); (ii) additive description, for example, (2), (4), (20); (iii) description of logical outcome, for example, (21); (iv) adequate interpretation, for example, (22), (23)

The corpus data reveal that among utterances with the clause-initial /kɔ̂ɔ/, those of the turn-internal type (e.g., (2), (4), (19) to (23)) [11 tokens] are as frequent as those of the turn-initial type (e.g., (6) to (16)) [12 tokens]. Whereas turn-initial uses of the clause-initial /kɔ̂ɔ/ are interpersonal, turninternal uses are text-procedural. To be specific, the latter turn-internal type functions as a conjunction or conjunctive adverb (clause linker) in narrative discourse, as elaborated in what follows. The clause-linker uses of /kɔ̂ɔ/ are related to such procedural meanings as “the description of phenomenal consequence (or temporal sequence)” (19), “additive description” (20), “the description of logical outcome” (21), and “adequate interpretation” (22) and (23). (19) mɛ̂ɛ bɔ̀ɔk wâa khít sáɂ wâa yùu yîipùn lɛ́ɛw kan lûuk mother tell comp think prtcl comp stay Japan prtcl child kɔ̂ɔ ləəy phim pay bɔ̀ɔk phɯ̂an wâa . . . prtcl therefore Pim(proper.name) go tell friend comp “(My) mother said, ‘Think of staying in Japan, my dear daughter.’ So, Pim (= I) told (my) friend that [...].” [turn-internal and clause-initial, description of phenomenal consequence] In (19) (description of phenomenal consequence), the narrator uses /kɔ̂ɔ ləəy/ “so,” which consists of /kɔ̂ɔ/ and the linkage morpheme /ləəy/ “therefore,” to describe a resultant action originating from the event described previously. (20) mây mii taŋ càay kháw neg exist money pay pron man tɔ̂ŋ ɂaw lûuk kɛ̂ɛw wêetmon pay lɛ̂ɛk pron must take glass.marble magic go exchange

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 207 lɛ́ɛw kɔ̂ɔ phɔɔ câw khɔ̌ɔŋ ráan hěn . . . then prtcl as.soon.as owner shop see “(She = a character in the movie) does not have money to pay them (= the shop owner). She must take the magic marble to make an exchange (with it). In addition, as soon as the shop owner saw (it), [...].” [turninternal and clause-initial, additive description] In (20) (additive description), the narrator describes an additional story by using /lɛ́ɛw kɔ̂ɔ/ “in addition,” in which the linkage morpheme /lɛ́ɛw/ “then” is combined with /kɔ̂ɔ/. (21) mii phûu lên cèt khon sɔ̌ɔŋ fàay fàay láɂ cèt khon exist player seven clf two clf clf per seven clf khɛ̀ŋ kan tɔ̂ŋ càp kɔ̂ɔ khɯɯ síkkəə khɔ̌ɔŋ tɛ̀ɛ láɂ compete recp must catch prtcl namely Seeker of each thiim tɔ̂ŋ càp koondên sanít hây dây team must catch Golden.Snitch without.fail “There are seven players in both the two teams. Each team has seven players. (They) compete with each other. (They) must catch (the Golden Snitch). That is, each team’s Seeker must catch the Golden Snitch without fail.” [turn-internal and clause-initial, description of logical outcome] In (21) (description of logical outcome), the narrator uses /kɔ̂ɔ khɯɯ/ “that is,” which is composed of /kɔ̂ɔ/ and the linkage morpheme /khɯɯ/ “namely,” to summarize what she has said so far, in order for the interlocutor to comprehend it. (22) thîi tɔɔn rɛ̂ɛk hɛɛrîi man yùu bâan chây máy at.first Harry pron stay house prtcl kɔ̂ɔ thúk pii man càɂ tɔ̂ŋ klàp maa thîi bâan prtcl every year pron irr must return come at house “At first, regarding Harry, he stayed at home, right? (It was so because) every year he must come back home.” [turn-internal and clause-initial, adequate interpretation] (23) lɛ́ɛw thîi nîi kɔ̂ɔ bɛ̀ɛp luŋ man mây yàak hây klàp thîi then now prtcl just.like uncle pron neg want comp return at rooŋ rian school “Then, now, in a manner of speaking, regarding the uncle, he does not want (him = Harry) to go back to the school.” [turn-internal and clauseinitial, adequate interpretation] In (22) and (23) (adequate interpretation), the narrator tries to interpret the information she wants to convey to the interlocutor as adequately as possible, in order to make herself fully understood. /kɔ̂ɔ bɛ̀ɛp/ “in a manner of

208

Kiyoko Takahashi

speaking” in (23) comprises /kɔ̂ɔ/ and the linkage morpheme /bɛ̀ɛp/ “just like.” 3.4

The system of diferent uses of /kɔ̂ɔ/

Table 8.8 shows a two-dimensional classification of speech acts pertaining to uses of /kɔ̂ɔ/ attested in the corpus data. All the speech-act types listed come under the major category of “assertion.”13 Table 8.8 All the attested speech-act types pertaining to uses of /kɔ̂ɔ/ Functional domain Text-procedural uses: Narrative type or conversational Syntactic position turn-internal type

Interpersonal uses: Conversational turn-initial type

Predicate-initial type

Highlighting

Collaborative completion

Clause-initial type

Conjunction i. Description of phenomenal consequence ii. Additive description iii. Description of logical outcome iv. Adequate interpretation

Reaction i. Response i-a. Reply i-a.1. Straight reply i-a.2. Hedging reply i-b. Skew reply (explanation) ii. Association ii-a. Concurring ii-a.1. Description of phenomenal consequence ii-a.2. Description of logical outcome ii-a.3. Description of logical outcome and Hortative suggestion ii-b. Contesting ii-b.1. Contesting explanation ii-b.2. Concessive description

From Table 8.8 we can see that the functional domain of a use of /kɔ̂ɔ/ (text-procedural or interpersonal)14 is not determined solely by its syntactic position (predicate- or clause-initial). The syntactic position is indeed an indexical factor, but not the absolute factor determining the functional domain that each use of /kɔ̂ɔ/ applies to. The distinction between interpersonal uses, on the one hand, and text-procedural uses, on the other hand, depends exclusively on whether or not they are interlocutor-oriented. With the speaker’s intention to react to his/her interlocutor, the use of /kɔ̂ɔ/ will be interpersonal regardless of whether its syntactic position is predicate-initial or clause-initial. The predicate-initial /kɔ̂ɔ/ mostly occurs turn-internally and functions in the text-procedural domain as a highlighting marker. When appearing turn-initially, it functions in the interpersonal domain and executes the act

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 209 of collaborative completion. In a similar vein, when the clause-initial /kɔ̂ɔ/ occurs turn-internally, it functions in the text-procedural domain as a kind of conjunction; when it occurs turn-initially, it functions in the interpersonal domain as a reaction marker. The category of reaction speech acts encompasses a number of subcategories. I would like to clarify the diferences between the description of phenomenal consequence and that of logical outcome in the left column of Table 8.8 (the turn-internal type of the clause-initial subtype, for example, (24)=(19) and (26)=(21)) and those in the right column (the turn-initial type of the clause-initial subtype, for example, (25)=(14) and (27)=(15)). The former are discourse-oriented, text-procedural senses (viz. conjunction), while the latter are interlocutor-oriented, interpersonal senses (viz. reaction). (24) mɛ̂ɛ bɔ̀ɔk wâa khít sáɂ wâa yùu yîipùn lɛ́ɛw kan lûuk mother tell comp think prtcl comp stay Japan prtcl child kɔ̂ɔ ləəy phim pay bɔ̀ɔk phɯ ˆ an wâa . . . prtcl therefore Pim(proper.name) go tell friend comp “(My) mother said, ‘Think of staying in Japan, my dear daughter.’ So, Pim (= I) told (my) friend that.” [turn-internal and clause-initial, description of phenomenal consequence, narration] (25) A: man phə̂əm khɯ ˆ n rɯ ˆ ay rɯ ˆ ay pron increase inc continually “It (= the reported intensity-value of the earthquake in the Chiangmai district) increased continually.” B: kɔ̂ɔ ləəy ŋoŋ prtcl therefore be.puzzled “Well (that being the case), therefore (we) were puzzled.” [turninitial and clause-initial, description of phenomenal consequence, reaction] In both (24) and (25), /kɔ̂ɔ ləəy/ “therefore” is used for the sense of the description of phenomenal consequence. However, the two descriptions differ in the description of the cause event. The narrator of (24) describes the cause event by herself, whereas Speaker B of (25) makes use of Speaker A’s utterance and takes it as the description of the cause event. Put diferently, the consequence-description led by /kɔ̂ɔ ləəy/ in (24) derives from the speaker’s own cause-description before /kɔ̂ɔ ləəy/; that in (25) derives from the interlocutor’s previous cause-description. (26) mii phûu lên exist player khɛ̀ŋ kan compete recp

cèt seven tɔ̂ŋ must

khon sɔ̌ɔŋ fàay fàay clf two clf clf càp kɔ̂ɔ khɯɯ catch prtcl namely

láɂ cèt khon per seven clf síkkəə khɔ̌ɔŋ tɛ̀ɛ láɂ Seeker of each

210

Kiyoko Takahashi thiim tɔ̂ŋ càp koondên sanít hây dây team must catch Golden.Snitch without.fail “There are seven players in both the two teams. Each team has seven players. (They) compete with each other. (They) must catch (the Golden Snitch). That is, each team’s Seeker must catch the Golden Snitch without fail.” [turn-internal and clause-initial, description of logical outcome, narration]

(27) A: dây rɔ́ɔy hâa sìp tɛ̂m obtain 150 score “(If they can catch the Golden Snitch, they will) get 150 points.” B: ɂɔ̌ɔ kɔ̂ɔ khɯɯ chanáɂ ŋíi rə̌ə prtcl prtcl namely win like.this prtcl “Oh, well (that being the case), that is, (they) won like this, yes?” [turn-initial and clause-initial, description of logical outcome, reaction] Likewise, /kɔ̂ɔ khɯɯ/ “that is” in (26) and (27) is used for the sense of the description of logical outcome or, more precisely, for the sense of summing up. Nonetheless, the two instances diferently introduce the presupposition based on which the logical outcome was brought forth. While the narrator of (26) describes the presupposition by herself, Speaker B of (27) takes Speaker A’s previous utterance as the presupposition. In conversations (25) and (27), the speaker B takes the preceding utterance by the interlocutor as the presupposition for his/her utterance. These conversations are intertwined conversations, so to speak. An intertwined conversation is a dynamic mutual product of the two dialogic partners. 4

Conclusion

The findings of the corpus-based research are summarized as follows. The pragmatic particle /kɔ̂ɔ/ appears mostly predicate-initially (105 tokens, about 82% of all the tokens) and functions as a contextual operator with connotations of emphatic assertion in narrative discourse (highlighting marker). Sometimes it occurs clause-initially (23 tokens, about 18%) and helps express a variety of assertive speech acts. What is important is that in either of the two syntactic positions (predicate- or clause-initial position), it may function interpersonally, provided that the speaker uses it turn-initially to take a turn while making use of the interlocutor’s previous utterance as an interactively given presupposition. The attested number of turn-initial uses (13 tokens, about 10%) is much less than that of turn-internal uses (115 tokens, about 90%), though. This chapter concentrated on an examination of social interactive uses of /kɔ̂ɔ/. But /kɔ̂ɔ/ is multifunctional; it has other functions besides social ones. Based on the results of the corpus analysis, I consider the system of the multifunctionality of /kɔ̂ɔ/ as follows. The basic function of /kɔ̂ɔ/ is

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 211 text-procedural (description of phenomenal consequence). Functioning as a sequential indicator, it organizes the flow of discourse. This basic function has extended into the cognitive-logical domain. Cognitive-logical uses of /kɔ̂ɔ/ signify the speaker’s construal of causal relation between two propositional contents concatenated. Since /kɔ̂ɔ/ marks the consequence content, its use connotes the speaker’s resultative description or conclusive judgment. On the basis of these implicit cognitive-logical meanings, the functions of /kɔ̂ɔ/ have developed into the communicative-interactional domain. In conversational discourse, the speaker may use it as the marking of reaction or hortative suggestion to the interlocutor, considering the interlocutor’s prior utterance as the conversational presupposition for his/her posterior utterance. Despite the limited number of instances, this case study demonstrated the validity of a corpus-driven approach for understanding the nature of pragmatic particles. Abbreviations clf cont inc neg pron rea

classifier continuous inchoative negative pronoun realization

comp cop irr pos prtcl recp

complementizer copula irrealis possibility particle reciprocal

Notes 1 Acknowledgments: This chapter was derived from the ILCAA Joint Research Project titled “Semantics of discourse particles in East and Southeast Asian languages” (PI: Elin McCready), which was supported by the Research Institute for Languages and Cultures of Asia and Africa (ILCAA), the Tokyo University of Foreign Studies (TUFS), Japan. Examples in this chapter are taken from the TUFS Spoken Thai Corpus, which consists of 11 naturally occurring casual conversations by college students in Bangkok. The corpus is a part of the products of the Japan–Thai collaborative project titled “Studies in discourse cohesiveness based on corpus data,” which was supported by JSPS Grant-in-Aid for Scientific Research (A) grant number 25244017 (PI: Makoto Minegishi). An earlier version of this chapter was presented at the 154th General Meeting of the Linguistic Society of Japan, Tokyo Metropolitan University, June 24–25, 2017. I would like to thank the audience for useful and helpful comments. My special thanks are due to Hiroki Nomoto and an anonymous reviewer for critical yet constructive comments and valuable advice. I am indebted to Isra Wongsarnpigoon for stylistic suggestions. Any remaining errors are mine. 2 See Takahashi (2016) for a detailed review of previous studies on Thai pragmatic particles of the phrase-final type. 3 Before the eighteenth century, /cɯŋ/ also could occur at both syntactic positions, viz. in front of a predicate and/or a clause (Takahashi 2004: 205). 4 “Subjectification” is “the semasiological process whereby SP/Ws [speakers/writers] come over time to develop meanings for Ls [lexemes] that encode or externalize their perspectives and attitudes as constrained by the communicative world

212

5 6 7

8 9 10 11

12 13 14

Kiyoko Takahashi

of the speech event, rather than by the so-called ‘real-world’ characteristics of the event or situation referred to” (Traugott & Dasher 2002: 30). “Intersubjectification” is the process “where meanings come explicitly to index and acknowledge SP/W’s [speaker’s/writer’s] attitude toward AD/R [addressee/ reader] in the here and now of the speech event” (Traugott & Dasher 2002: 31). This view, with which I agree, difers from Tsunoda’s (2018) claim that /kɔ̂ɔ/ is a subordinator. All the three clauses in (3) do not have an overt subject noun phrase in front of the predicate or the predicate-initial /kɔ̂ɔ/. The referents of the respective subject noun phrases, however, can be readily retrieved based on the discourse context. It is not uncommon that nominal arguments of verbs are not named in Thai discourse. Example (5) is from Iwasaki and Ingkaphirom (2005: 176). For convenience’s sake, I have changed its phonetic alphabets representing colloquial pronunciation into those denoting orthographic spelling. For details see footnote 11. Interjectional particles (e.g., /ɂɔ̌ɔ/ “oh”) may occur before the turn-initial /kɔ̂ɔ/, cf. example (15). In the corpus data, there are a number of idiomatic coalescent units containing /kɔ̂ɔ/ which have a conjunctive function (e.g., /tɛ̀ɛ kɔ̂ɔ/ “even so,” /lɛ́ɛw kɔ̂ɔ/ “in addition,” /kɔ̂ɔ bɛ̀ɛp/ “in a manner of speaking,” /kɔ̂ɔ ləəy/ “so,” /kɔ̂ɔ khɯɯ/ “that is”; see Section 3.3.2). Though they have a more specific meaning than /kɔ̂ɔ/, in this study I consider those as variants of /kɔ̂ɔ/ on the grounds that they entail /kɔ̂ɔ/’s essential, procedural function (i.e., to manage sequences of talk). If (13A) is uttered sarcastically as an assertive expression (“It is impossible for you to stay in Japan”), the turn-initial /kɔ̂ɔ/ in (13B) should be regarded as the concurring type (see Section 3.2.2). Admittedly, we have to examine a much larger number of instances of /kɔ̂ɔ/ to see whether there are other speech-act types. One way or the other, all uses of /kɔ̂ɔ/ are cognitive and logical, and therefore, “cognitive-logical uses” are not particularly referred to in Table 8.8.

References Bhamoraput, Amara. 1972. Final Particles in Thai. Master’s thesis, Brown University. ‘Burusphat, Somsonge’. 2008. An etymological speculation on the sequential indicator /kɔ̂ɔ/ in Thai narrative. In Anthony V. N. Diller, Jerry Edmondson & Yong Xian Luo (eds.), The Tai-Kadai Languages, Chapter 15, 431–444. London: Routledge. Chodchoey, Supa W. 1986. Strategies in Thai Oral Discourse. Ph.D. thesis, Pennsylvania State University. Cooke, Joseph R. 1989. Thai sentence particles: Forms, meanings and formal-semantic variations. In Papers in South-East Asian Linguistics, No. 12: Thai Sentence Particles and Other Topics (Pacific Linguistics, A-80), 1–90. Canberra: Department of Linguistics, Research School of Pacific Studies, Australian National University. Iwasaki, Shoichi & Preeya Ingkaphirom. 2005. A Reference Grammar of Thai. Cambridge: Cambridge University Press. Lerner, Gene H. 1991. On the syntax of sentences-in-progress. Language in Society 20. 441–458. Noss, Richard B. 1964. Thai Reference Grammar. Washington, DC: Foreign Service Institute. Peinukachon, Vichai. 2002. Taigo ni okeru setuzokusi /kɔ̂ɔ/ ni tuite no iti kōsatu [A study of the conjunction /kɔ̂ɔ/ in Thai]. Southeast Asian Studies, Tokyo University of Foreign Studies 7. 112–126.

Interpersonal uses of the pragmatic particle /kɔ̂ɔ/ in Thai 213 Peyasantiwong, Ratcharin. 1981. A Study of Final Particles in Conversational Thai. Ph.D. thesis, University of Michigan. Pittayaporn, Pittayawat & Pirachula Chulanon. 2012. Syntactically naughty?: Prosody of final particles in Thai. In Tadao Miyamoto, Naoyuki Ono, Kingkarn Thepkanjana & Satoshi Uehara (eds.), Typological Studies on Languages in Thailand and Japan, Chapter 1, 13–28. Tokyo: Hituzi Syobo. Sa-anwong, Wiyada. 1981. Kaan sadɛɛŋ khwaam maˇay khɔ̌ɔŋ kham kɔ̂ɔ nay phaasaˇa thay [The Use of the Expression /kɔ̂ɔ/ in the Thai Language]. M.A. thesis, Silpakorn University. Selkirk, Elizabeth. 1984. Phonology and Syntax: The Relation between Sound and Structure. Cambridge, MA: MIT Press. Singhabhandhu, Panomporn. 1983. Wíkhrɔ́ɂ kaan cháy kham kɔ̂ɔ nay phaasaˇa thay [An Analysis of the Word /kɔ̂ɔ/ in Thai]. M.A. thesis, Chulalongkorn University. Takahashi, Kiyoko. 2004. Logical resultative construction in Thai. Journal of Kanda University of International Studies 16. 203–224. Takahashi, Kiyoko. 2016. Taigo no goyōronteki shōji [Pragmatic particles in Thai]. Journal of Kanda University of International Studies 28. 289–309. Traugott, Elizabeth C. & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Tsunoda, Tasaku. 2018. Five levels in Thai. In Tasaku Tsunoda (ed.), Levels in Clause Linkage: A Crosslinguistic Survey, 615–675. Berlin & Boston: De Gruyter Mouton. Wilkinson, Sue & Celia Kitzinger. 2011. Conversation analysis. In Ken Hyland & Brian Paltridge (eds.), The Continuum Companion to Discourse Analysis, Chapter 2, 22–37. London & New York: Continuum International. Corpus data The TUFS Corpus of Spoken Thai, which belongs to the TUFS Multilingual Oral Discourse Corpus compiled by the Language Institute of the Tokyo University of Foreign Studies (TUFS).

9

A scalar semantics for the Vietnamese sentence-final particle cơ Anne Nguyen

1

Introduction1

This chapter considers the distribution and the semantics of the Vietnamese sentence-final particle cơ. I discuss its presence in various constructions and contexts and generalize the semantic contribution of the particle. I claim that cơ is a scalar particle, reflecting a conflict between speaker belief and expectations which arise from the preceding conversation. In addition, the proposition marked by the particle must be stronger than its contrasting alternatives on a given scale. Many in the use and distribution of cơ are not captured in the previous work. Nguyen (1997) only describes its use in a specific type of context, “expressing preference.” Le (2015) says that cơ has an “emphasis function,” without a precise definition. Nguyen (2017) proposes the first formal semantics for cơ, which can account for its corrective and skeptical uses. However, that account disregards many other uses of the particle and overlooks the scalar aspect of its meaning. This chapter studies cơ in a wider range of contexts and proposes an analysis that can capture the scalarity in the meaning of the particle. The chapter is structured as follows. I will first give an overview of cơ’s functions by looking at the basic facts about the particle. I will discuss its use in both declaratives and polar questions as well as contextual constraints on its use. In Section 3, I will present a formal analysis which can account for the discussed behaviors of the particle. The analysis in this section makes use of scales, focus, contrastive topics, and expectations. Section 4 will focus on the relationship of cơ to English even, because cơ, as shown later, is also a focus particle like even. Finally, the conclusion section summarizes the key points. 2

Basic facts about cơ

In this section, I will discuss the use of cơ in various contexts. Descriptive functions of the particle are labeled according to the contexts where it appears. The final goal is to isolate the features common to all cases in order to generalize the meaning of the particle. We will see through the discussion that the meaning of cơ is defined by its use conditions rather DOI: 10.4324/9781351057752-10

Scalar semantics for the Vietnamese sentence-final particle cơ

215

than truth conditions. Attention is restricted only to the association of cơ with declaratives and polar questions.2 2.1

Declaratives

Cơ is often observed in declaratives. Specific contexts where a declarative with cơ can be used are varied. (1) shows its use in two types of correction discourse. In context (1), B is trying to correct the prior claim of A. In context (2), what is corrected is an implication from A’s biased question. (1)

B: Không phải. Cậu ấy mua táo cơ. not true sg-3 buy apple cơ “Not true. He bought apples.” (Nguyen 2017) a. √ Context 1: A says Minh bought apples. B then says (1) to A b. √ Context 2: A asks B if it is right that Sam bought apples. B then says (1) to A.

In this kind of situations, cơ is optionally used. However, when present, it gives the impression that the speaker is confident about the information they are imparting. Nguyen (1997) points out that a declarative with cơ is also seen in contexts in which the speaker expresses their preference. The presence of cơ signals the speaker’s insistence on their choice. (2)

(Nó không thích cái sơ-mi đỏ.) Nó thích cái sơ-mi màu xanh kia3/cơ. sg-3 like cl shirt color blue kia/cơ “(He does not like the red shirt.) He like the blue shirt.” (Nguyen 1997)

This kind of insistence is also observed when a declarative with cơ is used to argue for the speaker’s choice when they are asked to do what they do not want. (3)

A: Con {nên/ phải} học NTU sg-2 should must study ntu “You should/must go to NTU.” B: Không. Con sẽ học NUS cơ. no sg-1 fut study nus cơ “No. I will go to NUS.”

A common feature in those contexts is that the speaker always expresses some sort of disagreement over some contrasting information mentioned in the preceding conversation. In the examples we have seen, the contrasting information is linguistically realized. It is also possible for it to arise non-linguistically in the course of the conversation.

216 (4)

Anne Nguyen A group of kids gets lost in a jungle and sees mysterious people. The kids are scared, so they run away. They find a shelter and stay there the whole night. When they get up in the morning, they see strange symbols and a knife sticking out of a tree. They think these things have just appeared and certainly left by the mysterious people. The kids are scared and worried that those people are following them. To calm them down, one of the kids says: Tớ nhìn thấy chúng từ hôm qua cơ. sg-1 see pl-3 since yesterday cơ “I already saw them yesterday.” (Nguyen 2012)

With the cơ sentence, the kid who speaks expresses her disagreement over the idea that those scary things have just appeared, thus suggesting that they are not from the mysterious people. That idea is not spoken out loud by anyone present there. It is simply that kid’s inference from the context. The discussion earlier points out that for the use of cơ, there exists some contrasting information about which the speaker disagrees with the addressee. Now, we will see that the lack of such information will makes cơ infelicitous. In particular, a declarative with cơ is bad when it is uttered out of the blue, at the beginning of a conversation, or as a direct answer to a wh-question, as in (5). (5)

Q: How many books did Minh buy? A: Nó mua năm quyển (#cơ). sg-3 buy five cl cơ “He bought five books.”

In such context, no contrasting information is found in the preceding conversation. The existence of some contrasting information is thus necessary to license the use of cơ. So far, we have seen that the speaker always explicitly expresses some disagreement when using cơ. However, it can be implicit in the below situation in which the speaker is making a comparison. (6)

A and B are playing a card game. The rule is that each player gets three cards and the person who has the highest card is the winner. A says that A has a queen. B then says: Tớ có cây K cơ. sg-1 have card king cơ “I have a king.”

Here, A and B are competing to be the one who has the highest card. By making the claim of having a queen, A implies that A has the highest card. Then, with the cơ sentence, B not only asserts that B has a king but also suggests B has a higher card, arguing against A’s implication of having the highest one. That is how cơ is licensed here. B’s disagreement is not explicitly or directly expressed. The scenario in (7) is slightly diferent, with no direct comparison being made.

Scalar semantics for the Vietnamese sentence-final particle cơ (7)

217

Context: In a secondhand watch shop, a customer C has been asking for the prices of many watches but shows no intention of buying any. The seller S is getting irritated. C: How much is this blue watch? S: It’s $80. C: Too expensive. How about this yellow one? S: Cái đồng hồ đấy giá $100 cơ. cl watch that cost $100 cơ “That watch costs $100.”

When the customer asks for the price of the yellow watch, it is not necessary that they want to compare the blue and the yellow. However, for the seller, the use of cơ indicates that they are making an implicit comparison. Given the customer’s comment in bold, it is possible for the seller to assume that the customer expects the yellow watch is cheaper than the blue one. The sentence with cơ then informs the customer the price of the yellow and rejects the customer’s expectation. The main diference between the two scenarios in (6) and (7) is that cơ is not always licensed in those like (7). The comparison purpose in (6) is shared by both participants, where in (7) it may not be assumed by both participants. Note that the comparison is the basis for the speaker to make an inference about the addressee’s expectation, which forms the contrasting information that licenses cơ. When the addressee has no intention of making any comparison, the speaker cannot make an inference about the addressee’s expectation; the use of cơ is inappropriate. In (8), the customer’s comment on the price of the blue watch together with the background information about the discourse may not suggest that the customer expects the yellow watch costs less than $80. It thus may be inappropriate for the seller to use cơ to imply that the actual price is higher than the customer expects. (8)

In a secondhand watch shop, a customer C is talking to the seller S. C: How much is this blue watch? S: It’s $80. C: Too expensive. How about this yellow one? S: Cái đồng hồ đấy giá $100 (?cơ). cl watch that cost $100 cơ “That watch costs $100.”

Let’s consider another important feature of cơ: the proposition marked by cơ is always ordered higher than its contrasting alternative(s) on a scale. Specifically, in (6) earlier we see that cơ goes with B’s claim about his card, which is ranked higher than A’s card on the scale of card values. Similarly, in (7), cơ marks the claim about the more expensive watch. In contrast, (9) and (10) in the following show that cơ is infelicitous if the sentence modified by cơ is ranked lower. (9)

A: Tớ có cây K. sg-1 have card king “I have a king.”

218

Anne Nguyen B: Tớ có cây Q (#cơ). sg-1 have card queen cơ “I have a queen.”

(10) In a secondhand watch shop, a customer C has been asking for the prices of many watches but shows no intention of buying any. The seller S is getting irritated. C: How much is this blue watch? S: It’s $80. C: Too expensive. How about this yellow one? S: Cái đồng hồ đấy giá $60 (#cơ). cl watch that cost $60 cơ “That watch costs $100.” This requirement is not restricted only to the comparison use. When the speaker expresses their preference, as in (11), or corrects the addressee’s assumption, as in (12), the cơ sentence is also required to be ranked higher than its contrasting information (in bold) on a scale. (11) A asks B if it is right that B wants to buy five books. B then says: Không. Em muốn mua {# ba / √ tám} quyển cơ. no sg-1 want buy three eight cl cơ “No. I want to buy five books.” (12) A says that Kim got the second prize. B then says: Không phải. Nó đạt giải {# ba/ √ nhất} cơ. not true sg-3 get prize third first cơ “Not true. She got the first prize.” Note that for many examples, we can immediately identify a relevant scale. Nevertheless, it is not always straightforward, such as those in (1)–(4). I will further discuss scales in the next section. The main conclusion in this subsection is that cơ is licensed in a declarative if there exists some contrasting linguistic or non-linguistic information in the preceding conversation and the cơ sentence is ranked higher than the propositions that represent the contrasting information on a given scale. 2.2

Polar questions

The use of cơ in polar questions is also very common. In terms of its contribution, cơ is claimed to always reflect the speaker’s surprise. (13) Minh is only an average student in the class. He took a contest a month ago. When the result is released, A says to B that Minh got the third prize. B then says:

Scalar semantics for the Vietnamese sentence-final particle cơ Nó được giải ba sg-3 get prize third “He really got the third prize?”

cơ cơ

219

à? à

(13) In (13), the background information in bold suggests that B may hold a lower expectation, that Minh might not get a prize. Thus, B is surprised when hearing the news from A. In (14), the background information is modified, making cơ infelicitous. (14) Minh is a newcomer, so no one in the class knows whether he is a good student. He took a contest a month ago. The result has been released. Many students in the class got the third or second prizes. A says to B that Minh got the third prize. B then says: cơ à? # Nó được giải ba sg-3 get prize third cơ à “He really got the third prize?” In this context, B cannot hold any prior belief regarding Minh’s achievement. That B does not know much about Minh makes B unable to predict the prize Minh might get. Any possibility could occur, including that Minh gets the third prize. B cannot be surprised at the rank of the prize that Minh got either, because many students in the class got that prize. The contrast between (13) and (14) indicates that the existence of the speaker’s prior expectation/belief is required for the use of cơ in polar questions. Note that what matters for the felicity of cơ here is the speaker’s prior belief rather than the beliefs of other participants. The speaker’s belief can be the same or different from others’. Consider (15). (15) Minh cao 1.7m cơ à? Minh tall 1.7m cơ q “Minh is really 1.7m tall?” a. √ Context 1: Minh was born in a family in which all members are very short. So far, no one is taller than 1.5 m. B and C know that, but they have never met Minh. A is Minh’s friend. A tells B and C that Minh is 1.7 m tall. B then says (15). b. √ Context 2: Minh was born in a family in which all members are very short. So far, no one is taller than 1.5 m. B and C know that. B has never met Minh. C met him two years ago. At that time, Minh looked like a normal 15-year-old boy who was about 1.6 m. A is Minh’s friend. A tells B and C that Minh is 1.7 m tall. B then says (15). c. # Context 3: Minh was born in a family in which all members are very short. So far, no one is taller than 1.5 m. B and C know that. C has never met Minh. B met him two years ago. At that time,

220

Anne Nguyen Minh looked like a normal 15-year-old boy who was about 1.6 m. A is Minh’s friend. A tells B and C that Minh is 1.7 m tall. B then says (15).

In context 1, both B and C assume that Minh should be very short, given the information about the heights of people in his family, and the use of cơ is allowed. In context 2, B, but not C, holds this belief, and B’s use of cơ is still felicitous. However, in context 3, in which B does not think that Minh should be very short, cơ becomes odd. This contrast suggests that as long as it is the prior belief of the person who utters the cơ question, cơ can be licensed. In other words, cơ in polar question only reflects the speaker’s surprise. So far, for all the examples of cơ in polar questions, the polar questions echo the prior claim of the addressee. But this is not necessary. (16) A: Minh has bought twenty packs of chocolate. B: Nó thích sô-cô-la thế cơ à? sg-3 like chocolate that-much cơ q “He really like chocolate that much?” (17) A is showing B his new watch. B thinks it is nice and is curious about the price. Also, B knows that watches normally cost $100–$200. A then shows five fingers. Cái đồng hồ giá $500 cơ à? cl watch cost $500 cơ q “The watch really costs $500?” In (16), the propositional content of the question is inferred from A’s claim. In (17), the addressee even did not say anything. The polar question is the result of B’s inference from A’s act of showing five fingers. We thus claim that for the speaker to ask a cơ polar question, they need to obtain some (linguistic or non-linguistic) evidence for the propositional content of the question. In the absence of such evidence, the speaker will not use cơ, as shown in (18). (18) B has no idea how much a watch is. A is showing B his new watch. B thinks it is nice and is curious about the price. A then shows five fingers. Cái đồng hồ giá $500 (#cơ) à? cl watch cost $500 cơ q “The watch really costs $500?” The reason for the infelicity of cơ in (18) is clear. It is due to the lack of the speaker’s prior belief about the price of the watch. Without cơ, the question is simply to confirm the price inferred from A’s action. We have seen that cơ in polar questions can be licensed if there exists evidence for the propositional content of the question and the speaker holds a prior contrasting belief. But this is not enough. Recall that cơ in declaratives requires

Scalar semantics for the Vietnamese sentence-final particle cơ

221

the marked sentence to be ranked higher than its contrasting alternative(s) on a given scale. This requirement is also imposed on cơ in polar questions. (19) A tells B that Minh bought five books. B then says: Năm quyển cơ à? Tớ tưởng nó mua  {√ ba/ # tám} quyển. five cl cơ q sg-1 think sg-3 buy three eight cl “five books? I thought he bought {three/ eight}.” (20) A tells B that Minh got the second prize. B then says: Giải nhì cơ à? Tớ tưởng nó đạt giải {√ ba/ # nhất}. prize second cơ q sg-1 think sg-3 get prize third first “The second prize? I thought he got the {third/ first} prize.” The contrasts in (19) and (20) clearly show that the propositional content of the polar question containing cơ must be ranked higher than the propositions that represent the speaker’s contrasting prior belief. 2.3

Summary

The discussion about the use of cơ in a wide range of contexts so far can be summarized as follows: • In declaratives, cơ reflects the existence of a prior contrasting information from the preceding conversation, and the declarative with cơ is ranked higher than the propositions that represent the contrasting information on a given scale. • In polar questions, cơ requires the existence of evidence for the propositional content of the question and the existence of the speaker’s prior contrasting belief. The propositional content of the question is stronger than the propositions that represent the speaker’s contrasting belief on a given scale. If we refer to the propositional content of the sentence containing cơ as its prejacent and the other contrasting proposition(s) to which the prejacent is compared as its alternative(s), the two earlier generalizations look very similar. Both require the existence of a set of contrasting alternatives to the prejacent and that the prejacent is stronger than them on a given scale. This is the core meaning of cơ. In the following section, I will present a formal account for the use of cơ in both declaratives and polar questions. 3

A formal semantics for cơ

This section concentrates on formalizing the meaning of cơ, which has been generalized earlier. In the following subsections, I will present a formal

222

Anne Nguyen

semantics for cơ and discuss in detail many relevant issues, including constraints on the form of the contrasting alternatives, scales, and expectations. 3.1

Proposal

I propose that cơ reflects the conflict between speaker belief and conversational expectation. The prejacent p (i) is supported by one and (ii) is stronger than alternatives supported by the other, according to a given scale. The proposal is as follows: • Let BSpkr be the set of propositions compatible with speaker belief. • Let ExpC (expectations from the conversation) be the set of propositions whose truth is supported by public (linguistic and non-linguistic) moves made in the preceding conversation. (21) In declaratives: [p cơ] is felicitous if: (i) BSpkr ⊨ p; and (ii) On a given scale σ, ∀q ∊ Alt ⋂ ExpC, p >σ q (22) In questions: [p cơ Q] is felicitous if: (i) ExpC ⊨ p; and (ii) On a given scale σ, ∀q ∊ Alt ⋂ BSpkr, p > σ q Alt = ⟦ p ⟧F, CT, a set of alternative propositions obtained by replacing all the F-marked (Rooth 1985, 1992) and CT-marked (Buring 2003) constituents in p with contextually determined alternatives. The formulas in (21)–(22) both require the prejacent to be stronger than its alternative(s) on a given scale. In addition, where cơ associates with a declarative, the prejacent is compatible with speaker belief, whereas its alternatives, whose form is constrained by focus (and contrastive topic) information, is compatible with expectations arising in the preceding conversation. This is reversed in polar questions, with the prejacent being compatible with conversational expectations and its alternative(s) with speaker belief. 3.2

Focus and contrastive topic

Both cơ in declaratives and polar questions require the existence of contrasting alternative(s). As discussed in Section 2, for cơ in declaratives, the contrasting alternative(s) are explicitly realized by a claim in the preceding conversation or implicitly inferred by the speaker from the current situation. When it is in polar questions, the contrasting alternatives are always related to the speaker’s beliefs. Such beliefs may be explicitly mentioned. But in many cases, the speakers do not

Scalar semantics for the Vietnamese sentence-final particle cơ

223

need to say anything. As the presence of the contrasting alternatives can be overt or covert, it is not always easy to identify their form. Therefore, it is necessary to point out constraints on the form of these propositions. Collected data shows that their form depends on the position of focus, and contrastive topic (CT) if present.4 Let’s first consider examples with focus, which are taken from Nguyen (2017). (23) Minh mua con mèo [này]F ở cửa hàng này cơ. Minh buy cl cat this at store this cơ “Minh bought this cat at this store.” ↝ “Minh is thought to buy some other cat at this store.” (24) Minh mua con mèo này ở cửa hàng [này]F cơ. Minh buy cl cat this at store this cơ “Minh bought this cat at this store.” ↝ “Minh is thought to buy the cat at some other store.” The preceding sentences assert the same proposition that Minh bought a certain cat at a certain store. However, we get diferent implications due to their diference in the placement of focus. This contrast shows that cơ is focussensitive. The focus structure can help specify the form of the contrasting alternatives which licenses cơ. In (23)–(24), the alternatives are respectively of the forms Minh bought cat x at this store and Minh bought this cat at store y, in which x and y refer to some other cats and some other stores. Note that cơ requires its prejacent to be stronger than its contrasting alternatives on a given scale. We may wonder in which sense this requirement is satisfied in (23)–(24). This will be discussed further in the subsection about scales. For the current purpose of illustration, let’s consider examples of another scale. (25) Cậu ấy đạt [ba]F giải nhì cơ. sg-3 get three prize second cơ “He got three second prizes.” ↝ “He got more second prizes than the addressee thought.” (26) Cậu ấy đạt ba giải [nhì]F cơ. sg-3 got three prize second cơ “He got three second prizes.” ↝ “He got higher prizes than the addressee thought.” (27) Cậu ấy đạt [ba]F giải nhì cơ à? sg-3 get three prize second cơ q “He really got three second prizes?” ↝ “He got more second prizes than the speaker thought.”

224

Anne Nguyen

(28) Cậu ấy đạt ba giải [nhì]F cơ à? sg-3 got three prize second cơ q “He really got three second prizes?” ↝ “He got higher prizes than the speaker thought.” In (25) and (27), the number of the prize is focused. The contrasting alternatives thus refer to the number less than three, thus of the form he got x second prizes, with x < 3. In contrast, in (26) and (28), the placement of focus is shifted to the rank of the prize. The contrasting alternatives, accordingly, are restricted to the prizes lower than the second. They thus have the form he got three y prizes, with y < the second prize. The data here shows that the scalar component of cơ is sensitive to focus information. When a contrastive topic is present, it also contributes to the identification of the contrasting alternative(s). The scalar meaning of cơ, however, is not affected.5 Consider (29): (29) The result of a contest has been released. A tells B that the average student got two second prizes. B then says: [Bạn học sinh khá]CT đat [ba]F giải nhì cơ. cl student good get three prize second cơ “The good student even got three second prizes.” If B’s sentence is uttered out of context, we will get the implication that some student other than the good student got less second prizes than the good student did. This suggests that the use of cơ in (29) can be licensed by A’s claim. If only focus is needed for the form of the contrasting alternative, the expected form would be that the good student got x second prize, with x < 3. A’s claim obviously does not have this form but still can license cơ. This suggests that when present, CT information also helps specify the form of the contrasting alternatives. However, CT information does not regulate the scalar component of cơ. In (30) in the following, the alternative proposition refers to the individual, the excellent student, which is better than the individual referred to in the prejacent, but cơ is still felicitous. (30) The result of a contest has been released. A says to B that the excellent student got two second prizes. B then says: [Bạn học sinh khá]CT đat [ba]F giải nhì cơ. cl student good get three prize second cơ “The good student even got three second prizes.” The main observation in this subsection is that cơ is focus- (and CT-) sensitive.6 Due to this, the contrasting alternatives are required to belong to ALT, as stated earlier in the proposal, which is defined as a set of alternative propositions which vary in the position of focus (and CT).7

Scalar semantics for the Vietnamese sentence-final particle cơ 3.3

225

Scales

A part in the core meaning of cơ pertains to the ordering of the prejacent with respect to its alternative propositions. This subsection is concerned with the classification of scales which cơ can refer to. The classification is determined by the nature of the scalar relation. The first type of scale is based on entailment and thus is labeled a logical scale. We have seen many examples of this type in Section 2. Let’s consider them in more detail. (31) A tells B that Minh bought five books. B then says: (Không phải.) Nó mua tám quyển cơ. not true sg-3 buy eight book cơ “Not true. He bought eight books.” (32) A tells B that Minh bought eight books. B then says: Tám quyển cơ à? Tớ tưởng nó mua năm quyển. eight cl cơ q sg-1 think sg-3 buy five cl “eight books? I thought he bought five.” In (31)–(32), the entailment relation between the prejacent Minh bought eight books and the contrasting alternative Minh bought five books is clear. The situation becomes more complicated when CT is involved. (33) A tells B that Kim bought five books. B then says: [Minh]CT mua [tám]F quyển sách cơ. Minh buy eight cl book cơ “Minh bought eight books.” The prejacent Minh bought eight books here does not entail the antecedent Kim bought five books. However, as discussed earlier, the scalar component is solely determined by the position of focus. In this narrow sense, we can maintain that the comparison between the prejacent and its antecedent depends on an entailment relation regulated by the focus information. Note that the contrasting alternative(s) in the case of logical scales cannot be false, as it is entailed by the prejacent. The negative response không phải “not true” in (31) does not target the contrasting proposition Minh bought five books but the conversational implicature that Minh did not buy more than five books, that is, the exhaustified interpretation of A’s assertion. In (33), which is involved with CT, it is impossible to say không phải right before the cơ sentence. Assume that the implication that Minh did not buy more than five books may be implied from A’s assertion as well. This implication is not the strengthened proposition of A’s assertion. That might be the reason that không phải is not allowed here. The unacceptability of a

226

Anne Nguyen

negative response, however, is independent from the use of cơ, as removing cơ does not make any changes. The second type of scale, which I refer to as a categorical scale, deals with non-logical relations. Consider (34)–(35). (34) A asks B if it is right that Minh has a jack. B then says: (Không phải.) Nó có [cây Q ]F cơ. not true sg-3 have card queen cơ “Not true. He has a queen.” (35) A says to B that Minh has a queen. B then says: Nó có cây Q cơ à? Tớ tưởng nó có cây J. sg-3 have card queen cơ Q sg-1 think sg-3 have card jack. “He really has a queen? I thought he had a jack.” There is no logical relationship between the prejacent Minh has a queen and the contrasting alternative Minh has a jack. Instead, their ordering is determined by a convention about card values. A queen is conventionally regarded as better than a jack, making the proposition that Minh has a queen stronger than the one Minh has a jack. As there is no entailment relation between the prejacent and its contrasting alternatives, the alternatives can be false. This is how categorical scales are different from logical scales. The contrasting alternative in (34)–(35) Minh has a jack can be true or false. The two sentences can be uttered in the scenario in which Minh has both a queen and a jack as well as in the context in which Minh only has a queen. Moreover, the negative response in (34) is ambiguous. The expression may target A’s strengthened proposition, negating the proposition that the best card Minh has is a jack. It can also negate A’s assertion, saying that Minh has a jack is false. When CT is present, a negative response is not allowed. (36) A says that Kim has a jack. B then says: (#Không phải.) Minh có [cây Q ]F cơ. not true Minh have card queen cơ “Not true. Minh has a queen.” For both logical and categorical scales, exchanging the prejacent and the contrasting alternatives, which makes the prejacent weaker than its contrasting alternatives, results in the infelicity of cơ. See the contrast between (31) and (35) and (37) and (40) in the following. (37) A tells B that Minh bought eight books. B then says: (Không phải.) Nó mua năm quyển (#cơ). not true sg-3 buy five book cơ “Not true. He bought five books.”

Scalar semantics for the Vietnamese sentence-final particle cơ

227

(38) A tells B that Minh bought five books. B then says: Năm quyển (#cơ) à? Tớ tưởng nó mua tám quyển. five cl cơ q sg-1 think sg-3 buy eight cl “five books? I thought he bought eight.” (39) A asks B if it is right that Minh has a queen. B then says: (Không phải.) Nó có [cây J]F (#cơ). not true sg-3 have card jack cơ “Not true. He has a jack.” (40) A tells B that Minh has a jack. B then says: Nó có cây J (#cơ) à? Tớ tưởng nó có cây Q. sg-3 have card jack cơ q sg-1 think sg-3 have card queen “He really has a queen? I thought he had a jack.” There is a set of data which does not seem to follow the requirement that the prejacent is stronger than its contrasting alternatives. They include examples (1)–(8) and (23)–(24) discussed earlier. For these examples, switching the prejacent and its contrasting alternatives is acceptable. Consider (1), repeated here as (41), and its reversed version in (42). (41) A says Minh bought oranges. B then says: (42) A says Minh bought apples. B then says: Không phải. Nó mua táo cơ. Không phải. Nó mua cam cơ not true sg-3 buy apple cơ not true sg-3 buy orange cơ “Not true. He bought apples.” “Not true. He bought oranges.” I argue that this is not an exception to the generalization that the prejacent must be stronger than its alternative. There is a diference between the ordering of the two propositions Minh bought oranges and Minh bought apples in (41) and (42). In (41), B is confident that Minh bought apples. In (42), the confidence is placed on Minh buying oranges. I suggest that B’s confidence is supported by B’s evidence for their claim. Furthermore, if B has evidence, then A can have evidence as well. So each claim is supported by some evidence. Then, the reason the prejacent – B’s claim – is stronger than its alternative – A’s claim – is that B thinks his claim is supported by morereliable evidence. As long as the speaker sees their evidence as superior, the use of cơ is felicitous. This provides an explanation for the exchange of the prejacent and its alternative in (41)–(42). As the relevant scale here operates at an abstract level, concerning the evidence which each participant has, I call it a meta-evidential scale. On the other hand, the exchange is not allowed in the case of logical and categorical scales because the ordering between the prejacent and its alternatives is inherent and based on their content. It is worth mentioning that conversations related to meta-evidential scale is always about the truth of claims, specifically, with the question of which

228

Anne Nguyen

claim is true. If the speaker thinks his claim is true because it is supported by more reliable evidence, he also thinks the other’s claim supported by lessreliable evidence is false. In other words, when the ordering is based on a meta-evidential scale, the contrasting alternative must be false. However, we must be very careful in identifying the relevant set of alternatives. Consider the following example. (43) A tells B that Minh bought oranges. B then says: Nó mua cả táo cơ. sg-3 buy also/even apple cơ “He also bought apples.” I want to focus on the ordering of the prejacent and its alternative, putting aside the detailed computation of the propositions. The ordering here is over propositions regarding pluralities of objects, which, to some extent, can be put under the category of logical scales. On this scale, the prejacent that Minh also bought apples is explicitly marked as stronger than its alternative, Minh bought oranges, by the additive particle cả (also/even). Therefore, cơ is felicitous here. We may argue that B thinks his claim is supported by more reliable evidence than A’s claim. Accordingly, the use of cơ makes reference to a meta-evidential scale. The contrasting alternative – A’s claim – is not false. Moreover, it must be true to license the use of cả (also/even). This is the opposite of what we have stated about the falsehood of the contrasting alternatives in the case of meta-evidential scales. A possible explanation is that the ordering in (43) is actually not based on a meta-evidential scale but the scale of pluralities, even though a meta-evidential scale is available. In other words, a logical scale comparing diferent (pluralities of) objects is chosen in (43). In many examples collected, when a logical or categorical scale and a meta-evidential scale are all available, it is usually the logical or categorical scale which will be selected. But this is not always the case. There are certain contexts where a meta-evidential scale can win in the competition. (44) A and B are arguing with each other. No one accepts the other’s opinion. A: We will buy eight books. A: No. Eight books. B: Too many. We will only buy three. B: Không. Ba quyển cơ. no three cl cơ “No. Three books.” Even though numerals are involved in (44), this cannot be the logical scale use, because on this scale, the prejacent we will buy three books is not stronger than its alternative, we will buy eight books. The felicity of cơ here suggests the presence of another scale – a meta-evidential scale. As (44) is a heated argument, the participants care more about which option is accepted or which

Scalar semantics for the Vietnamese sentence-final particle cơ

229

proposition is true. In this context, the efect of a logical scale is ignored, allowing a meta-evidential scale to determine the ordering of the prejacent and its alternative(s). 3.4

Speaker belief and conversational expectation

Now, we will discuss the notions of conversational expectation and speaker belief in more detail. Let’s start with cơ in declaratives by reconsidering example (7), repeated here as (45). (45) In a secondhand watch shop, a customer C has been asking for the prices of many watches but shows no intention of buying any. The seller S is getting irritated. C: How much is this blue watch? C: Too expensive. How about this yellow one? S: It’s $80. S: Cái đồng hồ đấy giá $100 cơ. cl watch that cost $100 cơ “That watch costs $100.” The prejacent is that the yellow watch costs $100. This proposition is compatible with speaker belief, as in this situation, the seller knows the prices of the watches in his store. This belief of the speaker is not shared with the addressee before the cơ sentence is uttered. Regarding conversational expectation, as discussed in the previous section, the seller infers that the customer expects the yellow watch to be less than $80. This inference is based on the customer’s comment about the price of the blue watch and the fact that he is asking for the price of the yellow watch. In other words, the current situation gives rise to a conversational expectation that the yellow watch is less than $80, which denotes a set of propositions of the form the yellow watch costs x with x < $80. The conversational expectation therefore defines the set of contrasting alternatives that are required to license the use of cơ in (45). For cơ in polar questions, let’s reconsider (18), repeated here as (46). (46) A shows B his new watch. B thinks it is nice, and is curious about the price. Also, B knows that watches normally cost $100–$200. A then shows five fingers. Cái đồng hồ giá $500 cơ à? cl watch cost $500 cơ q “The watch really costs $500?” The most salient information in the current situation which gives rise to a conversational expectation is A’s act of showing five fingers. B relies on this and

230

Anne Nguyen

guesses that the watch is $50, $50,000, or any number starting with 5. B additionally considers his knowledge about standard prices for a watch to narrow down the range of the predicted price to the price of $500. The conversation expectation in this case is represented by the proposition the watch costs $500, which is also the propositional content of the cơ question. Regarding prior belief, B’s prior belief about the price of the watch is based on his knowledge about watch prices, represented by a set of propositions of the form the watch costs x, with x ∈ ($100, $200). This is the set of contrasting alternatives that license cơ in (46). The discussion about speaker belief and conversational expectation in two typical examples of cơ indicates an important point about the diference between the two notions. Particularly, prior to the moment a cơ sentence is uttered, conversational expectation is based on public information, which is available to all participants, whereas speaker belief is private. There are many kinds of public information that can help form a conversational expectation. Public information can be linguistically realized, such as the addressee’s claim in (1), the addressee’s question in (2), the addressee’s comment in (7), or common knowledge in (17). Conversational expectations can also be rooted from a belief of the speaker when it becomes public. (47) A and B think a watch looks nice. A: I’m guessing it’s $300. I’ll go find out. [A goes and asks the seller how much the watch is and comes back.] A: Nó giá $400 cơ. sg-3 cost $400 cơ “It is $400.” The conversational expectation in this situation is that the watch in question costs $300. Note that no other discourse participants’ beliefs are relevant for licensing the use of cơ in (47). The use of cơ here reflects a contrast between prior expectations based on A’s earlier public belief and A’s new information. B’s beliefs or the seller’s attitudes do not regulate the use of cơ. The fact that speaker belief is private whereas conversational expectation is rooted from public information can account for the descriptive generalizations at the end of Section 2, which summarizes the diferences between cơ in declaratives and in polar questions. Specifically, a cơ declarative requires the existence of a conversational expectation in the preceding discourse. A cơ polar question is licensed if there exists information in the preceding conversation that supports the propositional content of the question and the speaker’s belief. Let’s take a closer look at these diferences. First, cơ in declaratives does not seem to have any requirement regarding the speaker’s belief. I claim that it is only an illusion. The speaker of a declarative always believes its propositional content. Therefore, that requirement of cơ in declaratives is always satisfied and is often ignored. For cơ in polar questions, the speaker’s belief can

Scalar semantics for the Vietnamese sentence-final particle cơ

231

be explicitly mentioned, as in (19)–(20), through a statement following the questions. The speaker’s belief can also be accommodated because it is private. Whichever case it is, for polar questions, the existence of such a belief is not taken for granted, and thus the requirement needs to be stated. Another point in the descriptive generalizations of cơ is that there seems no requirement of cơ in polar questions regarding conversational expectation. I argue that there is such a requirement. It is the requirement regarding the evidence for the propositional content of the question. For cơ in polar questions, the proposition that represents conversational expectation and the propositional content of the question are the same, and they are supported by the (public) evidence in the prior context. The formal characterizations in (21)–(22) successfully eliminate all those confusions and show that both cơ in declaratives and in polar questions have requirements regarding conversational expectation and speaker belief. 4

Cơ and even

Focus particles commonly include exclusives (only), additives (also), and scalar additives (even).8 The sentence-final particle cơ, to a certain extent, has been shown to be focus-sensitive and has a scalar semantics. It is then interesting to see how cơ is diferent from other focus particles. This section will focus on the comparison of cơ with English even. First, let’s talk about the types of scales to which each particle can make reference to. In the semantics of even, the ordering between the prejacent and its alternatives is proposed to be based on, for example, likelihood (Horn 1969; Karttunen & Peters 1979; but see also Greenberg 2016 and references there). This classification depends on the semantic domain of scales. According to the classification in this chapter which is pertaining to the nature of the ordering, whether it is logical or non-logical, even makes reference to both logical and categorical scales, just like cơ. This can account for the observation that, in many cases where English even is felicitous, cơ is also allowed: (48) A says to B that Kim can jump five feet. B then says: a. [Minh]CT can even jump [six]F feet. b. Minh có thể nhảy được 1.8m cơ. Minh can jump able 1.8m cơ “Minh can even jump 1.8 m.” (49) A says to B that Minh bought a car. B then says: a. Minh even bought a plane. b. Minh mua một chiếc máy bay cơ. Minh buy one cl plane cơ “Minh even bought a plane.”

232

Anne Nguyen

In addition, cơ and even can cooccur. A counterpart of even in Vietnamese is thậm chí.9 The combination of cơ and thậm chí is common. (50) A says to B that Kim bought five books. B then says: Minh thậm chí mua tám quyển cơ. Minh even buy eight cl cơ “Minh even bought eight books.” (51) A says to B that Kim can buy a car. B then says: Minh thậm chí có thể mua được một chiếc máy bay cơ Minh even can buy able one cl plane cơ “Minh can even buy a plane.” (52) English is assumed to be easier than French. A tells B Minh cannot speak French. B says: Nó thậm chí không nói được tiếng Anh cơ. sg-3 even not speak able English cơ “He cannot even speak English.” We have seen an overlap in the interpretation of cơ and even when the propositions are ordered on logical or categorical scales. Such an overlap between the two particles, however, disappears if the ordering is determined, given a meta-evidential scale. Unlike cơ, even cannot make reference to this type of scale. Given the assumption that books and notebooks have equal status, consider (53). (53) A tells B that Minh will buy a book. B then says: a. No. #He will even buy a notebook. b. Không phải. Nó sẽ mua một quyển vở cơ. not true sg-3 fut buy one cl notebook cơ “Not true. He will buy a notebook.” The assumption suggests that the categorical scale of status is blocked here. Cơ in (53) is allowed because it makes reference to a meta-evidential scale. If even was also involved with this type of scale, (53a) would be felicitous. Its infelicity here shows that even does not make reference to meta-evidential scales. As mentioned at the beginning of the section, even is defined as a scalar additive particle even though its additive component is quite controversial. The discussion about cơ so far shows that cơ is purely a scalar particle. This semantic diference between cơ and even may account for many examples where the behaviors of cơ and even come apart. Consider the contrast between cơ and even in the following. It is assumed French is harder to learn than English and Vietnamese.

Scalar semantics for the Vietnamese sentence-final particle cơ

233

(54) A tells B Minh can speak Vietnamese and asks if it is right that he can speak English. B then says: a. No. But, he can even speak French. b. Of course. He can even speak French. (55) A tells B Minh can speak Vietnamese and asks if it is right that he can speak English. B then says: a. Không. Nó nói được tiếng Pháp cơ. no sg-3 speak able cl French cơ “No. He can speak French.” b. #Dĩ nhiên.Nó nói được tiếng Pháp cơ. of-course sg-3 speak able cl French cơ “Of course. He can speak French.” However, if thậm chí is added in (55b), the sentence becomes grammatical, as shown in (56): (56) A tells B Minh can speak Vietnamese and asks if it is right that he can speak English. B then says: Dĩ nhiên. Nó thậm chí nói được tiếng Pháp cơ. of.course sg-3 even speak able cl French cơ “Of course. He can even speak French.” The assumption that even projects an additive inference but cơ does not can accounts for the contrast in (54) and (55). In particular, in (54b), the proposition Minh can speak English is confirmed true and thus is compatible with the additivity of even. In contrast, the meaning of cơ does not contain an additive component. That makes cơ unable to follow the confirmation of the proposition Minh can speak English in (55b). One way to make cơ in (55b) felicitous is to add an additive particle like thậm chí, as in (56), or còn, as in (57). (57) A tells B Minh can speak Vietnamese, and asks if it is right that he can speak English. B then says: Dĩ nhiên. Nó #(còn) nói được tiếng Pháp cơ. of.course sg-3 also speak able cl French cơ “Of course. He can also speak French.” Note that, as mentioned earlier, it is controversial regarding the additive component of even. For (54a), we may think that the additive inference of even is concerned with the proposition Minh can speak Vietnamese. However, when the information Minh can speak Vietnamese is deleted from the context, even is still accepted, as long as it is assumed that French is harder to learn than English. It is rather puzzling how the additive requirement of even is met here.

234

Anne Nguyen

(58) A: Can Minh speak English?

B: No. But he can even speak French.

Another point to note is that even though the (a) sentences in (54)– (56) – those in which even does not seem to be additive if the informative Minh can speak Vietnamese is omitted – are all accepted, there is a slight diference between them in the use of conjunctions like nhưng “but.” But is needed in the example of even in (54a), but its Vietnamese counterpart, nhưng, is not in the case of cơ in (55a). On the other hand, if conjunctions like but åre removed from those like (54a), they are degraded, as shown in (59)–(60). (59) A asks if it is right that Minh can speak English. B then says: a. No? (But) he can even speak French. b. Không. ?(Nhưng) nó thậm chí nói được tiếng Pháp. no but sg-3 even speak able cl French “No. But he can even speak French.” In contrast, the addition of but to (55a) does not seem to afect the sentence with cơ, as shown in (60). (60) A tells B Minh can speak Vietnamese, and asks if it is right that he can speak English. B says: Không. Nhưng nó nói được tiếng Pháp cơ. no but sg-3 speak able cl French cơ “No. But he can speak French.” It is interesting to know why but is needed for the felicity of even but not for the use of cơ and whether the additive part of even plays a role here. However, I have no explanation for these observations yet and will leave it for future work. 5

Conclusion

This chapter has considered the use-conditional meaning of the Vietnamese sentence-final particle cơ. The proposed formal semantics for the particle is drawn from the examination of its occurrence in a wide range of contexts. In the sense that the meaning of cơ refers to explicit or implicit contrasting alternatives in the preceding context, the particle serves as a discourse device. I propose that the particle reflects a conflict between speaker belief and conversational expectations. This behavior of cơ requires that speakers keep track of expectations from the conversation. As a whole, this understanding of cơ can help refine our theory of speakers’ mental representations of discourse. In addition, the various uses of cơ show that scalar particles may make reference to diferent types of scales and emphasizes the role of abstract notions like scales and scale strength in natural language.

Scalar semantics for the Vietnamese sentence-final particle cơ

235

Abbreviations 1 2 3 cl

first person second person third person classifier

fut pl q sg

future plural question singular

Notes 1 Acknowledgments: This research is supported by the National University of Singapore under grant WBS R-103–000–178–133. 2 Cơ also appears in some wh-questions and exclamative. See Nguyen (2017) for further discussion of cơ in wh-questions. 3 Cơ is mainly used in the North of Vietnam. A variant of cơ is kia, which is used in other regions of Vietnam. They are interchangeable in terms of their functions/ meaning. 4 On the phonetic realization of focus in Vietnamese, see Jannedy (2007). 5 See Zimmermann (2015) for further discussion about the unequal contribution of focus and contrastive topic information to the interpretation of scalar particles. 6 There is a theory that focus-sensitivity is actually always about the choice of relevant (implicit) question in the discourse (Beaver & Clark 2008 and work based on this). If the semantics of cơ is dependent on the prejacent and the contrasting alternatives being answers to the same question, the apparent focus- (and CT-) sensitivity could be derived from this requirement that they be in the same question (or discourse strategy, that is, a family of questions). 7 See Rooth (1985, 1992) and Büring (2003) on the computation of alternatives. 8 The additive presupposition of even is controversial. Refer to Wagner (2013) for further discussion. 9 Zimmerman (2015) treats the combination thậm chí . . . cũng as discontinuous scalar-additive particles in which thậm chí contributes the scalarity and cũng the additive component. Such combination, however, is required only when the subject is focused. In other cases, thậm chí and even behave similarly.

References Beck, Sigrid. 2006. Intervention efects follow from focus interpretation. Natural Language Semantics 14. 1–56. Büring, Daniel. 2003. On D-trees, beans and B-accents. Linguistics and Philosophy 26. 511–545. Greenberg, Yael. 2016. A novel problem for the likelihood-based semantics of even. Semantics & Pragmatics 9(2). 1–28. Horn, Larry. 1969. A presuppositional analysis of only and even. In Papers from the Fifth Regional Meeting, Chicago Linguistic Society, 97–108. Chicago: University of Chicago. Karttunen, Lauri & Stanley Peters. 1979. Conventional implicature. Syntax and Semantics 11. 1–56. Le, Ha Giang. 2015. Vietnamese Sentence Final Particles. MA thesis, University of Southern California.

236

Anne Nguyen

Nguyen, Anne. 2017. Unifying Correction and Skepticism in Vietnamese cơ. Manuscript, National University of Singapore. Nguyen, Dinh Hoa. 1997. Vietnamese. Amsterdam & Philadelphia, PA: John Benjamins. Nguyen, Nhat Anh. 2012. Kinh Van Hoa. Hanoi: Kim Dong Publishing House. Rooth, Mats. 1985. Association with Focus. Ph.D. dissertation, University of Massachusetts, Amherst. Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1. 75–116. Wagner, Michael. 2013. Additivity and the Syntax of Even. Manuscript, McGill University, September 2013. Zimmermann, Malte. 2015. Scalar particles and contrastive topics in English and Vietnamese. In Elitzur A. Bar-Asher Siegal (ed.), Proceeding of IALT 2015, 123–152. Cambridge, MA: Department of Linguistics and Philosophy, Massachusetts Institute of Technology.

10 Syntax-information structure interface in Vietnamese Thuan Tran

1

Introduction1

Grammatical relations in Vietnamese, a language without inflectional morphology, are expressed primarily by word order. As demonstrated in (1), the pronoun họ “they” and the noun phrase cô gái ấy “that girl” serving as subjects in (1a) and (1c) bear the same form as their object counterparts in (1b) and (1d). Similarly, the pronoun tôi “me” and the noun phrase sinh viên này “this student” functioning as objects in (1a) and (1c) are identical to those in (1b) and (1d), where they function as subjects. These examples suggest that a nominal expression depends on its relative position to the verb for its syntactic function: it functions as a subject if occurring in a pre-verbal position, and as an object if surfacing in a post-verbal position. (1)

a. Họ thich ́ tôi. they like I “They like me.” b. Tôi thích ho.̣ I like they “I like them.” c. Cô gái ấy thích sinh viên này. girl that like studen this “That girl likes this student.” d. Sinh viên này thích cô gái ấy. student this like girl that “This student likes that girl.”

Word order, as such, figures prominently in Vietnamese, with SVO being the canonical word order. However, this word order is not always observed: under certain semantic-pragmatic conditions, a sentence can emerge in a non-canonical word order of OSV. Illustrative examples are given in (2), where the object is fronted- or left-dislocated, leaving behind a trace, as notated. Left-dislocated elements are normally, but not obligatorily, followed by the particle thì, which the literature on Vietnamese linguistics characterizes either as a topic or as a DOI: 10.4324/9781351057752-11

238

Thuan Tran

contrastive topic marker. Nevertheless, there is evidence indicating that this particle functions, at best, as a discourse template marker (Needleman & van de Koot 2008) rather than a topic or contrastive topic marker, because, as will be discussed in Section 2.2, its use is not restricted to topic-marking contexts. For ease of exposition, in what follows it is glossed as PRT. (2)

a. Sinh viên nàyi thì cô gái ấy student this prt girl that “This student, that girl likes.” b. Cô gái ấyi thì sinh viên này girl that prt student this “That girl, this student likes.”

thich ́ ti. like thich ́  ti. like

The canonical word order sentences in (1c, d) and the non-canonical word order ones in (2a, b) are truth-conditional equivalent, respectively, but difer in terms of information structure. While the canonical sentences in (1c, d), depending on the context, can include information structural categories, such as topic and focus, the non-canonical ones in (2a, b) arguably involve, in addition to topic and focus, an extra component. It is obviously thanks to this extra component that (2a), not (1c), naturally occurs as a continuation of a preceding utterance, such as “I don’t know who likes that man, but . . .” Contrastingly, (2b), not (1d), is a felicitous follow-up utterance of “this student is grumpy and unable to get along with many people, but . . .” I assume that this component is contrast and propose, in line with Molnár (2006), Frey (2006), Neeleman and Vermeulen (2012), among others, that topic, focus, and contrast are autonomous information structural categories. The main claim, to be specific, is that in Vietnamese, topic and focus are two basic information structural categories that surface in situ, whereas contrast is an autonomous information structural category that triggers topic/focus left dislocation. The chapter is organized as follows. Section 2 discusses the two basic information structural categories, topic and focus, in Vietnamese, and the assumed topicmarking particle. The contrast component is explored separately in Section 3. Section 4 provides a formal analysis of contrastive topic and contrastive focus. Section 5 concludes the chapter. 2 2.1

Information structure in Vietnamese Topic

The information structural category of topic is notoriously controversial regarding its defining features, as reflected in the literature (Chafe 1976; Reinhart 1981; Vallduví 1992; Lambrecht 1994; Casielles-Suarez 2004, among others). In this chapter, I follow Reinhart (1981) in assuming that the topic of a sentence is what the sentence is about: to quote Reinhart verbatim, “topic of expresses the relation of being about.” Thus, the term “topic” used in this

Syntax-information structure interface in Vietnamese

239

chapter is interchangeable with the term “aboutness topic.” For a working definition, I adopt Krifka’s (2006) specification of topicality: “[t]he topic constituent identifies the entity or the set of entities under which the information expressed in the comment constituent should be stored in the Common Ground content.” A topic construed as such must be referential, normally represented by a referring expression whose function is to identify an individual, an e-type semantic object in the sense of Heim and Kratzer (1998). This referentiality constraint, according to Li and Thompson (1976), Gundel (1988), among others, is observed to hold across languages. Vietnamese is an illustrative language in that a topic in Vietnamese is referential, represented by either a definite or specific indefinite expression. Furthermore, Vietnamese can be characterized as a topic-prominent language in the spirit of Li and Thompson (1976) in that it tends to construe the definite NP subject of a canonical sentence as the topic of a discourse-initial sentence.2 It is unsurprising that a subject in Vietnamese is preferably referential. As demonstrated in (3), indefinite NPs are not allowed in subject position.3 (3)

a. *Một người phụ nữ làm giàu rât́ nhanh. one clf woman make rich very fast “A woman becomes rich very fast.” a’. Nam nói chuyện với môṭ người phụ nữ. Nam say story with one clf woman “Nam talked to a woman.”

Sentence (3a) is ill-formed with the indefinite NP “a woman” in the subject position, as opposed to the well-formedness of (3a’), where the same indefinite NP is in the object position. To express the proposition intended by (3a), an existential structure should be used (3b). b. Có một người phụ nữ làm giàu rât́ nhanh. exist one clf woman make rich very fast “A woman becomes rich very fast.” The referentiality constraint also explains why indefinite NPs are not allowed to undergo left dislocation, as illustrated in (4). (4)

a. *Một cái ghếi Nam mới mua ti. one clf chair Nam just buy “Nam just bought a chair.” b. Cái ghế nàyi Nam mới mua ti. clf chair this, Nam just buy “This chair, Nam just bought.”

The dislocation of the indefinite NP in (4a) results in ungrammaticality, but that of the definite NP in (4b) is grammatical.4 However, Cao (1991: 99)

240

Thuan Tran

argues that there currently appears a tendency that allows the use of indefinite noun phrases as topics, and provides the following examples. For ease of exposition, the supposed topics are underlined (the gloss and the translation are mine, with COMP standing for complementizer, and ASP for aspect marker). (5)

a. Vừa luć âý một người cao lớn bước vaò phòng. right moment that one person tall big step enter room “Right at that moment, a tall man stepped into the room.” b. Một số tác giả cho rằng vấn đề đã one number author assume COMP problem asp giải quyêt́ xong. solve already “Some authors assume that the problem has been solved.”

Before discussion, a few words on definiteness in Vietnamese are necessary. In Vietnamese, a bare noun phrase, that is, a noun without any modifying material, can be interpreted as definite or indefinite, depending on the context. Similarly, a non-bare noun phrase, namely, a noun phrase formed by a numeral, a classifier, and a noun, can be contextually construed definite or indefinite. This ambiguity can be resolved by adding a demonstrative to the noun phrase, giving rise to the definiteness reading. However, a non-bare noun phrase consisting of the numeral one, a classifier, and a noun is unambiguously indefinite: một quyên ̉ sach ́ (one CLF book), meaning “a book.” Forcing a definite reading by inserting a demonstrative into a noun phrase of this form results in illformedness: *một quyên ̉ sach ́ naỳ (one CLF book this), literally “a book this.” Back to the examples in (5a, b), if we take Götze et al.’s (2007: 165) tests for determining the aboutness topic of an utterance, it is obvious that the underlined NP in (5a) is not an aboutness topic. It is odd to make an announcement like, “Let me tell you something about a tall man,” to introduce (5a), while it is completely appropriate to precede (5b) with, “Let me tell you something about some authors.”5 Clearly, the noun phrase một số tác giả “some authors” in (5b) difers from the noun phrase một người cao lớn “a tall man” in (5a), in that the former is a specific indefinite NP that yields a partitive reading, as is well-known in the literature (Szabolsci 1997, among others).6 Vietnamese exhibits a pattern well-known across languages, where a topic is not necessarily in sentence-initial or pre-verbal position (see Casielles-Suarez 2004, and references therein), in that it allows a non-subject topic to remain in its base position, overriding the tendency to preferably construe a definite subject as topic, as illustrated in the following examples. (6)

Context: Let me tell you something about Tan.

(7)

a. Công ty A đã nhận Tân. company A pst accept Tan “Company A accepted Tan.”

Syntax-information structure interface in Vietnamese

241

b. Tâni thì công ty A đã nhận ti. Tan prt company A pst accept “Company A accepted Tan.” It appears that both (7a) and (7b) are felicitous in context (6), where the entity “Tan” is established as a topic. The non-subject topic remains in the base position in (7a) but is left-dislocated in (7b). It seems plausible to conclude that topic dislocation is optional. However, this is only an approximation. There is evidence suggesting that topic preferably remains in the base position: it is more natural if the non-dislocated topic (7a), not the dislocated topic (7b), is followed by a continuation, as in (7c). c. Và thật ra công ty A đã nhận tất cả người and in fact company A asp accept all person nộp đơn. submit application “And in fact, company A accepted all the applicants.” The data in (7) indicates that left dislocating is not optional but is associated with an interpretative efect of some sort. Let us try to work out what it is. Consider the contrastive topic context given in (8). For ease of exposition, a subscripted CT is appended to the contrastive topic contextually established. (8)

A asks B after a shopping done by a group of friends, pointing at the commodities on the table: Who bought what? Who bought the tea? Who bought the cofee? B replies:

(9)

a. Tôi không biết ai mua trà CT, nhưng . . . I not know who buy tea, but . . . “I don’t know who bought the tea, but . . .” b. #Nam mua cà phê CT. Nam buy cofee “Nam bought the cofee.” c. Cà phê i CT thì Nam mua ti. cofee prt Nam buy “As for cofee, Nam bought it.”

Both (9b) and (9c) are truth functionally the same, yet (9c) is a natural continuation of (9a), not (9b). The fact that (9c), with the contrastive topic being left-dislocated, is felicitous, while (9b), where the contrastive topic is in the base position, is not suggests that it is contrastive topic, not topic, that undergoes left dislocating. Since what distinguishes a contrastive topic from a topic is its obligatory dislocation, we can assume that, without further substantiation at this point, it is the feature “contrast” that triggers left dislocation. A discussion of contrast is provided in Section 3. In brief, a topic in Vietnamese

242

Thuan Tran

needs to be referential and is not required to be left-dislocated. A contrastive topic is, however, obligatory to undergo left dislocation. 2.2

Thì is not a (contrastive) topic marker

We have observed in the previous section that a topic is preferably followed by the particle thì, especially when it undergoes left dislocation. This particle is described in school grammar as a topic marker (Cao 1991). Formal analyses by Dufeld (2007) and Tran (2009) take it as the overt head of the topic phrase, glossed as TOP in the illustrative sentences that follow. The analysis of Nguyen and Lee (2004) and Lee (2017) is descriptively more adequate that thì is a contrastive topic marker; however, they fail to recognize that this particle can also associate with focus, yielding a contrastive focus reading, as illustrated by examples (2a, b) in the introduction. Another particle that can occur in place of thì is la.̀ In the absence of the overt particle, a null particle realized by a pause is in use. (10) a. Nó (thì) thích kẹo chanh nhất. he top like candy lemon best “He likes lemon candy best.” b. Bać Nguyên ̃ (thi)̀ đã qua đời. uncle Nguyen top asp pass life “Uncle Nguyen passed away.” (Cao 1991: 97–98; my gloss and translation) I assume, ignoring dialectal variation and stylistic preferences, that the two particles thì and là are interchangeable, and look at the more frequently used particle, thì. It is unexpected that this assumed topic head/topic marker, thì, is not always associated with topic.7 In the following example, the particle follows a wh-phrase, a highly inadequate candidate for topichood. The question in (11a) displays the non-canonical wh-question in that the object wh-phrase ai “who” is dislocated from the base position, the post-verbal position. Note that Vietnamese is a wh-in-situ language but licenses wh-ex-situ under certain semantic-pragmatic conditions. Essentially, while it is normal to start an exchange with a canonical wh-in-situ question, a wh-ex-situ question like (11a) cannot be used at the beginning of a conversation and is felicitous only when it is preceded by a denial or a series of denials of a propositional content of the form “you help X.” (11) a. Aii thì anh giúp ti? who prt you help “Who will you help?” b. Nami thì tôi giuṕ  ti. Nam prt I help “I will help Nam.”

Syntax-information structure interface in Vietnamese

243

It is not felicitous to answer a non-canonical wh-question with a canonical word order sentence. For instance, as a reply to the question in (11a), it is preferable to use a non-canonical sentence in (11b), where the fronted object NP “Nam,” followed by the particle, is a focus. Interestingly, (11b) can also serve as a natural continuation of a mini-discourse, such as “As for generosity, it is not my nature to help anyone, yet . . .” A topic-marking particle would not be expected to be felicitous in these contexts. Furthermore, if this particle were to mark topic, it would be difcult to explain why the particle chooses to accompany a wh-phrase, not a topic, in cases where both are left-dislocated, as shown in the following. Consider a bargaining context in which the seller asks the buyer how much he/she is happy to pay for the merchandise after the customer’s rejections of the price ofers given by the seller. This context ascertains that the shirt is the topic. It would be more natural to drop the topic in (12b). (12) a. Cái áo nàyi bao nhiêuj thì anh mua ti tj? clf shirt this how much prt you buy “How much will you buy/pay for this shirt?” b. Bao nhiêuj thì anh mua cái áo này tj? how much prt you buy clf shirt this “How much will you buy/pay for this shirt?” c. *Cái áo nàyi thì bao nhiêuj anh mua ti tj? clf shirt this prt how much you buy “How much will you buy/pay for this shirt?” If thì were a topic marker, we would expect (12a, b) to be ungrammatical, and (12c) to be grammatical, contrary to fact.8 It is therefore plausible to reject the assumption that the particle in question is a topic marker and assume at this point that it is a discourse template marker, following Needleman and van de Koot (2008). Its function is not to mark topic or focus but to mark the remaining material, namely, the material that resulted from the dislocation of topic or focus, as comment or background, respectively. In other words, the particle is related to dislocation. This analysis is indirectly supported by the fact that thì is not allowed in sentence-focus context, where the whole sentence is in focus, hence no dislocation. Cao (1991: 135) observes that thì cannot be used with sentences used to respond to questions that trigger allnew focus, such as “What happened?” or “What’s wrong?” It is presumably the case that thì is a contrast-marking particle, as explicated in footnote 8, but for ease of presentation, the gloss for thì is simply PRT “particle.” 2.3

Focus

I assume, in line with the current literature, that the information structural category focus refers to the part of the sentence that corresponds to the whphrase of the wh-question to which the sentence is a felicitous answer (Beaver & Clark 2008; among others). The contribution of focus is to provide a set

244

Thuan Tran

of alternative elements of the same type as the focus. Replacing the focus with the alternative elements will generate a set of propositions alternative to the expressed proposition. The set of the alternative propositions is the focus value of the sentence; the expressed proposition is its ordinary value (Rooth 1992, 1999). Focus is a pragmatic notion with linguistic reflex. It is well-known that intonation languages like English or German employ phonological means for focusing. Interestingly, Vietnamese, a tone language, also realizes focus by the same device. Cao (1991) observes that in Vietnamese, focus is generally expressed via prosodic marking, namely, by placing the stress on the focused element, with the focus being in situ. This observation is supported by Jannedy’s (2007) experimental study on focus in Vietnamese. Following the standard convention, a focused element is placed in squared brackets with a subscripted F. (13) a. Anh sẽ mua cái gì? you fut buy clf what “What will you buy?” b. Tôi sẽ mua [sách]F. I fut buy book “I will buy books.” Vietnamese also makes available ex situ focus with extra semantic-pragmatic efect, as illustrated in example (14). (14) a. Cái gì thì anh mua? clf what prt you buy “What will you buy?” b. [Sách]F thì tôi mua. book prt I buy “I will buy books.” The examples in (14), as well as the similar example in (11), suggest that ex situ focus does not occur freely: it cannot surface in sentences uttered out of the blue but needs to be licensed, for instance, by a non-canonical question, as in (14a), not by a canonical one, as in (13a). Also, diferent from in situ focus, ex situ focus implicates the denial of a proposition alternative to the focused proposition. For example, (14b) implicates a denial of the proposition “I will buy X,” where X is a constituent alternative to the focused element. Ex situ focus also brings about a temporal efect (see Tran 2021): the non-past interpretation as indicated by the future translation of (14b) in the absence of the future marker (cf. (13b)). The same temporal efect is observed in the non-canonical wh-question (14a). However, this chapter does not delve into this temporal issue due to its scope. In summary, Vietnamese realizes focus in situ and licenses ex situ focus. Ex situ focus difers from in situ focus in that it implies a negation of (at least) an alternative proposition. I take this

Syntax-information structure interface in Vietnamese

245

implication as being associated with the contrast component and propose that ex-situ focus is correlated with the information structural notion “contrast,” to be discussed in the next section. 3

Contrast

The previous section showed that the information structural categories, topic and focus, in Vietnamese normally surface in situ, namely, in the canonical word order, but can appear in a non-canonical word order through left dislocation, giving rise to a contrastive interpretative efect. This section will address the question of whether contrast is an autonomous category, since a positive answer to this question will help explain the diference in the syntactic behavior of focus and contrastive focus as well as that of topic and contrastive topic. 3.1

Contrast as a dependent component

Topic and focus are standardly established in the literature as autonomous categories of information structure. The notion of contrast, however, is normally taken as being contingent on topic and focus, as indicated by the use of technical terms like contrastive topic, and contrastive focus as opposed to topic, and focus. Contrast is not analyzed independently from focus and topic but is instead treated either as “the result of general cognitive processes referred to as ‘conversational implicatures’” (Lambrecht 1994: 291), or as a feature that helps subdivide the category focus or topic. For instance, according to Chafe (1976: 33), a sentence like “Rónald made the hamburgers,” where the acute accent mark indicates the highest pitch and stress on the stressed syllable of “Ronald,” involves either contrastive or non-contrastive focus. The sentence with the non-contrastive focus answers the question “Who made the hamburgers?” with the non-contrastive focus indicating the referent as novel, newly activated. Using a sentence with the contrastive focus, the speaker assumes that the hearer has a limited number of candidates, and that the hearer believes that the predicate “made the hamburgers” holds one of them. The speaker then informs the hearer of his/her knowledge of the hearer’s assumption and clarifies that it was Ronald and no one else who made the hamburgers. A similar distinction between contrastive focus and non-contrastive focus or focus is made in Zimmermann (2008), where contrast is analyzed as amounting to “emphasis,” bearing hardly any semantic efect. According to Zimmermann, a speaker marks a constituent as contrastive focus to prepare for a swifter update of the common ground because that constituent is less expected to the hearer. As illustrated by the example that follows, taken from Zimmermann (2008: 354), the need for contrastive focus marking is to indicate that the information provided is unexpected to the hearer. (15) a. Q: What did you eat in Russia? A: We ate pelmeni. b. A: Surely, you ate pelmeni! B: No, caviar, we ate!/No, we ate caviar!

246

Thuan Tran

In the answer A to the question Q in (15a), the focused element, the expression in italics, corresponding to the wh-phrase of the wh-question, involves no contrastive focus marking. However, the correcting statement given by B, as a response to A in (15b), includes contrastive focus marking. In this statement, the contrastive focus element, the italicized expression, is marked either by left dislocating or by a raised pitch, indicated by the upper arrow. The speaker chooses to mark contrastive focus in (15b) because he/she assumes that the hearer will find it unexpected that caviar is the food to be consumed in Russia. He/she does not mark contrastive focus in (15a) on the assumption that pelmeni is a normal food to eat in Russia. It is in the work of Vallduví and Vilkuna (1998), where “contrast” is spelled out as “kontrast,” that the term “contrast” is likely to refer to an independent component. However, it seems that their conception of contrast is basically parallel to Rooth’s notion of focus in the sense that both Vallduví and Vilkuna’s contrast and Rooth’s focus invoke a set of alternatives. Overall, the question of whether contrast is an information structure category, independent from focus and topic, has not been given adequate attention. Research on contrast as an autonomous notion is not as widespread as that on topic and focus and, in general, views contrast as a feature that pragmatically enriches the two categories, focus and topic, but is semantically vacuous (but see Molnár 2006, Frey 2006, among others, for analyses of contrast as a semantic notion).9 It is far more often to come across research concerned with the pair contrastive topic–contrastive focus. For example, Lee (2006, 2017) proposes, based on a discourse model of information structure, that contrastive focus difers from contrastive topic in that the former is initiated by a disjunctive question while the latter is by a conjunctive question. According to Lee, a response as in (16b) to the question in (16a), where the underlined expression subscripted with CT is a contrastive topic, presupposes a conjunctive question like “What did the little boy and the rest eat?” (16) a. What did these boys eat? b. The little boy CT ate cookies. By contrast, an answer to an alternative disjunctive question as in (16c) involves contrastive focus (16d), where the contrastive focus is subscripted with CF. (16) c. Did John eat banana or mango? d. John ate banana CF. 3.2

Contrast as an independent component

In Section 2, we witnessed the interpretative efect ensuing from the left dislocation of the information structural categories topic and focus and the

Syntax-information structure interface in Vietnamese

247

suggestion that this interpretation is associated with contrast. This subsection elaborates on the notion of contrast, with a view to accounting for the Vietnamese data. Crucially, it is argued, in line with Neeleman and Vermeulen (2012), that contrast is an independent category that semantically functions as a quantifier and syntactically licenses left dislocation. According to these authors, topic, focus, and contrast are autonomous notions. Topic and focus are basic notions that can be enriched by contrast to yield a contrastive interpretation. Accordingly, a contrastive topic is an aboutness topic interpreted contrastively, and a contrastive focus is a new information focus that receives a contrastive interpretation. What does it mean by contrastive interpretation? To be specific, what is the contribution of contrast in contrastive focus? On Neeleman and Vermeulen’s (2012) analysis, the partition of a sentence into focus-background as a result of focusing yields two sets: one is the set of alternative elements to the focus, and the other is the set represented by the background. The role of contrast is to give information about the relation between the two sets: at least one member of the set of (contextually relevant) alternatives is not contained in the set denoted by the background. Contrast as such functions as a quantifier, given that quantifiers express relations between sets. Contrast is linguistically relevant since the effect it brings about cannot be canceled. Illustrative examples are given next, taken from Neeleman and Vermeulen (2012: 12–23). Consider the following context. Dad and Mum know that John must read five books to prepare for the exam, and are discussing which book John has read. The exchanges in this context are developed around the implicit question: Which book has John read? (Neeleman and Vermeulen indicate the focus constituent with small capital letters.) (17) Mum: John’s read The Selfish Gene. Dad: Yes, I know. THE SELFISH GENE he’s read. (i) But THE EXTENDED PHENOTYPE he hasn’t read. (ii) #In fact, he’s read all five books on the reading list. Dad’s reply in (17) involves contrastive focus, as indicated by the dislocated NP The Selfish Gene, giving rise to the interpretation that (17) includes a positive statement that one member of the set of books, The Selfish Gene, is also a member of the set of things that John read, and a negative statement that there is at least one relevant book that John has not read. This interpretation is not cancelable, as shown by the oddness of (ii) as a continuation of dad’s reply, where a claim is made that there is no such relevant book, and by the felicity of (i) as a continuation, in which a negative statement about John’s reading of a relevant book is asserted. The assumption that the negative statement is derived from contrast is further supported by the unavailability of the negative statement in non-contrastive focus. The following exchanges are in the same context as those in (17). Dad’s reply is expressed by a normal focus (the focus stays in situ). The continuation

248

Thuan Tran

as in (ii) is felicitous, which makes no negative statement about John’s reading of the relevant books. (18) Mum: John’s read The Selfish Gene. Dad: Yes, I know. He’s read THE SELFISH GENE. (i) But THE EXTENDED PHENOTYPE he hasn’t read. (ii) In fact, he’s read all five books on the reading list. The preceding examples show that the positive statement and the alternatives to the focus, The Selfish Gene, derive from the normal interpretation of focus. The negative statement about an alternative, however, is part of the semantics of contrast. We observed the role of contrast in contrastive focus: as a quantifier and a licenser of left dislocation. In what follows we see that contrast performs a parallel role in contrastive topic in that a left-dislocated topic must be interpreted contrastively. Consider the exchanges given in (19, 20), taken from Neeleman and Vermeulen (2012: 20), where the contrastive topic is in boldfaced italics and double-underlined. (19) A: What about the popstars? Who showed them around? B: Well, the female popstars, Bill gave a tour. A asks B a question about the popstars; B’s reply involves the female popstars, a subset of the topic introduced by the context question, and is therefore in opposition to the complement set, the male popstars. B’s reply in (19) is felicitous. However, without a licensing context, the use of contrastive topic is not appropriate, as demonstrated in (20), where the context does not provide a contrastive interpretation: there is no complement set such that the set of female popstars stands in opposition to. (20) A: Tell me about the female popstars. B: #Well, the female popstars, Bill gave a tour. Contrastive topic and contrastive focus display similar syntactic behavior in that they both license left dislocation. Yet their semantic contributions are diferent. Unlike the answer that involves contrastive focus, B’s reply in (19) does not lead to a negative statement that there is an alternative to the female popstars such that Bill did not give a tour to this alternative. It instead suggests that the speaker, for some reason, cannot make an inclusive statement. The reason might be because B does not obtain sufcient information about the male popstars to answer with a claim that includes them, or because B does not want to make a claim involving the male popstars even though B is knowledgeable about them. Overall, contrastive topic is unlike contrastive focus in that the former does not operate at the propositional level but at a higher level, the speech-act level. We will come back to this issue in section 4.

Syntax-information structure interface in Vietnamese

249

Importantly, contrastive topic is like contrastive focus in that the negative efect implied by contrastive topic is linguistically encoded. This is illustrated in (21), where the dislocated topic must be construed contrastively. (21) Context: John and Bill (and nobody else) came to town. A: So you met John. How about Bill? B: (i) # Bill, I met. (ii) Bill, I didn’t meet. Given the context that of the two persons, John and Bill, B met John, B’s response in (i) is infelicitous because it implies that there is an alternative to Bill about whom the speaker is unable to make the same assertion. This implication is odd, because speaker B met that person. The felicity of (ii) is natural since it implies that there is a person alternative to Bill, John, that the speaker could not assert that he/she did not meet that person/John. Back to Vietnamese, it is plausible to assume that topic/focus left dislocation in Vietnamese is licensed by contrast, and that contrast is quantificational in the sense of Neeleman and Vermeulen (2012), that is, contrast expresses a relation between two sets, the alternative set triggered by focus or topic and the set resulted from left dislocation. This position is supported by examples given in the following. Consider the case of contrastive focus first. Suppose that Nam is required to read five books and that the conversation is built around the question of which book Nam read. A informs B of the book that Nam has read. B confirms the information provided by A by uttering either (a) or (b) and adds more information about Nam’s reading by using either (i) or (ii). Adhering to the notational format in Neeleman and Vermeulen’s (2012) examples, in the following examples, focus elements are in small capitals. (22) A: Nam đã đọc QUÊ NGƯỜI. Nam asp read Que nguoi “Nam has read Que nguoi.” B: Anh biết. I know “I know.” a. QUÊ NGƯỜI i thì Nam đã đọc ti. Que nguoi prt Nam asp read “Nam has read Que nguoi.” b. Nam đã đọc QUÊ NGƯỜI. Nam asp read Que nguoi “Nam has read Que nguoi.”  i. nhưng O CHUỘTi thì Nam chưa đọc ti. but O chuot prt Nam not.yet read “but Nam has not read O chuot yet.” ii. và thật ra, Nam đã đọc hết năm quyển sách and in fact Nam asp read finish five clf book

250

Thuan Tran theo yêu cầu follow requirement “and in fact, Nam has read all the five books as required.”

In this context, (ii) is odd as a continuation of (22a), where the focus is left-dislocated, but it is not as a continuation of (22b), where the focus is in situ. This is because the contrast component, licensed by left dislocation, in (22a) provides a negative statement that there is at least one alternative to Quê Người that does not belong to the set of things that Nam has read. This statement is in conflict with (ii), where the assertion is made that Nam has read all the books required. With the focus in situ, (22b) does not convey any negative statement regarding Nam’s reading, hence the felicity of (ii) as a continuation of (22b). Along the same line, the felicity of sentence (22i) as a continuation of (22a, b) is obvious. The negative component of (22i) states that there is at least one alternative to O Chuột such that this alternative is not contained in the set of things that Nam has not read. This statement is in agreement with both (22a) and (22b). Consider next another contrastive focus context. (23) A tells B about Nam’s shopping habit: what he buys and does not buy. (a) Nam does not buy pens, Nam does not buy notebooks, . . . but (b) Nam buys cookies, Nam buys chocolate, . . . and (i) SÁCHi thì Nam mua ti. book prt Nam buy “Nam buys books.” (ii) Nam mua SÁCH. Nam buy book “Nam buys books.” The non-canonical word order sentence in (23i), with the focus being leftdislocated, is felicitous as a continuation of (23a), where the negative statements about the alternative (pens, notebooks .  .  .) are provided. Negative statements like this are not available in (23b), hence the infelicity of (23i) as its continuation. By contrast, the canonical word order sentence in (23ii), where the focus object “books” stays in situ, is felicitous as a continuation of (23b), not of (23a). Now, let us look at the case of contrastive topic. Consider the exchanges in (24) in a context that triggers contrastive topic. For ease of exposition, the focus is written in small capitals, and the contrastive topic is underlined. (24) A: Tell me who will help whom. For instance, who will help Mai? B: I don’t know who will help Mai, but . . . a. #BA sẽ giúp Nam. Ba fut help Nam “Ba will help Nam.”

Syntax-information structure interface in Vietnamese b. Nami thì BA sẽ Nam prt Ba fut “Ba will help Nam.”

251

giúp ti. help

The two sentences in (24a, b) convey the same propositional content, namely, a help-relation is held between the two arguments, represented by two proper names, Nam and Ba, where Ba is the helper and Nam is the helpee. It is obvious from the context in (24) that B is not able to make an assertion about Mai, an alternative to Nam. In this context, (24a), where the CT object stays in the base position, is not felicitous, while (24b), where the CT object undergoes left dislocation, is. It is plausible to assume that similar to the case of contrastive focus, it is the contrast component in contrastive topic that licenses left dislocation. However, unlike contrastive focus, the contrast component in contrastive topic indicates that the speaker, for some reason, cannot make a claim about alternative topics. As a result of the contrast component efect, upon hearing an utterance that involves contrastive topic, the hearer is still in the dark with respect to the information about a topic alternative to the expressed topic. In summary, Vietnamese provides evidence in support of the hypothesis that contrast is an independent information structural category. Contrast affects the syntactic form of the utterance by left-dislocating topic or focus. The semantic contribution of contrast is to add a negative component to the meaning of the utterance. The next section is devoted to a formal analysis of the contrastive component. 4 4.1

Contrastive focus and contrastive topic Contrastive focus vs. focus

Section 2.3 showed that Vietnamese realizes focus in situ and licenses ex situ focus. Ex situ focus, as demonstrated in the previous subsection, involves the contrast component, hence the term contrastive focus. We also witnessed, in Section 3, evidence indicating that focus and contrastive focus are not semantically identical. This subsection is a formal analysis of their interpretative dissimilarity. Consider first how the semantic contribution of focus is formalized. According to standard focus semantics, that is, Rooth (1992), focus triggers a set of entities alternative to the focused constituent and makes available two semantic values, the ordinary value and the focus value. For illustration, examine the focus interpretation of (25b), as represented in (26), using the notational device by Neeleman and Vermeulen (2012). The representation in (26) denotes a triplet that consists of a function, the focus, and a set of alternatives to focus. (25) a. Nam sẽ giúp ai? Nam fut help who “Who will Nam help?”

252

Thuan Tran b. Nam sẽ giúp [Mai]F. Nam fut help Mai “Nam will help Mai.”

(26) The first element of the triplet, the function, introduced by the lambda operator, is equivalent to the background, as characterized in structured meanings (Krifka 2006); the second is the focus; and the third denotes a set of alternatives to focus. Applying the function to the focus yields the ordinary value, the asserted proposition, and applying it to the alternatives generates the focus value, the propositions alternative to the asserted one. Now, let us look at the semantics of contrastive focus. For instance, consider a sentence containing a contrastive focus as in (23i), repeated in the following. We concede that (23i) implicates a negative statement of the alternatives to the focus: there is at least one thing that Nam will not buy. In addition to this negative statement, however, it is obvious that (23i) also supplies the information about what Nam will buy, which is what focus or information focus does. Therefore, Neeleman and Vermeulen (2012) propose that the contrastive focus semantics involves two components, the focus component and the contrast component. (23) (i) [Sách]i CF thì Nam mua ti. book prt Nam buy “Nam will buy books.” The contrastive focus semantics contributed by (23i) can be represented as in (27), with (27a) being the focus-meaning component and (27b) being the contrast-meaning component, namely, the negative statement of at least one alternative to the focus. What (27b) expresses, informally, is that there is at least one alternative y to the focus books such that it is not the case that Nam will buy y.



(27) a. b. y [y{pens, notebooks, pencils}& [x Nam will buy x] (y)] In brief, contrast focus difers semantically from focus in that while the latter involves a focus component, the former consists of a focus component and a contrast component that expresses a negative statement of at least one alternative to the focus.10 4.1

Contrastive topic vs. topic

Again, for simplicity, the two technical terms, topic and aboutness topic, are used interchangeably to indicate the entity or the set of entities that holds an aboutness relation with the proposition denoted by the comment. We know that focus

Syntax-information structure interface in Vietnamese

253

gives rise to a set of alternatives, and that focus yields two semantic values, the ordinary value and the focus value. Then, what is the semantic contribution of topic? This issue is far from being settled, but within the scope of this chapter, it sufces to embrace the idea that topic introduces an operator that works at the speech-act level (Krifka 2001, 2003; Portner 2007; Tomioka 2010). To be specific, Portner (2007: 413) treats the aboutness relation between topic and comment as a separate performative. According to Portner, the topic Maria in the sentence Maria, I like her very much gives rise to a performative, a reporting act, informally expressed as “(I report that) my mental representation of Maria is active.” Neeleman and Vermeulen (2012) take a further step in assuming that intuitively, when a speaker makes an utterance, he/she performs two speech acts: (i) the act of considering the topic (out of a set of possible topics) and (ii) the act of asserting the expressed proposition. The authors also hold, in line with Tomioka (2010), that focus is a notion of the proposition level, while topic is a notion of the utterance level. Accordingly, the topic meaning of a sentence is represented, similar to the focus representation in the previous subsection, by a triplet that consists of a function, corresponding to the comment, the topic, and a set of alternatives to the topic. The function involves an ASSERT operator, to be read as “I assert that . . .” The topic semantics of an utterance includes the ordinary value and the topic value, the compositional procedure of which is similar to that of focus: applying the function to the topic yields the ordinary value, an utterance, and applying the function to the set of alternatives to the topic generates the topic value of the sentence, a set of utterances. For illustration, consider how the meaning of the sentence with an aboutness topic given in (7a) in the context (6), repeated in the following as (29), (28), respectively, is composed in the spirit of Neeleman and Vermeulen (2012). (28) Context: Let me tell you something about Tan. (29) Công ty A đã nhận Tân. company A pst accept Tan “Company A accepted Tan.” (30) < λx ASSERT [Company A accepted x], Tan, {Mai, Nam, Trung . . . }> As represented in (30), the topic meaning of (29) is represented as a triplet, consisting of the function introduced by the ASSERT operator, the topic, and the alternative to the topic. The ordinary value of (29), composed by applying the function to the topic, can be paraphrased as follows: considering Tan, I assert that Company A accepted Tan. The topic value is composed by applying the function to the alternative topics, giving rise to a set of alternative utterances, expressed informally: {Considering Mai, I assert that Company A accepted Mai, Considering Nam, I assert that Company A accepted Nam, Considering Trung, I assert that Company A accepted Trung . . .}. The interpretation of a contrastive topic sentence is derived in a procedure parallel to that

254

Thuan Tran

of contrastive focus sentence. However, given that focus and topic operate at diferent levels, the former at the propositional level, the latter at the utterance level, their interpretative efects are diferent with respect to the contrast component. While contrastive focus expresses that there is at least one alternative proposition such that it is negated, contrastive topic indicates that the speaker is unwilling or epistemically unable to make an alternative utterance. This is illustrated by example (9), section 2, where the preceding (9a), which expresses that the speaker is not sure about the alternative topic, tea, facilitates the dislocation contrastive topic (9c). The following exchange provides further support. (31) Nam prepared two dishes, a sandwich and a fried chicken (and nothing else). A asks: I knew that Bac ate the sandwich. What about the fried chicken? B replies: (a) #[Món gà chiên] CT thì Bắc ăn. dish chicken roast prt Bac eat “Bac ate the fried chicken.” (b) [Món gà chiên] CT thì Bắc không ăn. dish chicken roast prt Bac not eat “Bac did not eat the fried chicken.” (b’) # Bắc không ăn [món gà chiên] CT. Bac not eat dish chichen roast Given the context that of the two dishes, the sandwich and the fried chicken, Bac ate the sandwich, B’s response in (a) is infelicitous, but not contradictory, because it implies that there is an alternative to the fried chicken about which the speaker is unable to make the same assertion. That is, (a) indicates that B is epistemically unable to assert that Bac ate the sandwich. This implication is odd because B learns from A that Bac ate the sandwich. If contrastive topic was operative at the propositional level, (a) would be contradictory since it would convey that there is a dish alternative to the fried chicken such that Bac did not eat it. The felicity of (b) follows naturally since the implication is that B is not willing to assert that there is some dish alternative to the fried chicken such that Bac did not eat it. Again, if contrastive topic operated at the propositional level, (b) would be deemed to be redundant, since it would suggest that there is a dish alternative to the fried chicken such that it is not true that Bac did not eat it. Leaving the contrastive topic in situ will be infelicitous.11 Overall, a contrastive topic sentence as in (32b) in the context (32a), where the contrastive topic is underlined, will have a semantic representation, as in (33). (32) a. Context: Bac helped Mai. What about Tan? Who helped him? b. Tân i thì Nam giúp ti. Tan prt Nam help “Nam helped Tan.” (33) a. b.  y [y {Mai, Trung, Hoa . . . }& ¬ [λx ASSERT [Nam helped x] (y)]]

Syntax-information structure interface in Vietnamese

255

The ordinary value is generated by applying the ASSERT operator to the topic, yielding the utterance “Considering Tan, I assert that Nam helped Tan.” The topic value is derived by applying the ASSERT operator to the alternative topics, giving rise to a set of utterances: {“Considering Tan, I assert that Nam helped Tan, Considering Mai I assert that Nam helped Mai,” “Considering Trung, I assert that Nam helped Trung” .  .  .}. The contrast component as represented in (32b) states that there is at least one alternative such that the speaker cannot make or finds it difcult to make an assertion about. In the example given, it can be paraphrased as: ‘There is someone alternative to the entity denoted by the topic such that I cannot assert that Nam helped that person.” 4

Conclusion

Vietnamese strongly supports the status of contrast as an autonomous category of information structure and confirms that contrast is the primary player in the syntax-information structure interface. Basically, the two information structural categories, topic and focus, are syntactically inert and remain in situ. Contrast is semantically operative in that it functions as a quantifier, operating over entities of diferent levels, proposition and speech-act levels, depending on the information category it associates with. Contrast afects syntactic structure by triggering left dislocation of the contrast-marked element. Topic and focus are pragmatic since their status is determined by context. Focus involves a two-component semantics, including ordinary value semantics, which is the proposition denoted by the expressed sentence, and focus value semantics, a set of propositional alternatives to the meaning of the expressed sentence. Likewise, the two-component semantics of topic includes ordinary value semantics and topic value semantics. However, unlike focus, topic is operative at the utterance level. Hence, the ordinary value of topic is the utterance, and its topic value is a set of alternative utterances. Abbreviations asp cf clf ct comp

contrastive focus contrastive topic noun phrase aspect classifier

np prg prt pst top

compelentizer progressive particle past topic

Notes 1 Acknowledgments: I would like to express my gratitude to Malte Zimmermann, Hiroki Nomoto, Gisbert Fanselow, and two anonymous reviewers for their insightful comments and suggestions in various stages of the work. All shortcomings are obviously mine. The chapter was written when I was a researcher at the University of Potsdam, working on the project titled “Structure and Interpretation in the

256

Thuan Tran

Left Periphery in Vietnamese,” funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 1479/1–1. 2 It is well-known across languages that being definite is just a necessary condition for a noun phrase to qualify as a topic, considering that an article headline such as “Trump’s acquittal will pose problems for Democrats in the 2020 elections” is construed as all-new focus or sentence focus, and as such, the subject of the sentence cannot be topical, albeit being definite. The subject NP of the Vietnamese sentence corresponding to this English headline is not topical either. 3 A reviewer wonders whether an indefinite NP is acceptable in a specific indefinite reading context. The answer is positive. For instance, as an argument of the intensional verb “look for,” the NP một người phụ nữ “a woman” can obtain either de re or de dicto reading. (i) Nam đang tìm một người phụ nữ. Nam prg look.for one clf woman “Nam is looking for a woman.” 4 Note that (4a) is grammatical if the dislocated NP is focused. In that case, the preverbal particle mới functions as an exclusive particle, and as such, (4a) expresses a non-episodic/habitual, pragmatically weird, meaning, that Nam buys only one chair (when it comes to chair-buying) (see Tran 2021). 5 According to this test, a noun phrase X is the aboutness topic of a sentence S containing X if (a) S would be a natural continuation to the announcement Let me tell you something about X, or (b) S could be naturally transformed into the sentence Concerning X, S’, where S’ difers from S only insofar as X has been replaced by a suitable pronoun. 6 I am sympathetic with a reviewer that the Let me tell you something about X test can be used for the identification of not only sentence topic but also discourse topic. However, discourse topics, namely, non-discourse-initial anaphors, are not the concern of this chapter. 7 According to Nguyen and Lee (2004), a contrastive topic-marked sentence always induces implicatures: descriptive opposite implicature and denial-of-expectation implicature. The following example is illustrative. There is no diacritic marker in their example (numbered (8a)); CT stands for contrastive topic. (i) a. Mai thi an tao. Mai CT eat apple “As for Mai, she ate apples.” Imp: “As for other person, he/she ate other things.” (Nguyen & Lee 2004: 327) 8 An anonymous reviewer wonders how to account for the data in (12). I follow Forminyam and Tran (2019) in assuming that thì is a contrastive particle that requires the information structural category associated with it be contrastive. In this bargaining context, it is the price ofer that is contrastive, not the shirt. Therefore, associating the non-contrastive topic “this shirt” with the contrastive particle gives rise to the ill-formedness of (12c). 9 As far as I am aware of, Frey’s conception of contrast is in close proximity to the notion of contrast adopted in this chapter. Frey (2006) argues that in German, long A-bar movement of a topic induces a contrastive interpretation of the topic, where the notion of contrast is defined as follows: if an expression a in a declarative sentence S is contrastively interpreted, a set M of expressions, |M| 32, is assumed to exist, which contains a and the expressions which denote salient alternatives to the denotation of a. The utterance of a declarative clause S containing a contrastively interpreted expression a has the implicature that S is not true if a is replaced by any xÎ M, x 1a (Frey 2006: 246).

Syntax-information structure interface in Vietnamese

257

10 A reviewer provides the following examples, arguing that if (i) is accepted, and so is (ii), then (ii) would have a weaker implication than that given in the text regarding (23i), namely, that there is at least one thing that Nam will not buy. (i) He hates chocolates. He hates cookies. Nhưng sách thì nó thích. but book prt he like “But he likes books.” (ii) I do not know what he hates, nhưng sách thì nó thích. but book prt he like “but he likes books.” The last sentence in example (i) naturally implicates that there is at least one thing alternative to the contrastive focus “books,” explicitly provided in the first and second sentences, that the predicate “like,” or synonymously “not hate,” does not hold (i.e., he hates chocolates). Similar reasoning applies for (ii). Embracing the indirect question analysis of Karttunen and Peters (1976), I assume that the alternative to the contrastive focus in (ii) is derived from the presupposition of the first conjunct that he hates something, which in turn licenses the implication of the second conjunct: there is at least one thing alternative to “books” that the predicate “not hate” does not hold. If the first conjunct does not make available such a presupposition, the second conjunct becomes infelicitous. We find a parallel pattern in English, as shown by the anomalousness of the discourse “I do not know whether he hates anything, but he likes books.” 11 A reviewer claims that the same contrast seems to hold without left dislocation, as in (31b’). In fact, (31b’) is arguably a case of correction focus, analogous to the English example of correction focus in (i). (i) Context: John gave the CDs to Chris. No, he gave the books to Mary.

(Lopez 2009: 123)

According to Lopez, it is the negation particle preceding the sentence that gives rise to the contrastive interpretation; the focused constituents “the books” and “Mary” are plain information focus. (31b’) is infelicitous, because staying in situ does not allow the object “the fried chicken” to obtain the contrastive topic reading triggered by the context. Another reason is because, in terms of inquisitive semantics (Ciardello et al. 2019), in uttering (31b’), speaker B does not respond to what speaker A wonders about but corrects speaker A’s belief that Bac ate the fried chicken. This is not relevant, since in this context speaker A wonders what happened to the fried chicken.

References Beaver, David & Brady Clark. 2008. Sense and Sensitivity: How Focus Determines Meaning. Oxford: Wiley-Blackwell. Cao, Xuan Hao. 1991. Tiếng Viêt. ̣ Sơ thaỏ Ngữ phaṕ Chức năng. Nhà xuât́ ban ̉ Khoa học xã hôi. ̣ Thanh ̀ phố Hồ Chí Minh. [Vietnamese. A Functional Analysis]. Ho Chi Minh City: Social Sciences Publishing House. Cao, Xuan Hao. 1998. Tiếng Việt. Mấy vấn đề Ngữ âm, Ngữ pháp, Ngữ nghĩa. Nhà xuất bản giáo dục. Hà nội. [Remarks on Vietnamese Phonology, Grammar and Semantics]. Hanoi: Education Publisher. Casielles-Suarez, Eugenia. 2004. The Syntax-Information Structure Interface. Evidence from Spanish and English. New York & London: Routledge.

258

Thuan Tran

Chafe, Wallace. 1976. Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Charles N. Li (ed.), Subject and Topic, 25–56. New York: Academic Press. Dufeld, Nigel. 2007. Aspects of Vietnamese clausal structure: Separating tense from assertion. Linguistics 45(4). 765–814. Forminyam, Henry & Thuan Tran. 2019. Beware of discourse particles. In J.M.M. Brown, Andreas Schmidt & Marta Wierzba (eds.), Of Trees and Birds. A Festschrift for Gisbert Fanselow, 255–272. Potsdam: Potsdam University Press. Frey, Werner. 2006. Contrast and movement to the German prefield. In Valéria Molnár & Susanne Winkler (eds.), The Architecture of Focus, 197–233. Berlin & New York: Mouton de Gruyter. Götze, Michael, Thomas Weskott, et al. 2007. Information Structure. In Dipper Stefanie, Michael Götze & Stavaros Skopeteas (eds.), Interdisplinary Studies on Information Structure, Vol. 7, 147–187. Potsdam: Potsdam University Press. Gundel, Jeanette Koch. 1988. Universals of topic-comment structure. In Michael Hammond, Edith A. Moravcsik & Jessica R. Wirth (eds.), Studies in Syntactic Typology, 209–239. Amsterdam & Philadelphia: John Benjamins. Heim, Irene & Angelika Kratzer. 1998. Semantics in Generative Grammar. Malden & Oxford: Blackwell. Jannedy, Stefanie. 2007. Prosodic focus in Vietnamese. In Shinichiro Ishihara, Stefanie Jannedy & Anne Schwarz (eds.), Interdisciplinary Studies on Information Structure 08, 209–230. Potsdam: Potsdam University Press. Karttunen, Lauri & Stanley Peters. 1976. What indirect questions conventionally implicate. In CLS 12: Papers from the Twelfth Regional Meeting, 352–368. Chicago: Chicago Linguistics Society. Krifka, Manfred. 2001. Quantifying into question acts. Natural Language Semantics 9. 1–40. Krifka, Manfred. 2003. Quantifiers in questions. Korean Journal of English Language and Linguistics 3. 499–526. Krifka, Manfred. 2006. Association with focus phrases. In Valerie Molnar & Susanne Winkler (eds.), The Architecture of Focus, 105–136. Berlin: Mouton de Gruyter. Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lee, Chungmin. 2006. Contrastive (predicate) topic, intonation, and scalar meanings. In Chungmin Lee, Matt Gordon & Daniel Büring (eds.), Topic and Focus: CrossLinguistic Perspectives on Meaning and Intonation, 151–175. Dordrecht: Springer. Lee, Chungmin. 2017. Contrastive topic, contrastive focus, alternatives, and scalar implicature. In Chungmin Lee et al. (eds.), Contrastiveness in Information Structure, Alternatives and Scalar Implicature, Studies in Natural Language and Linguistics, 3–22. Dordrecht: Springer. http://doi.org/10.1007/978-3-319-10106-4-1 Li, Charles & Sandra Thompson. 1976. Subject and topic: A new typology of language. In Charles Li (ed.), Subject and Topic, 457–489. London & New York: Academic Press. Lopez, Luis. 2009. A Derivational Syntax for Information Structure. Oxford: Oxford University Press. Molnár, Valéria. 2006. On diferent kinds of contrast. In Valéria Molnár & Susanne Winkler (eds.), The Architecture of Focus, 197–233. Berlin & New York: Mouton de Gruyter. Neeleman, Ad & Hans van de Koot. 2008. Dutch scrambling and the nature of discourse templates. Journal of Comparative German Linguistics 11. 137–189.

Syntax-information structure interface in Vietnamese

259

Neeleman, Ad & Reiko Vermeulen. 2012. The syntactic expression of information structure. In Henk van Riemsdijk et al. (eds.), The Syntax of Topic, Focus, and Contrast. Studies in Generative Grammar, 113, 1–38. Berlin: De Gruyter Mouton. Nguyen, Hoai Thu Ba & Chungmin Lee. 2004. Contrastive topic in Vietnamese. Annual Conference on Human Language and Technology. 323–328. Portner, Paul. 2007. Instructions for interpretation as separate performatives. In Kerstin Schwabe & Susanne Winkler (eds.), On Information Structure, Meaning and Form, 407–426. Amsterdam & Philadelphia: John Benjamins. Reinhart, Tanya. 1981. Pragmatics and linguistics: An analysis of sentence topics. Philosophica 27. 53–94. Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1. 75–116. Rooth, Mats. 1999. Association with focus or association with presupposition? In Peter Bosch & Rob van der Sandt (eds.), Focus: Linguistic, Cognitive, and Computational Perspective, 232–244. Cambridge: Cambridge University Press. Szabolcsi, Anna. 1997. Strategies for scope taking. In Anna Szabolcsi (ed.), Ways of Scope Taking, 109–154. Dordrecht: Kluwer. Tomioka, Satoshi. 2010. Contrastive topics operate on Speech Acts. In Malte Zimmermann & Caroline Féry (eds.), Information Structure: Theoretical, Typological, and Experimental Perspectives, 115–138. Oxford: Oxford University Press. Tran, Thuan. 2009. Wh-Quantification in Vietnamese. PhD dissertation, University of Delaware. Tran, Thuan. 2021. Non-canonical word order and temporal reference in Vietnamese. Linguistics 59(1). 1–34. Vallduví, Enric. 1992. The Informational Component. New York: Garland. Vallduví, Eric & Maria Vilkuna. 1998. On Rheme and Kontrast. In Peter Culicover & Louise McNally (eds.), Syntax and Semantics 29. The Limits of Syntax, 79–108. Leiden: Brill. Zimmermann, Malte. 2008. Contrastive focus and emphasis. Acta Linguistica Hungarica 55. 347–360.

Index

Note: Boldface page references indicate tables. Italic references indicate figures. le Mandarin SFP, semantics a (discourse particle) 37 a (sentence-final particle) 41–45; frequency (conversational corpus) 42 aa3 (particle) 3 additive description 206 additive inference, even projection 233 additive particle (juga), pun (comparison) 155 additive pun, translation 149–152; 1739980116 Foc1739980116 P 150; punadditive, semantics 150–151 afrmative clauses, final pun (occurrence) 156–157 already: accomplishment p 97; habitual state p 94; positions, semantic scope (impact) 104–105; state p 97 already (Singlish) 91–99; syntax 102–107 already (Singlish sentence final): development 106–107; reanalysis, syntactic universal (impact) 107–112; relexification 105–107 AnderBois, Scott 4, 9 annoyance, result 30 A-NOT-A questions 3 aspect marker (ASP) 240 assertions: contradictions 76–77; überhaupt, attachment 179 ASSERT operator, usage 253 atelic description 98 attested speech-act types, kɔ̂ɔ (pragmatic particle) 208

biased question, mirative pala (usage) 18–19 canonical sentence, definite NP subject 239 canonical word order sentences 238 Cantonese laa, scope 101 Chinese cognate SFPs 92 Chinese languages, clausal spine (SpellOut domains) 110 Chinese SFPs (le / liau / laa): position, hypotheses 100–102; syntax 99–102 Chinese SFPs, positions 109 classifier (CLF) 186–187, 239–240 clausal negation 73 clausal spine, Spell-Out domains (Chinese languages) 110 clause-linker, /kɔ̂ɔ/ uses 206 clause, syntactic transitivity 39 clitics: classes 71; linear order 71; pronominal clitics 11; secondposition clitics 9, 28, 68–72 cơ (Vietnamese sentence-final particle): background information, modification 219; contrasting alternative 225–226; contrastive topic 222–224; conversational expectation 229–231; declaratives 215–218; disagreement, expression 216; entailment relation 225; entailment relation, absence 226; even, relationship 231–234; facts 214–221; felicity 219–220; focus 222–224; formal semantics 221–231; formal semantics, proposal 222;

Index insistence, observation 215; interpretation 232; linguistic/ non-linguistic evidence 220–221; meta-evidential scale 227–228; polar questions 218–221, 229; prejacent/alternative ordering, focus 228; prejacent, exception 227–228; prejacent, strength requirement 223–224; scalar meaning 224; scalar semantics 214; scales 225–229; scenarios 217; speaker belief 229–231; speaker declarative 230–231 cognate Chinese SFPs, Singlish already (hierarchical position, contrast) 92 collaborative completion (turn-initial /kɔ̂ɔ/) 205 colloquial Malay (CM): dah (discourse marker) 121; dah (discourse marker), co-occurrence restrictions (patterns) 124; postverbal dah, analysis 131–132; sentence-final dah, analysis 131–132; West Malaysia examples 121–122 Common Ground content 239 complementizer (COMP) 240 complementizer (no) 2 completive already 98 concurring replies (turn-initial /kɔ̂ɔ/) 202–204 confirm, usage 93 conjunction types (turn-initial /kɔ̂ɔ/) 206–208 contact-induced grammatical changes 92–93 contexts, impact 74–77 contextual subquestion, definition 182 contract language, syntactic universal 91 contrast: dependent component 245–246; focus, diferences 252; independent component 246–251; role 247 contrast-based mirativity strategies 28–29 contrastive focus 251–255; contrast, role 248; negative efect 249; semantics 252; syntactic behavior 248–249 contrastive topic 251–255; negative efect 249; syntactic behavior 248–249; topic, contrast 252–255 contrastive topic-contrastive focus 246 conversational corpus, SFPs (frequency) 42

261

conversation corpus (Tagalog) 39–40, 40 conversation, pragmatic particle kɔ̂ɔ (interpersonal uses) 191 co-occurrence, obligatoriness 85 co-occurrence restriction patterns: colloquial Malay dah 124; dah, quantifiers (relationship) 125–129; sudah (Sabah Malay) 139–141 corpus-based analyses 195–210 correction 80 correspondences, establishment 86 counterexpectation 43; encoding, absence 25–26; mirative attitude conception 12 dah: implications/conclusions 141–143; ke (particle), ordering/scope restriction (explanation) 137–138; multiple appearances, sentence usage 128–129; postverbal dah, semantics 131–138; post-verbal dah, syntax 130; preverbal dah, post-verbal/sentencefinal dah (contrast) 126–127; pre-verbal dah, syntax 129–130; sentence-final dah, semantics 131–138; sentence-final dah, syntax 130; syntax, assumptions/ preliminary analyses 129–130; usage 5 dah, colloquial Malay (CM) 121; cooccurrence restrictions, patterns 124; quantifiers, co-occurrence restriction patterns 125–129 dati’ 72 daw (enclitic) 38–39 declaratives: cơ (Vietnamese sentencefinal particle) 215–218; mirative pala, usage 14–18; pala, illocutionary uses 24–25; pala, illocutionary uses (propositional/ illocutionary contexts) 24 de, syntax 2 diba 45–46 din naman 48 discourse markers 1, 40; dah, colloquial Malay (CM) 121; sudah (Sabah Malay) 121 discourse particles 1; additional meaning, ofering 174; contributions, cross-linguistic perspective 86; definition, clarification 40; e, usage 37, 56; speculative

262

Index

characteristics 174; unified analysis (Thai) 173 discourse particles (Tagalog) 37, 40–48; background 38–40 doch (contrast component) 86; particle equivalent, absence 68; usage 80 downward-entailing quantifiers, nondownward-entailing quantifiers (contrast) 125 Dusunic subgroup 67 e (discourse particle) 37, 56 e (sentence-final particle) 41–45; analysis, case study 48–61; description 60; frequency (conversational corpus) 42; polysemy approach 59–61; prosody 49–51, 50; reality, ideal/ expected situations (situating) 56–59; stances, contrasts 51–56; utterances, connection 51 e-marked utterances 55 enclitic (daw) 38–39 enclitic particles 12; nga 60 epistemic-based authority 30 Erlewine, Michael Yoshitaka 4, 91 evaluative particles 71 even: additive component 233–234; additive inference, projection 233; cơ , relationship 231–234; scalar additive particle 232–233; scales, types 231 event-sequence pun 160–166; alternative analysis 162–164; complexity 165; descriptions/problems 160–162; examples 164–166; individual/propositional additives, usage 162–164 event-sequence reading (propositional additive pun) 164 evidential attitude 18 exclamatory use 83–85 explicitness 29 expressive meaning 83–85 expressive reduplication 86 Final-over-Final Constraint (FOFC) 92; domains 109–110; FOFC-driven reanalysis 112 Final-over-Final Constraint (FOFC)107; predictions 107–108 final pun 156–159; afrmative clause occurrence 156–157; development 158–159; negative clause occurrence 156

Finnish wh-questions, word orders 108 focus-aspect clitic 71 focus, notion (Rooth) 246 gaa3 (particle) 3 ge3 (particle) 3 GEN clitic 71 gima 4; appearance 79; expressive meaning/exclamatory use 83–85; indication 74; interchangeability 81; meaning 68; occurrence 78; scolding/complaint use 82; surprise use 81; usage, avoidance 67–68; use, belief indications 76 gima, usage 77–83; avoidance, contexts (impact) 74–77 got (Singlish invariant) 94 grammatical correlate 742 ha (discourse particle) 37 ha (sentence-final particle) 41–45; frequency (conversational corpus) 42; prosody 45 Hara, Yurie 3 head-final VP over head-initial DP (German) 108–109 head movement 160 hedging replies (turn-initial /kɔ̂ɔ/) 201 ho2 (interrogative) 3 Hokkien (Southern Min) liau, scope 101 illocutionary context 24 illocutionary-force types, comparison 200 imperatives: co-occurrence 19–22; context 20–22; issuance 21 implicit question 247 impoliteness, result 30 inchoative already 98 indirect evidentials 14, 19 individual additives, usage 162–164 information: correction 80; discovery, surprise 80–82; scolding, impact 82–83; structure (Vietnamese language) 238–245; transmission, mechanisms 175 initial interjection 27 initial pun 159–160; single head, reanalysis 160; syntactic change 159–160 interjectional type, occurrence 191 interrogative flip, mirative pala (usage) 18 interrogatives, pala (usage) 22–23 Ito, Satomi 3

Index ja (proposition, active consideration) 158 ja (uncontroversiality) 86 ja, felicitous use 75 ja, usage 80 juga (additive particle), pun (comparison) 155 kami (pronominal clitic) 11 katoy, usage 76 kaya 47 kɔ̂ɔ (pragmatic particle): attested speechact types 208; characteristics 193–195; corpus-based analyses 195–210; interpersonal uses 191; predicate-initial kɔ̂ɔ, occurrence 208–209; turn-initial /kɔ̂ɔ/, attested usage types 196–200; turn-initial /kɔ̂ɔ/, less-frequentusage types 204–208; turn-initial /kɔ̂ɔ/, syntactic subtypes 205, 206; turn-initial /kɔ̂ɔ/, usage frequencies 199; turn-initial /kɔ̂ɔ/, usage types 196; types 195; usage 6; usage, plausible extension 193; uses, system 208–210 kɔ̂ɔ ləəy, usage 209–210 ke (particle): dah, ordering/scope restriction (explanation) 137–138; occurrence 134–135; syntax, assumptions 136–137 khá, female-presenting speaker usage 180 khâ, female-presenting speaker usage 180 khráp, male-presenting speaker usage 180 Kimaragang status particle 67, 73 Kroeger, Paul R. 4, 67 lâ: question, indication 183; usage 181–182; use, appropriateness 183 laa (Cantonese): constraint 92; scope 101 Lai, Regine 3 Lantubon, Janama 68 1739980129 là, 1739980130 thì (interchangeability) 242 laɁ 183–184; resolution operative, notion 184 le: cognates 4–5; constraint 92 Lee, Soo-Hwan 3 left-dislocated elements 237–238 le / liau / laa (Chinese SFPs): position, hypotheses 100–102; syntax 99–102

263

le (Mandarin), scope 100–101; negative subject quantifier, appearance 101–102 LF, usage 103–104 liau (Hokkien: Southern Min): constraint 92; scope 101 logical outcome, description 203, 206, 207 Luk, Zoe Pei-sui 3 lέɛw (discourse particle) 173–174 Malay additive (pun), non-additive use 147 Mandarin Chinese, description 37 Mandarin le, scope 100–101 mari 72 markers, usage (motivation) 86 Maximize-Q-Anaphoricity 163 McCready, Elin 1, 5, 173 me1 (interrogative) 3 mirative attitude, concepts 12 mirative pala 12; context 14–15; prior expectations, meeting (context) 15–16; speaker, remembering (context) 16–18; usage 14–25 mirative proper 14 miratives, analysis 26–27 mirativity 9; contrast-based mirativity strategies 28–29; cross-linguistic analysis 12–14; notion 2; prototypical use 25–26; strategies 10; term, inception 12 modal particles 1 modelpartikeln 71 morphosyntactic encoding 28 m-performative 13 n1739980132 á 186–187 Nagaya, Naonori 4, 37 narrative tense 72 ná (Thai) utterance 176 n1739980132 âɁ 184–185 n1739980132 âɁ-marked utterance 185 negative clauses, final pun (occurrence) 156 negative subject quantifier, appearance 101–102 negators, scope 100–101 neo-Davidsonian event semantics, usage 155–156 new information: encoding, absence 25–26; mirative attitude conception 12 news/out-of-the-blue statements, breaking 75–76

264

Index

nga (enclitic particle) 60 nga (encoding) 31 nga (second-position enclitic particle) 29 nga, interaction 29–31 Nguyen, Anne 6, 214 n1739980132 îɁ 185–186; utterance 186 no da construction 2 NOM clitic 71 nominative NP, actor role 39 Nomoto, Hiroki 5, 147 non-additive use, pun (Malay additive) 147 non-canonical word order sentence 250 non-downward-entailing quantifiers, downward-entailing quantifiers (contrast) 125 o (discourse particle) 37 o (sentence-final particle) 41–45; frequency (conversational corpus) 42 object position, pun in(side) occurrence 154 object-verb-subject (OVS), Vietnamese (language) 237–238 oo nga e 59 orthographic conventions 45 Oshima, David Y. 2 Outer Circle English variety, contactinduced grammatical changes 92–93 outside the speech act interpretation 19 pa (temporal clitic) 28 pala (second-position particle) 4; meanings 10 pala (Tagalog) 9; elements, interactions 25–31; felicity 16; illocutionary uses (declaratives) 24–26; imperatives, co-occurrence 19–22; interrogative use 22–23; mirative pala, usage 14–18; verb-objectsubject (VOS) word order 11 pala (second-position clitic), usage 18 particle ke: occurrence 134–135; syntax, assumptions 136–137 particles: QUDs, relationship 178; term, usage 173 particles (Thai) 180–181 perfective description 98 phenomenal consequence (turn-initial / kɔ̂ɔ/): description 202, 206; logical outcome, diferences 209

Philippines, culture (prevalence) 54 phonological variants, approach 187 phrase-final type, occurrence 19 polar questions: cơ (Vietnamese sentence-final particle) 218–221; types 3 politeness, marking 1 polysemy approach (e) 59–61 post-verbal dah: assumptions 131; distribution (explanation), quantifiers (usage) 132–134; distribution, yes/no questions (usage) 135–136; pre-verbal/ sentence-final dah, contrast 126–127; proposal 131–132; semantics 131–138; support 134–138; syntax 130; usage 132–134 Praat analysis 44 pragmatic markers 40 pragmatic particle (kɔ̂ɔ): characteristics 193–195; corpus-based analyses 195–210; interpersonal uses 191; types 195; usage, plausible extension 193 pragmatic particles 1, 40; interpersonal uses 191; types 191 predicate-final possibility modal (Thai) 111–112 predicate-initial kɔ̂ɔ, occurrence 208–209 predicate-initial type, occurrence 191 pre-verbal dah: post-verbal/sentence-final dah, contrast 126–127; syntax 129–130 pronominal clitics 11–12 proposition, active consideration 158 propositional additive pun: eventsequence reading 164; non-eventsequence uses 166–169 propositional additives, usage 162–164 propositional context 24 prosody (e) 49–51, 50 prosody (ha) 45 pun (Malay additive): additive pun, translation 149–152; at-issue content 151, 156; eventsequence pun 160–166; final pun 156–159; initial pun 159–160; in(side) occurrence 154; juga (additive particle), comparison 155; meanings 148–160; neoDavidsonian event semantics, usage 155–156; non-additive use 147; PartP 157; presupposition

Index 151, 156; rhetorical question usage 157; scalar pun, translation 152–156; second-position pun 149–156; temporal meaning 5–6; uses, classification 148; XTOPIC pun 151–152 pun (propositional additive): eventsequence reading 164; non-eventsequence uses 166–169 punadditive, semantics 150–151 punfinal, semantics 158 punscalar, usage 155 pure remembering cases 17 pure remembering scenario, felicirty 17 Q-A congruence 163 Q-Givenness 163 quantifiers: dah, co-occurrence restriction patterns (relationship) 125–129; downward-entailing quantifiers, non-downwardentailing quantifiers (contrast) 125; sudah usage 140–141; usage 132–134 quantity hedge 68, 77 question: answering 76; contextual subquestion, definition 182; semantics 182 question under discussion (QUD) 5–6, 164, 175–180; analysis, usage 176; concept 175; direct approaches 178–179; immediate QUD 166–167; implicitness 176; indirect approaches 179–180; involvement 167–168; mechanism 177; particles, relationship 178; QUD-based story 185; shift 182–183; sub-QUD, addressing, assertion (usage) 177; sub-QUD, structure 177; super-QUD, structure 177; theory 163; tree 168; uses 176–178 reaction speech subtypes (turn-initial /kɔ̂ɔ/) 204 reason clauses 78–79 reason-efect relation 54 reason use, stance-justifying (diference) 53 referentiality constraint 239 relevance (identification), QUD mechanism (usage) 177 relexification 91–94; ; Singlish (already) development 112; Singlish

265

sentence-final (already) 98–99, 105 reply types (turn-initial /kɔ̂ɔ/) 201 resolution operative 184 response types 200 rhetorical question, pun (occurrence) 157 Rieser, Lukas 2 Sabah Malay, sudah (discourse marker) 121, 139–141; co-occurrence restriction patterns 139–141 s1739980140 agay 84 scalar additve particle (even) 232–233 scalar pun: example 166; final pun derivation 158–159; ForceP 160; semantics 155 scalar pun, translation 152–156; 1739980140 Foc1739980140 P 153 scalar semantics 214 schematic semantics 183 scolding 82–83 scopeless meaning 67 second-position (2P) clitics 9, 28, 68–72; inventory 67; ordering, template 70; pala, usage 18; personal pronouns/particles 39; sentence-medial particles, relationship 46–47; stance indication 39 second position, defining 69, 71 second-position enclitic particle (nga) 29 second-position enclitic particles 12 second-position pun 149–156 semantic scope, impact 104–105 sentence-final dah: assumptions 131; distribution (explanation), quantifiers (usage) 132–134; distribution, yes/no questions (usage) 135–136; post-verbal/ per-verbal dah, contrast 126–127; proposal 131–132; semantics 131–138; support 134–138; syntax 130; usage 132–134 sentence-final discourse particles 42 sentence-final particles (SFPs) 1, 27, 41–45; frequency 42; sentence-final (already), semantics/ development 94–99 sentence-initial discourse particles 42 sentence-initial particles (SIPs) 45–46 sentence-medial particles 46–48; secondposition clitics, relationship 46–47

266

Index

sentence types, mirative pala (usage) 18–25 shi ... de sentences 2 Singlish (already) 91; cognate Chinese SFPs, hierarchical position (contrast) 92; development, relexification/FOFC-driven reanalysis 112; syntax 102–107 Singlish, grammar 92–94 Singlish sentence-final (already): adverb, function (relexification) 105; development 106–107; reanalysis, syntactic universal (impact) 107–112; relexification 98–99; semantics/development 94–99 Soh, Hooi Ling 2, 5, 121 speaker-oriented assertive (turn-initial /kɔ̂ɔ/) 203 speaker, prior expectation (meeting): context 15–16; felicity 15 speaker, remembering 16–18 speech act, mirative pala (usage) 19 Spell-Out domains 109–110; clausal spine 110; verbal inflection, inclusion 111 stance-justifying, reason use (diference) 53 stances: contrasts 51–56; indication 39; intersubjective manipulation 61 Standard Formal Malay (SFM): examples 121 Standard Formal Malay (SFM), aspectual auxiliary sudah 121 status particles 67, 71–72 story, narrator description 207 straight replies (turn-initial /kɔ̂ɔ/) 201 subject-verb-object (SVO), Vietnamese (language) 237–238 subordinate clause, impact 69 sub-QUD: addressing, assertion (usage) 177; structure 177 substrate cognates (positions), semantic scope (impact) 104–105 sudah: aspectual auxiliary sudah, Standard Formal Malay (SFM) 121; implications/conclusions 141–143; usage 5 sudah, discourse marker (Sabah Malay) 121, 139–141; co-occurrence restriction patterns 139–140; cooccurrence restriction patterns, quantifiers (usage) 140–141 sudden discovery/revelation/realization (mirative attitude conception) 12 sudden inference cases 17

sudden inference scenario, felicity 17 super-QUD, structure 177 surprisal 13 surprise 80–82; encoding 25; mirative attitude conception 12; pala encoding 15 syntactic positions 191 syntactic transitivity 39 syntactic universal 91; impact 107–112 syntax, dah (assumptions/preliminary analyses) 129–130 syntax-information structure interface (Vietnamese) 237 Tagalog: background 11–12; context 14; conversation corpus 39–40, 40; discourse particles 37, 40–48; explicitness 29; language profile 38–39; miratives, analysis 26–27; morphosyntactic encoding 28 Tagalog pala 9 Takahashi, Kiyoko 6, 191 Tawilapakul, Upsorn 5, 173 telic description 98 temporal clitic (pa) 28 Thai (language): conversation, pragmatic particle kɔ̂ɔ (interpersonal uses) 191; discourse particles, unified analysis 173; khá, female-presenting speaker usage 180; khâ, female-presenting speaker usage 180; khráp, malepresenting speaker usage 180; lâ, usage 181–182; laɁ 183–184; n1739980144 á 186–187; n1739980144 âɁ 184–185; n1739980144 îɁ 185–186; particles 180–181, 185–187; predicate-final possibility modal 111–112 1739980144 thì (contrastive topic marker) 242–243; bargaining context 243; 1739980144 là, interchangeability 242 toomod 72 topic: contrastive topic, contrast 252–255; information structural category 238–239 topicality, specification 239 Tran, Thuan 6, 237 truth-conditional meaning, useconditional (contrast) 73–74 turn-initial /kɔ̂ɔ/: attested usage types 196–200; collaborative

Index completion 205–206; concurring replies 202–204; conjunction types 206–208; hedging replies 201; less-frequent-usage types 204–208; logical outcome, description 203; phenomenal consequence, description 202–203; phenomenal consequence, logical outcome (diferences) 209; reaction speech act, subtypes 204; reply types 201; response types 200; speakeroriented assertive 203; straight replies 201; syntactic subtypes 205, 206; usage frequencies 199; usage types 196, 201–204 überhaupt, attachment 179 unprepared mind (mirative attitude conception) 12 use-conditional, truth-conditional meaning (contrast) 73–74 utterances 43; connecting 51; e-marked utterances 55; marking 51–52; propositional content 58; redundancy, marking 86 u (Hokkien), uses 94 V-ed (telic/perfective description) 98 verbal inclusion, Spell-Out domains 111 verb-object-subject (VOS) word order 11 Vietnamese (language): canonical word order sentences 238; contrast 245–251; contrast, dependent component 245–246; contrast, independent component 246–251; contrastive focus 245, 251–255; contrastive topic 245, 251–255; contrastive topic, contextual establishment 241–242; focus 243–244; information structure 238–245; non-canonical word order sentence 250; noncontrastive focus 245; pattern 240–241; prosodic marking 244; referentiality constraint 239; sentences, propositional content 251; SVO/OSV word order 237–238; syntax-information

267

structure interface 237; thì (contrastive topic marker) 242–243; topic 238–242; topic dislocation 241; topic/focus left dislocation 249 Vietnamese sentence-final particle (cơ): background information, modification 219; contrasting alternative 225–226; contrastive topic 222–224; conversational expectation 229–231; declaratives 215–218; declarative speaker 230–231; disagreement, expression 216; entailment relation 225; entailment relation, absence 226; even, relationship 231–234; facts 214–221; felicity 219–220; focus 222–224; formal semantics 221–231; formal semantics, proposal 222; insistence, observation 215; interpretation 232; linguistic/ non-linguistic evidence 220–221; meta-evidential scale 227–228; prejacent/alternative, ordering (focus) 228; prejacent, generalization 227–228; prejacent strength, requirement 223–224; scalar meaning 224; scalar semantics 214; scales 225–229; scenarios, diference 217; speaker belief 229–231 V-ing (atelic description) 98 wh-questions (Finnish), word orders 108 Winterstein, Grégoire 3 word orders (Finnish wh-questions) 108 woy (expressive particle) 83 x punindividual_additive P, semantics 162–163 XTOPIC pun 151–152 yes/no questions: post-verbal/sentencefinal dah, distribution 135–136 yes/no questions (usage), particle ke (occurrence) 134–135; postverbal/sentence-final dah, distribution 135–136 Yucatec Maya bakáan 22